diff options
Diffstat (limited to 'src')
402 files changed, 21567 insertions, 16543 deletions
diff --git a/src/egl/drivers/demo/demo.c b/src/egl/drivers/demo/demo.c index aea4894448b..0933c0bdaa5 100644 --- a/src/egl/drivers/demo/demo.c +++ b/src/egl/drivers/demo/demo.c @@ -177,7 +177,7 @@ demoCreatePixmapSurface(_EGLDriver *drv, _EGLDisplay *dpy, _EGLConfig *conf, Nat } } - if (conf->Attrib[EGL_SURFACE_TYPE - FIRST_ATTRIB] == 0) { + if (GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE) == 0) { _eglError(EGL_BAD_MATCH, "eglCreatePixmapSurface"); return NULL; } diff --git a/src/egl/drivers/xdri/egl_xdri.c b/src/egl/drivers/xdri/egl_xdri.c index 518091a2d14..d2affc66dd4 100644 --- a/src/egl/drivers/xdri/egl_xdri.c +++ b/src/egl/drivers/xdri/egl_xdri.c @@ -48,6 +48,7 @@ #include "glapi/glapi.h" /* for glapi functions */ #include "eglconfig.h" +#include "eglconfigutil.h" #include "eglcontext.h" #include "egldisplay.h" #include "egldriver.h" @@ -104,6 +105,7 @@ struct xdri_egl_config _EGLConfig Base; /**< base class */ const __GLcontextModes *mode; /**< corresponding GLX mode */ + EGLint window_render_buffer; }; @@ -162,46 +164,76 @@ get_drawable_size(Display *dpy, Drawable d, uint *width, uint *height) } +static EGLBoolean +convert_config(_EGLConfig *conf, EGLint id, const __GLcontextModes *m) +{ + static const EGLint all_apis = (EGL_OPENGL_ES_BIT | + EGL_OPENGL_ES2_BIT | + EGL_OPENVG_BIT | + EGL_OPENGL_BIT); + EGLint val; + + _eglInitConfig(conf, id); + if (!_eglConfigFromContextModesRec(conf, m, all_apis, all_apis)) + return EGL_FALSE; + + if (m->doubleBufferMode) { + /* pixmap and pbuffer surfaces are always single-buffered */ + val = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE); + val &= ~(EGL_PIXMAP_BIT | EGL_PBUFFER_BIT); + SET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE, val); + } + else { + /* EGL requires OpenGL ES context to be double-buffered */ + val = GET_CONFIG_ATTRIB(conf, EGL_RENDERABLE_TYPE); + val &= ~(EGL_OPENGL_ES_BIT | EGL_OPENGL_ES2_BIT); + SET_CONFIG_ATTRIB(conf, EGL_RENDERABLE_TYPE, val); + } + /* skip "empty" config */ + if (!val) + return EGL_FALSE; + + val = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE); + if (!(val & EGL_PBUFFER_BIT)) { + /* bind-to-texture cannot be EGL_TRUE without pbuffer bit */ + SET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGB, EGL_FALSE); + SET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGBA, EGL_FALSE); + } + + /* EGL_NATIVE_RENDERABLE is a boolean */ + val = GET_CONFIG_ATTRIB(conf, EGL_NATIVE_RENDERABLE); + if (val != EGL_TRUE) + SET_CONFIG_ATTRIB(conf, EGL_NATIVE_RENDERABLE, EGL_FALSE); + + return _eglValidateConfig(conf, EGL_FALSE); +} + + /** * Produce a set of EGL configs. */ static EGLint create_configs(_EGLDisplay *disp, const __GLcontextModes *m, EGLint first_id) { - static const EGLint all_apis = (EGL_OPENGL_ES_BIT | - EGL_OPENGL_ES2_BIT | - EGL_OPENVG_BIT | - EGL_OPENGL_BIT); int id = first_id; for (; m; m = m->next) { - /* add double buffered visual */ - if (m->doubleBufferMode) { - struct xdri_egl_config *config = CALLOC_STRUCT(xdri_egl_config); - - _eglInitConfig(&config->Base, id++); - - SET_CONFIG_ATTRIB(&config->Base, EGL_BUFFER_SIZE, m->rgbBits); - SET_CONFIG_ATTRIB(&config->Base, EGL_RED_SIZE, m->redBits); - SET_CONFIG_ATTRIB(&config->Base, EGL_GREEN_SIZE, m->greenBits); - SET_CONFIG_ATTRIB(&config->Base, EGL_BLUE_SIZE, m->blueBits); - SET_CONFIG_ATTRIB(&config->Base, EGL_ALPHA_SIZE, m->alphaBits); - SET_CONFIG_ATTRIB(&config->Base, EGL_DEPTH_SIZE, m->depthBits); - SET_CONFIG_ATTRIB(&config->Base, EGL_STENCIL_SIZE, m->stencilBits); - SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLES, m->samples); - SET_CONFIG_ATTRIB(&config->Base, EGL_SAMPLE_BUFFERS, m->sampleBuffers); - SET_CONFIG_ATTRIB(&config->Base, EGL_NATIVE_VISUAL_ID, m->visualID); - SET_CONFIG_ATTRIB(&config->Base, EGL_NATIVE_VISUAL_TYPE, m->visualType); - SET_CONFIG_ATTRIB(&config->Base, EGL_CONFORMANT, all_apis); - SET_CONFIG_ATTRIB(&config->Base, EGL_RENDERABLE_TYPE, all_apis); - SET_CONFIG_ATTRIB(&config->Base, EGL_SURFACE_TYPE, EGL_WINDOW_BIT); - - /* XXX possibly other things to init... */ - - /* Ptr from EGL config to GLcontextMode. Used in CreateContext(). */ - config->mode = m; - - _eglAddConfig(disp, &config->Base); + struct xdri_egl_config *xdri_conf; + _EGLConfig conf; + EGLint rb; + + if (!convert_config(&conf, id, m)) + continue; + + rb = (m->doubleBufferMode) ? EGL_BACK_BUFFER : EGL_SINGLE_BUFFER; + + xdri_conf = CALLOC_STRUCT(xdri_egl_config); + if (xdri_conf) { + memcpy(&xdri_conf->Base, &conf, sizeof(conf)); + xdri_conf->mode = m; + xdri_conf->window_render_buffer = rb; + _eglAddConfig(disp, &xdri_conf->Base); + id++; } } @@ -363,6 +395,9 @@ xdri_eglCreateContext(_EGLDriver *drv, _EGLDisplay *dpy, _EGLConfig *conf, return NULL; } + /* the config decides the render buffer for the context */ + xdri_ctx->Base.WindowRenderBuffer = xdri_config->window_render_buffer; + xdri_ctx->driContext = psc->driScreen->createContext(psc, xdri_config->mode, diff --git a/src/egl/main/eglconfig.c b/src/egl/main/eglconfig.c index d47b99eed4b..2c8d1c4055b 100644 --- a/src/egl/main/eglconfig.c +++ b/src/egl/main/eglconfig.c @@ -15,302 +15,706 @@ #define MIN2(A, B) (((A) < (B)) ? (A) : (B)) - - -void -_eglSetConfigAttrib(_EGLConfig *config, EGLint attr, EGLint val) -{ - assert(attr >= FIRST_ATTRIB); - assert(attr < FIRST_ATTRIB + MAX_ATTRIBS); - config->Attrib[attr - FIRST_ATTRIB] = val; -} +#define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0])) /** * Init the given _EGLconfig to default values. * \param id the configuration's ID. + * + * Note that id must be positive for the config to be valid. + * It is also recommended that when there are N configs, their + * IDs are from 1 to N respectively. */ void _eglInitConfig(_EGLConfig *config, EGLint id) { memset(config, 0, sizeof(*config)); - config->Handle = (EGLConfig) _eglUIntToPointer((unsigned int) id); - _eglSetConfigAttrib(config, EGL_CONFIG_ID, id); - _eglSetConfigAttrib(config, EGL_BIND_TO_TEXTURE_RGB, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_BIND_TO_TEXTURE_RGBA, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_CONFIG_CAVEAT, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_NATIVE_RENDERABLE, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_NATIVE_VISUAL_TYPE, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_MIN_SWAP_INTERVAL, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_MAX_SWAP_INTERVAL, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_SURFACE_TYPE, EGL_WINDOW_BIT); - _eglSetConfigAttrib(config, EGL_TRANSPARENT_TYPE, EGL_NONE); - _eglSetConfigAttrib(config, EGL_TRANSPARENT_RED_VALUE, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_TRANSPARENT_GREEN_VALUE, EGL_DONT_CARE); - _eglSetConfigAttrib(config, EGL_TRANSPARENT_BLUE_VALUE, EGL_DONT_CARE); + + /* some attributes take non-zero default values */ + SET_CONFIG_ATTRIB(config, EGL_CONFIG_ID, id); + SET_CONFIG_ATTRIB(config, EGL_CONFIG_CAVEAT, EGL_NONE); + SET_CONFIG_ATTRIB(config, EGL_TRANSPARENT_TYPE, EGL_NONE); + SET_CONFIG_ATTRIB(config, EGL_NATIVE_VISUAL_TYPE, EGL_NONE); #ifdef EGL_VERSION_1_2 - _eglSetConfigAttrib(config, EGL_COLOR_BUFFER_TYPE, EGL_RGB_BUFFER); - _eglSetConfigAttrib(config, EGL_RENDERABLE_TYPE, EGL_OPENGL_ES_BIT); + SET_CONFIG_ATTRIB(config, EGL_COLOR_BUFFER_TYPE, EGL_RGB_BUFFER); #endif /* EGL_VERSION_1_2 */ } /** - * Return the public handle for an internal _EGLConfig. - * This is the inverse of _eglLookupConfig(). + * Link a config to a display and return the handle of the link. + * The handle can be passed to client directly. + * + * Note that we just save the ptr to the config (we don't copy the config). */ EGLConfig -_eglGetConfigHandle(_EGLConfig *config) +_eglAddConfig(_EGLDisplay *dpy, _EGLConfig *conf) { - return config ? config->Handle : 0; + _EGLConfig **configs; + + /* sanity check */ + assert(GET_CONFIG_ATTRIB(conf, EGL_CONFIG_ID) > 0); + + configs = dpy->Configs; + if (dpy->NumConfigs >= dpy->MaxConfigs) { + EGLint new_size = dpy->MaxConfigs + 16; + assert(dpy->NumConfigs < new_size); + + configs = realloc(dpy->Configs, new_size * sizeof(dpy->Configs[0])); + if (!configs) + return (EGLConfig) NULL; + + dpy->Configs = configs; + dpy->MaxConfigs = new_size; + } + + conf->Display = dpy; + dpy->Configs[dpy->NumConfigs++] = conf; + + return (EGLConfig) conf; } -/** - * Given an EGLConfig handle, return the corresponding _EGLConfig object. - * This is the inverse of _eglGetConfigHandle(). - */ -_EGLConfig * -_eglLookupConfig(EGLConfig config, _EGLDisplay *disp) +#ifndef _EGL_SKIP_HANDLE_CHECK + + +EGLBoolean +_eglCheckConfigHandle(EGLConfig config, _EGLDisplay *dpy) { + _EGLConfig *conf = NULL; EGLint i; - for (i = 0; i < disp->NumConfigs; i++) { - if (disp->Configs[i]->Handle == config) { - return disp->Configs[i]; + + for (i = 0; dpy && i < dpy->NumConfigs; i++) { + conf = dpy->Configs[i]; + if (conf == (_EGLConfig *) config) { + assert(conf->Display == dpy); + break; } } - return NULL; + + return (conf != NULL); } -/** - * Add the given _EGLConfig to the given display. - * Note that we just save the ptr to the config (we don't copy the config). - */ -_EGLConfig * -_eglAddConfig(_EGLDisplay *display, _EGLConfig *config) +#endif /* _EGL_SKIP_HANDLE_CHECK */ + + +enum { + /* types */ + ATTRIB_TYPE_INTEGER, + ATTRIB_TYPE_BOOLEAN, + ATTRIB_TYPE_BITMASK, + ATTRIB_TYPE_ENUM, + ATTRIB_TYPE_PSEUDO, /* non-queryable */ + ATTRIB_TYPE_PLATFORM, /* platform-dependent */ + /* criteria */ + ATTRIB_CRITERION_EXACT, + ATTRIB_CRITERION_ATLEAST, + ATTRIB_CRITERION_MASK, + ATTRIB_CRITERION_SPECIAL, + ATTRIB_CRITERION_IGNORE, +}; + + +/* EGL spec Table 3.1 and 3.4 */ +static const struct { + EGLint attr; + EGLint type; + EGLint criterion; + EGLint default_value; +} _eglValidationTable[] = { - _EGLConfig **newConfigs; - EGLint n; - - /* do some sanity checks on the config's attribs */ - assert(GET_CONFIG_ATTRIB(config, EGL_CONFIG_ID) > 0); - assert(GET_CONFIG_ATTRIB(config, EGL_RENDERABLE_TYPE) != 0x0); - assert(GET_CONFIG_ATTRIB(config, EGL_SURFACE_TYPE) != 0x0); - assert(GET_CONFIG_ATTRIB(config, EGL_RED_SIZE) > 0); - assert(GET_CONFIG_ATTRIB(config, EGL_GREEN_SIZE) > 0); - assert(GET_CONFIG_ATTRIB(config, EGL_BLUE_SIZE) > 0); - - n = display->NumConfigs; - - /* realloc array of ptrs */ - newConfigs = (_EGLConfig **) realloc(display->Configs, - (n + 1) * sizeof(_EGLConfig *)); - if (newConfigs) { - display->Configs = newConfigs; - display->Configs[n] = config; - display->NumConfigs++; - return config; - } - else { - return NULL; - } -} + { EGL_BUFFER_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_RED_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_GREEN_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_BLUE_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_LUMINANCE_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_ALPHA_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_ALPHA_MASK_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_BIND_TO_TEXTURE_RGB, ATTRIB_TYPE_BOOLEAN, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_BIND_TO_TEXTURE_RGBA, ATTRIB_TYPE_BOOLEAN, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_COLOR_BUFFER_TYPE, ATTRIB_TYPE_ENUM, + ATTRIB_CRITERION_EXACT, + EGL_RGB_BUFFER }, + { EGL_CONFIG_CAVEAT, ATTRIB_TYPE_ENUM, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_CONFIG_ID, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_CONFORMANT, ATTRIB_TYPE_BITMASK, + ATTRIB_CRITERION_MASK, + 0 }, + { EGL_DEPTH_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_LEVEL, ATTRIB_TYPE_PLATFORM, + ATTRIB_CRITERION_EXACT, + 0 }, + { EGL_MAX_PBUFFER_WIDTH, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_IGNORE, + 0 }, + { EGL_MAX_PBUFFER_HEIGHT, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_IGNORE, + 0 }, + { EGL_MAX_PBUFFER_PIXELS, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_IGNORE, + 0 }, + { EGL_MAX_SWAP_INTERVAL, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_MIN_SWAP_INTERVAL, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_NATIVE_RENDERABLE, ATTRIB_TYPE_BOOLEAN, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_NATIVE_VISUAL_ID, ATTRIB_TYPE_PLATFORM, + ATTRIB_CRITERION_IGNORE, + 0 }, + { EGL_NATIVE_VISUAL_TYPE, ATTRIB_TYPE_PLATFORM, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_RENDERABLE_TYPE, ATTRIB_TYPE_BITMASK, + ATTRIB_CRITERION_MASK, + EGL_OPENGL_ES_BIT }, + { EGL_SAMPLE_BUFFERS, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_SAMPLES, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_STENCIL_SIZE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_ATLEAST, + 0 }, + { EGL_SURFACE_TYPE, ATTRIB_TYPE_BITMASK, + ATTRIB_CRITERION_MASK, + EGL_WINDOW_BIT }, + { EGL_TRANSPARENT_TYPE, ATTRIB_TYPE_ENUM, + ATTRIB_CRITERION_EXACT, + EGL_NONE }, + { EGL_TRANSPARENT_RED_VALUE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_TRANSPARENT_GREEN_VALUE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + { EGL_TRANSPARENT_BLUE_VALUE, ATTRIB_TYPE_INTEGER, + ATTRIB_CRITERION_EXACT, + EGL_DONT_CARE }, + /* these are not real attributes */ + { EGL_MATCH_NATIVE_PIXMAP, ATTRIB_TYPE_PSEUDO, + ATTRIB_CRITERION_SPECIAL, + EGL_NONE }, + { EGL_PRESERVED_RESOURCES, ATTRIB_TYPE_PSEUDO, + ATTRIB_CRITERION_IGNORE, + 0 }, + { EGL_NONE, ATTRIB_TYPE_PSEUDO, + ATTRIB_CRITERION_IGNORE, + 0 } +}; /** - * Parse the attrib_list to fill in the fields of the given _eglConfig - * Return EGL_FALSE if any errors, EGL_TRUE otherwise. + * Return true if a config is valid. When for_matching is true, + * EGL_DONT_CARE is accepted as a valid attribute value, and checks + * for conflicting attribute values are skipped. + * + * Note that some attributes are platform-dependent and are not + * checked. */ EGLBoolean -_eglParseConfigAttribs(_EGLConfig *config, const EGLint *attrib_list) +_eglValidateConfig(const _EGLConfig *conf, EGLBoolean for_matching) { - EGLint i; - - /* set all config attribs to EGL_DONT_CARE */ - for (i = 0; i < MAX_ATTRIBS; i++) { - config->Attrib[i] = EGL_DONT_CARE; - } - - /* by default choose windows unless otherwise specified */ - config->Attrib[EGL_SURFACE_TYPE - FIRST_ATTRIB] = EGL_WINDOW_BIT; - - for (i = 0; attrib_list && attrib_list[i] != EGL_NONE; i++) { - const EGLint attr = attrib_list[i]; - if (attr >= EGL_BUFFER_SIZE && - attr <= EGL_MAX_SWAP_INTERVAL) { - EGLint k = attr - FIRST_ATTRIB; - assert(k >= 0); - assert(k < MAX_ATTRIBS); - config->Attrib[k] = attrib_list[++i]; - } -#ifdef EGL_VERSION_1_2 - else if (attr == EGL_COLOR_BUFFER_TYPE) { - EGLint bufType = attrib_list[++i]; - if (bufType != EGL_RGB_BUFFER && bufType != EGL_LUMINANCE_BUFFER) { - _eglError(EGL_BAD_ATTRIBUTE, "eglChooseConfig"); - return EGL_FALSE; + EGLint i, attr, val; + EGLBoolean valid = EGL_TRUE; + EGLint red_size = 0, green_size = 0, blue_size = 0, luminance_size = 0; + EGLint alpha_size = 0, buffer_size = 0; + + /* all attributes should have been listed */ + assert(ARRAY_SIZE(_eglValidationTable) == _EGL_CONFIG_NUM_ATTRIBS); + + /* check attributes by their types */ + for (i = 0; i < ARRAY_SIZE(_eglValidationTable); i++) { + EGLint mask; + + attr = _eglValidationTable[i].attr; + val = GET_CONFIG_ATTRIB(conf, attr); + + switch (_eglValidationTable[i].type) { + case ATTRIB_TYPE_INTEGER: + switch (attr) { + case EGL_CONFIG_ID: + /* config id must be positive */ + if (val <= 0) + valid = EGL_FALSE; + break; + case EGL_SAMPLE_BUFFERS: + /* there can be at most 1 sample buffer */ + if (val > 1) + valid = EGL_FALSE; + break; + case EGL_RED_SIZE: + red_size = val; + break; + case EGL_GREEN_SIZE: + green_size = val; + break; + case EGL_BLUE_SIZE: + blue_size = val; + break; + case EGL_LUMINANCE_SIZE: + luminance_size = val; + break; + case EGL_ALPHA_SIZE: + alpha_size = val; + break; + case EGL_BUFFER_SIZE: + buffer_size = val; + break; } - _eglSetConfigAttrib(config, EGL_COLOR_BUFFER_TYPE, bufType); - } - else if (attr == EGL_RENDERABLE_TYPE) { - EGLint renType = attrib_list[++i]; - if (renType & ~(EGL_OPENGL_ES_BIT | EGL_OPENGL_ES2_BIT | EGL_OPENVG_BIT)) { - _eglError(EGL_BAD_ATTRIBUTE, "eglChooseConfig"); - return EGL_FALSE; + if (val < 0) + valid = EGL_FALSE; + break; + case ATTRIB_TYPE_BOOLEAN: + if (val != EGL_TRUE && val != EGL_FALSE) + valid = EGL_FALSE; + break; + case ATTRIB_TYPE_ENUM: + switch (attr) { + case EGL_CONFIG_CAVEAT: + if (val != EGL_NONE && val != EGL_SLOW_CONFIG && + val != EGL_NON_CONFORMANT_CONFIG) + valid = EGL_FALSE; + break; + case EGL_TRANSPARENT_TYPE: + if (val != EGL_NONE && val != EGL_TRANSPARENT_RGB) + valid = EGL_FALSE; + break; + case EGL_COLOR_BUFFER_TYPE: + if (val != EGL_RGB_BUFFER && val != EGL_LUMINANCE_BUFFER) + valid = EGL_FALSE; + break; + default: + assert(0); + break; } - _eglSetConfigAttrib(config, EGL_RENDERABLE_TYPE, renType); - } - else if (attr == EGL_ALPHA_MASK_SIZE || - attr == EGL_LUMINANCE_SIZE) { - EGLint value = attrib_list[++i]; - _eglSetConfigAttrib(config, attr, value); + break; + case ATTRIB_TYPE_BITMASK: + switch (attr) { + case EGL_SURFACE_TYPE: + mask = EGL_PBUFFER_BIT | + EGL_PIXMAP_BIT | + EGL_WINDOW_BIT | + EGL_VG_COLORSPACE_LINEAR_BIT | + EGL_VG_ALPHA_FORMAT_PRE_BIT | + EGL_MULTISAMPLE_RESOLVE_BOX_BIT | + EGL_SWAP_BEHAVIOR_PRESERVED_BIT; + break; + case EGL_RENDERABLE_TYPE: + case EGL_CONFORMANT: + mask = EGL_OPENGL_ES_BIT | + EGL_OPENVG_BIT | + EGL_OPENGL_ES2_BIT | + EGL_OPENGL_BIT; + break; + default: + assert(0); + break; + } + if (val & ~mask) + valid = EGL_FALSE; + break; + case ATTRIB_TYPE_PLATFORM: + /* unable to check platform-dependent attributes here */ + break; + case ATTRIB_TYPE_PSEUDO: + /* pseudo attributes should not be set */ + if (val != 0) + valid = EGL_FALSE; + break; + default: + assert(0); + break; } -#endif /* EGL_VERSION_1_2 */ - else { - _eglError(EGL_BAD_ATTRIBUTE, "eglChooseConfig"); - return EGL_FALSE; + + if (!valid && for_matching) { + /* accept EGL_DONT_CARE as a valid value */ + if (val == EGL_DONT_CARE) + valid = EGL_TRUE; + if (_eglValidationTable[i].criterion == ATTRIB_CRITERION_SPECIAL) + valid = EGL_TRUE; } + if (!valid) + break; } - return EGL_TRUE; -} + /* any invalid attribute value should have been catched */ + if (!valid || for_matching) + return valid; + + /* now check for conflicting attribute values */ + + switch (GET_CONFIG_ATTRIB(conf, EGL_COLOR_BUFFER_TYPE)) { + case EGL_RGB_BUFFER: + if (luminance_size) + valid = EGL_FALSE; + if (red_size + green_size + blue_size + alpha_size != buffer_size) + valid = EGL_FALSE; + break; + case EGL_LUMINANCE_BUFFER: + if (red_size || green_size || blue_size) + valid = EGL_FALSE; + if (luminance_size + alpha_size != buffer_size) + valid = EGL_FALSE; + break; + } -#define EXACT 1 -#define ATLEAST 2 -#define MASK 3 -#define SMALLER 4 -#define SPECIAL 5 -#define NONE 6 + val = GET_CONFIG_ATTRIB(conf, EGL_SAMPLE_BUFFERS); + if (!val && GET_CONFIG_ATTRIB(conf, EGL_SAMPLES)) + valid = EGL_FALSE; + + val = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE); + if (!(val & EGL_WINDOW_BIT)) { + if (GET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_ID) != 0 || + GET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_TYPE) != EGL_NONE) + valid = EGL_FALSE; + } + if (!(val & EGL_PBUFFER_BIT)) { + if (GET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGB) || + GET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGBA)) + valid = EGL_FALSE; + } + + return valid; +} -struct sort_info { - EGLint Attribute; - EGLint MatchCriteria; - EGLint SortOrder; -}; -/* This encodes the info from Table 3.5 of the EGL spec, ordered by - * Sort Priority. +/** + * Return true if a config matches the criteria. This and + * _eglParseConfigAttribList together implement the algorithm + * described in "Selection of EGLConfigs". * - * XXX To do: EGL 1.2 attribs + * Note that attributes that are special (currently, only + * EGL_MATCH_NATIVE_PIXMAP) are ignored. */ -static struct sort_info SortInfo[] = { - { EGL_CONFIG_CAVEAT, EXACT, SPECIAL }, - { EGL_RED_SIZE, ATLEAST, SPECIAL }, - { EGL_GREEN_SIZE, ATLEAST, SPECIAL }, - { EGL_BLUE_SIZE, ATLEAST, SPECIAL }, - { EGL_ALPHA_SIZE, ATLEAST, SPECIAL }, - { EGL_BUFFER_SIZE, ATLEAST, SMALLER }, - { EGL_SAMPLE_BUFFERS, ATLEAST, SMALLER }, - { EGL_SAMPLES, ATLEAST, SMALLER }, - { EGL_DEPTH_SIZE, ATLEAST, SMALLER }, - { EGL_STENCIL_SIZE, ATLEAST, SMALLER }, - { EGL_NATIVE_VISUAL_TYPE, EXACT, SPECIAL }, - { EGL_CONFIG_ID, EXACT, SMALLER }, - { EGL_BIND_TO_TEXTURE_RGB, EXACT, NONE }, - { EGL_BIND_TO_TEXTURE_RGBA, EXACT, NONE }, - { EGL_LEVEL, EXACT, NONE }, - { EGL_NATIVE_RENDERABLE, EXACT, NONE }, - { EGL_MAX_SWAP_INTERVAL, EXACT, NONE }, - { EGL_MIN_SWAP_INTERVAL, EXACT, NONE }, - { EGL_SURFACE_TYPE, MASK, NONE }, - { EGL_TRANSPARENT_TYPE, EXACT, NONE }, - { EGL_TRANSPARENT_RED_VALUE, EXACT, NONE }, - { EGL_TRANSPARENT_GREEN_VALUE, EXACT, NONE }, - { EGL_TRANSPARENT_BLUE_VALUE, EXACT, NONE }, - { 0, 0, 0 } -}; +EGLBoolean +_eglMatchConfig(const _EGLConfig *conf, const _EGLConfig *criteria) +{ + EGLint attr, val, i; + EGLBoolean matched = EGL_TRUE; + + for (i = 0; i < ARRAY_SIZE(_eglValidationTable); i++) { + EGLint cmp; + if (_eglValidationTable[i].criterion == ATTRIB_CRITERION_IGNORE) + continue; + + attr = _eglValidationTable[i].attr; + cmp = GET_CONFIG_ATTRIB(criteria, attr); + if (cmp == EGL_DONT_CARE) + continue; + + val = GET_CONFIG_ATTRIB(conf, attr); + switch (_eglValidationTable[i].criterion) { + case ATTRIB_CRITERION_EXACT: + if (val != cmp) + matched = EGL_FALSE; + break; + case ATTRIB_CRITERION_ATLEAST: + if (val < cmp) + matched = EGL_FALSE; + break; + case ATTRIB_CRITERION_MASK: + if ((val & cmp) != cmp) + matched = EGL_FALSE; + break; + case ATTRIB_CRITERION_SPECIAL: + /* ignored here */ + break; + default: + assert(0); + break; + } + + if (!matched) + break; + } + + return matched; +} /** - * Return EGL_TRUE if the attributes of c meet or exceed the minimums - * specified by min. + * Initialize a criteria config from the given attribute list. + * Return EGL_FALSE if any of the attribute is invalid. */ -static EGLBoolean -_eglConfigQualifies(const _EGLConfig *c, const _EGLConfig *min) +EGLBoolean +_eglParseConfigAttribList(_EGLConfig *conf, const EGLint *attrib_list) { - EGLint i; - for (i = 0; SortInfo[i].Attribute != 0; i++) { - const EGLint mv = GET_CONFIG_ATTRIB(min, SortInfo[i].Attribute); - if (mv != EGL_DONT_CARE) { - const EGLint cv = GET_CONFIG_ATTRIB(c, SortInfo[i].Attribute); - if (SortInfo[i].MatchCriteria == EXACT) { - if (cv != mv) { - return EGL_FALSE; - } - } - else if (SortInfo[i].MatchCriteria == ATLEAST) { - if (cv < mv) { - return EGL_FALSE; - } - } - else { - assert(SortInfo[i].MatchCriteria == MASK); - if ((mv & cv) != mv) { - return EGL_FALSE; - } + EGLint attr, val, i; + EGLint config_id = 0, level = 0; + EGLBoolean has_native_visual_type = EGL_FALSE; + EGLBoolean has_transparent_color = EGL_FALSE; + + /* reset to default values */ + for (i = 0; i < ARRAY_SIZE(_eglValidationTable); i++) { + attr = _eglValidationTable[i].attr; + val = _eglValidationTable[i].default_value; + SET_CONFIG_ATTRIB(conf, attr, val); + } + + /* parse the list */ + for (i = 0; attrib_list && attrib_list[i] != EGL_NONE; i += 2) { + EGLint idx; + + attr = attrib_list[i]; + val = attrib_list[i + 1]; + + idx = _eglIndexConfig(conf, attr); + if (idx < 0) + return EGL_FALSE; + conf->Storage[idx] = val; + + /* rememeber some attributes for post-processing */ + switch (attr) { + case EGL_CONFIG_ID: + config_id = val; + break; + case EGL_LEVEL: + level = val; + break; + case EGL_NATIVE_VISUAL_TYPE: + has_native_visual_type = EGL_TRUE; + break; + case EGL_TRANSPARENT_RED_VALUE: + case EGL_TRANSPARENT_GREEN_VALUE: + case EGL_TRANSPARENT_BLUE_VALUE: + has_transparent_color = EGL_TRUE; + break; + default: + break; + } + } + + if (!_eglValidateConfig(conf, EGL_TRUE)) + return EGL_FALSE; + + /* the spec says that EGL_LEVEL cannot be EGL_DONT_CARE */ + if (level == EGL_DONT_CARE) + return EGL_FALSE; + + /* ignore other attributes when EGL_CONFIG_ID is given */ + if (config_id > 0) { + _eglResetConfigKeys(conf, EGL_DONT_CARE); + SET_CONFIG_ATTRIB(conf, EGL_CONFIG_ID, config_id); + } + else { + if (has_native_visual_type) { + val = GET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE); + if (!(val & EGL_WINDOW_BIT)) + SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_TYPE, EGL_DONT_CARE); + } + + if (has_transparent_color) { + val = GET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_TYPE); + if (val == EGL_NONE) { + SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_RED_VALUE, + EGL_DONT_CARE); + SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_GREEN_VALUE, + EGL_DONT_CARE); + SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_BLUE_VALUE, + EGL_DONT_CARE); } } } + return EGL_TRUE; } /** - * Compare configs 'a' and 'b' and return -1 if a belongs before b, - * 1 if a belongs after b, or 0 if they're equal. - * Used by qsort(). + * Decide the ordering of conf1 and conf2, under the given criteria. + * When compare_id is true, this implements the algorithm described + * in "Sorting of EGLConfigs". When compare_id is false, + * EGL_CONFIG_ID is not compared. + * + * It returns a negative integer if conf1 is considered to come + * before conf2; a positive integer if conf2 is considered to come + * before conf1; zero if the ordering cannot be decided. + * + * Note that EGL_NATIVE_VISUAL_TYPE is platform-dependent and is + * ignored here. */ -static int -_eglCompareConfigs(const void *a, const void *b) +EGLint +_eglCompareConfigs(const _EGLConfig *conf1, const _EGLConfig *conf2, + const _EGLConfig *criteria, EGLBoolean compare_id) { - const _EGLConfig *aConfig = (const _EGLConfig *) a; - const _EGLConfig *bConfig = (const _EGLConfig *) b; + const EGLint compare_attribs[] = { + EGL_BUFFER_SIZE, + EGL_SAMPLE_BUFFERS, + EGL_SAMPLES, + EGL_DEPTH_SIZE, + EGL_STENCIL_SIZE, + EGL_ALPHA_MASK_SIZE, + }; + EGLint val1, val2; + EGLBoolean rgb_buffer; EGLint i; - for (i = 0; SortInfo[i].Attribute != 0; i++) { - const EGLint aVal = GET_CONFIG_ATTRIB(aConfig, SortInfo[i].Attribute); - const EGLint bVal = GET_CONFIG_ATTRIB(bConfig, SortInfo[i].Attribute); - if (SortInfo[i].SortOrder == SMALLER) { - if (aVal < bVal) - return -1; - else if (aVal > bVal) - return 1; - /* else, continue examining attribute values */ - } - else if (SortInfo[i].SortOrder == SPECIAL) { - if (SortInfo[i].Attribute == EGL_CONFIG_CAVEAT) { - /* values are EGL_NONE, SLOW_CONFIG, or NON_CONFORMANT_CONFIG */ - if (aVal < bVal) - return -1; - else if (aVal > bVal) - return 1; + if (conf1 == conf2) + return 0; + + /* the enum values have the desired ordering */ + assert(EGL_NONE < EGL_SLOW_CONFIG); + assert(EGL_SLOW_CONFIG < EGL_NON_CONFORMANT_CONFIG); + val1 = GET_CONFIG_ATTRIB(conf1, EGL_CONFIG_CAVEAT); + val2 = GET_CONFIG_ATTRIB(conf2, EGL_CONFIG_CAVEAT); + if (val1 != val2) + return (val1 - val2); + + /* the enum values have the desired ordering */ + assert(EGL_RGB_BUFFER < EGL_LUMINANCE_BUFFER); + val1 = GET_CONFIG_ATTRIB(conf1, EGL_COLOR_BUFFER_TYPE); + val2 = GET_CONFIG_ATTRIB(conf2, EGL_COLOR_BUFFER_TYPE); + if (val1 != val2) + return (val1 - val2); + rgb_buffer = (val1 == EGL_RGB_BUFFER); + + if (criteria) { + val1 = val2 = 0; + if (rgb_buffer) { + if (GET_CONFIG_ATTRIB(criteria, EGL_RED_SIZE) > 0) { + val1 += GET_CONFIG_ATTRIB(conf1, EGL_RED_SIZE); + val2 += GET_CONFIG_ATTRIB(conf2, EGL_RED_SIZE); } - else if (SortInfo[i].Attribute == EGL_RED_SIZE || - SortInfo[i].Attribute == EGL_GREEN_SIZE || - SortInfo[i].Attribute == EGL_BLUE_SIZE || - SortInfo[i].Attribute == EGL_ALPHA_SIZE) { - if (aVal > bVal) - return -1; - else if (aVal < bVal) - return 1; + if (GET_CONFIG_ATTRIB(criteria, EGL_GREEN_SIZE) > 0) { + val1 += GET_CONFIG_ATTRIB(conf1, EGL_GREEN_SIZE); + val2 += GET_CONFIG_ATTRIB(conf2, EGL_GREEN_SIZE); } - else { - assert(SortInfo[i].Attribute == EGL_NATIVE_VISUAL_TYPE); - if (aVal < bVal) - return -1; - else if (aVal > bVal) - return 1; + if (GET_CONFIG_ATTRIB(criteria, EGL_BLUE_SIZE) > 0) { + val1 += GET_CONFIG_ATTRIB(conf1, EGL_BLUE_SIZE); + val2 += GET_CONFIG_ATTRIB(conf2, EGL_BLUE_SIZE); } } else { - assert(SortInfo[i].SortOrder == NONE); - /* continue examining attribute values */ + if (GET_CONFIG_ATTRIB(criteria, EGL_LUMINANCE_SIZE) > 0) { + val1 += GET_CONFIG_ATTRIB(conf1, EGL_LUMINANCE_SIZE); + val2 += GET_CONFIG_ATTRIB(conf2, EGL_LUMINANCE_SIZE); + } + } + if (GET_CONFIG_ATTRIB(criteria, EGL_ALPHA_SIZE) > 0) { + val1 += GET_CONFIG_ATTRIB(conf1, EGL_ALPHA_SIZE); + val2 += GET_CONFIG_ATTRIB(conf2, EGL_ALPHA_SIZE); } } + else { + /* assume the default criteria, which gives no specific ordering */ + val1 = val2 = 0; + } + + /* for color bits, larger one is preferred */ + if (val1 != val2) + return (val2 - val1); + + for (i = 0; i < ARRAY_SIZE(compare_attribs); i++) { + val1 = GET_CONFIG_ATTRIB(conf1, compare_attribs[i]); + val2 = GET_CONFIG_ATTRIB(conf2, compare_attribs[i]); + if (val1 != val2) + return (val1 - val2); + } + + /* EGL_NATIVE_VISUAL_TYPE cannot be compared here */ + + if (compare_id) { + val1 = GET_CONFIG_ATTRIB(conf1, EGL_CONFIG_ID); + val2 = GET_CONFIG_ATTRIB(conf2, EGL_CONFIG_ID); + assert(val1 != val2); + } + else { + val1 = val2 = 0; + } + + return (val1 - val2); +} + + +static INLINE +void _eglSwapConfigs(const _EGLConfig **conf1, const _EGLConfig **conf2) +{ + const _EGLConfig *tmp = *conf1; + *conf1 = *conf2; + *conf2 = tmp; +} + + +/** + * Quick sort an array of configs. This differs from the standard + * qsort() in that the compare function accepts an additional + * argument. + */ +void +_eglSortConfigs(const _EGLConfig **configs, EGLint count, + EGLint (*compare)(const _EGLConfig *, const _EGLConfig *, + void *), + void *priv_data) +{ + const EGLint pivot = 0; + EGLint i, j; + + if (count <= 1) + return; + + _eglSwapConfigs(&configs[pivot], &configs[count / 2]); + i = 1; + j = count - 1; + do { + while (i < count && compare(configs[i], configs[pivot], priv_data) < 0) + i++; + while (compare(configs[j], configs[pivot], priv_data) > 0) + j--; + if (i < j) { + _eglSwapConfigs(&configs[i], &configs[j]); + i++; + j--; + } + else if (i == j) { + i++; + j--; + break; + } + } while (i <= j); + _eglSwapConfigs(&configs[pivot], &configs[j]); - /* all attributes identical */ - return 0; + _eglSortConfigs(configs, j, compare, priv_data); + _eglSortConfigs(configs + i, count - i, compare, priv_data); +} + + +static int +_eglFallbackCompare(const _EGLConfig *conf1, const _EGLConfig *conf2, + void *priv_data) +{ + const _EGLConfig *criteria = (const _EGLConfig *) priv_data; + return _eglCompareConfigs(conf1, conf2, criteria, EGL_TRUE); } @@ -324,33 +728,33 @@ _eglChooseConfig(_EGLDriver *drv, _EGLDisplay *disp, const EGLint *attrib_list, _EGLConfig **configList, criteria; EGLint i, count; - /* parse the attrib_list to initialize criteria */ - if (!_eglParseConfigAttribs(&criteria, attrib_list)) { - return EGL_FALSE; - } + if (!num_configs) + return _eglError(EGL_BAD_PARAMETER, "eglChooseConfigs"); + + _eglInitConfig(&criteria, 0); + if (!_eglParseConfigAttribList(&criteria, attrib_list)) + return _eglError(EGL_BAD_ATTRIBUTE, "eglChooseConfig"); /* allocate array of config pointers */ - configList = (_EGLConfig **) malloc(config_size * sizeof(_EGLConfig *)); - if (!configList) { - _eglError(EGL_BAD_CONFIG, "eglChooseConfig(out of memory)"); - return EGL_FALSE; - } + configList = (_EGLConfig **) + malloc(disp->NumConfigs * sizeof(_EGLConfig *)); + if (!configList) + return _eglError(EGL_BAD_ALLOC, "eglChooseConfig(out of memory)"); - /* make array of pointers to qualifying configs */ - for (i = count = 0; i < disp->NumConfigs && count < config_size; i++) { - if (_eglConfigQualifies(disp->Configs[i], &criteria)) { + /* perform selection of configs */ + count = 0; + for (i = 0; i < disp->NumConfigs; i++) { + if (_eglMatchConfig(disp->Configs[i], &criteria)) configList[count++] = disp->Configs[i]; - } } - /* sort array of pointers */ - qsort(configList, count, sizeof(_EGLConfig *), _eglCompareConfigs); - - /* copy config handles to output array */ - if (configs) { - for (i = 0; i < count; i++) { - configs[i] = configList[i]->Handle; - } + /* perform sorting of configs */ + if (configs && count) { + _eglSortConfigs((const _EGLConfig **) configList, count, + _eglFallbackCompare, (void *) &criteria); + count = MIN2(count, config_size); + for (i = 0; i < count; i++) + configs[i] = _eglGetConfigHandle(configList[i]); } free(configList); @@ -361,6 +765,28 @@ _eglChooseConfig(_EGLDriver *drv, _EGLDisplay *disp, const EGLint *attrib_list, } +static INLINE EGLBoolean +_eglIsConfigAttribValid(_EGLConfig *conf, EGLint attr) +{ + if (_eglIndexConfig(conf, attr) < 0) + return EGL_FALSE; + + /* there are some holes in the range */ + switch (attr) { + case EGL_PRESERVED_RESOURCES: + case EGL_NONE: +#ifdef EGL_VERSION_1_4 + case EGL_MATCH_NATIVE_PIXMAP: +#endif + return EGL_FALSE; + default: + break; + } + + return EGL_TRUE; +} + + /** * Fallback for eglGetConfigAttrib. */ @@ -368,15 +794,13 @@ EGLBoolean _eglGetConfigAttrib(_EGLDriver *drv, _EGLDisplay *dpy, _EGLConfig *conf, EGLint attribute, EGLint *value) { - const EGLint k = attribute - FIRST_ATTRIB; - if (k >= 0 && k < MAX_ATTRIBS) { - *value = conf->Attrib[k]; - return EGL_TRUE; - } - else { - _eglError(EGL_BAD_ATTRIBUTE, "eglGetConfigAttrib"); - return EGL_FALSE; - } + if (!_eglIsConfigAttribValid(conf, attribute)) + return _eglError(EGL_BAD_ATTRIBUTE, "eglGetConfigAttrib"); + if (!value) + return _eglError(EGL_BAD_PARAMETER, "eglGetConfigAttrib"); + + *value = GET_CONFIG_ATTRIB(conf, attribute); + return EGL_TRUE; } @@ -387,12 +811,14 @@ EGLBoolean _eglGetConfigs(_EGLDriver *drv, _EGLDisplay *disp, EGLConfig *configs, EGLint config_size, EGLint *num_config) { + if (!num_config) + return _eglError(EGL_BAD_PARAMETER, "eglGetConfigs"); + if (configs) { EGLint i; *num_config = MIN2(disp->NumConfigs, config_size); - for (i = 0; i < *num_config; i++) { - configs[i] = disp->Configs[i]->Handle; - } + for (i = 0; i < *num_config; i++) + configs[i] = _eglGetConfigHandle(disp->Configs[i]); } else { /* just return total number of supported configs */ diff --git a/src/egl/main/eglconfig.h b/src/egl/main/eglconfig.h index 36ed96ae956..6b8a259984e 100644 --- a/src/egl/main/eglconfig.h +++ b/src/egl/main/eglconfig.h @@ -2,27 +2,93 @@ #define EGLCONFIG_INCLUDED +#include <assert.h> #include "egltypedefs.h" -#include <GLES/gl.h> -#define MAX_ATTRIBS 128 -#define FIRST_ATTRIB EGL_BUFFER_SIZE +#define _EGL_CONFIG_FIRST_ATTRIB EGL_BUFFER_SIZE +#define _EGL_CONFIG_LAST_ATTRIB EGL_CONFORMANT +#define _EGL_CONFIG_NUM_ATTRIBS \ + (_EGL_CONFIG_LAST_ATTRIB - _EGL_CONFIG_FIRST_ATTRIB + 1) + +#define _EGL_CONFIG_STORAGE_SIZE _EGL_CONFIG_NUM_ATTRIBS struct _egl_config { - EGLConfig Handle; /* the public/opaque handle which names this config */ - EGLint Attrib[MAX_ATTRIBS]; + _EGLDisplay *Display; + EGLint Storage[_EGL_CONFIG_STORAGE_SIZE]; }; -#define SET_CONFIG_ATTRIB(CONF, ATTR, VAL) \ - assert((ATTR) - FIRST_ATTRIB < MAX_ATTRIBS); \ - ((CONF)->Attrib[(ATTR) - FIRST_ATTRIB] = VAL) +#define SET_CONFIG_ATTRIB(CONF, ATTR, VAL) _eglSetConfigKey(CONF, ATTR, VAL) +#define GET_CONFIG_ATTRIB(CONF, ATTR) _eglGetConfigKey(CONF, ATTR) + + +/** + * Given a key, return an index into the storage of the config. + * Return -1 if the key is invalid. + */ +static INLINE EGLint +_eglIndexConfig(const _EGLConfig *conf, EGLint key) +{ + (void) conf; + if (key >= _EGL_CONFIG_FIRST_ATTRIB && + key < _EGL_CONFIG_FIRST_ATTRIB + _EGL_CONFIG_NUM_ATTRIBS) + return key - _EGL_CONFIG_FIRST_ATTRIB; + else + return -1; +} + + +/** + * Reset all keys in the config to a given value. + */ +static INLINE void +_eglResetConfigKeys(_EGLConfig *conf, EGLint val) +{ + EGLint i; + for (i = 0; i < _EGL_CONFIG_NUM_ATTRIBS; i++) + conf->Storage[i] = val; +} -#define GET_CONFIG_ATTRIB(CONF, ATTR) ((CONF)->Attrib[(ATTR) - FIRST_ATTRIB]) +/** + * Update a config for a given key. + */ +static INLINE void +_eglSetConfigKey(_EGLConfig *conf, EGLint key, EGLint val) +{ + EGLint idx = _eglIndexConfig(conf, key); + assert(idx >= 0); + conf->Storage[idx] = val; +} + + +/** + * Return the value for a given key. + */ +static INLINE EGLint +_eglGetConfigKey(const _EGLConfig *conf, EGLint key) +{ + EGLint idx = _eglIndexConfig(conf, key); + assert(idx >= 0); + return conf->Storage[idx]; +} + + +/** + * Set a given attribute. + * + * Because _eglGetConfigAttrib is already used as a fallback driver + * function, this function is not considered to have a good name. + * SET_CONFIG_ATTRIB is preferred over this function. + */ +static INLINE void +_eglSetConfigAttrib(_EGLConfig *conf, EGLint attr, EGLint val) +{ + SET_CONFIG_ATTRIB(conf, attr, val); +} extern void @@ -30,19 +96,76 @@ _eglInitConfig(_EGLConfig *config, EGLint id); extern EGLConfig -_eglGetConfigHandle(_EGLConfig *config); +_eglAddConfig(_EGLDisplay *dpy, _EGLConfig *conf); + + +#ifndef _EGL_SKIP_HANDLE_CHECK + + +extern EGLBoolean +_eglCheckConfigHandle(EGLConfig config, _EGLDisplay *dpy); + + +#else + +static INLINE EGLBoolean +_eglCheckConfigHandle(EGLConfig config, _EGLDisplay *dpy) +{ + _EGLConfig *conf = (_EGLConfig *) config; + return (dpy && conf && conf->Display == dpy); +} + + +#endif /* _EGL_SKIP_HANDLE_CHECK */ -extern _EGLConfig * -_eglLookupConfig(EGLConfig config, _EGLDisplay *dpy); + +/** + * Lookup a handle to find the linked config. + * Return NULL if the handle has no corresponding linked config. + */ +static INLINE _EGLConfig * +_eglLookupConfig(EGLConfig config, _EGLDisplay *dpy) +{ + _EGLConfig *conf = (_EGLConfig *) config; + if (!_eglCheckConfigHandle(config, dpy)) + conf = NULL; + return conf; +} + + +/** + * Return the handle of a linked config, or NULL. + */ +static INLINE EGLConfig +_eglGetConfigHandle(_EGLConfig *conf) +{ + return (EGLConfig) ((conf && conf->Display) ? conf : NULL); +} + + +extern EGLBoolean +_eglValidateConfig(const _EGLConfig *conf, EGLBoolean for_matching); -extern _EGLConfig * -_eglAddConfig(_EGLDisplay *display, _EGLConfig *config); +extern EGLBoolean +_eglMatchConfig(const _EGLConfig *conf, const _EGLConfig *criteria); extern EGLBoolean -_eglParseConfigAttribs(_EGLConfig *config, const EGLint *attrib_list); +_eglParseConfigAttribList(_EGLConfig *conf, const EGLint *attrib_list); + + +extern EGLint +_eglCompareConfigs(const _EGLConfig *conf1, const _EGLConfig *conf2, + const _EGLConfig *criteria, EGLBoolean compare_id); + + +extern void +_eglSortConfigs(const _EGLConfig **configs, EGLint count, + EGLint (*compare)(const _EGLConfig *, const _EGLConfig *, + void *), + void *priv_data); extern EGLBoolean @@ -57,8 +180,4 @@ extern EGLBoolean _eglGetConfigs(_EGLDriver *drv, _EGLDisplay *dpy, EGLConfig *configs, EGLint config_size, EGLint *num_config); -extern void -_eglSetConfigAttrib(_EGLConfig *config, EGLint attr, EGLint val); - - #endif /* EGLCONFIG_INCLUDED */ diff --git a/src/egl/main/eglconfigutil.c b/src/egl/main/eglconfigutil.c index c9d00e79826..a5fcdcd2876 100644 --- a/src/egl/main/eglconfigutil.c +++ b/src/egl/main/eglconfigutil.c @@ -52,6 +52,78 @@ _eglConfigToContextModesRec(const _EGLConfig *config, __GLcontextModes *mode) } +/** + * Convert a __GLcontextModes object to an _EGLConfig. + */ +EGLBoolean +_eglConfigFromContextModesRec(_EGLConfig *conf, const __GLcontextModes *m, + EGLint conformant, EGLint renderable_type) +{ + EGLint config_caveat, surface_type; + + /* must be RGBA */ + if (!m->rgbMode || !(m->renderType & GLX_RGBA_BIT)) + return EGL_FALSE; + + config_caveat = EGL_NONE; + if (m->visualRating == GLX_SLOW_CONFIG) + config_caveat = EGL_SLOW_CONFIG; + + if (m->visualRating == GLX_NON_CONFORMANT_CONFIG) + conformant &= ~EGL_OPENGL_BIT; + if (!(conformant & EGL_OPENGL_ES_BIT)) + config_caveat = EGL_NON_CONFORMANT_CONFIG; + + surface_type = 0; + if (m->drawableType & GLX_WINDOW_BIT) + surface_type |= EGL_WINDOW_BIT; + if (m->drawableType & GLX_PIXMAP_BIT) + surface_type |= EGL_PIXMAP_BIT; + if (m->drawableType & GLX_PBUFFER_BIT) + surface_type |= EGL_PBUFFER_BIT; + + SET_CONFIG_ATTRIB(conf, EGL_BUFFER_SIZE, m->rgbBits); + SET_CONFIG_ATTRIB(conf, EGL_RED_SIZE, m->redBits); + SET_CONFIG_ATTRIB(conf, EGL_GREEN_SIZE, m->greenBits); + SET_CONFIG_ATTRIB(conf, EGL_BLUE_SIZE, m->blueBits); + SET_CONFIG_ATTRIB(conf, EGL_ALPHA_SIZE, m->alphaBits); + + SET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGB, m->bindToTextureRgb); + SET_CONFIG_ATTRIB(conf, EGL_BIND_TO_TEXTURE_RGBA, m->bindToTextureRgba); + SET_CONFIG_ATTRIB(conf, EGL_COLOR_BUFFER_TYPE, EGL_RGB_BUFFER); + SET_CONFIG_ATTRIB(conf, EGL_CONFIG_CAVEAT, config_caveat); + + SET_CONFIG_ATTRIB(conf, EGL_CONFORMANT, conformant); + SET_CONFIG_ATTRIB(conf, EGL_DEPTH_SIZE, m->depthBits); + SET_CONFIG_ATTRIB(conf, EGL_LEVEL, m->level); + SET_CONFIG_ATTRIB(conf, EGL_MAX_PBUFFER_WIDTH, m->maxPbufferWidth); + SET_CONFIG_ATTRIB(conf, EGL_MAX_PBUFFER_HEIGHT, m->maxPbufferHeight); + SET_CONFIG_ATTRIB(conf, EGL_MAX_PBUFFER_PIXELS, m->maxPbufferPixels); + + SET_CONFIG_ATTRIB(conf, EGL_NATIVE_RENDERABLE, m->xRenderable); + SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_ID, m->visualID); + SET_CONFIG_ATTRIB(conf, EGL_NATIVE_VISUAL_TYPE, m->visualType); + SET_CONFIG_ATTRIB(conf, EGL_RENDERABLE_TYPE, renderable_type); + SET_CONFIG_ATTRIB(conf, EGL_SAMPLE_BUFFERS, m->sampleBuffers); + SET_CONFIG_ATTRIB(conf, EGL_SAMPLES, m->samples); + SET_CONFIG_ATTRIB(conf, EGL_STENCIL_SIZE, m->stencilBits); + + SET_CONFIG_ATTRIB(conf, EGL_SURFACE_TYPE, surface_type); + + /* what to do with GLX_TRANSPARENT_INDEX? */ + if (m->transparentPixel == GLX_TRANSPARENT_RGB) { + SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_TYPE, EGL_TRANSPARENT_RGB); + SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_RED_VALUE, m->transparentRed); + SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_GREEN_VALUE, m->transparentGreen); + SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_BLUE_VALUE, m->transparentBlue); + } + else { + SET_CONFIG_ATTRIB(conf, EGL_TRANSPARENT_TYPE, EGL_NONE); + } + + return EGL_TRUE; +} + /** * Creates a set of \c _EGLConfigs that a driver will expose. diff --git a/src/egl/main/eglconfigutil.h b/src/egl/main/eglconfigutil.h index c477b94737d..ad850798908 100644 --- a/src/egl/main/eglconfigutil.h +++ b/src/egl/main/eglconfigutil.h @@ -1,12 +1,10 @@ - #ifndef EGLCONFIGUTIL_INCLUDED #define EGLCONFIGUTIL_INCLUDED -#include "eglconfig.h" + +#include "GLES/gl.h" #include "GL/internal/glcore.h" -#if (!defined(WIN32) && !defined(_WIN32_WCE)) -#include "stdint.h" -#endif +#include "eglconfig.h" extern void @@ -14,6 +12,11 @@ _eglConfigToContextModesRec(const _EGLConfig *config, __GLcontextModes *mode); extern EGLBoolean +_eglConfigFromContextModesRec(_EGLConfig *conf, const __GLcontextModes *m, + EGLint conformant, EGLint renderable_type); + + +extern EGLBoolean _eglFillInConfigs( _EGLConfig *configs, EGLenum fb_format, EGLenum fb_type, const uint8_t * depth_bits, const uint8_t * stencil_bits, diff --git a/src/egl/main/eglcontext.c b/src/egl/main/eglcontext.c index b094f49bfc1..ee4b1b59f5b 100644 --- a/src/egl/main/eglcontext.c +++ b/src/egl/main/eglcontext.c @@ -45,6 +45,7 @@ _eglInitContext(_EGLDriver *drv, _EGLContext *ctx, ctx->DrawSurface = EGL_NO_SURFACE; ctx->ReadSurface = EGL_NO_SURFACE; ctx->ClientAPI = api; + ctx->WindowRenderBuffer = EGL_NONE; return EGL_TRUE; } @@ -87,6 +88,24 @@ _eglDestroyContext(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *ctx) } +#ifdef EGL_VERSION_1_2 +static EGLint +_eglQueryContextRenderBuffer(_EGLContext *ctx) +{ + _EGLSurface *surf = ctx->DrawSurface; + EGLint rb; + + if (!surf) + return EGL_NONE; + if (surf->Type == EGL_WINDOW_BIT && ctx->WindowRenderBuffer != EGL_NONE) + rb = ctx->WindowRenderBuffer; + else + rb = surf->RenderBuffer; + return rb; +} +#endif /* EGL_VERSION_1_2 */ + + EGLBoolean _eglQueryContext(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *c, EGLint attribute, EGLint *value) @@ -94,22 +113,29 @@ _eglQueryContext(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *c, (void) drv; (void) dpy; + if (!value) + return _eglError(EGL_BAD_PARAMETER, "eglQueryContext"); + switch (attribute) { case EGL_CONFIG_ID: *value = GET_CONFIG_ATTRIB(c->Config, EGL_CONFIG_ID); - return EGL_TRUE; + break; + case EGL_CONTEXT_CLIENT_VERSION: + *value = c->ClientVersion; + break; #ifdef EGL_VERSION_1_2 case EGL_CONTEXT_CLIENT_TYPE: *value = c->ClientAPI; - return EGL_TRUE; + break; + case EGL_RENDER_BUFFER: + *value = _eglQueryContextRenderBuffer(c); + break; #endif /* EGL_VERSION_1_2 */ - case EGL_CONTEXT_CLIENT_VERSION: - *value = c->ClientVersion; - return EGL_TRUE; default: - _eglError(EGL_BAD_ATTRIBUTE, "eglQueryContext"); - return EGL_FALSE; + return _eglError(EGL_BAD_ATTRIBUTE, "eglQueryContext"); } + + return EGL_TRUE; } diff --git a/src/egl/main/eglcontext.h b/src/egl/main/eglcontext.h index 647f24488ff..45c7b4717b7 100644 --- a/src/egl/main/eglcontext.h +++ b/src/egl/main/eglcontext.h @@ -24,6 +24,9 @@ struct _egl_context EGLint ClientAPI; /**< EGL_OPENGL_ES_API, EGL_OPENGL_API, EGL_OPENVG_API */ EGLint ClientVersion; /**< 1 = OpenGLES 1.x, 2 = OpenGLES 2.x */ + + /* The real render buffer when a window surface is bound */ + EGLint WindowRenderBuffer; }; diff --git a/src/egl/main/egldisplay.h b/src/egl/main/egldisplay.h index 6575fdf198f..ea4e35a8b3f 100644 --- a/src/egl/main/egldisplay.h +++ b/src/egl/main/egldisplay.h @@ -44,6 +44,7 @@ struct _egl_display EGLint NumScreens; _EGLScreen **Screens; /* array [NumScreens] */ + EGLint MaxConfigs; EGLint NumConfigs; _EGLConfig **Configs; /* array [NumConfigs] of ptr to _EGLConfig */ diff --git a/src/gallium/SConscript b/src/gallium/SConscript index 89c69d7205e..8be84cddbe7 100644 --- a/src/gallium/SConscript +++ b/src/gallium/SConscript @@ -23,6 +23,7 @@ SConscript([ 'auxiliary/pipebuffer/SConscript', 'auxiliary/indices/SConscript', 'auxiliary/rbug/SConscript', + 'auxiliary/vl/SConscript', ]) for driver in env['drivers']: diff --git a/src/gallium/auxiliary/draw/draw_pt_post_vs.c b/src/gallium/auxiliary/draw/draw_pt_post_vs.c index 00d7197b132..e25f16c354a 100644 --- a/src/gallium/auxiliary/draw/draw_pt_post_vs.c +++ b/src/gallium/auxiliary/draw/draw_pt_post_vs.c @@ -104,7 +104,7 @@ static boolean post_vs_cliptest_viewport_gl( struct pt_post_vs *pvs, unsigned clipped = 0; unsigned j; - if (0) debug_printf("%s\n"); + if (0) debug_printf("%s\n", __FUNCTION__); for (j = 0; j < count; j++) { float *position = out->data[pos]; diff --git a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c index 109ac7c9d63..d01f8666222 100644 --- a/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c +++ b/src/gallium/auxiliary/pipebuffer/pb_buffer_fenced.c @@ -542,7 +542,7 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list) debug_printf("%10p %7u %7u\n", fenced_buf, fenced_buf->base.base.size, - fenced_buf->base.base.reference.count); + p_atomic_read(&fenced_buf->base.base.reference.count)); curr = next; next = curr->next; } @@ -556,7 +556,7 @@ fenced_buffer_list_dump(struct fenced_buffer_list *fenced_list) debug_printf("%10p %7u %7u %10p %s\n", fenced_buf, fenced_buf->base.base.size, - fenced_buf->base.base.reference.count, + p_atomic_read(&fenced_buf->base.base.reference.count), fenced_buf->fence, signaled == 0 ? "y" : "n"); curr = next; diff --git a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c index 1b4df28c707..6e3214ca9c9 100644 --- a/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c +++ b/src/gallium/auxiliary/pipebuffer/pb_bufmgr_debug.c @@ -350,7 +350,7 @@ pb_debug_manager_dump(struct pb_debug_manager *mgr) buf = LIST_ENTRY(struct pb_debug_buffer, curr, head); debug_printf("buffer = %p\n", buf); - debug_printf(" .size = %p\n", buf->base.base.size); + debug_printf(" .size = 0x%x\n", buf->base.base.size); debug_backtrace_dump(buf->create_backtrace, PB_DEBUG_CREATE_BACKTRACE); curr = next; diff --git a/src/gallium/auxiliary/tgsi/tgsi_sse2.c b/src/gallium/auxiliary/tgsi/tgsi_sse2.c index 501fc05e729..5f6b83b2362 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_sse2.c +++ b/src/gallium/auxiliary/tgsi/tgsi_sse2.c @@ -1445,11 +1445,11 @@ fetch_texel( struct tgsi_sampler **sampler, { float rgba[NUM_CHANNELS][QUAD_SIZE]; (*sampler)->get_samples(*sampler, - &store[0], - &store[4], - &store[8], - 0.0f, /*store[12], lodbias */ - rgba); + &store[0], /* s */ + &store[4], /* t */ + &store[8], /* r */ + store[12], /* lodbias */ + rgba); /* results */ memcpy( store, rgba, 16 * sizeof(float)); } @@ -1855,7 +1855,6 @@ emit_instruction( break; case TGSI_OPCODE_RCP: - /* TGSI_OPCODE_RECIP */ FETCH( func, *inst, 0, 0, CHAN_X ); emit_rcp( func, 0, 0 ); FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { @@ -1864,7 +1863,6 @@ emit_instruction( break; case TGSI_OPCODE_RSQ: - /* TGSI_OPCODE_RECIPSQRT */ FETCH( func, *inst, 0, 0, CHAN_X ); emit_abs( func, 0 ); emit_rsqrt( func, 1, 0 ); @@ -1962,7 +1960,6 @@ emit_instruction( break; case TGSI_OPCODE_DP3: - /* TGSI_OPCODE_DOT3 */ FETCH( func, *inst, 0, 0, CHAN_X ); FETCH( func, *inst, 1, 1, CHAN_X ); emit_mul( func, 0, 1 ); @@ -1980,7 +1977,6 @@ emit_instruction( break; case TGSI_OPCODE_DP4: - /* TGSI_OPCODE_DOT4 */ FETCH( func, *inst, 0, 0, CHAN_X ); FETCH( func, *inst, 1, 1, CHAN_X ); emit_mul( func, 0, 1 ); @@ -2051,17 +2047,14 @@ emit_instruction( break; case TGSI_OPCODE_SLT: - /* TGSI_OPCODE_SETLT */ emit_setcc( func, inst, cc_LessThan ); break; case TGSI_OPCODE_SGE: - /* TGSI_OPCODE_SETGE */ emit_setcc( func, inst, cc_NotLessThan ); break; case TGSI_OPCODE_MAD: - /* TGSI_OPCODE_MADD */ FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 0, 0, chan_index ); FETCH( func, *inst, 1, 1, chan_index ); @@ -2291,7 +2284,7 @@ emit_instruction( break; case TGSI_OPCODE_SEQ: - return 0; + emit_setcc( func, inst, cc_Equal ); break; case TGSI_OPCODE_SFL: @@ -2299,7 +2292,7 @@ emit_instruction( break; case TGSI_OPCODE_SGT: - return 0; + emit_setcc( func, inst, cc_NotLessThanEqual ); break; case TGSI_OPCODE_SIN: @@ -2311,11 +2304,11 @@ emit_instruction( break; case TGSI_OPCODE_SLE: - return 0; + emit_setcc( func, inst, cc_LessThanEqual ); break; case TGSI_OPCODE_SNE: - return 0; + emit_setcc( func, inst, cc_NotEqual ); break; case TGSI_OPCODE_STR: @@ -2379,7 +2372,6 @@ emit_instruction( break; case TGSI_OPCODE_SSG: - /* TGSI_OPCODE_SGN */ FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( func, *inst, 0, 0, chan_index ); emit_sgn( func, 0, 0 ); diff --git a/src/gallium/auxiliary/util/Makefile b/src/gallium/auxiliary/util/Makefile index ae8d330a787..1d8bb55bbd6 100644 --- a/src/gallium/auxiliary/util/Makefile +++ b/src/gallium/auxiliary/util/Makefile @@ -10,6 +10,7 @@ C_SOURCES = \ u_debug_stack.c \ u_blit.c \ u_cache.c \ + u_cpu_detect.c \ u_draw_quad.c \ u_format.c \ u_format_access.c \ diff --git a/src/gallium/auxiliary/util/SConscript b/src/gallium/auxiliary/util/SConscript index 28a5ab42569..2187935fa40 100644 --- a/src/gallium/auxiliary/util/SConscript +++ b/src/gallium/auxiliary/util/SConscript @@ -24,6 +24,7 @@ util = env.ConvenienceLibrary( 'u_bitmask.c', 'u_blit.c', 'u_cache.c', + 'u_cpu_detect.c', 'u_debug.c', 'u_debug_dump.c', 'u_debug_memory.c', diff --git a/src/gallium/auxiliary/util/u_cpu_detect.c b/src/gallium/auxiliary/util/u_cpu_detect.c index d9f2f8fc288..ecfb96138d6 100644 --- a/src/gallium/auxiliary/util/u_cpu_detect.c +++ b/src/gallium/auxiliary/util/u_cpu_detect.c @@ -24,23 +24,21 @@ * **************************************************************************/ -/* - * Based on the work of Eric Anholt <[email protected]> +/** + * @file + * CPU feature detection. + * + * @author Dennis Smit + * @author Based on the work of Eric Anholt <[email protected]> */ -/* FIXME: clean this entire file up */ +#include "pipe/p_config.h" +#include "u_debug.h" #include "u_cpu_detect.h" -#ifdef __linux__ -#define OS_LINUX -#endif -#ifdef WIN32 -#define OS_WIN32 -#endif - -#if defined(ARCH_POWERPC) -#if defined(OS_DARWIN) +#if defined(PIPE_ARCH_PPC) +#if defined(PIPE_OS_DARWIN) #include <sys/sysctl.h> #else #include <signal.h> @@ -48,137 +46,140 @@ #endif #endif -#if defined(OS_NETBSD) || defined(OS_OPENBSD) +#if defined(PIPE_OS_NETBSD) || defined(PIPE_OS_OPENBSD) #include <sys/param.h> #include <sys/sysctl.h> #include <machine/cpu.h> #endif -#if defined(OS_FREEBSD) +#if defined(PIPE_OS_FREEBSD) #include <sys/types.h> #include <sys/sysctl.h> #endif -#if defined(OS_LINUX) +#if defined(PIPE_OS_LINUX) #include <signal.h> #endif -#if defined(OS_WIN32) -#include <windows.h> +#ifdef PIPE_OS_UNIX +#include <unistd.h> #endif -#include <stdio.h> -#include <stdlib.h> -#include <unistd.h> -#include <string.h> +#if defined(PIPE_OS_WINDOWS) +#include <windows.h> +#endif -static struct cpu_detect_caps __cpu_detect_caps; -static int __cpu_detect_initialized = 0; +struct util_cpu_caps util_cpu_caps; static int has_cpuid(void); static int cpuid(unsigned int ax, unsigned int *p); +#if defined(PIPE_ARCH_X86) + /* The sigill handlers */ -#if defined(ARCH_X86) /* x86 (linux katmai handler check thing) */ -#if defined(OS_LINUX) && defined(_POSIX_SOURCE) && defined(X86_FXSR_MAGIC) -static void sigill_handler_sse(int signal, struct sigcontext sc) +#if defined(PIPE_OS_LINUX) //&& defined(_POSIX_SOURCE) && defined(X86_FXSR_MAGIC) +static void +sigill_handler_sse(int signal, struct sigcontext sc) { - /* Both the "xorps %%xmm0,%%xmm0" and "divps %xmm0,%%xmm1" - * instructions are 3 bytes long. We must increment the instruction - * pointer manually to avoid repeated execution of the offending - * instruction. - * - * If the SIGILL is caused by a divide-by-zero when unmasked - * exceptions aren't supported, the SIMD FPU status and control - * word will be restored at the end of the test, so we don't need - * to worry about doing it here. Besides, we may not be able to... - */ - sc.eip += 3; - - __cpu_detect_caps.hasSSE=0; + /* Both the "xorps %%xmm0,%%xmm0" and "divps %xmm0,%%xmm1" + * instructions are 3 bytes long. We must increment the instruction + * pointer manually to avoid repeated execution of the offending + * instruction. + * + * If the SIGILL is caused by a divide-by-zero when unmasked + * exceptions aren't supported, the SIMD FPU status and control + * word will be restored at the end of the test, so we don't need + * to worry about doing it here. Besides, we may not be able to... + */ + sc.eip += 3; + + util_cpu_caps.has_sse=0; } -static void sigfpe_handler_sse(int signal, struct sigcontext sc) +static void +sigfpe_handler_sse(int signal, struct sigcontext sc) { - if (sc.fpstate->magic != 0xffff) { - /* Our signal context has the extended FPU state, so reset the - * divide-by-zero exception mask and clear the divide-by-zero - * exception bit. - */ - sc.fpstate->mxcsr |= 0x00000200; - sc.fpstate->mxcsr &= 0xfffffffb; - } else { - /* If we ever get here, we're completely hosed. - */ - } + if (sc.fpstate->magic != 0xffff) { + /* Our signal context has the extended FPU state, so reset the + * divide-by-zero exception mask and clear the divide-by-zero + * exception bit. + */ + sc.fpstate->mxcsr |= 0x00000200; + sc.fpstate->mxcsr &= 0xfffffffb; + } else { + /* If we ever get here, we're completely hosed. + */ + } } -#endif -#endif /* OS_LINUX && _POSIX_SOURCE && X86_FXSR_MAGIC */ +#endif /* PIPE_OS_LINUX && _POSIX_SOURCE && X86_FXSR_MAGIC */ -#if defined(OS_WIN32) -LONG CALLBACK win32_sig_handler_sse(EXCEPTION_POINTERS* ep) +#if defined(PIPE_OS_WINDOWS) +static LONG CALLBACK +win32_sig_handler_sse(EXCEPTION_POINTERS* ep) { - if(ep->ExceptionRecord->ExceptionCode==EXCEPTION_ILLEGAL_INSTRUCTION){ - ep->ContextRecord->Eip +=3; - __cpu_detect_caps.hasSSE=0; - return EXCEPTION_CONTINUE_EXECUTION; - } - return EXCEPTION_CONTINUE_SEARCH; + if(ep->ExceptionRecord->ExceptionCode==EXCEPTION_ILLEGAL_INSTRUCTION){ + ep->ContextRecord->Eip +=3; + util_cpu_caps.has_sse=0; + return EXCEPTION_CONTINUE_EXECUTION; + } + return EXCEPTION_CONTINUE_SEARCH; } -#endif /* OS_WIN32 */ +#endif /* PIPE_OS_WINDOWS */ + +#endif /* PIPE_ARCH_X86 */ -#if defined(ARCH_POWERPC) && !defined(OS_DARWIN) +#if defined(PIPE_ARCH_PPC) && !defined(PIPE_OS_DARWIN) static sigjmp_buf __lv_powerpc_jmpbuf; static volatile sig_atomic_t __lv_powerpc_canjump = 0; -static void sigill_handler (int sig); - -static void sigill_handler (int sig) +static void +sigill_handler(int sig) { - if (!__lv_powerpc_canjump) { - signal (sig, SIG_DFL); - raise (sig); - } + if (!__lv_powerpc_canjump) { + signal (sig, SIG_DFL); + raise (sig); + } - __lv_powerpc_canjump = 0; - siglongjmp(__lv_powerpc_jmpbuf, 1); + __lv_powerpc_canjump = 0; + siglongjmp(__lv_powerpc_jmpbuf, 1); } -static void check_os_altivec_support(void) +static void +check_os_altivec_support(void) { -#if defined(OS_DARWIN) - int sels[2] = {CTL_HW, HW_VECTORUNIT}; - int has_vu = 0; - int len = sizeof (has_vu); - int err; - - err = sysctl(sels, 2, &has_vu, &len, NULL, 0); - - if (err == 0) { - if (has_vu != 0) { - __cpu_detect_caps.hasAltiVec = 1; - } - } -#else /* !OS_DARWIN */ - /* no Darwin, do it the brute-force way */ - /* this is borrowed from the libmpeg2 library */ - signal(SIGILL, sigill_handler); - if (sigsetjmp(__lv_powerpc_jmpbuf, 1)) { - signal(SIGILL, SIG_DFL); - } else { - __lv_powerpc_canjump = 1; - - __asm __volatile - ("mtspr 256, %0\n\t" - "vand %%v0, %%v0, %%v0" - : - : "r" (-1)); - - signal(SIGILL, SIG_DFL); - __cpu_detect_caps.hasAltiVec = 1; - } +#if defined(PIPE_OS_DARWIN) + int sels[2] = {CTL_HW, HW_VECTORUNIT}; + int has_vu = 0; + int len = sizeof (has_vu); + int err; + + err = sysctl(sels, 2, &has_vu, &len, NULL, 0); + + if (err == 0) { + if (has_vu != 0) { + util_cpu_caps.has_altivec = 1; + } + } +#else /* !PIPE_OS_DARWIN */ + /* no Darwin, do it the brute-force way */ + /* this is borrowed from the libmpeg2 library */ + signal(SIGILL, sigill_handler); + if (sigsetjmp(__lv_powerpc_jmpbuf, 1)) { + signal(SIGILL, SIG_DFL); + } else { + __lv_powerpc_canjump = 1; + + __asm __volatile + ("mtspr 256, %0\n\t" + "vand %%v0, %%v0, %%v0" + : + : "r" (-1)); + + signal(SIGILL, SIG_DFL); + util_cpu_caps.has_altivec = 1; + } #endif } #endif @@ -189,318 +190,312 @@ static void check_os_altivec_support(void) * and RedHat patched 2.2 kernels that have broken exception handling * support for user space apps that do SSE. */ -static void check_os_katmai_support(void) +static void +check_os_katmai_support(void) { -#if defined(ARCH_X86) -#if defined(OS_FREEBSD) - int has_sse=0, ret; - int len = sizeof (has_sse); - - ret = sysctlbyname("hw.instruction_sse", &has_sse, &len, NULL, 0); - if (ret || !has_sse) - __cpu_detect_caps.hasSSE=0; - -#elif defined(OS_NETBSD) || defined(OS_OPENBSD) - int has_sse, has_sse2, ret, mib[2]; - int varlen; - - mib[0] = CTL_MACHDEP; - mib[1] = CPU_SSE; - varlen = sizeof (has_sse); - - ret = sysctl(mib, 2, &has_sse, &varlen, NULL, 0); - if (ret < 0 || !has_sse) { - __cpu_detect_caps.hasSSE = 0; - } else { - __cpu_detect_caps.hasSSE = 1; - } - - mib[1] = CPU_SSE2; - varlen = sizeof (has_sse2); - ret = sysctl(mib, 2, &has_sse2, &varlen, NULL, 0); - if (ret < 0 || !has_sse2) { - __cpu_detect_caps.hasSSE2 = 0; - } else { - __cpu_detect_caps.hasSSE2 = 1; - } - __cpu_detect_caps.hasSSE = 0; /* FIXME ?!?!? */ - -#elif defined(OS_WIN32) - LPTOP_LEVEL_EXCEPTION_FILTER exc_fil; - if (__cpu_detect_caps.hasSSE) { - exc_fil = SetUnhandledExceptionFilter(win32_sig_handler_sse); - __asm __volatile ("xorps %xmm0, %xmm0"); - SetUnhandledExceptionFilter(exc_fil); - } -#elif defined(OS_LINUX) - struct sigaction saved_sigill; - struct sigaction saved_sigfpe; - - /* Save the original signal handlers. - */ - sigaction(SIGILL, NULL, &saved_sigill); - sigaction(SIGFPE, NULL, &saved_sigfpe); - - signal(SIGILL, (void (*)(int))sigill_handler_sse); - signal(SIGFPE, (void (*)(int))sigfpe_handler_sse); - - /* Emulate test for OSFXSR in CR4. The OS will set this bit if it - * supports the extended FPU save and restore required for SSE. If - * we execute an SSE instruction on a PIII and get a SIGILL, the OS - * doesn't support Streaming SIMD Exceptions, even if the processor - * does. - */ - if (__cpu_detect_caps.hasSSE) { - __asm __volatile ("xorps %xmm1, %xmm0"); - } - - /* Emulate test for OSXMMEXCPT in CR4. The OS will set this bit if - * it supports unmasked SIMD FPU exceptions. If we unmask the - * exceptions, do a SIMD divide-by-zero and get a SIGILL, the OS - * doesn't support unmasked SIMD FPU exceptions. If we get a SIGFPE - * as expected, we're okay but we need to clean up after it. - * - * Are we being too stringent in our requirement that the OS support - * unmasked exceptions? Certain RedHat 2.2 kernels enable SSE by - * setting CR4.OSFXSR but don't support unmasked exceptions. Win98 - * doesn't even support them. We at least know the user-space SSE - * support is good in kernels that do support unmasked exceptions, - * and therefore to be safe I'm going to leave this test in here. - */ - if (__cpu_detect_caps.hasSSE) { - // test_os_katmai_exception_support(); - } - - /* Restore the original signal handlers. - */ - sigaction(SIGILL, &saved_sigill, NULL); - sigaction(SIGFPE, &saved_sigfpe, NULL); +#if defined(PIPE_ARCH_X86) +#if defined(PIPE_OS_FREEBSD) + int has_sse=0, ret; + int len = sizeof (has_sse); + + ret = sysctlbyname("hw.instruction_sse", &has_sse, &len, NULL, 0); + if (ret || !has_sse) + util_cpu_caps.has_sse=0; + +#elif defined(PIPE_OS_NETBSD) || defined(PIPE_OS_OPENBSD) + int has_sse, has_sse2, ret, mib[2]; + int varlen; + + mib[0] = CTL_MACHDEP; + mib[1] = CPU_SSE; + varlen = sizeof (has_sse); + + ret = sysctl(mib, 2, &has_sse, &varlen, NULL, 0); + if (ret < 0 || !has_sse) { + util_cpu_caps.has_sse = 0; + } else { + util_cpu_caps.has_sse = 1; + } + + mib[1] = CPU_SSE2; + varlen = sizeof (has_sse2); + ret = sysctl(mib, 2, &has_sse2, &varlen, NULL, 0); + if (ret < 0 || !has_sse2) { + util_cpu_caps.has_sse2 = 0; + } else { + util_cpu_caps.has_sse2 = 1; + } + util_cpu_caps.has_sse = 0; /* FIXME ?!?!? */ + +#elif defined(PIPE_OS_WINDOWS) + LPTOP_LEVEL_EXCEPTION_FILTER exc_fil; + if (util_cpu_caps.has_sse) { + exc_fil = SetUnhandledExceptionFilter(win32_sig_handler_sse); +#if defined(PIPE_CC_GCC) + __asm __volatile ("xorps %xmm0, %xmm0"); +#elif defined(PIPE_CC_MSVC) + __asm { + xorps xmm0, xmm0 // executing SSE instruction + } +#else +#error Unsupported compiler +#endif + SetUnhandledExceptionFilter(exc_fil); + } +#elif defined(PIPE_OS_LINUX) + struct sigaction saved_sigill; + struct sigaction saved_sigfpe; + + /* Save the original signal handlers. + */ + sigaction(SIGILL, NULL, &saved_sigill); + sigaction(SIGFPE, NULL, &saved_sigfpe); + + signal(SIGILL, (void (*)(int))sigill_handler_sse); + signal(SIGFPE, (void (*)(int))sigfpe_handler_sse); + + /* Emulate test for OSFXSR in CR4. The OS will set this bit if it + * supports the extended FPU save and restore required for SSE. If + * we execute an SSE instruction on a PIII and get a SIGILL, the OS + * doesn't support Streaming SIMD Exceptions, even if the processor + * does. + */ + if (util_cpu_caps.has_sse) { + __asm __volatile ("xorps %xmm1, %xmm0"); + } + + /* Emulate test for OSXMMEXCPT in CR4. The OS will set this bit if + * it supports unmasked SIMD FPU exceptions. If we unmask the + * exceptions, do a SIMD divide-by-zero and get a SIGILL, the OS + * doesn't support unmasked SIMD FPU exceptions. If we get a SIGFPE + * as expected, we're okay but we need to clean up after it. + * + * Are we being too stringent in our requirement that the OS support + * unmasked exceptions? Certain RedHat 2.2 kernels enable SSE by + * setting CR4.OSFXSR but don't support unmasked exceptions. Win98 + * doesn't even support them. We at least know the user-space SSE + * support is good in kernels that do support unmasked exceptions, + * and therefore to be safe I'm going to leave this test in here. + */ + if (util_cpu_caps.has_sse) { + // test_os_katmai_exception_support(); + } + + /* Restore the original signal handlers. + */ + sigaction(SIGILL, &saved_sigill, NULL); + sigaction(SIGFPE, &saved_sigfpe, NULL); #else - /* We can't use POSIX signal handling to test the availability of - * SSE, so we disable it by default. - */ - __cpu_detect_caps.hasSSE = 0; + /* We can't use POSIX signal handling to test the availability of + * SSE, so we disable it by default. + */ + util_cpu_caps.has_sse = 0; #endif /* __linux__ */ #endif + +#if defined(PIPE_ARCH_X86_64) + util_cpu_caps.has_sse = 1; +#endif } static int has_cpuid(void) { -#if defined(ARCH_X86) - int a, c; - - __asm __volatile - ("pushf\n" - "popl %0\n" - "movl %0, %1\n" - "xorl $0x200000, %0\n" - "push %0\n" - "popf\n" - "pushf\n" - "popl %0\n" - : "=a" (a), "=c" (c) - : - : "cc"); - - return a != c; +#if defined(PIPE_ARCH_X86) +#if defined(PIPE_OS_GCC) + int a, c; + + __asm __volatile + ("pushf\n" + "popl %0\n" + "movl %0, %1\n" + "xorl $0x200000, %0\n" + "push %0\n" + "popf\n" + "pushf\n" + "popl %0\n" + : "=a" (a), "=c" (c) + : + : "cc"); + + return a != c; +#else + /* FIXME */ + return 1; +#endif +#elif defined(PIPE_ARCH_X86_64) + return 1; #else - return 0; + return 0; #endif } -static int cpuid(unsigned int ax, unsigned int *p) +static INLINE int +cpuid(unsigned int ax, unsigned int *p) { -#if defined(ARCH_X86) - unsigned int flags; - - __asm __volatile - ("movl %%ebx, %%esi\n\t" - "cpuid\n\t" - "xchgl %%ebx, %%esi" - : "=a" (p[0]), "=S" (p[1]), - "=c" (p[2]), "=d" (p[3]) - : "0" (ax)); - - return 0; -#else - return -1; + int ret = -1; + +#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) +#if defined(PIPE_CC_GCC) + __asm __volatile + ("movl %%ebx, %%esi\n\t" + "cpuid\n\t" + "xchgl %%ebx, %%esi" + : "=a" (p[0]), "=S" (p[1]), + "=c" (p[2]), "=d" (p[3]) + : "0" (ax)); + + ret = 0; +#elif defined(PIPE_CC_MSVC) + __cpuid(ax, p); + + ret = 0; +#endif #endif + + return ret; } -void cpu_detect_initialize() +void +util_cpu_detect(void) { - unsigned int regs[4]; - unsigned int regs2[4]; - - int mib[2], ncpu; - int len; - - memset(&__cpu_detect_caps, 0, sizeof (struct cpu_detect_caps)); - - /* Check for arch type */ -#if defined(ARCH_MIPS) - __cpu_detect_caps.type = CPU_DETECT_TYPE_MIPS; -#elif defined(ARCH_ALPHA) - __cpu_detect_caps.type = CPU_DETECT_TYPE_ALPHA; -#elif defined(ARCH_SPARC) - __cpu_detect_caps.type = CPU_DETECT_TYPE_SPARC; -#elif defined(ARCH_X86) - __cpu_detect_caps.type = CPU_DETECT_TYPE_X86; -#elif defined(ARCH_POWERPC) - __cpu_detect_caps.type = CPU_DETECT_TYPE_POWERPC; + static boolean util_cpu_detect_initialized = FALSE; + + if(util_cpu_detect_initialized) + return; + + memset(&util_cpu_caps, 0, sizeof util_cpu_caps); + + /* Check for arch type */ +#if defined(PIPE_ARCH_MIPS) + util_cpu_caps.arch = UTIL_CPU_ARCH_MIPS; +#elif defined(PIPE_ARCH_ALPHA) + util_cpu_caps.arch = UTIL_CPU_ARCH_ALPHA; +#elif defined(PIPE_ARCH_SPARC) + util_cpu_caps.arch = UTIL_CPU_ARCH_SPARC; +#elif defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) + util_cpu_caps.arch = UTIL_CPU_ARCH_X86; +#elif defined(PIPE_ARCH_PPC) + util_cpu_caps.arch = UTIL_CPU_ARCH_POWERPC; #else - __cpu_detect_caps.type = CPU_DETECT_TYPE_OTHER; + util_cpu_caps.arch = UTIL_CPU_ARCH_UNKNOWN; #endif - /* Count the number of CPUs in system */ -#if !defined(OS_WIN32) && !defined(OS_UNKNOWN) && defined(_SC_NPROCESSORS_ONLN) - __cpu_detect_caps.nrcpu = sysconf(_SC_NPROCESSORS_ONLN); - if (__cpu_detect_caps.nrcpu == -1) - __cpu_detect_caps.nrcpu = 1; - -#elif defined(OS_NETBSD) || defined(OS_FREEBSD) || defined(OS_OPENBSD) + /* Count the number of CPUs in system */ +#if !defined(PIPE_OS_WINDOWS) && !defined(PIPE_OS_UNKNOWN) && defined(_SC_NPROCESSORS_ONLN) + util_cpu_caps.nr_cpus = sysconf(_SC_NPROCESSORS_ONLN); + if (util_cpu_caps.nr_cpus == -1) + util_cpu_caps.nr_cpus = 1; - mib[0] = CTL_HW; - mib[1] = HW_NCPU; +#elif defined(PIPE_OS_NETBSD) || defined(PIPE_OS_FREEBSD) || defined(PIPE_OS_OPENBSD) + { + int mib[2], ncpu; + int len; - len = sizeof (ncpu); - sysctl(mib, 2, &ncpu, &len, NULL, 0); - __cpu_detect_caps.nrcpu = ncpu; + mib[0] = CTL_HW; + mib[1] = HW_NCPU; + len = sizeof (ncpu); + sysctl(mib, 2, &ncpu, &len, NULL, 0); + util_cpu_caps.nr_cpus = ncpu; + } #else - __cpu_detect_caps.nrcpu = 1; + util_cpu_caps.nr_cpus = 1; #endif -#if defined(ARCH_X86) - /* No cpuid, old 486 or lower */ - if (has_cpuid() == 0) - return; +#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) + if (has_cpuid()) { + unsigned int regs[4]; + unsigned int regs2[4]; - __cpu_detect_caps.cacheline = 32; + util_cpu_caps.cacheline = 32; - /* Get max cpuid level */ - cpuid(0x00000000, regs); + /* Get max cpuid level */ + cpuid(0x00000000, regs); - if (regs[0] >= 0x00000001) { - unsigned int cacheline; + if (regs[0] >= 0x00000001) { + unsigned int cacheline; - cpuid (0x00000001, regs2); + cpuid (0x00000001, regs2); - __cpu_detect_caps.x86cpuType = (regs2[0] >> 8) & 0xf; - if (__cpu_detect_caps.x86cpuType == 0xf) - __cpu_detect_caps.x86cpuType = 8 + ((regs2[0] >> 20) & 255); /* use extended family (P4, IA64) */ + util_cpu_caps.x86_cpu_type = (regs2[0] >> 8) & 0xf; + if (util_cpu_caps.x86_cpu_type == 0xf) + util_cpu_caps.x86_cpu_type = 8 + ((regs2[0] >> 20) & 255); /* use extended family (P4, IA64) */ - /* general feature flags */ - __cpu_detect_caps.hasTSC = (regs2[3] & (1 << 8 )) >> 8; /* 0x0000010 */ - __cpu_detect_caps.hasMMX = (regs2[3] & (1 << 23 )) >> 23; /* 0x0800000 */ - __cpu_detect_caps.hasSSE = (regs2[3] & (1 << 25 )) >> 25; /* 0x2000000 */ - __cpu_detect_caps.hasSSE2 = (regs2[3] & (1 << 26 )) >> 26; /* 0x4000000 */ - __cpu_detect_caps.hasSSE3 = (regs2[2] & (1)); /* 0x0000001 */ - __cpu_detect_caps.hasSSSE3 = (regs2[2] & (1 << 9 )) >> 9; /* 0x0000020 */ - __cpu_detect_caps.hasMMX2 = __cpu_detect_caps.hasSSE; /* SSE cpus supports mmxext too */ + /* general feature flags */ + util_cpu_caps.has_tsc = (regs2[3] & (1 << 8 )) >> 8; /* 0x0000010 */ + util_cpu_caps.has_mmx = (regs2[3] & (1 << 23 )) >> 23; /* 0x0800000 */ + util_cpu_caps.has_sse = (regs2[3] & (1 << 25 )) >> 25; /* 0x2000000 */ + util_cpu_caps.has_sse2 = (regs2[3] & (1 << 26 )) >> 26; /* 0x4000000 */ + util_cpu_caps.has_sse3 = (regs2[2] & (1)); /* 0x0000001 */ + util_cpu_caps.has_ssse3 = (regs2[2] & (1 << 9 )) >> 9; /* 0x0000020 */ + util_cpu_caps.has_sse4_1 = (regs2[2] & (1 << 19)) >> 19; + util_cpu_caps.has_mmx2 = util_cpu_caps.has_sse; /* SSE cpus supports mmxext too */ - cacheline = ((regs2[1] >> 8) & 0xFF) * 8; - if (cacheline > 0) - __cpu_detect_caps.cacheline = cacheline; - } + cacheline = ((regs2[1] >> 8) & 0xFF) * 8; + if (cacheline > 0) + util_cpu_caps.cacheline = cacheline; + } - cpuid(0x80000000, regs); + cpuid(0x80000000, regs); - if (regs[0] >= 0x80000001) { + if (regs[0] >= 0x80000001) { - cpuid(0x80000001, regs2); + cpuid(0x80000001, regs2); - __cpu_detect_caps.hasMMX |= (regs2[3] & (1 << 23 )) >> 23; /* 0x0800000 */ - __cpu_detect_caps.hasMMX2 |= (regs2[3] & (1 << 22 )) >> 22; /* 0x400000 */ - __cpu_detect_caps.has3DNow = (regs2[3] & (1 << 31 )) >> 31; /* 0x80000000 */ - __cpu_detect_caps.has3DNowExt = (regs2[3] & (1 << 30 )) >> 30; - } + util_cpu_caps.has_mmx |= (regs2[3] & (1 << 23 )) >> 23; /* 0x0800000 */ + util_cpu_caps.has_mmx2 |= (regs2[3] & (1 << 22 )) >> 22; /* 0x400000 */ + util_cpu_caps.has_3dnow = (regs2[3] & (1 << 31 )) >> 31; /* 0x80000000 */ + util_cpu_caps.has_3dnow_ext = (regs2[3] & (1 << 30 )) >> 30; + } - if (regs[0] >= 0x80000006) { - cpuid(0x80000006, regs2); - __cpu_detect_caps.cacheline = regs2[2] & 0xFF; - } + if (regs[0] >= 0x80000006) { + cpuid(0x80000006, regs2); + util_cpu_caps.cacheline = regs2[2] & 0xFF; + } +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_FREEBSD) || defined(PIPE_OS_NETBSD) || defined(PIPE_OS_CYGWIN) || defined(PIPE_OS_OPENBSD) + if (util_cpu_caps.has_sse) + check_os_katmai_support(); -#if defined(OS_LINUX) || defined(OS_FREEBSD) || defined(OS_NETBSD) || defined(OS_CYGWIN) || defined(OS_OPENBSD) - if (__cpu_detect_caps.hasSSE) - check_os_katmai_support(); - - if (!__cpu_detect_caps.hasSSE) { - __cpu_detect_caps.hasSSE2 = 0; - __cpu_detect_caps.hasSSE3 = 0; - __cpu_detect_caps.hasSSSE3 = 0; - } + if (!util_cpu_caps.has_sse) { + util_cpu_caps.has_sse2 = 0; + util_cpu_caps.has_sse3 = 0; + util_cpu_caps.has_ssse3 = 0; + } #else - __cpu_detect_caps.hasSSE = 0; - __cpu_detect_caps.hasSSE2 = 0; - __cpu_detect_caps.hasSSE3 = 0; - __cpu_detect_caps.hasSSSE3 = 0; + util_cpu_caps.has_sse = 0; + util_cpu_caps.has_sse2 = 0; + util_cpu_caps.has_sse3 = 0; + util_cpu_caps.has_ssse3 = 0; +#endif + } +#endif /* PIPE_ARCH_X86 || PIPE_ARCH_X86_64 */ + +#if defined(PIPE_ARCH_PPC) + check_os_altivec_support(); +#endif /* PIPE_ARCH_PPC */ + +#ifdef DEBUG + debug_printf("util_cpu_caps.arch = %i\n", util_cpu_caps.arch); + debug_printf("util_cpu_caps.nr_cpus = %u\n", util_cpu_caps.nr_cpus); + + debug_printf("util_cpu_caps.x86_cpu_type = %u\n", util_cpu_caps.x86_cpu_type); + debug_printf("util_cpu_caps.cacheline = %u\n", util_cpu_caps.cacheline); + + debug_printf("util_cpu_caps.has_tsc = %u\n", util_cpu_caps.has_tsc); + debug_printf("util_cpu_caps.has_mmx = %u\n", util_cpu_caps.has_mmx); + debug_printf("util_cpu_caps.has_mmx2 = %u\n", util_cpu_caps.has_mmx2); + debug_printf("util_cpu_caps.has_sse = %u\n", util_cpu_caps.has_sse); + debug_printf("util_cpu_caps.has_sse2 = %u\n", util_cpu_caps.has_sse2); + debug_printf("util_cpu_caps.has_sse3 = %u\n", util_cpu_caps.has_sse3); + debug_printf("util_cpu_caps.has_ssse3 = %u\n", util_cpu_caps.has_ssse3); + debug_printf("util_cpu_caps.has_sse4_1 = %u\n", util_cpu_caps.has_sse4_1); + debug_printf("util_cpu_caps.has_3dnow = %u\n", util_cpu_caps.has_3dnow); + debug_printf("util_cpu_caps.has_3dnow_ext = %u\n", util_cpu_caps.has_3dnow_ext); + debug_printf("util_cpu_caps.has_altivec = %u\n", util_cpu_caps.has_altivec); #endif -#endif /* ARCH_X86 */ - -#if defined(ARCH_POWERPC) - check_os_altivec_support(); -#endif /* ARCH_POWERPC */ - - __cpu_detect_initialized = 1; -} - -struct cpu_detect_caps *cpu_detect_get_caps() -{ - return &__cpu_detect_caps; -} - -/* The getters and setters for feature flags */ -int cpu_detect_get_tsc() -{ - return __cpu_detect_caps.hasTSC; -} - -int cpu_detect_get_mmx() -{ - return __cpu_detect_caps.hasMMX; -} - -int cpu_detect_get_mmx2() -{ - return __cpu_detect_caps.hasMMX2; -} - -int cpu_detect_get_sse() -{ - return __cpu_detect_caps.hasSSE; -} - -int cpu_detect_get_sse2() -{ - return __cpu_detect_caps.hasSSE2; -} - -int cpu_detect_get_sse3() -{ - return __cpu_detect_caps.hasSSE3; -} - -int cpu_detect_get_ssse3() -{ - return __cpu_detect_caps.hasSSSE3; -} - -int cpu_detect_get_3dnow() -{ - return __cpu_detect_caps.has3DNow; -} - -int cpu_detect_get_3dnow2() -{ - return __cpu_detect_caps.has3DNowExt; -} -int cpu_detect_get_altivec() -{ - return __cpu_detect_caps.hasAltiVec; + util_cpu_detect_initialized = TRUE; } - diff --git a/src/gallium/auxiliary/util/u_cpu_detect.h b/src/gallium/auxiliary/util/u_cpu_detect.h index 1612d49286a..7ea0121c07f 100644 --- a/src/gallium/auxiliary/util/u_cpu_detect.h +++ b/src/gallium/auxiliary/util/u_cpu_detect.h @@ -24,55 +24,53 @@ * ***************************************************************************/ -/* - * Based on the work of Eric Anholt <[email protected]> +/** + * @file + * CPU feature detection. + * + * @author Dennis Smit + * @author Based on the work of Eric Anholt <[email protected]> */ -#ifndef _CPU_DETECT_H -#define _CPU_DETECT_H +#ifndef _UTIL_CPU_DETECT_H +#define _UTIL_CPU_DETECT_H + +#include "pipe/p_compiler.h" -typedef enum { - CPU_DETECT_TYPE_MIPS, - CPU_DETECT_TYPE_ALPHA, - CPU_DETECT_TYPE_SPARC, - CPU_DETECT_TYPE_X86, - CPU_DETECT_TYPE_POWERPC, - CPU_DETECT_TYPE_OTHER -} cpu_detect_type; +enum util_cpu_arch { + UTIL_CPU_ARCH_UNKNOWN = 0, + UTIL_CPU_ARCH_MIPS, + UTIL_CPU_ARCH_ALPHA, + UTIL_CPU_ARCH_SPARC, + UTIL_CPU_ARCH_X86, + UTIL_CPU_ARCH_POWERPC +}; -struct cpu_detect_caps { - cpu_detect_type type; - int nrcpu; +struct util_cpu_caps { + enum util_cpu_arch arch; + unsigned nr_cpus; - /* Feature flags */ - int x86cpuType; - int cacheline; + /* Feature flags */ + int x86_cpu_type; + unsigned cacheline; - int hasTSC; - int hasMMX; - int hasMMX2; - int hasSSE; - int hasSSE2; - int hasSSE3; - int hasSSSE3; - int has3DNow; - int has3DNowExt; - int hasAltiVec; + unsigned has_tsc:1; + unsigned has_mmx:1; + unsigned has_mmx2:1; + unsigned has_sse:1; + unsigned has_sse2:1; + unsigned has_sse3:1; + unsigned has_ssse3:1; + unsigned has_sse4_1:1; + unsigned has_3dnow:1; + unsigned has_3dnow_ext:1; + unsigned has_altivec:1; }; -/* prototypes */ -void cpu_detect_initialize(void); -struct cpu_detect_caps *cpu_detect_get_caps(void); +extern struct util_cpu_caps +util_cpu_caps; + +void util_cpu_detect(void); -int cpu_detect_get_tsc(void); -int cpu_detect_get_mmx(void); -int cpu_detect_get_mmx2(void); -int cpu_detect_get_sse(void); -int cpu_detect_get_sse2(void); -int cpu_detect_get_sse3(void); -int cpu_detect_get_ssse3(void); -int cpu_detect_get_3dnow(void); -int cpu_detect_get_3dnow2(void); -int cpu_detect_get_altivec(void); -#endif /* _CPU_DETECT_H */ +#endif /* _UTIL_CPU_DETECT_H */ diff --git a/src/gallium/auxiliary/util/u_debug.h b/src/gallium/auxiliary/util/u_debug.h index 1380d98d7ee..b82e7cb4d40 100644 --- a/src/gallium/auxiliary/util/u_debug.h +++ b/src/gallium/auxiliary/util/u_debug.h @@ -65,6 +65,11 @@ extern "C" { #define __FUNCTION__ "???" #endif +#if defined(__GNUC__) +#define _util_printf_format(fmt, list) __attribute__ ((format (printf, fmt, list))) +#else +#define _util_printf_format(fmt, list) +#endif void _debug_vprintf(const char *format, va_list ap); @@ -82,14 +87,17 @@ _debug_printf(const char *format, ...) /** * Print debug messages. * - * The actual channel used to output debug message is platform specific. To - * avoid misformating or truncation, follow these rules of thumb: + * The actual channel used to output debug message is platform specific. To + * avoid misformating or truncation, follow these rules of thumb: * - output whole lines - * - avoid outputing large strings (512 bytes is the current maximum length + * - avoid outputing large strings (512 bytes is the current maximum length * that is guaranteed to be printed in all platforms) */ #if !defined(PIPE_OS_HAIKU) static INLINE void +debug_printf(const char *format, ...) _util_printf_format(1,2); + +static INLINE void debug_printf(const char *format, ...) { #ifdef DEBUG diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c index f06c0e463d0..4e3d35f40e7 100644 --- a/src/gallium/auxiliary/util/u_gen_mipmap.c +++ b/src/gallium/auxiliary/util/u_gen_mipmap.c @@ -1515,6 +1515,17 @@ util_gen_mipmap(struct gen_mipmap_state *ctx, uint zslice = 0; uint offset; + /* The texture object should have room for the levels which we're + * about to generate. + */ + assert(lastLevel <= pt->last_level); + + /* If this fails, why are we here? */ + assert(lastLevel > baseLevel); + + assert(filter == PIPE_TEX_FILTER_LINEAR || + filter == PIPE_TEX_FILTER_NEAREST); + /* check if we can render in the texture's format */ if (!screen->is_format_supported(screen, pt->format, PIPE_TEXTURE_2D, PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) { diff --git a/src/gallium/auxiliary/util/u_math.h b/src/gallium/auxiliary/util/u_math.h index b428dc544c3..75b075f160d 100644 --- a/src/gallium/auxiliary/util/u_math.h +++ b/src/gallium/auxiliary/util/u_math.h @@ -283,6 +283,14 @@ util_fast_pow(float x, float y) return util_fast_exp2(util_fast_log2(x) * y); } +/* Note that this counts zero as a power of two. + */ +static INLINE boolean +util_is_power_of_two( unsigned v ) +{ + return (v & (v-1)) == 0; +} + /** * Floor(x), returned as int. @@ -463,6 +471,26 @@ util_logbase2(unsigned n) /** + * Returns the smallest power of two >= x + */ +static INLINE unsigned +util_next_power_of_two(unsigned x) +{ + unsigned i; + + if (x == 0) + return 1; + + --x; + + for (i = 1; i < sizeof(unsigned) * 8; i <<= 1) + x |= x >> i; + + return x + 1; +} + + +/** * Clamp X to [MIN, MAX]. * This is a macro to allow float, int, uint, etc. types. */ diff --git a/src/gallium/auxiliary/util/u_network.c b/src/gallium/auxiliary/util/u_network.c index bc4b7584067..6269c72e121 100644 --- a/src/gallium/auxiliary/util/u_network.c +++ b/src/gallium/auxiliary/util/u_network.c @@ -6,7 +6,7 @@ #if defined(PIPE_SUBSYSTEM_WINDOWS_USER) # include <winsock2.h> # include <windows.h> -#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) +#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_BSD) # include <sys/socket.h> # include <netinet/in.h> # include <unistd.h> @@ -54,7 +54,7 @@ u_socket_close(int s) if (s < 0) return; -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_BSD) shutdown(s, SHUT_RDWR); close(s); #elif defined(PIPE_SUBSYSTEM_WINDOWS_USER) @@ -169,7 +169,7 @@ u_socket_listen_on_port(uint16_t portnum) void u_socket_block(int s, boolean block) { -#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) +#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_BSD) int old = fcntl(s, F_GETFL, 0); if (old == -1) return; diff --git a/src/gallium/auxiliary/util/u_network.h b/src/gallium/auxiliary/util/u_network.h index 8c778f492ca..0aa898b9676 100644 --- a/src/gallium/auxiliary/util/u_network.h +++ b/src/gallium/auxiliary/util/u_network.h @@ -6,7 +6,7 @@ #if defined(PIPE_SUBSYSTEM_WINDOWS_USER) # define PIPE_HAVE_SOCKETS -#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) +#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_HAIKU) || defined(PIPE_OS_BSD) # define PIPE_HAVE_SOCKETS #endif diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c index 0d6489c26e4..8a22f584bee 100644 --- a/src/gallium/auxiliary/util/u_tile.c +++ b/src/gallium/auxiliary/util/u_tile.c @@ -1452,7 +1452,7 @@ pipe_put_tile_z(struct pipe_transfer *pt, case PIPE_FORMAT_S8Z24_UNORM: { uint *pDest = (uint *) (map + y * pt->stride + x*4); - assert(pt->usage == PIPE_TRANSFER_READ_WRITE); + assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE); for (i = 0; i < h; i++) { for (j = 0; j < w; j++) { /* convert 32-bit Z to 24-bit Z, preserve stencil */ @@ -1479,7 +1479,7 @@ pipe_put_tile_z(struct pipe_transfer *pt, case PIPE_FORMAT_Z24S8_UNORM: { uint *pDest = (uint *) (map + y * pt->stride + x*4); - assert(pt->usage == PIPE_TRANSFER_READ_WRITE); + assert((pt->usage & PIPE_TRANSFER_READ_WRITE) == PIPE_TRANSFER_READ_WRITE); for (i = 0; i < h; i++) { for (j = 0; j < w; j++) { /* convert 32-bit Z to 24-bit Z, preserve stencil */ diff --git a/src/gallium/auxiliary/vl/Makefile b/src/gallium/auxiliary/vl/Makefile new file mode 100644 index 00000000000..4314c1e8d69 --- /dev/null +++ b/src/gallium/auxiliary/vl/Makefile @@ -0,0 +1,13 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +LIBNAME = vl + +C_SOURCES = \ + vl_bitstream_parser.c \ + vl_mpeg12_mc_renderer.c \ + vl_compositor.c \ + vl_csc.c \ + vl_shader_build.c + +include ../../Makefile.template diff --git a/src/gallium/auxiliary/vl/SConscript b/src/gallium/auxiliary/vl/SConscript new file mode 100644 index 00000000000..aed69f5efed --- /dev/null +++ b/src/gallium/auxiliary/vl/SConscript @@ -0,0 +1,13 @@ +Import('*') + +vl = env.ConvenienceLibrary( + target = 'vl', + source = [ + 'vl_bitstream_parser.c', + 'vl_mpeg12_mc_renderer.c', + 'vl_compositor.c', + 'vl_csc.c', + 'vl_shader_build.c', + ]) + +auxiliaries.insert(0, vl) diff --git a/src/gallium/auxiliary/vl/vl_bitstream_parser.c b/src/gallium/auxiliary/vl/vl_bitstream_parser.c new file mode 100644 index 00000000000..3193ea5f41c --- /dev/null +++ b/src/gallium/auxiliary/vl/vl_bitstream_parser.c @@ -0,0 +1,167 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "vl_bitstream_parser.h" +#include <assert.h> +#include <limits.h> +#include <util/u_memory.h> + +static unsigned +grab_bits(unsigned cursor, unsigned how_many_bits, unsigned bitstream_elt) +{ + unsigned excess_bits = sizeof(unsigned) * CHAR_BIT - how_many_bits - cursor; + + assert(cursor < sizeof(unsigned) * CHAR_BIT); + assert(how_many_bits > 0 && how_many_bits <= sizeof(unsigned) * CHAR_BIT); + assert(cursor + how_many_bits <= sizeof(unsigned) * CHAR_BIT); + + return (bitstream_elt << excess_bits) >> (excess_bits + cursor); +} + +static unsigned +show_bits(unsigned cursor, unsigned how_many_bits, const unsigned *bitstream) +{ + unsigned cur_int = cursor / (sizeof(unsigned) * CHAR_BIT); + unsigned cur_bit = cursor % (sizeof(unsigned) * CHAR_BIT); + + assert(bitstream); + + if (cur_bit + how_many_bits > sizeof(unsigned) * CHAR_BIT) { + unsigned lower = grab_bits(cur_bit, sizeof(unsigned) * CHAR_BIT - cur_bit, + bitstream[cur_int]); + unsigned upper = grab_bits(0, cur_bit + how_many_bits - sizeof(unsigned) * CHAR_BIT, + bitstream[cur_int + 1]); + return lower | upper << (sizeof(unsigned) * CHAR_BIT - cur_bit); + } + else + return grab_bits(cur_bit, how_many_bits, bitstream[cur_int]); +} + +bool vl_bitstream_parser_init(struct vl_bitstream_parser *parser, + unsigned num_bitstreams, + const void **bitstreams, + const unsigned *sizes) +{ + assert(parser); + assert(num_bitstreams); + assert(bitstreams); + assert(sizes); + + parser->num_bitstreams = num_bitstreams; + parser->bitstreams = (const unsigned**)bitstreams; + parser->sizes = sizes; + parser->cur_bitstream = 0; + parser->cursor = 0; + + return true; +} + +void vl_bitstream_parser_cleanup(struct vl_bitstream_parser *parser) +{ + assert(parser); +} + +unsigned +vl_bitstream_parser_get_bits(struct vl_bitstream_parser *parser, + unsigned how_many_bits) +{ + unsigned bits; + + assert(parser); + + bits = vl_bitstream_parser_show_bits(parser, how_many_bits); + + vl_bitstream_parser_forward(parser, how_many_bits); + + return bits; +} + +unsigned +vl_bitstream_parser_show_bits(struct vl_bitstream_parser *parser, + unsigned how_many_bits) +{ + unsigned bits = 0; + unsigned shift = 0; + unsigned cursor; + unsigned cur_bitstream; + + assert(parser); + + cursor = parser->cursor; + cur_bitstream = parser->cur_bitstream; + + while (1) { + unsigned bits_left = parser->sizes[cur_bitstream] * CHAR_BIT - cursor; + unsigned bits_to_show = how_many_bits > bits_left ? bits_left : how_many_bits; + + bits |= show_bits(cursor, bits_to_show, + parser->bitstreams[cur_bitstream]) << shift; + + if (how_many_bits > bits_to_show) { + how_many_bits -= bits_to_show; + cursor = 0; + ++cur_bitstream; + shift += bits_to_show; + } + else + break; + } + + return bits; +} + +void vl_bitstream_parser_forward(struct vl_bitstream_parser *parser, + unsigned how_many_bits) +{ + assert(parser); + assert(how_many_bits); + + parser->cursor += how_many_bits; + + while (parser->cursor > parser->sizes[parser->cur_bitstream] * CHAR_BIT) { + parser->cursor -= parser->sizes[parser->cur_bitstream++] * CHAR_BIT; + assert(parser->cur_bitstream < parser->num_bitstreams); + } +} + +void vl_bitstream_parser_rewind(struct vl_bitstream_parser *parser, + unsigned how_many_bits) +{ + signed c; + + assert(parser); + assert(how_many_bits); + + c = parser->cursor - how_many_bits; + + while (c < 0) { + c += parser->sizes[parser->cur_bitstream--] * CHAR_BIT; + assert(parser->cur_bitstream < parser->num_bitstreams); + } + + parser->cursor = (unsigned)c; +} diff --git a/src/gallium/state_trackers/wgl/shared/stw_public.h b/src/gallium/auxiliary/vl/vl_bitstream_parser.h index 7fe9cfb3561..30ec743fa75 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_public.h +++ b/src/gallium/auxiliary/vl/vl_bitstream_parser.h @@ -1,8 +1,8 @@ /************************************************************************** - * - * Copyright 2009 VMware, Inc. + * + * Copyright 2009 Younes Manton. * All Rights Reserved. - * + * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the * "Software"), to deal in the Software without restriction, including @@ -10,11 +10,11 @@ * distribute, sub license, and/or sell copies of the Software, and to * permit persons to whom the Software is furnished to do so, subject to * the following conditions: - * + * * The above copyright notice and this permission notice (including the * next paragraph) shall be included in all copies or substantial portions * of the Software. - * + * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. @@ -22,52 +22,42 @@ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * + * **************************************************************************/ -#ifndef STW_PUBLIC_H -#define STW_PUBLIC_H +#ifndef vl_bitstream_parser_h +#define vl_bitstream_parser_h -#include <windows.h> +#include "pipe/p_compiler.h" -BOOL stw_copy_context( UINT_PTR hglrcSrc, - UINT_PTR hglrcDst, - UINT mask ); +struct vl_bitstream_parser +{ + unsigned num_bitstreams; + const unsigned **bitstreams; + const unsigned *sizes; + unsigned cur_bitstream; + unsigned cursor; +}; -UINT_PTR stw_create_layer_context( HDC hdc, - int iLayerPlane ); +bool vl_bitstream_parser_init(struct vl_bitstream_parser *parser, + unsigned num_bitstreams, + const void **bitstreams, + const unsigned *sizes); -BOOL stw_share_lists( UINT_PTR hglrc1, UINT_PTR hglrc2 ); +void vl_bitstream_parser_cleanup(struct vl_bitstream_parser *parser); -BOOL stw_delete_context( UINT_PTR hglrc ); +unsigned +vl_bitstream_parser_get_bits(struct vl_bitstream_parser *parser, + unsigned how_many_bits); -BOOL -stw_release_context( UINT_PTR dhglrc ); +unsigned +vl_bitstream_parser_show_bits(struct vl_bitstream_parser *parser, + unsigned how_many_bits); -UINT_PTR stw_get_current_context( void ); +void vl_bitstream_parser_forward(struct vl_bitstream_parser *parser, + unsigned how_many_bits); -HDC stw_get_current_dc( void ); +void vl_bitstream_parser_rewind(struct vl_bitstream_parser *parser, + unsigned how_many_bits); -BOOL stw_make_current( HDC hdc, UINT_PTR hglrc ); - -BOOL stw_swap_buffers( HDC hdc ); - -BOOL -stw_swap_layer_buffers( HDC hdc, UINT fuPlanes ); - -PROC stw_get_proc_address( LPCSTR lpszProc ); - -int stw_pixelformat_describe( HDC hdc, - int iPixelFormat, - UINT nBytes, - LPPIXELFORMATDESCRIPTOR ppfd ); - -int stw_pixelformat_get( HDC hdc ); - -BOOL stw_pixelformat_set( HDC hdc, - int iPixelFormat ); - -int stw_pixelformat_choose( HDC hdc, - CONST PIXELFORMATDESCRIPTOR *ppfd ); - -#endif +#endif /* vl_bitstream_parser_h */ diff --git a/src/gallium/auxiliary/vl/vl_compositor.c b/src/gallium/auxiliary/vl/vl_compositor.c new file mode 100644 index 00000000000..b36dbeb2088 --- /dev/null +++ b/src/gallium/auxiliary/vl/vl_compositor.c @@ -0,0 +1,532 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "vl_compositor.h" +#include <assert.h> +#include <pipe/p_context.h> +#include <pipe/p_inlines.h> +#include <tgsi/tgsi_parse.h> +#include <tgsi/tgsi_build.h> +#include <util/u_memory.h> +#include "vl_csc.h" +#include "vl_shader_build.h" + +struct vertex2f +{ + float x, y; +}; + +struct vertex4f +{ + float x, y, z, w; +}; + +struct vertex_shader_consts +{ + struct vertex4f dst_scale; + struct vertex4f dst_trans; + struct vertex4f src_scale; + struct vertex4f src_trans; +}; + +struct fragment_shader_consts +{ + float matrix[16]; +}; + +/* + * Represents 2 triangles in a strip in normalized coords. + * Used to render the surface onto the frame buffer. + */ +static const struct vertex2f surface_verts[4] = +{ + {0.0f, 0.0f}, + {0.0f, 1.0f}, + {1.0f, 0.0f}, + {1.0f, 1.0f} +}; + +/* + * Represents texcoords for the above. We can use the position values directly. + * TODO: Duplicate these in the shader, no need to create a buffer. + */ +static const struct vertex2f *surface_texcoords = surface_verts; + +static void +create_vert_shader(struct vl_compositor *c) +{ + const unsigned max_tokens = 50; + + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned ti; + + unsigned i; + + assert(c); + + tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token)); + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Vertex texcoords + */ + for (i = 0; i < 2; i++) { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling vector to scale vertex pos rect to destination size + * decl c1 ; Translation vector to move vertex pos rect into position + * decl c2 ; Scaling vector to scale texcoord rect to source size + * decl c3 ; Translation vector to move texcoord rect into position + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl o0 ; Vertex pos + * decl o1 ; Vertex texcoords + */ + for (i = 0; i < 2; i++) { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * mad o0, i0, c0, c1 ; Scale and translate unit output rect to destination size and pos + * mad o1, i1, c2, c3 ; Scale and translate unit texcoord rect to source size and pos + */ + for (i = 0; i < 2; ++i) { + inst = vl_inst4(TGSI_OPCODE_MAD, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i, TGSI_FILE_CONSTANT, i * 2, TGSI_FILE_CONSTANT, i * 2 + 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + assert(ti <= max_tokens); + + vs.tokens = tokens; + c->vertex_shader = c->pipe->create_vs_state(c->pipe, &vs); + FREE(tokens); +} + +static void +create_frag_shader(struct vl_compositor *c) +{ + const unsigned max_tokens = 50; + + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned ti; + + unsigned i; + + assert(c); + + tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token)); + *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); + header = (struct tgsi_header*)&tokens[1]; + *header = tgsi_build_header(); + *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* decl i0 ; Texcoords for s0 */ + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl c0-c3 ; CSC matrix c0-c3 + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0 */ + decl = vl_decl_temps(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl s0 ; Sampler for tex containing picture to display */ + decl = vl_decl_samplers(0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* tex2d t0, i0, s0 ; Read src pixel */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * dp4 o0.x, t0, c0 ; Multiply pixel by the color conversion matrix + * dp4 o0.y, t0, c1 + * dp4 o0.z, t0, c2 + * dp4 o0.w, t0, c3 + */ + for (i = 0; i < 4; ++i) { + inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i); + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + assert(ti <= max_tokens); + + fs.tokens = tokens; + c->fragment_shader = c->pipe->create_fs_state(c->pipe, &fs); + FREE(tokens); +} + +static bool +init_pipe_state(struct vl_compositor *c) +{ + struct pipe_sampler_state sampler; + + assert(c); + + c->fb_state.nr_cbufs = 1; + c->fb_state.zsbuf = NULL; + + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR; + sampler.compare_mode = PIPE_TEX_COMPARE_NONE; + sampler.compare_func = PIPE_FUNC_ALWAYS; + sampler.normalized_coords = 1; + /*sampler.prefilter = ;*/ + /*sampler.lod_bias = ;*/ + /*sampler.min_lod = ;*/ + /*sampler.max_lod = ;*/ + /*sampler.border_color[i] = ;*/ + /*sampler.max_anisotropy = ;*/ + c->sampler = c->pipe->create_sampler_state(c->pipe, &sampler); + + return true; +} + +static void cleanup_pipe_state(struct vl_compositor *c) +{ + assert(c); + + c->pipe->delete_sampler_state(c->pipe, c->sampler); +} + +static bool +init_shaders(struct vl_compositor *c) +{ + assert(c); + + create_vert_shader(c); + create_frag_shader(c); + + return true; +} + +static void cleanup_shaders(struct vl_compositor *c) +{ + assert(c); + + c->pipe->delete_vs_state(c->pipe, c->vertex_shader); + c->pipe->delete_fs_state(c->pipe, c->fragment_shader); +} + +static bool +init_buffers(struct vl_compositor *c) +{ + struct fragment_shader_consts fsc; + + assert(c); + + /* + * Create our vertex buffer and vertex buffer element + * VB contains 4 vertices that render a quad covering the entire window + * to display a rendered surface + * Quad is rendered as a tri strip + */ + c->vertex_bufs[0].stride = sizeof(struct vertex2f); + c->vertex_bufs[0].max_index = 3; + c->vertex_bufs[0].buffer_offset = 0; + c->vertex_bufs[0].buffer = pipe_buffer_create + ( + c->pipe->screen, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vertex2f) * 4 + ); + + memcpy + ( + pipe_buffer_map(c->pipe->screen, c->vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + surface_verts, + sizeof(struct vertex2f) * 4 + ); + + pipe_buffer_unmap(c->pipe->screen, c->vertex_bufs[0].buffer); + + c->vertex_elems[0].src_offset = 0; + c->vertex_elems[0].vertex_buffer_index = 0; + c->vertex_elems[0].nr_components = 2; + c->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* + * Create our texcoord buffer and texcoord buffer element + * Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices + */ + c->vertex_bufs[1].stride = sizeof(struct vertex2f); + c->vertex_bufs[1].max_index = 3; + c->vertex_bufs[1].buffer_offset = 0; + c->vertex_bufs[1].buffer = pipe_buffer_create + ( + c->pipe->screen, + 1, + PIPE_BUFFER_USAGE_VERTEX, + sizeof(struct vertex2f) * 4 + ); + + memcpy + ( + pipe_buffer_map(c->pipe->screen, c->vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + surface_texcoords, + sizeof(struct vertex2f) * 4 + ); + + pipe_buffer_unmap(c->pipe->screen, c->vertex_bufs[1].buffer); + + c->vertex_elems[1].src_offset = 0; + c->vertex_elems[1].vertex_buffer_index = 1; + c->vertex_elems[1].nr_components = 2; + c->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* + * Create our vertex shader's constant buffer + * Const buffer contains scaling and translation vectors + */ + c->vs_const_buf.buffer = pipe_buffer_create + ( + c->pipe->screen, + 1, + PIPE_BUFFER_USAGE_CONSTANT | PIPE_BUFFER_USAGE_DISCARD, + sizeof(struct vertex_shader_consts) + ); + + /* + * Create our fragment shader's constant buffer + * Const buffer contains the color conversion matrix and bias vectors + */ + c->fs_const_buf.buffer = pipe_buffer_create + ( + c->pipe->screen, + 1, + PIPE_BUFFER_USAGE_CONSTANT, + sizeof(struct fragment_shader_consts) + ); + + vl_csc_get_matrix(VL_CSC_COLOR_STANDARD_IDENTITY, NULL, true, fsc.matrix); + + vl_compositor_set_csc_matrix(c, fsc.matrix); + + return true; +} + +static void +cleanup_buffers(struct vl_compositor *c) +{ + unsigned i; + + assert(c); + + for (i = 0; i < 2; ++i) + pipe_buffer_reference(&c->vertex_bufs[i].buffer, NULL); + + pipe_buffer_reference(&c->vs_const_buf.buffer, NULL); + pipe_buffer_reference(&c->fs_const_buf.buffer, NULL); +} + +bool vl_compositor_init(struct vl_compositor *compositor, struct pipe_context *pipe) +{ + assert(compositor); + + memset(compositor, 0, sizeof(struct vl_compositor)); + + compositor->pipe = pipe; + + if (!init_pipe_state(compositor)) + return false; + if (!init_shaders(compositor)) { + cleanup_pipe_state(compositor); + return false; + } + if (!init_buffers(compositor)) { + cleanup_shaders(compositor); + cleanup_pipe_state(compositor); + return false; + } + + return true; +} + +void vl_compositor_cleanup(struct vl_compositor *compositor) +{ + assert(compositor); + + cleanup_buffers(compositor); + cleanup_shaders(compositor); + cleanup_pipe_state(compositor); +} + +void vl_compositor_render(struct vl_compositor *compositor, + /*struct pipe_texture *backround, + struct pipe_video_rect *backround_area,*/ + struct pipe_texture *src_surface, + enum pipe_mpeg12_picture_type picture_type, + /*unsigned num_past_surfaces, + struct pipe_texture *past_surfaces, + unsigned num_future_surfaces, + struct pipe_texture *future_surfaces,*/ + struct pipe_video_rect *src_area, + struct pipe_texture *dst_surface, + struct pipe_video_rect *dst_area, + /*unsigned num_layers, + struct pipe_texture *layers, + struct pipe_video_rect *layer_src_areas, + struct pipe_video_rect *layer_dst_areas*/ + struct pipe_fence_handle **fence) +{ + struct vertex_shader_consts *vs_consts; + + assert(compositor); + assert(src_surface); + assert(src_area); + assert(dst_surface); + assert(dst_area); + assert(picture_type == PIPE_MPEG12_PICTURE_TYPE_FRAME); + + compositor->fb_state.width = dst_surface->width[0]; + compositor->fb_state.height = dst_surface->height[0]; + compositor->fb_state.cbufs[0] = compositor->pipe->screen->get_tex_surface + ( + compositor->pipe->screen, + dst_surface, + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE + ); + + compositor->viewport.scale[0] = compositor->fb_state.width; + compositor->viewport.scale[1] = compositor->fb_state.height; + compositor->viewport.scale[2] = 1; + compositor->viewport.scale[3] = 1; + compositor->viewport.translate[0] = 0; + compositor->viewport.translate[1] = 0; + compositor->viewport.translate[2] = 0; + compositor->viewport.translate[3] = 0; + + compositor->pipe->set_framebuffer_state(compositor->pipe, &compositor->fb_state); + compositor->pipe->set_viewport_state(compositor->pipe, &compositor->viewport); + compositor->pipe->bind_sampler_states(compositor->pipe, 1, &compositor->sampler); + compositor->pipe->set_sampler_textures(compositor->pipe, 1, &src_surface); + compositor->pipe->bind_vs_state(compositor->pipe, compositor->vertex_shader); + compositor->pipe->bind_fs_state(compositor->pipe, compositor->fragment_shader); + compositor->pipe->set_vertex_buffers(compositor->pipe, 2, compositor->vertex_bufs); + compositor->pipe->set_vertex_elements(compositor->pipe, 2, compositor->vertex_elems); + compositor->pipe->set_constant_buffer(compositor->pipe, PIPE_SHADER_VERTEX, 0, &compositor->vs_const_buf); + compositor->pipe->set_constant_buffer(compositor->pipe, PIPE_SHADER_FRAGMENT, 0, &compositor->fs_const_buf); + + vs_consts = pipe_buffer_map + ( + compositor->pipe->screen, + compositor->vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD + ); + + vs_consts->dst_scale.x = dst_area->w / (float)compositor->fb_state.cbufs[0]->width; + vs_consts->dst_scale.y = dst_area->h / (float)compositor->fb_state.cbufs[0]->height; + vs_consts->dst_scale.z = 1; + vs_consts->dst_scale.w = 1; + vs_consts->dst_trans.x = dst_area->x / (float)compositor->fb_state.cbufs[0]->width; + vs_consts->dst_trans.y = dst_area->y / (float)compositor->fb_state.cbufs[0]->height; + vs_consts->dst_trans.z = 0; + vs_consts->dst_trans.w = 0; + + vs_consts->src_scale.x = src_area->w / (float)src_surface->width[0]; + vs_consts->src_scale.y = src_area->h / (float)src_surface->height[0]; + vs_consts->src_scale.z = 1; + vs_consts->src_scale.w = 1; + vs_consts->src_trans.x = src_area->x / (float)src_surface->width[0]; + vs_consts->src_trans.y = src_area->y / (float)src_surface->height[0]; + vs_consts->src_trans.z = 0; + vs_consts->src_trans.w = 0; + + pipe_buffer_unmap(compositor->pipe->screen, compositor->vs_const_buf.buffer); + + compositor->pipe->draw_arrays(compositor->pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4); + compositor->pipe->flush(compositor->pipe, PIPE_FLUSH_RENDER_CACHE, fence); + + pipe_surface_reference(&compositor->fb_state.cbufs[0], NULL); +} + +void vl_compositor_set_csc_matrix(struct vl_compositor *compositor, const float *mat) +{ + assert(compositor); + + memcpy + ( + pipe_buffer_map(compositor->pipe->screen, compositor->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + mat, + sizeof(struct fragment_shader_consts) + ); + + pipe_buffer_unmap(compositor->pipe->screen, compositor->fs_const_buf.buffer); +} diff --git a/src/gallium/auxiliary/vl/vl_compositor.h b/src/gallium/auxiliary/vl/vl_compositor.h new file mode 100644 index 00000000000..17e2afd3539 --- /dev/null +++ b/src/gallium/auxiliary/vl/vl_compositor.h @@ -0,0 +1,76 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef vl_compositor_h +#define vl_compositor_h + +#include <pipe/p_compiler.h> +#include <pipe/p_state.h> +#include <pipe/p_video_state.h> + +struct pipe_context; +struct pipe_texture; + +struct vl_compositor +{ + struct pipe_context *pipe; + + struct pipe_framebuffer_state fb_state; + void *sampler; + void *vertex_shader; + void *fragment_shader; + struct pipe_viewport_state viewport; + struct pipe_vertex_buffer vertex_bufs[2]; + struct pipe_vertex_element vertex_elems[2]; + struct pipe_constant_buffer vs_const_buf, fs_const_buf; +}; + +bool vl_compositor_init(struct vl_compositor *compositor, struct pipe_context *pipe); + +void vl_compositor_cleanup(struct vl_compositor *compositor); + +void vl_compositor_render(struct vl_compositor *compositor, + /*struct pipe_texture *backround, + struct pipe_video_rect *backround_area,*/ + struct pipe_texture *src_surface, + enum pipe_mpeg12_picture_type picture_type, + /*unsigned num_past_surfaces, + struct pipe_texture *past_surfaces, + unsigned num_future_surfaces, + struct pipe_texture *future_surfaces,*/ + struct pipe_video_rect *src_area, + struct pipe_texture *dst_surface, + struct pipe_video_rect *dst_area, + /*unsigned num_layers, + struct pipe_texture *layers, + struct pipe_video_rect *layer_src_areas, + struct pipe_video_rect *layer_dst_areas,*/ + struct pipe_fence_handle **fence); + +void vl_compositor_set_csc_matrix(struct vl_compositor *compositor, const float *mat); + +#endif /* vl_compositor_h */ diff --git a/src/gallium/auxiliary/vl/vl_csc.c b/src/gallium/auxiliary/vl/vl_csc.c new file mode 100644 index 00000000000..5ecc43a5fa3 --- /dev/null +++ b/src/gallium/auxiliary/vl/vl_csc.c @@ -0,0 +1,206 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "vl_csc.h" +#include <util/u_math.h> +#include <util/u_debug.h> + +/* + * Color space conversion formulas + * + * To convert YCbCr to RGB, + * vec4 ycbcr, rgb + * mat44 csc + * rgb = csc * ycbcr + * + * To calculate the color space conversion matrix csc with ProcAmp adjustments, + * mat44 csc, cstd, procamp, bias + * csc = cstd * (procamp * bias) + * + * Where cstd is a matrix corresponding to one of the color standards (BT.601, BT.709, etc) + * adjusted for the kind of YCbCr -> RGB mapping wanted (1:1, full), + * bias is a matrix corresponding to the kind of YCbCr -> RGB mapping wanted (1:1, full) + * + * To calculate procamp, + * mat44 procamp, hue, saturation, brightness, contrast + * procamp = brightness * (saturation * (contrast * hue)) + * Alternatively, + * procamp = saturation * (brightness * (contrast * hue)) + * + * contrast + * [ c, 0, 0, 0] + * [ 0, c, 0, 0] + * [ 0, 0, c, 0] + * [ 0, 0, 0, 1] + * + * brightness + * [ 1, 0, 0, b] + * [ 0, 1, 0, 0] + * [ 0, 0, 1, 0] + * [ 0, 0, 0, 1] + * + * saturation + * [ 1, 0, 0, 0] + * [ 0, s, 0, 0] + * [ 0, 0, s, 0] + * [ 0, 0, 0, 1] + * + * hue + * [ 1, 0, 0, 0] + * [ 0, cos(h), sin(h), 0] + * [ 0, -sin(h), cos(h), 0] + * [ 0, 0, 0, 1] + * + * procamp + * [ c, 0, 0, b] + * [ 0, c*s*cos(h), c*s*sin(h), 0] + * [ 0, -c*s*sin(h), c*s*cos(h), 0] + * [ 0, 0, 0, 1] + * + * bias + * [ 1, 0, 0, ybias] + * [ 0, 1, 0, cbbias] + * [ 0, 0, 1, crbias] + * [ 0, 0, 0, 1] + * + * csc + * [ c*cstd[ 0], c*cstd[ 1]*s*cos(h) - c*cstd[ 2]*s*sin(h), c*cstd[ 2]*s*cos(h) + c*cstd[ 1]*s*sin(h), cstd[ 3] + cstd[ 0]*(b + c*ybias) + cstd[ 1]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[ 2]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h))] + * [ c*cstd[ 4], c*cstd[ 5]*s*cos(h) - c*cstd[ 6]*s*sin(h), c*cstd[ 6]*s*cos(h) + c*cstd[ 5]*s*sin(h), cstd[ 7] + cstd[ 4]*(b + c*ybias) + cstd[ 5]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[ 6]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h))] + * [ c*cstd[ 8], c*cstd[ 9]*s*cos(h) - c*cstd[10]*s*sin(h), c*cstd[10]*s*cos(h) + c*cstd[ 9]*s*sin(h), cstd[11] + cstd[ 8]*(b + c*ybias) + cstd[ 9]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[10]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h))] + * [ c*cstd[12], c*cstd[13]*s*cos(h) - c*cstd[14]*s*sin(h), c*cstd[14]*s*cos(h) + c*cstd[13]*s*sin(h), cstd[15] + cstd[12]*(b + c*ybias) + cstd[13]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[14]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h))] + */ + +/* + * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [16,235] + */ +static const float bt_601[16] = +{ + 1.0f, 0.0f, 1.371f, 0.0f, + 1.0f, -0.336f, -0.698f, 0.0f, + 1.0f, 1.732f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f +}; + +/* + * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [0,255] + */ +static const float bt_601_full[16] = +{ + 1.164f, 0.0f, 1.596f, 0.0f, + 1.164f, -0.391f, -0.813f, 0.0f, + 1.164f, 2.018f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f +}; + +/* + * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [16,235] + */ +static const float bt_709[16] = +{ + 1.0f, 0.0f, 1.540f, 0.0f, + 1.0f, -0.183f, -0.459f, 0.0f, + 1.0f, 1.816f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f +}; + +/* + * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: + * Y is in [16,235], Cb and Cr are in [16,240] + * R, G, and B are in [0,255] + */ +static const float bt_709_full[16] = +{ + 1.164f, 0.0f, 1.793f, 0.0f, + 1.164f, -0.213f, -0.534f, 0.0f, + 1.164f, 2.115f, 0.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f +}; + +static const float identity[16] = +{ + 1.0f, 0.0f, 0.0f, 0.0f, + 0.0f, 1.0f, 0.0f, 0.0f, + 0.0f, 0.0f, 1.0f, 0.0f, + 0.0f, 0.0f, 0.0f, 1.0f +}; + +void vl_csc_get_matrix(enum VL_CSC_COLOR_STANDARD cs, + struct vl_procamp *procamp, + bool full_range, + float *matrix) +{ + float ybias = full_range ? -16.0f/255.0f : 0.0f; + float cbbias = -128.0f/255.0f; + float crbias = -128.0f/255.0f; + float c = procamp ? procamp->contrast : 1.0f; + float s = procamp ? procamp->saturation : 1.0f; + float b = procamp ? procamp->brightness : 0.0f; + float h = procamp ? procamp->hue : 0.0f; + const float *cstd; + + assert(matrix); + + switch (cs) { + case VL_CSC_COLOR_STANDARD_BT_601: + cstd = full_range ? &bt_601_full[0] : &bt_601[0]; + break; + case VL_CSC_COLOR_STANDARD_BT_709: + cstd = full_range ? &bt_709_full[0] : &bt_709[0]; + break; + case VL_CSC_COLOR_STANDARD_IDENTITY: + default: + assert(cs == VL_CSC_COLOR_STANDARD_IDENTITY); + memcpy(matrix, &identity[0], sizeof(float) * 16); + return; + } + + matrix[ 0] = c*cstd[ 0]; + matrix[ 1] = c*cstd[ 1]*s*cosf(h) - c*cstd[ 2]*s*sinf(h); + matrix[ 2] = c*cstd[ 2]*s*cosf(h) + c*cstd[ 1]*s*sinf(h); + matrix[ 3] = cstd[ 3] + cstd[ 0]*(b + c*ybias) + cstd[ 1]*(c*cbbias*s*cosf(h) + c*crbias*s*sinf(h)) + cstd[ 2]*(c*crbias*s*cosf(h) - c*cbbias*s*sinf(h)); + + matrix[ 4] = c*cstd[ 4]; + matrix[ 5] = c*cstd[ 5]*s*cosf(h) - c*cstd[ 6]*s*sinf(h); + matrix[ 6] = c*cstd[ 6]*s*cosf(h) + c*cstd[ 5]*s*sinf(h); + matrix[ 7] = cstd[ 7] + cstd[ 4]*(b + c*ybias) + cstd[ 5]*(c*cbbias*s*cosf(h) + c*crbias*s*sinf(h)) + cstd[ 6]*(c*crbias*s*cosf(h) - c*cbbias*s*sinf(h)); + + matrix[ 8] = c*cstd[ 8]; + matrix[ 9] = c*cstd[ 9]*s*cosf(h) - c*cstd[10]*s*sinf(h); + matrix[10] = c*cstd[10]*s*cosf(h) + c*cstd[ 9]*s*sinf(h); + matrix[11] = cstd[11] + cstd[ 8]*(b + c*ybias) + cstd[ 9]*(c*cbbias*s*cosf(h) + c*crbias*s*sinf(h)) + cstd[10]*(c*crbias*s*cosf(h) - c*cbbias*s*sinf(h)); + + matrix[12] = c*cstd[12]; + matrix[13] = c*cstd[13]*s*cos(h) - c*cstd[14]*s*sin(h); + matrix[14] = c*cstd[14]*s*cos(h) + c*cstd[13]*s*sin(h); + matrix[15] = cstd[15] + cstd[12]*(b + c*ybias) + cstd[13]*(c*cbbias*s*cos(h) + c*crbias*s*sin(h)) + cstd[14]*(c*crbias*s*cos(h) - c*cbbias*s*sin(h)); +} diff --git a/src/gallium/state_trackers/wgl/shared/stw_winsys.h b/src/gallium/auxiliary/vl/vl_csc.h index c0bf82c9ed7..722ca35f339 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_winsys.h +++ b/src/gallium/auxiliary/vl/vl_csc.h @@ -1,8 +1,8 @@ /************************************************************************** - * - * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * + * Copyright 2009 Younes Manton. * All Rights Reserved. - * + * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the * "Software"), to deal in the Software without restriction, including @@ -10,11 +10,11 @@ * distribute, sub license, and/or sell copies of the Software, and to * permit persons to whom the Software is furnished to do so, subject to * the following conditions: - * + * * The above copyright notice and this permission notice (including the * next paragraph) shall be included in all copies or substantial portions * of the Software. - * + * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. @@ -22,44 +22,32 @@ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * + * **************************************************************************/ -#ifndef STW_WINSYS_H -#define STW_WINSYS_H +#ifndef vl_csc_h +#define vl_csc_h -#include <windows.h> /* for HDC */ +#include <pipe/p_compiler.h> -#include "pipe/p_compiler.h" - -struct pipe_screen; -struct pipe_context; -struct pipe_surface; - -struct stw_winsys +struct vl_procamp { - struct pipe_screen * - (*create_screen)( void ); - - struct pipe_context * - (*create_context)( struct pipe_screen *screen ); - - void - (*flush_frontbuffer)( struct pipe_screen *screen, - struct pipe_surface *surf, - HDC hDC ); + float brightness; + float contrast; + float saturation; + float hue; }; -boolean -stw_init(const struct stw_winsys *stw_winsys); - -boolean -stw_init_thread(void); - -void -stw_cleanup_thread(void); +enum VL_CSC_COLOR_STANDARD +{ + VL_CSC_COLOR_STANDARD_IDENTITY, + VL_CSC_COLOR_STANDARD_BT_601, + VL_CSC_COLOR_STANDARD_BT_709 +}; -void -stw_cleanup(void); +void vl_csc_get_matrix(enum VL_CSC_COLOR_STANDARD cs, + struct vl_procamp *procamp, + bool full_range, + float *matrix); -#endif /* STW_WINSYS_H */ +#endif /* vl_csc_h */ diff --git a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c new file mode 100644 index 00000000000..6b3614821cc --- /dev/null +++ b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c @@ -0,0 +1,1654 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "vl_mpeg12_mc_renderer.h" +#include <assert.h> +#include <pipe/p_context.h> +#include <pipe/p_inlines.h> +#include <util/u_math.h> +#include <util/u_memory.h> +#include <tgsi/tgsi_parse.h> +#include <tgsi/tgsi_build.h> +#include "vl_shader_build.h" + +#define DEFAULT_BUF_ALIGNMENT 1 +#define MACROBLOCK_WIDTH 16 +#define MACROBLOCK_HEIGHT 16 +#define BLOCK_WIDTH 8 +#define BLOCK_HEIGHT 8 +#define ZERO_BLOCK_NIL -1.0f +#define ZERO_BLOCK_IS_NIL(zb) ((zb).x < 0.0f) + +struct vertex2f +{ + float x, y; +}; + +struct vertex4f +{ + float x, y, z, w; +}; + +struct vertex_shader_consts +{ + struct vertex4f denorm; +}; + +struct fragment_shader_consts +{ + struct vertex4f multiplier; + struct vertex4f div; +}; + +/* + * Muliplier renormalizes block samples from 16 bits to 12 bits. + * Divider is used when calculating Y % 2 for choosing top or bottom + * field for P or B macroblocks. + * TODO: Use immediates. + */ +static const struct fragment_shader_consts fs_consts = { + {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f}, + {0.5f, 2.0f, 0.0f, 0.0f} +}; + +struct vert_stream_0 +{ + struct vertex2f pos; + struct vertex2f luma_tc; + struct vertex2f cb_tc; + struct vertex2f cr_tc; +}; + +enum MACROBLOCK_TYPE +{ + MACROBLOCK_TYPE_INTRA, + MACROBLOCK_TYPE_FWD_FRAME_PRED, + MACROBLOCK_TYPE_FWD_FIELD_PRED, + MACROBLOCK_TYPE_BKWD_FRAME_PRED, + MACROBLOCK_TYPE_BKWD_FIELD_PRED, + MACROBLOCK_TYPE_BI_FRAME_PRED, + MACROBLOCK_TYPE_BI_FIELD_PRED, + + NUM_MACROBLOCK_TYPES +}; + +static void +create_intra_vert_shader(struct vl_mpeg12_mc_renderer *r) +{ + const unsigned max_tokens = 50; + + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned ti; + + unsigned i; + + assert(r); + + tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token)); + *(struct tgsi_version *) &tokens[0] = tgsi_build_version(); + header = (struct tgsi_header *) &tokens[1]; + *header = tgsi_build_header(); + *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma Cb texcoords + * decl i3 ; Chroma Cr texcoords + */ + for (i = 0; i < 4; i++) { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma Cb texcoords + * decl o3 ; Chroma Cr texcoords + */ + for (i = 0; i < 4; i++) { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * mov o0, i0 ; Move input vertex pos to output + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma Cb texcoords to output + * mov o3, i3 ; Move input chroma Cr texcoords to output + */ + for (i = 0; i < 4; ++i) { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + assert(ti <= max_tokens); + + vs.tokens = tokens; + r->i_vs = r->pipe->create_vs_state(r->pipe, &vs); + free(tokens); +} + +static void +create_intra_frag_shader(struct vl_mpeg12_mc_renderer *r) +{ + const unsigned max_tokens = 100; + + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned ti; + + unsigned i; + + assert(r); + + tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token)); + *(struct tgsi_version *) &tokens[0] = tgsi_build_version(); + header = (struct tgsi_header *) &tokens[1]; + *header = tgsi_build_header(); + *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Luma texcoords + * decl i1 ; Chroma Cb texcoords + * decl i2 ; Chroma Cr texcoords + */ + for (i = 0; i < 3; ++i) { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + */ + for (i = 0; i < 3; ++i) { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i2, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* mul o0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + assert(ti <= max_tokens); + + fs.tokens = tokens; + r->i_fs = r->pipe->create_fs_state(r->pipe, &fs); + free(tokens); +} + +static void +create_frame_pred_vert_shader(struct vl_mpeg12_mc_renderer *r) +{ + const unsigned max_tokens = 100; + + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned ti; + + unsigned i; + + assert(r); + + tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token)); + *(struct tgsi_version *) &tokens[0] = tgsi_build_version(); + header = (struct tgsi_header *) &tokens[1]; + *header = tgsi_build_header(); + *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma Cb texcoords + * decl i3 ; Chroma Cr texcoords + * decl i4 ; Ref surface top field texcoords + * decl i5 ; Ref surface bottom field texcoords (unused, packed in the same stream) + */ + for (i = 0; i < 6; i++) { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma Cb texcoords + * decl o3 ; Chroma Cr texcoords + * decl o4 ; Ref macroblock texcoords + */ + for (i = 0; i < 5; i++) { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * mov o0, i0 ; Move input vertex pos to output + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma Cb texcoords to output + * mov o3, i3 ; Move input chroma Cr texcoords to output + */ + for (i = 0; i < 4; ++i) { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* add o4, i0, i4 ; Translate vertex pos by motion vec to form ref macroblock texcoords */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 4); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + assert(ti <= max_tokens); + + vs.tokens = tokens; + r->p_vs[0] = r->pipe->create_vs_state(r->pipe, &vs); + free(tokens); +} + +static void +create_field_pred_vert_shader(struct vl_mpeg12_mc_renderer *r) +{ + assert(false); +} + +static void +create_frame_pred_frag_shader(struct vl_mpeg12_mc_renderer *r) +{ + const unsigned max_tokens = 100; + + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned ti; + + unsigned i; + + assert(r); + + tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token)); + *(struct tgsi_version *) &tokens[0] = tgsi_build_version(); + header = (struct tgsi_header *) &tokens[1]; + *header = tgsi_build_header(); + *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Luma texcoords + * decl i1 ; Chroma Cb texcoords + * decl i2 ; Chroma Cr texcoords + * decl i3 ; Ref macroblock texcoords + */ + for (i = 0; i < 4; ++i) { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0, t1 */ + decl = vl_decl_temps(0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for ref surface texture + */ + for (i = 0; i < 4; ++i) { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i2, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* tex2d t1, i3, s3 ; Read texel from ref macroblock */ + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 3, TGSI_FILE_SAMPLER, 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + assert(ti <= max_tokens); + + fs.tokens = tokens; + r->p_fs[0] = r->pipe->create_fs_state(r->pipe, &fs); + free(tokens); +} + +static void +create_field_pred_frag_shader(struct vl_mpeg12_mc_renderer *r) +{ + assert(false); +} + +static void +create_frame_bi_pred_vert_shader(struct vl_mpeg12_mc_renderer *r) +{ + const unsigned max_tokens = 100; + + struct pipe_shader_state vs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned ti; + + unsigned i; + + assert(r); + + tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token)); + *(struct tgsi_version *) &tokens[0] = tgsi_build_version(); + header = (struct tgsi_header *) &tokens[1]; + *header = tgsi_build_header(); + *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); + + ti = 3; + + /* + * decl i0 ; Vertex pos + * decl i1 ; Luma texcoords + * decl i2 ; Chroma Cb texcoords + * decl i3 ; Chroma Cr texcoords + * decl i4 ; First ref macroblock top field texcoords + * decl i5 ; First ref macroblock bottom field texcoords (unused, packed in the same stream) + * decl i6 ; Second ref macroblock top field texcoords + * decl i7 ; Second ref macroblock bottom field texcoords (unused, packed in the same stream) + */ + for (i = 0; i < 8; i++) { + decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl o0 ; Vertex pos + * decl o1 ; Luma texcoords + * decl o2 ; Chroma Cb texcoords + * decl o3 ; Chroma Cr texcoords + * decl o4 ; First ref macroblock texcoords + * decl o5 ; Second ref macroblock texcoords + */ + for (i = 0; i < 6; i++) { + decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * mov o0, i0 ; Move input vertex pos to output + * mov o1, i1 ; Move input luma texcoords to output + * mov o2, i2 ; Move input chroma Cb texcoords to output + * mov o3, i3 ; Move input chroma Cr texcoords to output + */ + for (i = 0; i < 4; ++i) { + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* + * add o4, i0, i4 ; Translate vertex pos by motion vec to form first ref macroblock texcoords + * add o5, i0, i6 ; Translate vertex pos by motion vec to form second ref macroblock texcoords + */ + for (i = 0; i < 2; ++i) { + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, (i + 2) * 2); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + assert(ti <= max_tokens); + + vs.tokens = tokens; + r->b_vs[0] = r->pipe->create_vs_state(r->pipe, &vs); + free(tokens); +} + +static void +create_field_bi_pred_vert_shader(struct vl_mpeg12_mc_renderer *r) +{ + assert(false); +} + +static void +create_frame_bi_pred_frag_shader(struct vl_mpeg12_mc_renderer *r) +{ + const unsigned max_tokens = 100; + + struct pipe_shader_state fs; + struct tgsi_token *tokens; + struct tgsi_header *header; + + struct tgsi_full_declaration decl; + struct tgsi_full_instruction inst; + + unsigned ti; + + unsigned i; + + assert(r); + + tokens = (struct tgsi_token *) malloc(max_tokens * sizeof(struct tgsi_token)); + *(struct tgsi_version *) &tokens[0] = tgsi_build_version(); + header = (struct tgsi_header *) &tokens[1]; + *header = tgsi_build_header(); + *(struct tgsi_processor *) &tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); + + ti = 3; + + /* + * decl i0 ; Luma texcoords + * decl i1 ; Chroma Cb texcoords + * decl i2 ; Chroma Cr texcoords + * decl i3 ; First ref macroblock texcoords + * decl i4 ; Second ref macroblock texcoords + */ + for (i = 0; i < 5; ++i) { + decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm + * decl c1 ; Constant 1/2 in .x channel to use as weight to blend past and future texels + */ + decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl o0 ; Fragment color */ + decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* decl t0-t2 */ + decl = vl_decl_temps(0, 2); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + + /* + * decl s0 ; Sampler for luma texture + * decl s1 ; Sampler for chroma Cb texture + * decl s2 ; Sampler for chroma Cr texture + * decl s3 ; Sampler for first ref surface texture + * decl s4 ; Sampler for second ref surface texture + */ + for (i = 0; i < 5; ++i) { + decl = vl_decl_samplers(i, i); + ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); + } + + /* + * tex2d t1, i0, s0 ; Read texel from luma texture + * mov t0.x, t1.x ; Move luma sample into .x component + * tex2d t1, i1, s1 ; Read texel from chroma Cb texture + * mov t0.y, t1.x ; Move Cb sample into .y component + * tex2d t1, i2, s2 ; Read texel from chroma Cr texture + * mov t0.z, t1.x ; Move Cr sample into .z component + */ + for (i = 0; i < 3; ++i) { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* mul t0, t0, c0 ; Rescale texel to correct range */ + inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* + * tex2d t1, i3, s3 ; Read texel from first ref macroblock + * tex2d t2, i4, s4 ; Read texel from second ref macroblock + */ + for (i = 0; i < 2; ++i) { + inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, i + 3); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + } + + /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ + inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); + inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; + inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ + inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + /* end */ + inst = vl_end(); + ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); + + assert(ti <= max_tokens); + + fs.tokens = tokens; + r->b_fs[0] = r->pipe->create_fs_state(r->pipe, &fs); + free(tokens); +} + +static void +create_field_bi_pred_frag_shader(struct vl_mpeg12_mc_renderer *r) +{ + assert(false); +} + +static void +xfer_buffers_map(struct vl_mpeg12_mc_renderer *r) +{ + unsigned i; + + assert(r); + + for (i = 0; i < 3; ++i) { + r->tex_transfer[i] = r->pipe->screen->get_tex_transfer + ( + r->pipe->screen, r->textures.all[i], + 0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0, + r->textures.all[i]->width[0], r->textures.all[i]->height[0] + ); + + r->texels[i] = r->pipe->screen->transfer_map(r->pipe->screen, r->tex_transfer[i]); + } +} + +static void +xfer_buffers_unmap(struct vl_mpeg12_mc_renderer *r) +{ + unsigned i; + + assert(r); + + for (i = 0; i < 3; ++i) { + r->pipe->screen->transfer_unmap(r->pipe->screen, r->tex_transfer[i]); + r->pipe->screen->tex_transfer_destroy(r->tex_transfer[i]); + } +} + +static bool +init_pipe_state(struct vl_mpeg12_mc_renderer *r) +{ + struct pipe_sampler_state sampler; + unsigned filters[5]; + unsigned i; + + assert(r); + + r->viewport.scale[0] = r->pot_buffers ? + util_next_power_of_two(r->picture_width) : r->picture_width; + r->viewport.scale[1] = r->pot_buffers ? + util_next_power_of_two(r->picture_height) : r->picture_height; + r->viewport.scale[2] = 1; + r->viewport.scale[3] = 1; + r->viewport.translate[0] = 0; + r->viewport.translate[1] = 0; + r->viewport.translate[2] = 0; + r->viewport.translate[3] = 0; + + r->fb_state.width = r->pot_buffers ? + util_next_power_of_two(r->picture_width) : r->picture_width; + r->fb_state.height = r->pot_buffers ? + util_next_power_of_two(r->picture_height) : r->picture_height; + r->fb_state.nr_cbufs = 1; + r->fb_state.zsbuf = NULL; + + /* Luma filter */ + filters[0] = PIPE_TEX_FILTER_NEAREST; + /* Chroma filters */ + if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_444 || + r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE) { + filters[1] = PIPE_TEX_FILTER_NEAREST; + filters[2] = PIPE_TEX_FILTER_NEAREST; + } + else { + filters[1] = PIPE_TEX_FILTER_LINEAR; + filters[2] = PIPE_TEX_FILTER_LINEAR; + } + /* Fwd, bkwd ref filters */ + filters[3] = PIPE_TEX_FILTER_LINEAR; + filters[4] = PIPE_TEX_FILTER_LINEAR; + + for (i = 0; i < 5; ++i) { + sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; + sampler.min_img_filter = filters[i]; + sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; + sampler.mag_img_filter = filters[i]; + sampler.compare_mode = PIPE_TEX_COMPARE_NONE; + sampler.compare_func = PIPE_FUNC_ALWAYS; + sampler.normalized_coords = 1; + /*sampler.prefilter = ; */ + /*sampler.shadow_ambient = ; */ + /*sampler.lod_bias = ; */ + sampler.min_lod = 0; + /*sampler.max_lod = ; */ + /*sampler.border_color[i] = ; */ + /*sampler.max_anisotropy = ; */ + r->samplers.all[i] = r->pipe->create_sampler_state(r->pipe, &sampler); + } + + return true; +} + +static void +cleanup_pipe_state(struct vl_mpeg12_mc_renderer *r) +{ + unsigned i; + + assert(r); + + for (i = 0; i < 5; ++i) + r->pipe->delete_sampler_state(r->pipe, r->samplers.all[i]); +} + +static bool +init_shaders(struct vl_mpeg12_mc_renderer *r) +{ + assert(r); + + create_intra_vert_shader(r); + create_intra_frag_shader(r); + create_frame_pred_vert_shader(r); + create_frame_pred_frag_shader(r); + create_frame_bi_pred_vert_shader(r); + create_frame_bi_pred_frag_shader(r); + + return true; +} + +static void +cleanup_shaders(struct vl_mpeg12_mc_renderer *r) +{ + assert(r); + + r->pipe->delete_vs_state(r->pipe, r->i_vs); + r->pipe->delete_fs_state(r->pipe, r->i_fs); + r->pipe->delete_vs_state(r->pipe, r->p_vs[0]); + r->pipe->delete_fs_state(r->pipe, r->p_fs[0]); + r->pipe->delete_vs_state(r->pipe, r->b_vs[0]); + r->pipe->delete_fs_state(r->pipe, r->b_fs[0]); +} + +static bool +init_buffers(struct vl_mpeg12_mc_renderer *r) +{ + struct pipe_texture template; + + const unsigned mbw = + align(r->picture_width, MACROBLOCK_WIDTH) / MACROBLOCK_WIDTH; + const unsigned mbh = + align(r->picture_height, MACROBLOCK_HEIGHT) / MACROBLOCK_HEIGHT; + + unsigned i; + + assert(r); + + r->macroblocks_per_batch = + mbw * (r->bufmode == VL_MPEG12_MC_RENDERER_BUFFER_PICTURE ? mbh : 1); + r->num_macroblocks = 0; + r->macroblock_buf = MALLOC(r->macroblocks_per_batch * sizeof(struct pipe_mpeg12_macroblock)); + + memset(&template, 0, sizeof(struct pipe_texture)); + template.target = PIPE_TEXTURE_2D; + /* TODO: Accomodate HW that can't do this and also for cases when this isn't precise enough */ + template.format = PIPE_FORMAT_R16_SNORM; + template.last_level = 0; + template.width[0] = r->pot_buffers ? + util_next_power_of_two(r->picture_width) : r->picture_width; + template.height[0] = r->pot_buffers ? + util_next_power_of_two(r->picture_height) : r->picture_height; + template.depth[0] = 1; + pf_get_block(template.format, &template.block); + template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_DYNAMIC; + + r->textures.individual.y = r->pipe->screen->texture_create(r->pipe->screen, &template); + + if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420) { + template.width[0] = r->pot_buffers ? + util_next_power_of_two(r->picture_width / 2) : + r->picture_width / 2; + template.height[0] = r->pot_buffers ? + util_next_power_of_two(r->picture_height / 2) : + r->picture_height / 2; + } + else if (r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_422) + template.height[0] = r->pot_buffers ? + util_next_power_of_two(r->picture_height / 2) : + r->picture_height / 2; + + r->textures.individual.cb = + r->pipe->screen->texture_create(r->pipe->screen, &template); + r->textures.individual.cr = + r->pipe->screen->texture_create(r->pipe->screen, &template); + + r->vertex_bufs.individual.ycbcr.stride = sizeof(struct vertex2f) * 4; + r->vertex_bufs.individual.ycbcr.max_index = 24 * r->macroblocks_per_batch - 1; + r->vertex_bufs.individual.ycbcr.buffer_offset = 0; + r->vertex_bufs.individual.ycbcr.buffer = pipe_buffer_create + ( + r->pipe->screen, + DEFAULT_BUF_ALIGNMENT, + PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_DISCARD, + sizeof(struct vertex2f) * 4 * 24 * r->macroblocks_per_batch + ); + + for (i = 1; i < 3; ++i) { + r->vertex_bufs.all[i].stride = sizeof(struct vertex2f) * 2; + r->vertex_bufs.all[i].max_index = 24 * r->macroblocks_per_batch - 1; + r->vertex_bufs.all[i].buffer_offset = 0; + r->vertex_bufs.all[i].buffer = pipe_buffer_create + ( + r->pipe->screen, + DEFAULT_BUF_ALIGNMENT, + PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_DISCARD, + sizeof(struct vertex2f) * 2 * 24 * r->macroblocks_per_batch + ); + } + + /* Position element */ + r->vertex_elems[0].src_offset = 0; + r->vertex_elems[0].vertex_buffer_index = 0; + r->vertex_elems[0].nr_components = 2; + r->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Luma, texcoord element */ + r->vertex_elems[1].src_offset = sizeof(struct vertex2f); + r->vertex_elems[1].vertex_buffer_index = 0; + r->vertex_elems[1].nr_components = 2; + r->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Chroma Cr texcoord element */ + r->vertex_elems[2].src_offset = sizeof(struct vertex2f) * 2; + r->vertex_elems[2].vertex_buffer_index = 0; + r->vertex_elems[2].nr_components = 2; + r->vertex_elems[2].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Chroma Cb texcoord element */ + r->vertex_elems[3].src_offset = sizeof(struct vertex2f) * 3; + r->vertex_elems[3].vertex_buffer_index = 0; + r->vertex_elems[3].nr_components = 2; + r->vertex_elems[3].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* First ref surface top field texcoord element */ + r->vertex_elems[4].src_offset = 0; + r->vertex_elems[4].vertex_buffer_index = 1; + r->vertex_elems[4].nr_components = 2; + r->vertex_elems[4].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* First ref surface bottom field texcoord element */ + r->vertex_elems[5].src_offset = sizeof(struct vertex2f); + r->vertex_elems[5].vertex_buffer_index = 1; + r->vertex_elems[5].nr_components = 2; + r->vertex_elems[5].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Second ref surface top field texcoord element */ + r->vertex_elems[6].src_offset = 0; + r->vertex_elems[6].vertex_buffer_index = 2; + r->vertex_elems[6].nr_components = 2; + r->vertex_elems[6].src_format = PIPE_FORMAT_R32G32_FLOAT; + + /* Second ref surface bottom field texcoord element */ + r->vertex_elems[7].src_offset = sizeof(struct vertex2f); + r->vertex_elems[7].vertex_buffer_index = 2; + r->vertex_elems[7].nr_components = 2; + r->vertex_elems[7].src_format = PIPE_FORMAT_R32G32_FLOAT; + + r->vs_const_buf.buffer = pipe_buffer_create + ( + r->pipe->screen, + DEFAULT_BUF_ALIGNMENT, + PIPE_BUFFER_USAGE_CONSTANT | PIPE_BUFFER_USAGE_DISCARD, + sizeof(struct vertex_shader_consts) + ); + + r->fs_const_buf.buffer = pipe_buffer_create + ( + r->pipe->screen, + DEFAULT_BUF_ALIGNMENT, + PIPE_BUFFER_USAGE_CONSTANT, sizeof(struct fragment_shader_consts) + ); + + memcpy + ( + pipe_buffer_map(r->pipe->screen, r->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), + &fs_consts, sizeof(struct fragment_shader_consts) + ); + + pipe_buffer_unmap(r->pipe->screen, r->fs_const_buf.buffer); + + return true; +} + +static void +cleanup_buffers(struct vl_mpeg12_mc_renderer *r) +{ + unsigned i; + + assert(r); + + pipe_buffer_reference(&r->vs_const_buf.buffer, NULL); + pipe_buffer_reference(&r->fs_const_buf.buffer, NULL); + + for (i = 0; i < 3; ++i) + pipe_buffer_reference(&r->vertex_bufs.all[i].buffer, NULL); + + for (i = 0; i < 3; ++i) + pipe_texture_reference(&r->textures.all[i], NULL); + + FREE(r->macroblock_buf); +} + +static enum MACROBLOCK_TYPE +get_macroblock_type(struct pipe_mpeg12_macroblock *mb) +{ + assert(mb); + + switch (mb->mb_type) { + case PIPE_MPEG12_MACROBLOCK_TYPE_INTRA: + return MACROBLOCK_TYPE_INTRA; + case PIPE_MPEG12_MACROBLOCK_TYPE_FWD: + return mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME ? + MACROBLOCK_TYPE_FWD_FRAME_PRED : MACROBLOCK_TYPE_FWD_FIELD_PRED; + case PIPE_MPEG12_MACROBLOCK_TYPE_BKWD: + return mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME ? + MACROBLOCK_TYPE_BKWD_FRAME_PRED : MACROBLOCK_TYPE_BKWD_FIELD_PRED; + case PIPE_MPEG12_MACROBLOCK_TYPE_BI: + return mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME ? + MACROBLOCK_TYPE_BI_FRAME_PRED : MACROBLOCK_TYPE_BI_FIELD_PRED; + default: + assert(0); + } + + /* Unreachable */ + return -1; +} + +/* XXX: One of these days this will have to be killed with fire */ +#define SET_BLOCK(vb, cbp, mbx, mby, unitx, unity, ofsx, ofsy, hx, hy, lm, cbm, crm, use_zb, zb) \ + do { \ + (vb)[0].pos.x = (mbx) * (unitx) + (ofsx); (vb)[0].pos.y = (mby) * (unity) + (ofsy); \ + (vb)[1].pos.x = (mbx) * (unitx) + (ofsx); (vb)[1].pos.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[2].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].pos.y = (mby) * (unity) + (ofsy); \ + (vb)[3].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].pos.y = (mby) * (unity) + (ofsy); \ + (vb)[4].pos.x = (mbx) * (unitx) + (ofsx); (vb)[4].pos.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[5].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].pos.y = (mby) * (unity) + (ofsy) + (hy); \ + \ + if (!use_zb || (cbp) & (lm)) \ + { \ + (vb)[0].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].luma_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[1].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[2].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].luma_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[3].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].luma_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[4].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[5].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + } \ + else \ + { \ + (vb)[0].luma_tc.x = (zb)[0].x; (vb)[0].luma_tc.y = (zb)[0].y; \ + (vb)[1].luma_tc.x = (zb)[0].x; (vb)[1].luma_tc.y = (zb)[0].y + (hy); \ + (vb)[2].luma_tc.x = (zb)[0].x + (hx); (vb)[2].luma_tc.y = (zb)[0].y; \ + (vb)[3].luma_tc.x = (zb)[0].x + (hx); (vb)[3].luma_tc.y = (zb)[0].y; \ + (vb)[4].luma_tc.x = (zb)[0].x; (vb)[4].luma_tc.y = (zb)[0].y + (hy); \ + (vb)[5].luma_tc.x = (zb)[0].x + (hx); (vb)[5].luma_tc.y = (zb)[0].y + (hy); \ + } \ + \ + if (!use_zb || (cbp) & (cbm)) \ + { \ + (vb)[0].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].cb_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[1].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[2].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].cb_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[3].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].cb_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[4].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[5].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + } \ + else \ + { \ + (vb)[0].cb_tc.x = (zb)[1].x; (vb)[0].cb_tc.y = (zb)[1].y; \ + (vb)[1].cb_tc.x = (zb)[1].x; (vb)[1].cb_tc.y = (zb)[1].y + (hy); \ + (vb)[2].cb_tc.x = (zb)[1].x + (hx); (vb)[2].cb_tc.y = (zb)[1].y; \ + (vb)[3].cb_tc.x = (zb)[1].x + (hx); (vb)[3].cb_tc.y = (zb)[1].y; \ + (vb)[4].cb_tc.x = (zb)[1].x; (vb)[4].cb_tc.y = (zb)[1].y + (hy); \ + (vb)[5].cb_tc.x = (zb)[1].x + (hx); (vb)[5].cb_tc.y = (zb)[1].y + (hy); \ + } \ + \ + if (!use_zb || (cbp) & (crm)) \ + { \ + (vb)[0].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].cr_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[1].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[2].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].cr_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[3].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].cr_tc.y = (mby) * (unity) + (ofsy); \ + (vb)[4].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + (vb)[5].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ + } \ + else \ + { \ + (vb)[0].cr_tc.x = (zb)[2].x; (vb)[0].cr_tc.y = (zb)[2].y; \ + (vb)[1].cr_tc.x = (zb)[2].x; (vb)[1].cr_tc.y = (zb)[2].y + (hy); \ + (vb)[2].cr_tc.x = (zb)[2].x + (hx); (vb)[2].cr_tc.y = (zb)[2].y; \ + (vb)[3].cr_tc.x = (zb)[2].x + (hx); (vb)[3].cr_tc.y = (zb)[2].y; \ + (vb)[4].cr_tc.x = (zb)[2].x; (vb)[4].cr_tc.y = (zb)[2].y + (hy); \ + (vb)[5].cr_tc.x = (zb)[2].x + (hx); (vb)[5].cr_tc.y = (zb)[2].y + (hy); \ + } \ + } while (0) + +static void +gen_macroblock_verts(struct vl_mpeg12_mc_renderer *r, + struct pipe_mpeg12_macroblock *mb, unsigned pos, + struct vert_stream_0 *ycbcr_vb, struct vertex2f **ref_vb) +{ + struct vertex2f mo_vec[2]; + + unsigned i; + + assert(r); + assert(mb); + assert(ycbcr_vb); + assert(pos < r->macroblocks_per_batch); + + switch (mb->mb_type) { + case PIPE_MPEG12_MACROBLOCK_TYPE_BI: + { + struct vertex2f *vb; + + assert(ref_vb && ref_vb[1]); + + vb = ref_vb[1] + pos * 2 * 24; + + mo_vec[0].x = mb->pmv[0][1][0] * 0.5f * r->surface_tex_inv_size.x; + mo_vec[0].y = mb->pmv[0][1][1] * 0.5f * r->surface_tex_inv_size.y; + + if (mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FRAME) { + for (i = 0; i < 24 * 2; i += 2) { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + } + } + else { + mo_vec[1].x = mb->pmv[1][1][0] * 0.5f * r->surface_tex_inv_size.x; + mo_vec[1].y = mb->pmv[1][1][1] * 0.5f * r->surface_tex_inv_size.y; + + for (i = 0; i < 24 * 2; i += 2) { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + vb[i + 1].x = mo_vec[1].x; + vb[i + 1].y = mo_vec[1].y; + } + } + + /* fall-through */ + } + case PIPE_MPEG12_MACROBLOCK_TYPE_FWD: + case PIPE_MPEG12_MACROBLOCK_TYPE_BKWD: + { + struct vertex2f *vb; + + assert(ref_vb && ref_vb[0]); + + vb = ref_vb[0] + pos * 2 * 24; + + if (mb->mb_type == PIPE_MPEG12_MACROBLOCK_TYPE_BKWD) { + mo_vec[0].x = mb->pmv[0][1][0] * 0.5f * r->surface_tex_inv_size.x; + mo_vec[0].y = mb->pmv[0][1][1] * 0.5f * r->surface_tex_inv_size.y; + + if (mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FIELD) { + mo_vec[1].x = mb->pmv[1][1][0] * 0.5f * r->surface_tex_inv_size.x; + mo_vec[1].y = mb->pmv[1][1][1] * 0.5f * r->surface_tex_inv_size.y; + } + } + else { + mo_vec[0].x = mb->pmv[0][0][0] * 0.5f * r->surface_tex_inv_size.x; + mo_vec[0].y = mb->pmv[0][0][1] * 0.5f * r->surface_tex_inv_size.y; + + if (mb->mo_type == PIPE_MPEG12_MOTION_TYPE_FIELD) { + mo_vec[1].x = mb->pmv[1][0][0] * 0.5f * r->surface_tex_inv_size.x; + mo_vec[1].y = mb->pmv[1][0][1] * 0.5f * r->surface_tex_inv_size.y; + } + } + + if (mb->mb_type == PIPE_MPEG12_MOTION_TYPE_FRAME) { + for (i = 0; i < 24 * 2; i += 2) { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + } + } + else { + for (i = 0; i < 24 * 2; i += 2) { + vb[i].x = mo_vec[0].x; + vb[i].y = mo_vec[0].y; + vb[i + 1].x = mo_vec[1].x; + vb[i + 1].y = mo_vec[1].y; + } + } + + /* fall-through */ + } + case PIPE_MPEG12_MACROBLOCK_TYPE_INTRA: + { + const struct vertex2f unit = + { + r->surface_tex_inv_size.x * MACROBLOCK_WIDTH, + r->surface_tex_inv_size.y * MACROBLOCK_HEIGHT + }; + const struct vertex2f half = + { + r->surface_tex_inv_size.x * (MACROBLOCK_WIDTH / 2), + r->surface_tex_inv_size.y * (MACROBLOCK_HEIGHT / 2) + }; + const bool use_zb = r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE; + + struct vert_stream_0 *vb = ycbcr_vb + pos * 24; + + SET_BLOCK(vb, mb->cbp, mb->mbx, mb->mby, + unit.x, unit.y, 0, 0, half.x, half.y, + 32, 2, 1, use_zb, r->zero_block); + + SET_BLOCK(vb + 6, mb->cbp, mb->mbx, mb->mby, + unit.x, unit.y, half.x, 0, half.x, half.y, + 16, 2, 1, use_zb, r->zero_block); + + SET_BLOCK(vb + 12, mb->cbp, mb->mbx, mb->mby, + unit.x, unit.y, 0, half.y, half.x, half.y, + 8, 2, 1, use_zb, r->zero_block); + + SET_BLOCK(vb + 18, mb->cbp, mb->mbx, mb->mby, + unit.x, unit.y, half.x, half.y, half.x, half.y, + 4, 2, 1, use_zb, r->zero_block); + + break; + } + default: + assert(0); + } +} + +static void +gen_macroblock_stream(struct vl_mpeg12_mc_renderer *r, + unsigned *num_macroblocks) +{ + unsigned offset[NUM_MACROBLOCK_TYPES]; + struct vert_stream_0 *ycbcr_vb; + struct vertex2f *ref_vb[2]; + unsigned i; + + assert(r); + assert(num_macroblocks); + + for (i = 0; i < r->num_macroblocks; ++i) { + enum MACROBLOCK_TYPE mb_type = get_macroblock_type(&r->macroblock_buf[i]); + ++num_macroblocks[mb_type]; + } + + offset[0] = 0; + + for (i = 1; i < NUM_MACROBLOCK_TYPES; ++i) + offset[i] = offset[i - 1] + num_macroblocks[i - 1]; + + ycbcr_vb = (struct vert_stream_0 *)pipe_buffer_map + ( + r->pipe->screen, + r->vertex_bufs.individual.ycbcr.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD + ); + + for (i = 0; i < 2; ++i) + ref_vb[i] = (struct vertex2f *)pipe_buffer_map + ( + r->pipe->screen, + r->vertex_bufs.individual.ref[i].buffer, + PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD + ); + + for (i = 0; i < r->num_macroblocks; ++i) { + enum MACROBLOCK_TYPE mb_type = get_macroblock_type(&r->macroblock_buf[i]); + + gen_macroblock_verts(r, &r->macroblock_buf[i], offset[mb_type], + ycbcr_vb, ref_vb); + + ++offset[mb_type]; + } + + pipe_buffer_unmap(r->pipe->screen, r->vertex_bufs.individual.ycbcr.buffer); + for (i = 0; i < 2; ++i) + pipe_buffer_unmap(r->pipe->screen, r->vertex_bufs.individual.ref[i].buffer); +} + +static void +flush(struct vl_mpeg12_mc_renderer *r) +{ + unsigned num_macroblocks[NUM_MACROBLOCK_TYPES] = { 0 }; + unsigned vb_start = 0; + struct vertex_shader_consts *vs_consts; + unsigned i; + + assert(r); + assert(r->num_macroblocks == r->macroblocks_per_batch); + + gen_macroblock_stream(r, num_macroblocks); + + r->fb_state.cbufs[0] = r->pipe->screen->get_tex_surface + ( + r->pipe->screen, r->surface, + 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE + ); + + r->pipe->set_framebuffer_state(r->pipe, &r->fb_state); + r->pipe->set_viewport_state(r->pipe, &r->viewport); + + vs_consts = pipe_buffer_map + ( + r->pipe->screen, r->vs_const_buf.buffer, + PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD + ); + + vs_consts->denorm.x = r->surface->width[0]; + vs_consts->denorm.y = r->surface->height[0]; + + pipe_buffer_unmap(r->pipe->screen, r->vs_const_buf.buffer); + + r->pipe->set_constant_buffer(r->pipe, PIPE_SHADER_VERTEX, 0, + &r->vs_const_buf); + r->pipe->set_constant_buffer(r->pipe, PIPE_SHADER_FRAGMENT, 0, + &r->fs_const_buf); + + if (num_macroblocks[MACROBLOCK_TYPE_INTRA] > 0) { + r->pipe->set_vertex_buffers(r->pipe, 1, r->vertex_bufs.all); + r->pipe->set_vertex_elements(r->pipe, 4, r->vertex_elems); + r->pipe->set_sampler_textures(r->pipe, 3, r->textures.all); + r->pipe->bind_sampler_states(r->pipe, 3, r->samplers.all); + r->pipe->bind_vs_state(r->pipe, r->i_vs); + r->pipe->bind_fs_state(r->pipe, r->i_fs); + + r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start, + num_macroblocks[MACROBLOCK_TYPE_INTRA] * 24); + vb_start += num_macroblocks[MACROBLOCK_TYPE_INTRA] * 24; + } + + if (num_macroblocks[MACROBLOCK_TYPE_FWD_FRAME_PRED] > 0) { + r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all); + r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems); + r->textures.individual.ref[0] = r->past; + r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all); + r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all); + r->pipe->bind_vs_state(r->pipe, r->p_vs[0]); + r->pipe->bind_fs_state(r->pipe, r->p_fs[0]); + + r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start, + num_macroblocks[MACROBLOCK_TYPE_FWD_FRAME_PRED] * 24); + vb_start += num_macroblocks[MACROBLOCK_TYPE_FWD_FRAME_PRED] * 24; + } + + if (false /*num_macroblocks[MACROBLOCK_TYPE_FWD_FIELD_PRED] > 0 */ ) { + r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all); + r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems); + r->textures.individual.ref[0] = r->past; + r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all); + r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all); + r->pipe->bind_vs_state(r->pipe, r->p_vs[1]); + r->pipe->bind_fs_state(r->pipe, r->p_fs[1]); + + r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start, + num_macroblocks[MACROBLOCK_TYPE_FWD_FIELD_PRED] * 24); + vb_start += num_macroblocks[MACROBLOCK_TYPE_FWD_FIELD_PRED] * 24; + } + + if (num_macroblocks[MACROBLOCK_TYPE_BKWD_FRAME_PRED] > 0) { + r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all); + r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems); + r->textures.individual.ref[0] = r->future; + r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all); + r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all); + r->pipe->bind_vs_state(r->pipe, r->p_vs[0]); + r->pipe->bind_fs_state(r->pipe, r->p_fs[0]); + + r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start, + num_macroblocks[MACROBLOCK_TYPE_BKWD_FRAME_PRED] * 24); + vb_start += num_macroblocks[MACROBLOCK_TYPE_BKWD_FRAME_PRED] * 24; + } + + if (false /*num_macroblocks[MACROBLOCK_TYPE_BKWD_FIELD_PRED] > 0 */ ) { + r->pipe->set_vertex_buffers(r->pipe, 2, r->vertex_bufs.all); + r->pipe->set_vertex_elements(r->pipe, 6, r->vertex_elems); + r->textures.individual.ref[0] = r->future; + r->pipe->set_sampler_textures(r->pipe, 4, r->textures.all); + r->pipe->bind_sampler_states(r->pipe, 4, r->samplers.all); + r->pipe->bind_vs_state(r->pipe, r->p_vs[1]); + r->pipe->bind_fs_state(r->pipe, r->p_fs[1]); + + r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start, + num_macroblocks[MACROBLOCK_TYPE_BKWD_FIELD_PRED] * 24); + vb_start += num_macroblocks[MACROBLOCK_TYPE_BKWD_FIELD_PRED] * 24; + } + + if (num_macroblocks[MACROBLOCK_TYPE_BI_FRAME_PRED] > 0) { + r->pipe->set_vertex_buffers(r->pipe, 3, r->vertex_bufs.all); + r->pipe->set_vertex_elements(r->pipe, 8, r->vertex_elems); + r->textures.individual.ref[0] = r->past; + r->textures.individual.ref[1] = r->future; + r->pipe->set_sampler_textures(r->pipe, 5, r->textures.all); + r->pipe->bind_sampler_states(r->pipe, 5, r->samplers.all); + r->pipe->bind_vs_state(r->pipe, r->b_vs[0]); + r->pipe->bind_fs_state(r->pipe, r->b_fs[0]); + + r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start, + num_macroblocks[MACROBLOCK_TYPE_BI_FRAME_PRED] * 24); + vb_start += num_macroblocks[MACROBLOCK_TYPE_BI_FRAME_PRED] * 24; + } + + if (false /*num_macroblocks[MACROBLOCK_TYPE_BI_FIELD_PRED] > 0 */ ) { + r->pipe->set_vertex_buffers(r->pipe, 3, r->vertex_bufs.all); + r->pipe->set_vertex_elements(r->pipe, 8, r->vertex_elems); + r->textures.individual.ref[0] = r->past; + r->textures.individual.ref[1] = r->future; + r->pipe->set_sampler_textures(r->pipe, 5, r->textures.all); + r->pipe->bind_sampler_states(r->pipe, 5, r->samplers.all); + r->pipe->bind_vs_state(r->pipe, r->b_vs[1]); + r->pipe->bind_fs_state(r->pipe, r->b_fs[1]); + + r->pipe->draw_arrays(r->pipe, PIPE_PRIM_TRIANGLES, vb_start, + num_macroblocks[MACROBLOCK_TYPE_BI_FIELD_PRED] * 24); + vb_start += num_macroblocks[MACROBLOCK_TYPE_BI_FIELD_PRED] * 24; + } + + r->pipe->flush(r->pipe, PIPE_FLUSH_RENDER_CACHE, r->fence); + pipe_surface_reference(&r->fb_state.cbufs[0], NULL); + + if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE) + for (i = 0; i < 3; ++i) + r->zero_block[i].x = ZERO_BLOCK_NIL; + + r->num_macroblocks = 0; +} + +static void +grab_frame_coded_block(short *src, short *dst, unsigned dst_pitch) +{ + unsigned y; + + assert(src); + assert(dst); + + for (y = 0; y < BLOCK_HEIGHT; ++y) + memcpy(dst + y * dst_pitch, src + y * BLOCK_WIDTH, BLOCK_WIDTH * 2); +} + +static void +grab_field_coded_block(short *src, short *dst, unsigned dst_pitch) +{ + unsigned y; + + assert(src); + assert(dst); + + for (y = 0; y < BLOCK_HEIGHT; ++y) + memcpy(dst + y * dst_pitch * 2, src + y * BLOCK_WIDTH, BLOCK_WIDTH * 2); +} + +static void +fill_zero_block(short *dst, unsigned dst_pitch) +{ + unsigned y; + + assert(dst); + + for (y = 0; y < BLOCK_HEIGHT; ++y) + memset(dst + y * dst_pitch, 0, BLOCK_WIDTH * 2); +} + +static void +grab_blocks(struct vl_mpeg12_mc_renderer *r, unsigned mbx, unsigned mby, + enum pipe_mpeg12_dct_type dct_type, unsigned cbp, short *blocks) +{ + unsigned tex_pitch; + short *texels; + unsigned tb = 0, sb = 0; + unsigned mbpx = mbx * MACROBLOCK_WIDTH, mbpy = mby * MACROBLOCK_HEIGHT; + unsigned x, y; + + assert(r); + assert(blocks); + + tex_pitch = r->tex_transfer[0]->stride / r->tex_transfer[0]->block.size; + texels = r->texels[0] + mbpy * tex_pitch + mbpx; + + for (y = 0; y < 2; ++y) { + for (x = 0; x < 2; ++x, ++tb) { + if ((cbp >> (5 - tb)) & 1) { + if (dct_type == PIPE_MPEG12_DCT_TYPE_FRAME) { + grab_frame_coded_block(blocks + sb * BLOCK_WIDTH * BLOCK_HEIGHT, + texels + y * tex_pitch * BLOCK_WIDTH + + x * BLOCK_WIDTH, tex_pitch); + } + else { + grab_field_coded_block(blocks + sb * BLOCK_WIDTH * BLOCK_HEIGHT, + texels + y * tex_pitch + x * BLOCK_WIDTH, + tex_pitch); + } + + ++sb; + } + else if (r->eb_handling != VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_NONE) { + if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ALL || + ZERO_BLOCK_IS_NIL(r->zero_block[0])) { + fill_zero_block(texels + y * tex_pitch * BLOCK_WIDTH + x * BLOCK_WIDTH, tex_pitch); + if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE) { + r->zero_block[0].x = (mbpx + x * 8) * r->surface_tex_inv_size.x; + r->zero_block[0].y = (mbpy + y * 8) * r->surface_tex_inv_size.y; + } + } + } + } + } + + /* TODO: Implement 422, 444 */ + assert(r->chroma_format == PIPE_VIDEO_CHROMA_FORMAT_420); + + mbpx /= 2; + mbpy /= 2; + + for (tb = 0; tb < 2; ++tb) { + tex_pitch = r->tex_transfer[tb + 1]->stride / r->tex_transfer[tb + 1]->block.size; + texels = r->texels[tb + 1] + mbpy * tex_pitch + mbpx; + + if ((cbp >> (1 - tb)) & 1) { + grab_frame_coded_block(blocks + sb * BLOCK_WIDTH * BLOCK_HEIGHT, texels, tex_pitch); + ++sb; + } + else if (r->eb_handling != VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_NONE) { + if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ALL || + ZERO_BLOCK_IS_NIL(r->zero_block[tb + 1])) { + fill_zero_block(texels, tex_pitch); + if (r->eb_handling == VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE) { + r->zero_block[tb + 1].x = (mbpx << 1) * r->surface_tex_inv_size.x; + r->zero_block[tb + 1].y = (mbpy << 1) * r->surface_tex_inv_size.y; + } + } + } + } +} + +static void +grab_macroblock(struct vl_mpeg12_mc_renderer *r, + struct pipe_mpeg12_macroblock *mb) +{ + assert(r); + assert(mb); + assert(r->num_macroblocks < r->macroblocks_per_batch); + + memcpy(&r->macroblock_buf[r->num_macroblocks], mb, + sizeof(struct pipe_mpeg12_macroblock)); + + grab_blocks(r, mb->mbx, mb->mby, mb->dct_type, mb->cbp, mb->blocks); + + ++r->num_macroblocks; +} + +bool +vl_mpeg12_mc_renderer_init(struct vl_mpeg12_mc_renderer *renderer, + struct pipe_context *pipe, + unsigned picture_width, + unsigned picture_height, + enum pipe_video_chroma_format chroma_format, + enum VL_MPEG12_MC_RENDERER_BUFFER_MODE bufmode, + enum VL_MPEG12_MC_RENDERER_EMPTY_BLOCK eb_handling, + bool pot_buffers) +{ + unsigned i; + + assert(renderer); + assert(pipe); + /* TODO: Implement other policies */ + assert(bufmode == VL_MPEG12_MC_RENDERER_BUFFER_PICTURE); + /* TODO: Implement this */ + /* XXX: XFER_ALL sampling issue at block edges when using bilinear filtering */ + assert(eb_handling != VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_NONE); + /* TODO: Non-pot buffers untested, probably doesn't work without changes to texcoord generation, vert shader, etc */ + assert(pot_buffers); + + memset(renderer, 0, sizeof(struct vl_mpeg12_mc_renderer)); + + renderer->pipe = pipe; + renderer->picture_width = picture_width; + renderer->picture_height = picture_height; + renderer->chroma_format = chroma_format; + renderer->bufmode = bufmode; + renderer->eb_handling = eb_handling; + renderer->pot_buffers = pot_buffers; + + if (!init_pipe_state(renderer)) + return false; + if (!init_shaders(renderer)) { + cleanup_pipe_state(renderer); + return false; + } + if (!init_buffers(renderer)) { + cleanup_shaders(renderer); + cleanup_pipe_state(renderer); + return false; + } + + renderer->surface = NULL; + renderer->past = NULL; + renderer->future = NULL; + for (i = 0; i < 3; ++i) + renderer->zero_block[i].x = ZERO_BLOCK_NIL; + renderer->num_macroblocks = 0; + + xfer_buffers_map(renderer); + + return true; +} + +void +vl_mpeg12_mc_renderer_cleanup(struct vl_mpeg12_mc_renderer *renderer) +{ + assert(renderer); + + xfer_buffers_unmap(renderer); + + cleanup_pipe_state(renderer); + cleanup_shaders(renderer); + cleanup_buffers(renderer); +} + +void +vl_mpeg12_mc_renderer_render_macroblocks(struct vl_mpeg12_mc_renderer + *renderer, + struct pipe_texture *surface, + struct pipe_texture *past, + struct pipe_texture *future, + unsigned num_macroblocks, + struct pipe_mpeg12_macroblock + *mpeg12_macroblocks, + struct pipe_fence_handle **fence) +{ + bool new_surface = false; + + assert(renderer); + assert(surface); + assert(num_macroblocks); + assert(mpeg12_macroblocks); + + if (renderer->surface) { + if (surface != renderer->surface) { + if (renderer->num_macroblocks > 0) { + xfer_buffers_unmap(renderer); + flush(renderer); + } + + new_surface = true; + } + + /* If the surface we're rendering hasn't changed the ref frames shouldn't change. */ + assert(surface != renderer->surface || renderer->past == past); + assert(surface != renderer->surface || renderer->future == future); + } + else + new_surface = true; + + if (new_surface) { + renderer->surface = surface; + renderer->past = past; + renderer->future = future; + renderer->fence = fence; + renderer->surface_tex_inv_size.x = 1.0f / surface->width[0]; + renderer->surface_tex_inv_size.y = 1.0f / surface->height[0]; + } + + while (num_macroblocks) { + unsigned left_in_batch = renderer->macroblocks_per_batch - renderer->num_macroblocks; + unsigned num_to_submit = MIN2(num_macroblocks, left_in_batch); + unsigned i; + + for (i = 0; i < num_to_submit; ++i) { + assert(mpeg12_macroblocks[i].base.codec == PIPE_VIDEO_CODEC_MPEG12); + grab_macroblock(renderer, &mpeg12_macroblocks[i]); + } + + num_macroblocks -= num_to_submit; + + if (renderer->num_macroblocks == renderer->macroblocks_per_batch) { + xfer_buffers_unmap(renderer); + flush(renderer); + xfer_buffers_map(renderer); + /* Next time we get this surface it may have new ref frames */ + renderer->surface = NULL; + } + } +} diff --git a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.h b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.h new file mode 100644 index 00000000000..5d2c1273ee3 --- /dev/null +++ b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.h @@ -0,0 +1,120 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef vl_mpeg12_mc_renderer_h +#define vl_mpeg12_mc_renderer_h + +#include <pipe/p_compiler.h> +#include <pipe/p_state.h> +#include <pipe/p_video_state.h> + +struct pipe_context; +struct pipe_video_surface; +struct pipe_macroblock; + +/* A slice is video-width (rounded up to a multiple of macroblock width) x macroblock height */ +enum VL_MPEG12_MC_RENDERER_BUFFER_MODE +{ + VL_MPEG12_MC_RENDERER_BUFFER_SLICE, /* Saves memory at the cost of smaller batches */ + VL_MPEG12_MC_RENDERER_BUFFER_PICTURE /* Larger batches, more memory */ +}; + +enum VL_MPEG12_MC_RENDERER_EMPTY_BLOCK +{ + VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ALL, /* Waste of memory bandwidth */ + VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE, /* Can only do point-filtering when interpolating subsampled chroma channels */ + VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_NONE /* Needs conditional texel fetch! */ +}; + +struct vl_mpeg12_mc_renderer +{ + struct pipe_context *pipe; + unsigned picture_width; + unsigned picture_height; + enum pipe_video_chroma_format chroma_format; + enum VL_MPEG12_MC_RENDERER_BUFFER_MODE bufmode; + enum VL_MPEG12_MC_RENDERER_EMPTY_BLOCK eb_handling; + bool pot_buffers; + unsigned macroblocks_per_batch; + + struct pipe_viewport_state viewport; + struct pipe_constant_buffer vs_const_buf; + struct pipe_constant_buffer fs_const_buf; + struct pipe_framebuffer_state fb_state; + struct pipe_vertex_element vertex_elems[8]; + + union + { + void *all[5]; + struct { void *y, *cb, *cr, *ref[2]; } individual; + } samplers; + + void *i_vs, *p_vs[2], *b_vs[2]; + void *i_fs, *p_fs[2], *b_fs[2]; + + union + { + struct pipe_texture *all[5]; + struct { struct pipe_texture *y, *cb, *cr, *ref[2]; } individual; + } textures; + + union + { + struct pipe_vertex_buffer all[3]; + struct { struct pipe_vertex_buffer ycbcr, ref[2]; } individual; + } vertex_bufs; + + struct pipe_texture *surface, *past, *future; + struct pipe_fence_handle **fence; + unsigned num_macroblocks; + struct pipe_mpeg12_macroblock *macroblock_buf; + struct pipe_transfer *tex_transfer[3]; + short *texels[3]; + struct { float x, y; } surface_tex_inv_size; + struct { float x, y; } zero_block[3]; +}; + +bool vl_mpeg12_mc_renderer_init(struct vl_mpeg12_mc_renderer *renderer, + struct pipe_context *pipe, + unsigned picture_width, + unsigned picture_height, + enum pipe_video_chroma_format chroma_format, + enum VL_MPEG12_MC_RENDERER_BUFFER_MODE bufmode, + enum VL_MPEG12_MC_RENDERER_EMPTY_BLOCK eb_handling, + bool pot_buffers); + +void vl_mpeg12_mc_renderer_cleanup(struct vl_mpeg12_mc_renderer *renderer); + +void vl_mpeg12_mc_renderer_render_macroblocks(struct vl_mpeg12_mc_renderer *renderer, + struct pipe_texture *surface, + struct pipe_texture *past, + struct pipe_texture *future, + unsigned num_macroblocks, + struct pipe_mpeg12_macroblock *mpeg12_macroblocks, + struct pipe_fence_handle **fence); + +#endif /* vl_mpeg12_mc_renderer_h */ diff --git a/src/gallium/auxiliary/vl/vl_shader_build.c b/src/gallium/auxiliary/vl/vl_shader_build.c new file mode 100644 index 00000000000..faa20a903cd --- /dev/null +++ b/src/gallium/auxiliary/vl/vl_shader_build.c @@ -0,0 +1,242 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "vl_shader_build.h" +#include <assert.h> +#include <tgsi/tgsi_parse.h> +#include <tgsi/tgsi_build.h> + +struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = name; + decl.Semantic.SemanticIndex = index; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_interpolated_input +( + unsigned int name, + unsigned int index, + unsigned int first, + unsigned int last, + int interpolation +) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + assert + ( + interpolation == TGSI_INTERPOLATE_CONSTANT || + interpolation == TGSI_INTERPOLATE_LINEAR || + interpolation == TGSI_INTERPOLATE_PERSPECTIVE + ); + + decl.Declaration.File = TGSI_FILE_INPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = name; + decl.Semantic.SemanticIndex = index; + decl.Declaration.Interpolate = interpolation;; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl.Declaration.File = TGSI_FILE_CONSTANT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = name; + decl.Semantic.SemanticIndex = index; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl.Declaration.File = TGSI_FILE_OUTPUT; + decl.Declaration.Semantic = 1; + decl.Semantic.SemanticName = name; + decl.Semantic.SemanticIndex = index; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_TEMPORARY; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last) +{ + struct tgsi_full_declaration decl = tgsi_default_full_declaration(); + + decl = tgsi_default_full_declaration(); + decl.Declaration.File = TGSI_FILE_SAMPLER; + decl.DeclarationRange.First = first; + decl.DeclarationRange.Last = last; + + return decl; +} + +struct tgsi_full_instruction vl_inst2 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src_file, + unsigned int src_index +) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = opcode; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = dst_file; + inst.FullDstRegisters[0].DstRegister.Index = dst_index; + inst.Instruction.NumSrcRegs = 1; + inst.FullSrcRegisters[0].SrcRegister.File = src_file; + inst.FullSrcRegisters[0].SrcRegister.Index = src_index; + + return inst; +} + +struct tgsi_full_instruction vl_inst3 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index +) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = opcode; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = dst_file; + inst.FullDstRegisters[0].DstRegister.Index = dst_index; + inst.Instruction.NumSrcRegs = 2; + inst.FullSrcRegisters[0].SrcRegister.File = src1_file; + inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; + inst.FullSrcRegisters[1].SrcRegister.File = src2_file; + inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; + + return inst; +} + +struct tgsi_full_instruction vl_tex +( + int tex, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index +) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = TGSI_OPCODE_TEX; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = dst_file; + inst.FullDstRegisters[0].DstRegister.Index = dst_index; + inst.Instruction.NumSrcRegs = 2; + inst.InstructionExtTexture.Texture = tex; + inst.FullSrcRegisters[0].SrcRegister.File = src1_file; + inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; + inst.FullSrcRegisters[1].SrcRegister.File = src2_file; + inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; + + return inst; +} + +struct tgsi_full_instruction vl_inst4 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index, + enum tgsi_file_type src3_file, + unsigned int src3_index +) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = opcode; + inst.Instruction.NumDstRegs = 1; + inst.FullDstRegisters[0].DstRegister.File = dst_file; + inst.FullDstRegisters[0].DstRegister.Index = dst_index; + inst.Instruction.NumSrcRegs = 3; + inst.FullSrcRegisters[0].SrcRegister.File = src1_file; + inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; + inst.FullSrcRegisters[1].SrcRegister.File = src2_file; + inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; + inst.FullSrcRegisters[2].SrcRegister.File = src3_file; + inst.FullSrcRegisters[2].SrcRegister.Index = src3_index; + + return inst; +} + +struct tgsi_full_instruction vl_end(void) +{ + struct tgsi_full_instruction inst = tgsi_default_full_instruction(); + + inst.Instruction.Opcode = TGSI_OPCODE_END; + inst.Instruction.NumDstRegs = 0; + inst.Instruction.NumSrcRegs = 0; + + return inst; +} diff --git a/src/gallium/auxiliary/vl/vl_shader_build.h b/src/gallium/auxiliary/vl/vl_shader_build.h new file mode 100644 index 00000000000..5da71f8e136 --- /dev/null +++ b/src/gallium/auxiliary/vl/vl_shader_build.h @@ -0,0 +1,88 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef vl_shader_build_h +#define vl_shader_build_h + +#include <pipe/p_shader_tokens.h> + +struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last); +struct tgsi_full_declaration vl_decl_interpolated_input +( + unsigned int name, + unsigned int index, + unsigned int first, + unsigned int last, + int interpolation +); +struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last); +struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last); +struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last); +struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last); +struct tgsi_full_instruction vl_inst2 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src_file, + unsigned int src_index +); +struct tgsi_full_instruction vl_inst3 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index +); +struct tgsi_full_instruction vl_tex +( + int tex, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index +); +struct tgsi_full_instruction vl_inst4 +( + int opcode, + enum tgsi_file_type dst_file, + unsigned int dst_index, + enum tgsi_file_type src1_file, + unsigned int src1_index, + enum tgsi_file_type src2_file, + unsigned int src2_index, + enum tgsi_file_type src3_file, + unsigned int src3_index +); +struct tgsi_full_instruction vl_end(void); + +#endif diff --git a/src/gallium/drivers/cell/ppu/cell_screen.c b/src/gallium/drivers/cell/ppu/cell_screen.c index 9161747fdbf..d185c6b8497 100644 --- a/src/gallium/drivers/cell/ppu/cell_screen.c +++ b/src/gallium/drivers/cell/ppu/cell_screen.c @@ -41,7 +41,7 @@ static const char * cell_get_vendor(struct pipe_screen *screen) { - return "Tungsten Graphics, Inc."; + return "VMware, Inc."; } diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c index 6a63a0e6ced..ae4c61efb3b 100644 --- a/src/gallium/drivers/cell/ppu/cell_texture.c +++ b/src/gallium/drivers/cell/ppu/cell_texture.c @@ -389,22 +389,14 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer) const uint texWidth = pt->width[level]; const uint texHeight = pt->height[level]; const uint stride = ct->stride[level]; - unsigned flags = 0x0; unsigned size; assert(transfer->texture); - if (transfer->usage != PIPE_TRANSFER_READ) { - flags |= PIPE_BUFFER_USAGE_CPU_WRITE; - } - - if (transfer->usage != PIPE_TRANSFER_WRITE) { - flags |= PIPE_BUFFER_USAGE_CPU_READ; - } - if (!ct->mapped) { /* map now */ - ct->mapped = pipe_buffer_map(screen, ct->buffer, flags); + ct->mapped = pipe_buffer_map(screen, ct->buffer, + pipe_transfer_buffer_flags(transfer)); } /* @@ -417,8 +409,7 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer) if (!ctrans->map) return NULL; /* out of memory */ - if (transfer->usage == PIPE_TRANSFER_READ || - transfer->usage == PIPE_TRANSFER_READ_WRITE) { + if (transfer->usage & PIPE_TRANSFER_READ) { /* need to untwiddle the texture to make a linear version */ const uint bpp = pf_get_size(ct->base.format); if (bpp == 4) { @@ -459,8 +450,7 @@ cell_transfer_unmap(struct pipe_screen *screen, PIPE_BUFFER_USAGE_CPU_READ); } - if (transfer->usage == PIPE_TRANSFER_WRITE || - transfer->usage == PIPE_TRANSFER_READ_WRITE) { + if (transfer->usage & PIPE_TRANSFER_WRITE) { /* The user wrote new texture data into the mapped buffer. * We need to convert the new linear data into the twiddled/tiled format. */ diff --git a/src/gallium/drivers/i915simple/i915_context.c b/src/gallium/drivers/i915simple/i915_context.c index b43f7352456..e745f3342d1 100644 --- a/src/gallium/drivers/i915simple/i915_context.c +++ b/src/gallium/drivers/i915simple/i915_context.c @@ -175,12 +175,19 @@ i915_is_buffer_referenced(struct pipe_context *pipe, static void i915_destroy(struct pipe_context *pipe) { struct i915_context *i915 = i915_context(pipe); + int i; draw_destroy(i915->draw); if(i915->batch) i915->iws->batchbuffer_destroy(i915->batch); + /* unbind framebuffer */ + for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) { + pipe_surface_reference(&i915->framebuffer.cbufs[i], NULL); + } + pipe_surface_reference(&i915->framebuffer.zsbuf, NULL); + FREE(i915); } diff --git a/src/gallium/drivers/i915simple/i915_prim_vbuf.c b/src/gallium/drivers/i915simple/i915_prim_vbuf.c index b3a7774fd6a..8a3e466c845 100644 --- a/src/gallium/drivers/i915simple/i915_prim_vbuf.c +++ b/src/gallium/drivers/i915simple/i915_prim_vbuf.c @@ -198,7 +198,7 @@ i915_vbuf_render_map_vertices(struct vbuf_render *render) struct intel_winsys *iws = i915->iws; if (i915->vbo_flushed) - debug_printf("%s bad vbo flush occured stalling on hw\n"); + debug_printf("%s bad vbo flush occured stalling on hw\n", __FUNCTION__); i915_render->vbo_ptr = iws->buffer_map(iws, i915_render->vbo, TRUE); @@ -389,14 +389,43 @@ i915_vbuf_render_draw_arrays(struct vbuf_render *render, uint nr) { struct i915_vbuf_render *i915_render = i915_vbuf_render(render); + struct i915_context *i915 = i915_render->i915; if (i915_render->fallback) { draw_arrays_fallback(render, start, nr); return; } - /* JB: TODO submit direct cmds */ - draw_arrays_fallback(render, start, nr); + if (i915->dirty) + i915_update_derived(i915); + + if (i915->hardware_dirty) + i915_emit_hardware_state(i915); + + if (!BEGIN_BATCH(2, 0)) { + FLUSH_BATCH(NULL); + + /* Make sure state is re-emitted after a flush: + */ + i915_update_derived(i915); + i915_emit_hardware_state(i915); + i915->vbo_flushed = 1; + + if (!BEGIN_BATCH(2, 0)) { + assert(0); + goto out; + } + } + + OUT_BATCH(_3DPRIMITIVE | + PRIM_INDIRECT | + PRIM_INDIRECT_SEQUENTIAL | + i915_render->hwprim | + nr); + OUT_BATCH(start); /* Beginning vertex index */ + +out: + return; } /** diff --git a/src/gallium/drivers/i915simple/i915_screen.c b/src/gallium/drivers/i915simple/i915_screen.c index a1dd43c1bcc..c66558c320e 100644 --- a/src/gallium/drivers/i915simple/i915_screen.c +++ b/src/gallium/drivers/i915simple/i915_screen.c @@ -46,7 +46,7 @@ static const char * i915_get_vendor(struct pipe_screen *screen) { - return "Tungsten Graphics, Inc."; + return "VMware, Inc."; } static const char * diff --git a/src/gallium/drivers/i915simple/i915_state.c b/src/gallium/drivers/i915simple/i915_state.c index 0087dfa410f..7d48e6e84d5 100644 --- a/src/gallium/drivers/i915simple/i915_state.c +++ b/src/gallium/drivers/i915simple/i915_state.c @@ -588,9 +588,17 @@ static void i915_set_framebuffer_state(struct pipe_context *pipe, const struct pipe_framebuffer_state *fb) { struct i915_context *i915 = i915_context(pipe); + int i; + draw_flush(i915->draw); - i915->framebuffer = *fb; /* struct copy */ + i915->framebuffer.width = fb->width; + i915->framebuffer.height = fb->height; + i915->framebuffer.nr_cbufs = fb->nr_cbufs; + for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) { + pipe_surface_reference(&i915->framebuffer.cbufs[i], fb->cbufs[i]); + } + pipe_surface_reference(&i915->framebuffer.zsbuf, fb->zsbuf); i915->dirty |= I915_NEW_FRAMEBUFFER; } diff --git a/src/gallium/drivers/i915simple/i915_texture.c b/src/gallium/drivers/i915simple/i915_texture.c index 6a6c6542717..286c9ace8e5 100644 --- a/src/gallium/drivers/i915simple/i915_texture.c +++ b/src/gallium/drivers/i915simple/i915_texture.c @@ -165,7 +165,7 @@ i915_scanout_layout(struct i915_texture *tex) struct pipe_texture *pt = &tex->base; if (pt->last_level > 0 || pt->block.size != 4) - return 0; + return FALSE; i915_miptree_set_level_info(tex, 0, 1, tex->base.width[0], @@ -191,6 +191,38 @@ i915_scanout_layout(struct i915_texture *tex) return TRUE; } +/** + * Special case to deal with shared textures. + */ +static boolean +i915_display_target_layout(struct i915_texture *tex) +{ + struct pipe_texture *pt = &tex->base; + + if (pt->last_level > 0 || pt->block.size != 4) + return FALSE; + + /* fallback to normal textures for small textures */ + if (tex->base.width[0] < 240) + return FALSE; + + i915_miptree_set_level_info(tex, 0, 1, + tex->base.width[0], + tex->base.height[0], + 1); + i915_miptree_set_image_offset(tex, 0, 0, 0, 0); + + tex->stride = power_of_two(tex->base.nblocksx[0] * pt->block.size); + tex->total_nblocksy = round_up(tex->base.nblocksy[0], 8); + tex->hw_tiled = INTEL_TILE_X; + + debug_printf("%s size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__, + tex->base.width[0], tex->base.height[0], pt->block.size, + tex->stride, tex->total_nblocksy, tex->stride * tex->total_nblocksy); + + return TRUE; +} + static void i915_miptree_layout_2d(struct i915_texture *tex) { @@ -201,6 +233,16 @@ i915_miptree_layout_2d(struct i915_texture *tex) unsigned nblocksx = pt->nblocksx[0]; unsigned nblocksy = pt->nblocksy[0]; + /* used for scanouts that need special layouts */ + if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_PRIMARY) + if (i915_scanout_layout(tex)) + return; + + /* for shared buffers we use some very like scanout */ + if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET) + if (i915_display_target_layout(tex)) + return; + tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4); tex->total_nblocksy = 0; @@ -351,6 +393,11 @@ i945_miptree_layout_2d(struct i915_texture *tex) if (i915_scanout_layout(tex)) return; + /* for shared buffers we use some very like scanout */ + if (tex->base.tex_usage & PIPE_TEXTURE_USAGE_DISPLAY_TARGET) + if (i915_display_target_layout(tex)) + return; + tex->stride = round_up(pt->nblocksx[0] * pt->block.size, 4); /* May need to adjust pitch to accomodate the placement of @@ -812,7 +859,7 @@ i915_transfer_map(struct pipe_screen *screen, char *map; boolean write = FALSE; - if (transfer->usage != PIPE_TRANSFER_READ) + if (transfer->usage & PIPE_TRANSFER_WRITE) write = TRUE; map = iws->buffer_map(iws, tex->buffer, write); diff --git a/src/gallium/drivers/i965simple/brw_screen.c b/src/gallium/drivers/i965simple/brw_screen.c index fb68fd624b3..4a84c4db23f 100644 --- a/src/gallium/drivers/i965simple/brw_screen.c +++ b/src/gallium/drivers/i965simple/brw_screen.c @@ -39,7 +39,7 @@ static const char * brw_get_vendor( struct pipe_screen *screen ) { - return "Tungsten Graphics, Inc."; + return "VMware, Inc."; } diff --git a/src/gallium/drivers/llvmpipe/lp_bld_arit.c b/src/gallium/drivers/llvmpipe/lp_bld_arit.c index 0b115fc9b07..d27ef0de041 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_arit.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_arit.c @@ -48,11 +48,13 @@ #include "util/u_memory.h" #include "util/u_debug.h" #include "util/u_string.h" +#include "util/u_cpu_detect.h" #include "lp_bld_type.h" #include "lp_bld_const.h" #include "lp_bld_intr.h" #include "lp_bld_logic.h" +#include "lp_bld_debug.h" #include "lp_bld_arit.h" @@ -71,30 +73,28 @@ lp_build_min_simple(struct lp_build_context *bld, /* TODO: optimize the constant case */ -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) if(type.width * type.length == 128) { if(type.floating) { - if(type.width == 32) + if(type.width == 32 && util_cpu_caps.has_sse) intrinsic = "llvm.x86.sse.min.ps"; - if(type.width == 64) + if(type.width == 64 && util_cpu_caps.has_sse2) intrinsic = "llvm.x86.sse2.min.pd"; } else { - if(type.width == 8 && !type.sign) + if(type.width == 8 && !type.sign && util_cpu_caps.has_sse2) intrinsic = "llvm.x86.sse2.pminu.b"; - if(type.width == 8 && type.sign) + if(type.width == 8 && type.sign && util_cpu_caps.has_sse4_1) intrinsic = "llvm.x86.sse41.pminsb"; - if(type.width == 16 && !type.sign) + if(type.width == 16 && !type.sign && util_cpu_caps.has_sse4_1) intrinsic = "llvm.x86.sse41.pminuw"; - if(type.width == 16 && type.sign) + if(type.width == 16 && type.sign && util_cpu_caps.has_sse2) intrinsic = "llvm.x86.sse2.pmins.w"; - if(type.width == 32 && !type.sign) + if(type.width == 32 && !type.sign && util_cpu_caps.has_sse4_1) intrinsic = "llvm.x86.sse41.pminud"; - if(type.width == 32 && type.sign) + if(type.width == 32 && type.sign && util_cpu_caps.has_sse4_1) intrinsic = "llvm.x86.sse41.pminsd"; } } -#endif if(intrinsic) return lp_build_intrinsic_binary(bld->builder, intrinsic, lp_build_vec_type(bld->type), a, b); @@ -119,30 +119,28 @@ lp_build_max_simple(struct lp_build_context *bld, /* TODO: optimize the constant case */ -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) if(type.width * type.length == 128) { if(type.floating) { - if(type.width == 32) + if(type.width == 32 && util_cpu_caps.has_sse) intrinsic = "llvm.x86.sse.max.ps"; - if(type.width == 64) + if(type.width == 64 && util_cpu_caps.has_sse2) intrinsic = "llvm.x86.sse2.max.pd"; } else { - if(type.width == 8 && !type.sign) + if(type.width == 8 && !type.sign && util_cpu_caps.has_sse2) intrinsic = "llvm.x86.sse2.pmaxu.b"; - if(type.width == 8 && type.sign) + if(type.width == 8 && type.sign && util_cpu_caps.has_sse4_1) intrinsic = "llvm.x86.sse41.pmaxsb"; - if(type.width == 16 && !type.sign) + if(type.width == 16 && !type.sign && util_cpu_caps.has_sse4_1) intrinsic = "llvm.x86.sse41.pmaxuw"; - if(type.width == 16 && type.sign) + if(type.width == 16 && type.sign && util_cpu_caps.has_sse2) intrinsic = "llvm.x86.sse2.pmaxs.w"; - if(type.width == 32 && !type.sign) + if(type.width == 32 && !type.sign && util_cpu_caps.has_sse4_1) intrinsic = "llvm.x86.sse41.pmaxud"; - if(type.width == 32 && type.sign) + if(type.width == 32 && type.sign && util_cpu_caps.has_sse4_1) intrinsic = "llvm.x86.sse41.pmaxsd"; } } -#endif if(intrinsic) return lp_build_intrinsic_binary(bld->builder, intrinsic, lp_build_vec_type(bld->type), a, b); @@ -204,15 +202,14 @@ lp_build_add(struct lp_build_context *bld, if(a == bld->one || b == bld->one) return bld->one; -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - if(type.width * type.length == 128 && + if(util_cpu_caps.has_sse2 && + type.width * type.length == 128 && !type.floating && !type.fixed) { if(type.width == 8) intrinsic = type.sign ? "llvm.x86.sse2.padds.b" : "llvm.x86.sse2.paddus.b"; if(type.width == 16) intrinsic = type.sign ? "llvm.x86.sse2.padds.w" : "llvm.x86.sse2.paddus.w"; } -#endif if(intrinsic) return lp_build_intrinsic_binary(bld->builder, intrinsic, lp_build_vec_type(bld->type), a, b); @@ -257,15 +254,14 @@ lp_build_sub(struct lp_build_context *bld, if(b == bld->one) return bld->zero; -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - if(type.width * type.length == 128 && + if(util_cpu_caps.has_sse2 && + type.width * type.length == 128 && !type.floating && !type.fixed) { if(type.width == 8) intrinsic = type.sign ? "llvm.x86.sse2.psubs.b" : "llvm.x86.sse2.psubus.b"; if(type.width == 16) intrinsic = type.sign ? "llvm.x86.sse2.psubs.w" : "llvm.x86.sse2.psubus.w"; } -#endif if(intrinsic) return lp_build_intrinsic_binary(bld->builder, intrinsic, lp_build_vec_type(bld->type), a, b); @@ -419,8 +415,7 @@ lp_build_mul(struct lp_build_context *bld, return bld->undef; if(!type.floating && !type.fixed && type.norm) { -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - if(type.width == 8 && type.length == 16) { + if(util_cpu_caps.has_sse2 && type.width == 8 && type.length == 16) { LLVMTypeRef i16x8 = LLVMVectorType(LLVMInt16Type(), 8); LLVMTypeRef i8x16 = LLVMVectorType(LLVMInt8Type(), 16); static LLVMValueRef ml = NULL; @@ -456,7 +451,6 @@ lp_build_mul(struct lp_build_context *bld, return ab; } -#endif /* FIXME */ assert(0); @@ -493,10 +487,8 @@ lp_build_div(struct lp_build_context *bld, if(LLVMIsConstant(a) && LLVMIsConstant(b)) return LLVMConstFDiv(a, b); -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - if(type.width == 32 && type.length == 4) + if(util_cpu_caps.has_sse && type.width == 32 && type.length == 4) return lp_build_mul(bld, a, lp_build_rcp(bld, b)); -#endif return LLVMBuildFDiv(bld->builder, a, b, ""); } @@ -606,8 +598,7 @@ lp_build_abs(struct lp_build_context *bld, return a; } -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - if(type.width*type.length == 128) { + if(type.width*type.length == 128 && util_cpu_caps.has_ssse3) { switch(type.width) { case 8: return lp_build_intrinsic_unary(bld->builder, "llvm.x86.ssse3.pabs.b.128", vec_type, a); @@ -617,7 +608,6 @@ lp_build_abs(struct lp_build_context *bld, return lp_build_intrinsic_unary(bld->builder, "llvm.x86.ssse3.pabs.d.128", vec_type, a); } } -#endif return lp_build_max(bld, a, LLVMBuildNeg(bld->builder, a, "")); } @@ -684,6 +674,8 @@ lp_build_round_sse41(struct lp_build_context *bld, assert(type.floating); assert(type.width*type.length == 128); + assert(lp_check_value(type, a)); + assert(util_cpu_caps.has_sse4_1); switch(type.width) { case 32: @@ -703,20 +695,45 @@ lp_build_round_sse41(struct lp_build_context *bld, LLVMValueRef -lp_build_round(struct lp_build_context *bld, +lp_build_trunc(struct lp_build_context *bld, LLVMValueRef a) { const struct lp_type type = bld->type; assert(type.floating); + assert(lp_check_value(type, a)); + + if(util_cpu_caps.has_sse4_1) + return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_TRUNCATE); + else { + LLVMTypeRef vec_type = lp_build_vec_type(type); + LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); + LLVMValueRef res; + res = LLVMBuildFPToSI(bld->builder, a, int_vec_type, ""); + res = LLVMBuildSIToFP(bld->builder, res, vec_type, ""); + return res; + } +} -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_NEAREST); -#endif - /* FIXME */ - assert(0); - return bld->undef; +LLVMValueRef +lp_build_round(struct lp_build_context *bld, + LLVMValueRef a) +{ + const struct lp_type type = bld->type; + + assert(type.floating); + assert(lp_check_value(type, a)); + + if(util_cpu_caps.has_sse4_1) + return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_NEAREST); + else { + LLVMTypeRef vec_type = lp_build_vec_type(type); + LLVMValueRef res; + res = lp_build_iround(bld, a); + res = LLVMBuildSIToFP(bld->builder, res, vec_type, ""); + return res; + } } @@ -728,13 +745,15 @@ lp_build_floor(struct lp_build_context *bld, assert(type.floating); -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_FLOOR); -#endif - - /* FIXME */ - assert(0); - return bld->undef; + if(util_cpu_caps.has_sse4_1) + return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_FLOOR); + else { + LLVMTypeRef vec_type = lp_build_vec_type(type); + LLVMValueRef res; + res = lp_build_ifloor(bld, a); + res = LLVMBuildSIToFP(bld->builder, res, vec_type, ""); + return res; + } } @@ -745,59 +764,143 @@ lp_build_ceil(struct lp_build_context *bld, const struct lp_type type = bld->type; assert(type.floating); + assert(lp_check_value(type, a)); + + if(util_cpu_caps.has_sse4_1) + return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_CEIL); + else { + LLVMTypeRef vec_type = lp_build_vec_type(type); + LLVMValueRef res; + res = lp_build_iceil(bld, a); + res = LLVMBuildSIToFP(bld->builder, res, vec_type, ""); + return res; + } +} -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_CEIL); -#endif - /* FIXME */ - assert(0); - return bld->undef; +/** + * Convert to integer, through whichever rounding method that's fastest, + * typically truncating to zero. + */ +LLVMValueRef +lp_build_itrunc(struct lp_build_context *bld, + LLVMValueRef a) +{ + const struct lp_type type = bld->type; + LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); + + assert(type.floating); + assert(lp_check_value(type, a)); + + return LLVMBuildFPToSI(bld->builder, a, int_vec_type, ""); } LLVMValueRef -lp_build_trunc(struct lp_build_context *bld, - LLVMValueRef a) +lp_build_iround(struct lp_build_context *bld, + LLVMValueRef a) { const struct lp_type type = bld->type; + LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); + LLVMValueRef res; assert(type.floating); + assert(lp_check_value(type, a)); -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_TRUNCATE); -#endif + if(util_cpu_caps.has_sse4_1) { + res = lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_NEAREST); + } + else { + LLVMTypeRef vec_type = lp_build_vec_type(type); + LLVMValueRef mask = lp_build_int_const_scalar(type, (unsigned long long)1 << (type.width - 1)); + LLVMValueRef sign; + LLVMValueRef half; + + /* get sign bit */ + sign = LLVMBuildBitCast(bld->builder, a, int_vec_type, ""); + sign = LLVMBuildAnd(bld->builder, sign, mask, ""); + + /* sign * 0.5 */ + half = lp_build_const_scalar(type, 0.5); + half = LLVMBuildBitCast(bld->builder, half, int_vec_type, ""); + half = LLVMBuildOr(bld->builder, sign, half, ""); + half = LLVMBuildBitCast(bld->builder, half, vec_type, ""); - /* FIXME */ - assert(0); - return bld->undef; + res = LLVMBuildAdd(bld->builder, a, half, ""); + } + + res = LLVMBuildFPToSI(bld->builder, res, int_vec_type, ""); + + return res; } -/** - * Convert to integer, through whichever rounding method that's fastest, - * typically truncating to zero. - */ LLVMValueRef -lp_build_int(struct lp_build_context *bld, - LLVMValueRef a) +lp_build_ifloor(struct lp_build_context *bld, + LLVMValueRef a) { const struct lp_type type = bld->type; LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); + LLVMValueRef res; assert(type.floating); + assert(lp_check_value(type, a)); - return LLVMBuildFPToSI(bld->builder, a, int_vec_type, ""); + if(util_cpu_caps.has_sse4_1) { + res = lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_FLOOR); + } + else { + /* Take the sign bit and add it to 1 constant */ + LLVMTypeRef vec_type = lp_build_vec_type(type); + unsigned mantissa = lp_mantissa(type); + LLVMValueRef mask = lp_build_int_const_scalar(type, (unsigned long long)1 << (type.width - 1)); + LLVMValueRef sign; + LLVMValueRef offset; + + /* sign = a < 0 ? ~0 : 0 */ + sign = LLVMBuildBitCast(bld->builder, a, int_vec_type, ""); + sign = LLVMBuildAnd(bld->builder, sign, mask, ""); + sign = LLVMBuildAShr(bld->builder, sign, lp_build_int_const_scalar(type, type.width - 1), ""); + + /* offset = -0.99999(9)f */ + offset = lp_build_const_scalar(type, -(double)(((unsigned long long)1 << mantissa) - 1)/((unsigned long long)1 << mantissa)); + offset = LLVMConstBitCast(offset, int_vec_type); + + /* offset = a < 0 ? -0.99999(9)f : 0.0f */ + offset = LLVMBuildAnd(bld->builder, offset, sign, ""); + offset = LLVMBuildBitCast(bld->builder, offset, vec_type, ""); + + res = LLVMBuildAdd(bld->builder, a, offset, ""); + } + + res = LLVMBuildFPToSI(bld->builder, res, int_vec_type, ""); + + return res; } LLVMValueRef -lp_build_ifloor(struct lp_build_context *bld, - LLVMValueRef a) +lp_build_iceil(struct lp_build_context *bld, + LLVMValueRef a) { - a = lp_build_floor(bld, a); - a = lp_build_int(bld, a); - return a; + const struct lp_type type = bld->type; + LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); + LLVMValueRef res; + + assert(type.floating); + assert(lp_check_value(type, a)); + + if(util_cpu_caps.has_sse4_1) { + res = lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_CEIL); + } + else { + assert(0); + res = bld->undef; + } + + res = LLVMBuildFPToSI(bld->builder, res, int_vec_type, ""); + + return res; } @@ -837,11 +940,9 @@ lp_build_rcp(struct lp_build_context *bld, if(LLVMIsConstant(a)) return LLVMConstFDiv(bld->one, a); - /* XXX: is this really necessary? */ -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - if(type.width == 32 && type.length == 4) + if(util_cpu_caps.has_sse && type.width == 32 && type.length == 4) + /* FIXME: improve precision */ return lp_build_intrinsic_unary(bld->builder, "llvm.x86.sse.rcp.ps", lp_build_vec_type(type), a); -#endif return LLVMBuildFDiv(bld->builder, bld->one, a, ""); } @@ -858,11 +959,8 @@ lp_build_rsqrt(struct lp_build_context *bld, assert(type.floating); - /* XXX: is this really necessary? */ -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - if(type.width == 32 && type.length == 4) + if(util_cpu_caps.has_sse && type.width == 32 && type.length == 4) return lp_build_intrinsic_unary(bld->builder, "llvm.x86.sse.rsqrt.ps", lp_build_vec_type(type), a); -#endif return lp_build_rcp(bld, lp_build_sqrt(bld, a)); } @@ -918,7 +1016,8 @@ lp_build_pow(struct lp_build_context *bld, { /* TODO: optimize the constant case */ if(LLVMIsConstant(x) && LLVMIsConstant(y)) - debug_printf("%s: inefficient/imprecise constant arithmetic\n"); + debug_printf("%s: inefficient/imprecise constant arithmetic\n", + __FUNCTION__); return lp_build_exp2(bld, lp_build_mul(bld, lp_build_log2(bld, x), y)); } @@ -972,7 +1071,8 @@ lp_build_polynomial(struct lp_build_context *bld, /* TODO: optimize the constant case */ if(LLVMIsConstant(x)) - debug_printf("%s: inefficient/imprecise constant arithmetic\n"); + debug_printf("%s: inefficient/imprecise constant arithmetic\n", + __FUNCTION__); for (i = num_coeffs; i--; ) { LLVMValueRef coeff = lp_build_const_scalar(type, coeffs[i]); @@ -1026,7 +1126,8 @@ lp_build_exp2_approx(struct lp_build_context *bld, if(p_exp2_int_part || p_frac_part || p_exp2) { /* TODO: optimize the constant case */ if(LLVMIsConstant(x)) - debug_printf("%s: inefficient/imprecise constant arithmetic\n"); + debug_printf("%s: inefficient/imprecise constant arithmetic\n", + __FUNCTION__); assert(type.floating && type.width == 32); @@ -1125,7 +1226,8 @@ lp_build_log2_approx(struct lp_build_context *bld, if(p_exp || p_floor_log2 || p_log2) { /* TODO: optimize the constant case */ if(LLVMIsConstant(x)) - debug_printf("%s: inefficient/imprecise constant arithmetic\n"); + debug_printf("%s: inefficient/imprecise constant arithmetic\n", + __FUNCTION__); assert(type.floating && type.width == 32); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_arit.h b/src/gallium/drivers/llvmpipe/lp_bld_arit.h index d68a97c4b87..095a8e1cabe 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_arit.h +++ b/src/gallium/drivers/llvmpipe/lp_bld_arit.h @@ -126,11 +126,18 @@ lp_build_trunc(struct lp_build_context *bld, LLVMValueRef a); LLVMValueRef -lp_build_int(struct lp_build_context *bld, - LLVMValueRef a); +lp_build_ifloor(struct lp_build_context *bld, + LLVMValueRef a); +LLVMValueRef +lp_build_iceil(struct lp_build_context *bld, + LLVMValueRef a); LLVMValueRef -lp_build_ifloor(struct lp_build_context *bld, +lp_build_iround(struct lp_build_context *bld, + LLVMValueRef a); + +LLVMValueRef +lp_build_itrunc(struct lp_build_context *bld, LLVMValueRef a); LLVMValueRef diff --git a/src/gallium/drivers/llvmpipe/lp_bld_conv.c b/src/gallium/drivers/llvmpipe/lp_bld_conv.c index 186cac70f62..20c8710214b 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_conv.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_conv.c @@ -63,6 +63,7 @@ #include "util/u_debug.h" #include "util/u_math.h" +#include "util/u_cpu_detect.h" #include "lp_bld_type.h" #include "lp_bld_const.h" @@ -334,8 +335,7 @@ lp_build_pack2(LLVMBuilderRef builder, assert(!src_type.floating); assert(!dst_type.floating); -#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) - if(src_type.width * src_type.length == 128) { + if(util_cpu_caps.has_sse2 && src_type.width * src_type.length == 128) { /* All X86 non-interleaved pack instructions all take signed inputs and * saturate them, so saturate beforehand. */ if(!src_type.sign && !clamped) { @@ -349,7 +349,7 @@ lp_build_pack2(LLVMBuilderRef builder, switch(src_type.width) { case 32: - if(dst_type.sign) + if(dst_type.sign || !util_cpu_caps.has_sse4_1) res = lp_build_intrinsic_binary(builder, "llvm.x86.sse2.packssdw.128", src_vec_type, lo, hi); else /* PACKUSDW is the only instrinsic with a consistent signature */ @@ -372,7 +372,6 @@ lp_build_pack2(LLVMBuilderRef builder, res = LLVMBuildBitCast(builder, res, dst_vec_type, ""); return res; } -#endif lo = LLVMBuildBitCast(builder, lo, dst_vec_type, ""); hi = LLVMBuildBitCast(builder, hi, dst_vec_type, ""); diff --git a/src/gallium/drivers/llvmpipe/lp_bld_depth.c b/src/gallium/drivers/llvmpipe/lp_bld_depth.c index 21c665c4d4c..98ec1cb1b9d 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_depth.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_depth.c @@ -210,7 +210,4 @@ lp_build_depth_test(LLVMBuilderRef builder, dst = lp_build_select(&bld, z_bitmask, src, dst); LLVMBuildStore(builder, dst, dst_ptr); } - - /* FIXME */ - assert(!state->occlusion_count); } diff --git a/src/gallium/drivers/llvmpipe/lp_bld_logic.c b/src/gallium/drivers/llvmpipe/lp_bld_logic.c index 6b6f8207697..db22a8028a6 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_logic.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_logic.c @@ -33,6 +33,8 @@ */ +#include "util/u_cpu_detect.h" + #include "lp_bld_type.h" #include "lp_bld_const.h" #include "lp_bld_intr.h" @@ -65,7 +67,7 @@ lp_build_cmp(struct lp_build_context *bld, #if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) if(type.width * type.length == 128) { - if(type.floating) { + if(type.floating && util_cpu_caps.has_sse) { LLVMValueRef args[3]; unsigned cc; boolean swap; @@ -114,7 +116,7 @@ lp_build_cmp(struct lp_build_context *bld, res = LLVMBuildBitCast(bld->builder, res, int_vec_type, ""); return res; } - else { + else if(util_cpu_caps.has_sse2) { static const struct { unsigned swap:1; unsigned eq:1; diff --git a/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c b/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c index 8ca1be6f1be..1a47ca32d2d 100644 --- a/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c +++ b/src/gallium/drivers/llvmpipe/lp_bld_sample_soa.c @@ -274,8 +274,8 @@ lp_build_sample_2d_linear_soa(struct lp_build_sample_context *bld, s_fpart = lp_build_sub(&bld->coord_bld, s, s_ipart); t_fpart = lp_build_sub(&bld->coord_bld, t, t_ipart); - x0 = lp_build_int(&bld->coord_bld, s_ipart); - y0 = lp_build_int(&bld->coord_bld, t_ipart); + x0 = lp_build_itrunc(&bld->coord_bld, s_ipart); + y0 = lp_build_itrunc(&bld->coord_bld, t_ipart); x0 = lp_build_sample_wrap(bld, x0, width, bld->static_state->pot_width, bld->static_state->wrap_s); y0 = lp_build_sample_wrap(bld, y0, height, bld->static_state->pot_height, bld->static_state->wrap_t); diff --git a/src/gallium/drivers/llvmpipe/lp_context.c b/src/gallium/drivers/llvmpipe/lp_context.c index a4b2bd8c2ad..202cb8ef439 100644 --- a/src/gallium/drivers/llvmpipe/lp_context.c +++ b/src/gallium/drivers/llvmpipe/lp_context.c @@ -107,11 +107,16 @@ static void llvmpipe_destroy( struct pipe_context *pipe ) if (llvmpipe->draw) draw_destroy( llvmpipe->draw ); - for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) + for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) { lp_destroy_tile_cache(llvmpipe->cbuf_cache[i]); + pipe_surface_reference(&llvmpipe->framebuffer.cbufs[i], NULL); + } + pipe_surface_reference(&llvmpipe->framebuffer.zsbuf, NULL); - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) + for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { lp_destroy_tex_tile_cache(llvmpipe->tex_cache[i]); + pipe_texture_reference(&llvmpipe->texture[i], NULL); + } for (i = 0; i < Elements(llvmpipe->constants); i++) { if (llvmpipe->constants[i].buffer) { diff --git a/src/gallium/drivers/llvmpipe/lp_jit.c b/src/gallium/drivers/llvmpipe/lp_jit.c index b4a22ff4a97..1126bf90b96 100644 --- a/src/gallium/drivers/llvmpipe/lp_jit.c +++ b/src/gallium/drivers/llvmpipe/lp_jit.c @@ -36,6 +36,7 @@ #include <llvm-c/Transforms/Scalar.h> #include "util/u_memory.h" +#include "util/u_cpu_detect.h" #include "lp_screen.h" #include "lp_bld_intr.h" #include "lp_jit.h" @@ -147,6 +148,19 @@ lp_jit_screen_init(struct llvmpipe_screen *screen) { char *error = NULL; + util_cpu_detect(); + +#if 0 + /* For simulating less capable machines */ + util_cpu_caps.has_sse3 = 0; + util_cpu_caps.has_sse4_1 = 0; +#endif + +#ifdef LLVM_NATIVE_ARCH + LLVMLinkInJIT(); + LLVMInitializeNativeTarget(); +#endif + screen->module = LLVMModuleCreateWithName("llvmpipe"); screen->provider = LLVMCreateModuleProviderForExistingModule(screen->module); @@ -163,8 +177,15 @@ lp_jit_screen_init(struct llvmpipe_screen *screen) LLVMAddTargetData(screen->target, screen->pass); /* These are the passes currently listed in llvm-c/Transforms/Scalar.h, * but there are more on SVN. */ + /* TODO: Add more passes */ LLVMAddConstantPropagationPass(screen->pass); - LLVMAddInstructionCombiningPass(screen->pass); + if(util_cpu_caps.has_sse4_1) { + /* FIXME: There is a bug in this pass, whereby the combination of fptosi + * and sitofp (necessary for trunc/floor/ceil/round implementation) + * somehow becomes invalid code. + */ + LLVMAddInstructionCombiningPass(screen->pass); + } LLVMAddPromoteMemoryToRegisterPass(screen->pass); LLVMAddGVNPass(screen->pass); LLVMAddCFGSimplificationPass(screen->pass); diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c index ff7ef8658a9..05189274589 100644 --- a/src/gallium/drivers/llvmpipe/lp_screen.c +++ b/src/gallium/drivers/llvmpipe/lp_screen.c @@ -82,7 +82,7 @@ llvmpipe_get_param(struct pipe_screen *screen, int param) case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: return 13; /* max 4Kx4K */ case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 8; /* max 128x128x128 */ + return 9; /* max 256x256x256 */ case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; /* max 4Kx4K */ case PIPE_CAP_TGSI_CONT_SUPPORTED: diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c index 9faed5a0b18..b00be0cc32a 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_fs.c +++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c @@ -401,7 +401,6 @@ generate_fragment(struct llvmpipe_context *lp, if(key->depth.enabled) { debug_printf("depth.func = %s\n", debug_dump_func(key->depth.func, TRUE)); debug_printf("depth.writemask = %u\n", key->depth.writemask); - debug_printf("depth.occlusion_count = %u\n", key->depth.occlusion_count); } if(key->alpha.enabled) { debug_printf("alpha.func = %s\n", debug_dump_func(key->alpha.func, TRUE)); @@ -582,6 +581,11 @@ generate_fragment(struct llvmpipe_context *lp, * Translate the LLVM IR into machine code. */ + if(LLVMVerifyFunction(variant->function, LLVMPrintMessageAction)) { + LLVMDumpValue(variant->function); + abort(); + } + LLVMRunFunctionPassManager(screen->pass, variant->function); #ifdef DEBUG @@ -589,11 +593,6 @@ generate_fragment(struct llvmpipe_context *lp, debug_printf("\n"); #endif - if(LLVMVerifyFunction(variant->function, LLVMPrintMessageAction)) { - LLVMDumpValue(variant->function); - abort(); - } - variant->jit_function = (lp_jit_frag_func)LLVMGetPointerToGlobal(screen->engine, variant->function); #ifdef DEBUG diff --git a/src/gallium/drivers/llvmpipe/lp_state_surface.c b/src/gallium/drivers/llvmpipe/lp_state_surface.c index 177a26b7b1f..2c29144c039 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_surface.c +++ b/src/gallium/drivers/llvmpipe/lp_state_surface.c @@ -56,7 +56,7 @@ llvmpipe_set_framebuffer_state(struct pipe_context *pipe, lp_flush_tile_cache(lp->cbuf_cache[i]); /* assign new */ - lp->framebuffer.cbufs[i] = fb->cbufs[i]; + pipe_surface_reference(&lp->framebuffer.cbufs[i], fb->cbufs[i]); /* update cache */ lp_tile_cache_set_surface(lp->cbuf_cache[i], fb->cbufs[i]); @@ -81,7 +81,7 @@ llvmpipe_set_framebuffer_state(struct pipe_context *pipe, } /* assign new */ - lp->framebuffer.zsbuf = fb->zsbuf; + pipe_surface_reference(&lp->framebuffer.zsbuf, fb->zsbuf); /* Tell draw module how deep the Z/depth buffer is */ if (lp->framebuffer.zsbuf) { diff --git a/src/gallium/drivers/llvmpipe/lp_test_format.c b/src/gallium/drivers/llvmpipe/lp_test_format.c index d8455e56490..7d83f899e6a 100644 --- a/src/gallium/drivers/llvmpipe/lp_test_format.c +++ b/src/gallium/drivers/llvmpipe/lp_test_format.c @@ -264,6 +264,11 @@ int main(int argc, char **argv) unsigned i; int ret; +#ifdef LLVM_NATIVE_ARCH + LLVMLinkInJIT(); + LLVMInitializeNativeTarget(); +#endif + for (i = 0; i < sizeof(test_cases)/sizeof(test_cases[0]); ++i) if(!test_format(&test_cases[i])) ret = 1; diff --git a/src/gallium/drivers/llvmpipe/lp_test_main.c b/src/gallium/drivers/llvmpipe/lp_test_main.c index 4592dc0b2d0..f07fa256f16 100644 --- a/src/gallium/drivers/llvmpipe/lp_test_main.c +++ b/src/gallium/drivers/llvmpipe/lp_test_main.c @@ -365,6 +365,11 @@ int main(int argc, char **argv) n = atoi(argv[i]); } +#ifdef LLVM_NATIVE_ARCH + LLVMLinkInJIT(); + LLVMInitializeNativeTarget(); +#endif + if(fp) { /* Warm up the caches */ test_some(0, NULL, 100); diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c index 724d4378336..08f0950d475 100644 --- a/src/gallium/drivers/llvmpipe/lp_texture.c +++ b/src/gallium/drivers/llvmpipe/lp_texture.c @@ -353,17 +353,9 @@ llvmpipe_transfer_map( struct pipe_screen *_screen, if(lpt->dt) { struct llvmpipe_winsys *winsys = screen->winsys; - unsigned flags = 0; - if (transfer->usage != PIPE_TRANSFER_READ) { - flags |= PIPE_BUFFER_USAGE_CPU_WRITE; - } - - if (transfer->usage != PIPE_TRANSFER_WRITE) { - flags |= PIPE_BUFFER_USAGE_CPU_READ; - } - - map = winsys->displaytarget_map(winsys, lpt->dt, flags); + map = winsys->displaytarget_map(winsys, lpt->dt, + pipe_transfer_buffer_flags(transfer)); if (map == NULL) return NULL; } @@ -373,7 +365,7 @@ llvmpipe_transfer_map( struct pipe_screen *_screen, /* May want to different things here depending on read/write nature * of the map: */ - if (transfer->texture && transfer->usage != PIPE_TRANSFER_READ) + if (transfer->texture && (transfer->usage & PIPE_TRANSFER_WRITE)) { /* Do something to notify sharing contexts of a texture change. * In llvmpipe, that would mean flushing the texture cache. diff --git a/src/gallium/drivers/llvmpipe/lp_tile_cache.c b/src/gallium/drivers/llvmpipe/lp_tile_cache.c index 2e576e6039d..0c06b659a1f 100644 --- a/src/gallium/drivers/llvmpipe/lp_tile_cache.c +++ b/src/gallium/drivers/llvmpipe/lp_tile_cache.c @@ -44,10 +44,53 @@ #include "lp_tile_cache.h" +#define MAX_WIDTH 4096 +#define MAX_HEIGHT 4096 + + +enum llvmpipe_tile_status +{ + LP_TILE_STATUS_UNDEFINED = 0, + LP_TILE_STATUS_CLEAR = 1, + LP_TILE_STATUS_DEFINED = 2 +}; + + +struct llvmpipe_cached_tile +{ + enum llvmpipe_tile_status status; + + /** color in SOA format */ + uint8_t *color; +}; + + +struct llvmpipe_tile_cache +{ + struct pipe_screen *screen; + struct pipe_surface *surface; /**< the surface we're caching */ + struct pipe_transfer *transfer; + void *transfer_map; + + struct llvmpipe_cached_tile entries[MAX_WIDTH/TILE_SIZE][MAX_HEIGHT/TILE_SIZE]; + + uint8_t clear_color[4]; /**< for color bufs */ + uint clear_val; /**< for z+stencil, or packed color clear value */ + + struct llvmpipe_cached_tile *last_tile; /**< most recently retrieved tile */ +}; + + struct llvmpipe_tile_cache * lp_create_tile_cache( struct pipe_screen *screen ) { struct llvmpipe_tile_cache *tc; + int maxLevels, maxTexSize; + + /* sanity checking: max sure MAX_WIDTH/HEIGHT >= largest texture image */ + maxLevels = screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS); + maxTexSize = 1 << (maxLevels - 1); + assert(MAX_WIDTH >= maxTexSize); tc = CALLOC_STRUCT( llvmpipe_tile_cache ); if(!tc) diff --git a/src/gallium/drivers/llvmpipe/lp_tile_cache.h b/src/gallium/drivers/llvmpipe/lp_tile_cache.h index 6d8ba5ece7a..161bab37991 100644 --- a/src/gallium/drivers/llvmpipe/lp_tile_cache.h +++ b/src/gallium/drivers/llvmpipe/lp_tile_cache.h @@ -33,42 +33,7 @@ #include "lp_tile_soa.h" -enum llvmpipe_tile_status -{ - LP_TILE_STATUS_UNDEFINED = 0, - LP_TILE_STATUS_CLEAR = 1, - LP_TILE_STATUS_DEFINED = 2 -}; - - -struct llvmpipe_cached_tile -{ - enum llvmpipe_tile_status status; - - /** color in SOA format */ - uint8_t *color; -}; - - -/** XXX move these */ -#define MAX_WIDTH 2048 -#define MAX_HEIGHT 2048 - - -struct llvmpipe_tile_cache -{ - struct pipe_screen *screen; - struct pipe_surface *surface; /**< the surface we're caching */ - struct pipe_transfer *transfer; - void *transfer_map; - - struct llvmpipe_cached_tile entries[MAX_WIDTH/TILE_SIZE][MAX_HEIGHT/TILE_SIZE]; - - uint8_t clear_color[4]; /**< for color bufs */ - uint clear_val; /**< for z+stencil, or packed color clear value */ - - struct llvmpipe_cached_tile *last_tile; /**< most recently retrieved tile */ -}; +struct llvmpipe_tile_cache; /* opaque */ extern struct llvmpipe_tile_cache * diff --git a/src/gallium/drivers/nv04/nv04_transfer.c b/src/gallium/drivers/nv04/nv04_transfer.c index 854b855d64a..6618660743d 100644 --- a/src/gallium/drivers/nv04/nv04_transfer.c +++ b/src/gallium/drivers/nv04/nv04_transfer.c @@ -13,22 +13,6 @@ struct nv04_transfer { bool direct; }; -static unsigned nv04_usage_tx_to_buf(unsigned tx_usage) -{ - switch (tx_usage) { - case PIPE_TRANSFER_READ: - return PIPE_BUFFER_USAGE_CPU_READ; - case PIPE_TRANSFER_WRITE: - return PIPE_BUFFER_USAGE_CPU_WRITE; - case PIPE_TRANSFER_READ_WRITE: - return PIPE_BUFFER_USAGE_CPU_READ_WRITE; - default: - assert(0); - } - - return -1; -} - static void nv04_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, struct pipe_texture *template) @@ -86,7 +70,7 @@ nv04_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->direct = true; tx->surface = pscreen->get_tex_surface(pscreen, pt, 0, 0, 0, - nv04_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); return &tx->base; } @@ -103,7 +87,7 @@ nv04_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->surface = pscreen->get_tex_surface(pscreen, tx_tex, face, level, zslice, - nv04_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); pipe_texture_reference(&tx_tex, NULL); @@ -114,7 +98,7 @@ nv04_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; } - if (usage != PIPE_TRANSFER_WRITE) { + if (usage & PIPE_TRANSFER_READ) { struct nv04_screen *nvscreen = nv04_screen(pscreen); struct pipe_surface *src; @@ -140,7 +124,7 @@ nv04_transfer_del(struct pipe_transfer *ptx) { struct nv04_transfer *tx = (struct nv04_transfer *)ptx; - if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) { + if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) { struct pipe_screen *pscreen = ptx->texture->screen; struct nv04_screen *nvscreen = nv04_screen(pscreen); struct pipe_surface *dst; @@ -170,7 +154,7 @@ nv04_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) struct nv04_surface *ns = (struct nv04_surface *)tx->surface; struct nv04_miptree *mt = (struct nv04_miptree *)tx->surface->texture; void *map = pipe_buffer_map(pscreen, mt->buffer, - nv04_usage_tx_to_buf(ptx->usage)); + pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + ptx->y * ns->pitch + ptx->x * ptx->block.size; diff --git a/src/gallium/drivers/nv10/nv10_transfer.c b/src/gallium/drivers/nv10/nv10_transfer.c index c06b8d34c72..8feb85e4bda 100644 --- a/src/gallium/drivers/nv10/nv10_transfer.c +++ b/src/gallium/drivers/nv10/nv10_transfer.c @@ -13,22 +13,6 @@ struct nv10_transfer { bool direct; }; -static unsigned nv10_usage_tx_to_buf(unsigned tx_usage) -{ - switch (tx_usage) { - case PIPE_TRANSFER_READ: - return PIPE_BUFFER_USAGE_CPU_READ; - case PIPE_TRANSFER_WRITE: - return PIPE_BUFFER_USAGE_CPU_WRITE; - case PIPE_TRANSFER_READ_WRITE: - return PIPE_BUFFER_USAGE_CPU_READ_WRITE; - default: - assert(0); - } - - return -1; -} - static void nv10_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, struct pipe_texture *template) @@ -86,7 +70,7 @@ nv10_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->direct = true; tx->surface = pscreen->get_tex_surface(pscreen, pt, 0, 0, 0, - nv10_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); return &tx->base; } @@ -103,7 +87,7 @@ nv10_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->surface = pscreen->get_tex_surface(pscreen, tx_tex, face, level, zslice, - nv10_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); pipe_texture_reference(&tx_tex, NULL); @@ -114,7 +98,7 @@ nv10_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; } - if (usage != PIPE_TRANSFER_WRITE) { + if (usage & PIPE_TRANSFER_READ) { struct nv10_screen *nvscreen = nv10_screen(pscreen); struct pipe_surface *src; @@ -140,7 +124,7 @@ nv10_transfer_del(struct pipe_transfer *ptx) { struct nv10_transfer *tx = (struct nv10_transfer *)ptx; - if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) { + if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) { struct pipe_screen *pscreen = ptx->texture->screen; struct nv10_screen *nvscreen = nv10_screen(pscreen); struct pipe_surface *dst; @@ -170,7 +154,7 @@ nv10_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) struct nv04_surface *ns = (struct nv04_surface *)tx->surface; struct nv10_miptree *mt = (struct nv10_miptree *)tx->surface->texture; void *map = pipe_buffer_map(pscreen, mt->buffer, - nv10_usage_tx_to_buf(ptx->usage)); + pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + ptx->y * ns->pitch + ptx->x * ptx->block.size; diff --git a/src/gallium/drivers/nv20/nv20_transfer.c b/src/gallium/drivers/nv20/nv20_transfer.c index 5018995596c..81b4f1a9177 100644 --- a/src/gallium/drivers/nv20/nv20_transfer.c +++ b/src/gallium/drivers/nv20/nv20_transfer.c @@ -13,22 +13,6 @@ struct nv20_transfer { bool direct; }; -static unsigned nv20_usage_tx_to_buf(unsigned tx_usage) -{ - switch (tx_usage) { - case PIPE_TRANSFER_READ: - return PIPE_BUFFER_USAGE_CPU_READ; - case PIPE_TRANSFER_WRITE: - return PIPE_BUFFER_USAGE_CPU_WRITE; - case PIPE_TRANSFER_READ_WRITE: - return PIPE_BUFFER_USAGE_CPU_READ_WRITE; - default: - assert(0); - } - - return -1; -} - static void nv20_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, struct pipe_texture *template) @@ -86,7 +70,7 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->direct = true; tx->surface = pscreen->get_tex_surface(pscreen, pt, 0, 0, 0, - nv20_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); return &tx->base; } @@ -103,7 +87,7 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->surface = pscreen->get_tex_surface(pscreen, tx_tex, face, level, zslice, - nv20_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); pipe_texture_reference(&tx_tex, NULL); @@ -114,7 +98,7 @@ nv20_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; } - if (usage != PIPE_TRANSFER_WRITE) { + if (usage & PIPE_TRANSFER_READ) { struct nv20_screen *nvscreen = nv20_screen(pscreen); struct pipe_surface *src; @@ -140,7 +124,7 @@ nv20_transfer_del(struct pipe_transfer *ptx) { struct nv20_transfer *tx = (struct nv20_transfer *)ptx; - if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) { + if (!tx->direct && (ptx->usage = PIPE_TRANSFER_WRITE)) { struct pipe_screen *pscreen = ptx->texture->screen; struct nv20_screen *nvscreen = nv20_screen(pscreen); struct pipe_surface *dst; @@ -170,7 +154,7 @@ nv20_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) struct nv04_surface *ns = (struct nv04_surface *)tx->surface; struct nv20_miptree *mt = (struct nv20_miptree *)tx->surface->texture; void *map = pipe_buffer_map(pscreen, mt->buffer, - nv20_usage_tx_to_buf(ptx->usage)); + pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + ptx->y * ns->pitch + ptx->x * ptx->block.size; diff --git a/src/gallium/drivers/nv30/nv30_transfer.c b/src/gallium/drivers/nv30/nv30_transfer.c index 23675718781..98011decf7c 100644 --- a/src/gallium/drivers/nv30/nv30_transfer.c +++ b/src/gallium/drivers/nv30/nv30_transfer.c @@ -13,22 +13,6 @@ struct nv30_transfer { bool direct; }; -static unsigned nv30_usage_tx_to_buf(unsigned tx_usage) -{ - switch (tx_usage) { - case PIPE_TRANSFER_READ: - return PIPE_BUFFER_USAGE_CPU_READ; - case PIPE_TRANSFER_WRITE: - return PIPE_BUFFER_USAGE_CPU_WRITE; - case PIPE_TRANSFER_READ_WRITE: - return PIPE_BUFFER_USAGE_CPU_READ_WRITE; - default: - assert(0); - } - - return -1; -} - static void nv30_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, struct pipe_texture *template) @@ -86,7 +70,7 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->direct = true; tx->surface = pscreen->get_tex_surface(pscreen, pt, face, level, zslice, - nv30_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); return &tx->base; } @@ -103,7 +87,7 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->surface = pscreen->get_tex_surface(pscreen, tx_tex, 0, 0, 0, - nv30_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); pipe_texture_reference(&tx_tex, NULL); @@ -114,7 +98,7 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; } - if (usage != PIPE_TRANSFER_WRITE) { + if (usage & PIPE_TRANSFER_READ) { struct nv30_screen *nvscreen = nv30_screen(pscreen); struct pipe_surface *src; @@ -140,7 +124,7 @@ nv30_transfer_del(struct pipe_transfer *ptx) { struct nv30_transfer *tx = (struct nv30_transfer *)ptx; - if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) { + if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) { struct pipe_screen *pscreen = ptx->texture->screen; struct nv30_screen *nvscreen = nv30_screen(pscreen); struct pipe_surface *dst; @@ -170,7 +154,7 @@ nv30_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) struct nv04_surface *ns = (struct nv04_surface *)tx->surface; struct nv30_miptree *mt = (struct nv30_miptree *)tx->surface->texture; void *map = pipe_buffer_map(pscreen, mt->buffer, - nv30_usage_tx_to_buf(ptx->usage)); + pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + ptx->y * ns->pitch + ptx->x * ptx->block.size; diff --git a/src/gallium/drivers/nv40/nv40_transfer.c b/src/gallium/drivers/nv40/nv40_transfer.c index 6d92ac3db9c..92caee6f382 100644 --- a/src/gallium/drivers/nv40/nv40_transfer.c +++ b/src/gallium/drivers/nv40/nv40_transfer.c @@ -13,22 +13,6 @@ struct nv40_transfer { bool direct; }; -static unsigned nv40_usage_tx_to_buf(unsigned tx_usage) -{ - switch (tx_usage) { - case PIPE_TRANSFER_READ: - return PIPE_BUFFER_USAGE_CPU_READ; - case PIPE_TRANSFER_WRITE: - return PIPE_BUFFER_USAGE_CPU_WRITE; - case PIPE_TRANSFER_READ_WRITE: - return PIPE_BUFFER_USAGE_CPU_READ_WRITE; - default: - assert(0); - } - - return -1; -} - static void nv40_compatible_transfer_tex(struct pipe_texture *pt, unsigned level, struct pipe_texture *template) @@ -86,7 +70,7 @@ nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->direct = true; tx->surface = pscreen->get_tex_surface(pscreen, pt, face, level, zslice, - nv40_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); return &tx->base; } @@ -103,7 +87,7 @@ nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, tx->surface = pscreen->get_tex_surface(pscreen, tx_tex, 0, 0, 0, - nv40_usage_tx_to_buf(usage)); + pipe_transfer_buffer_flags(&tx->base)); pipe_texture_reference(&tx_tex, NULL); @@ -114,7 +98,7 @@ nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; } - if (usage != PIPE_TRANSFER_WRITE) { + if (usage & PIPE_TRANSFER_READ) { struct nv40_screen *nvscreen = nv40_screen(pscreen); struct pipe_surface *src; @@ -140,7 +124,7 @@ nv40_transfer_del(struct pipe_transfer *ptx) { struct nv40_transfer *tx = (struct nv40_transfer *)ptx; - if (!tx->direct && ptx->usage != PIPE_TRANSFER_READ) { + if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) { struct pipe_screen *pscreen = ptx->texture->screen; struct nv40_screen *nvscreen = nv40_screen(pscreen); struct pipe_surface *dst; @@ -170,7 +154,7 @@ nv40_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) struct nv04_surface *ns = (struct nv04_surface *)tx->surface; struct nv40_miptree *mt = (struct nv40_miptree *)tx->surface->texture; void *map = pipe_buffer_map(pscreen, mt->buffer, - nv40_usage_tx_to_buf(ptx->usage)); + pipe_transfer_buffer_flags(ptx)); return map + ns->base.offset + ptx->y * ns->pitch + ptx->x * ptx->block.size; diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c index eb90d5e66f9..576d075318f 100644 --- a/src/gallium/drivers/nv50/nv50_program.c +++ b/src/gallium/drivers/nv50/nv50_program.c @@ -90,6 +90,10 @@ struct nv50_reg { int acc; /* instruction where this reg is last read (first insn == 1) */ }; +/* arbitrary limits */ +#define MAX_IF_DEPTH 4 +#define MAX_LOOP_DEPTH 4 + struct nv50_pc { struct nv50_program *p; @@ -121,6 +125,13 @@ struct nv50_pc { struct nv50_reg *iv_p; struct nv50_reg *iv_c; + struct nv50_program_exec *if_cond; + struct nv50_program_exec *if_insn[MAX_IF_DEPTH]; + struct nv50_program_exec *br_join[MAX_IF_DEPTH]; + struct nv50_program_exec *br_loop[MAX_LOOP_DEPTH]; /* for BRK branch */ + int if_lvl, loop_lvl; + unsigned loop_pos[MAX_LOOP_DEPTH]; + /* current instruction and total number of insns */ unsigned insn_cur; unsigned insn_nr; @@ -196,6 +207,10 @@ alloc_reg(struct nv50_pc *pc, struct nv50_reg *reg) assert(0); } +/* XXX: For shaders that aren't executed linearly (e.g. shaders that + * contain loops), we need to assign all hw regs to TGSI TEMPs early, + * lest we risk temp_temps overwriting regs alloc'd "later". + */ static struct nv50_reg * alloc_temp(struct nv50_pc *pc, struct nv50_reg *dst) { @@ -559,6 +574,22 @@ check_swap_src_0_1(struct nv50_pc *pc, } static void +set_src_0_restricted(struct nv50_pc *pc, struct nv50_reg *src, + struct nv50_program_exec *e) +{ + struct nv50_reg *temp; + + if (src->type != P_TEMP) { + temp = temp_temp(pc); + emit_mov(pc, temp, src); + src = temp; + } + + alloc_reg(pc, src); + e->inst[0] |= (src->hw << 9); +} + +static void set_src_0(struct nv50_pc *pc, struct nv50_reg *src, struct nv50_program_exec *e) { if (src->type == P_ATTR) { @@ -760,7 +791,11 @@ emit_flop(struct nv50_pc *pc, unsigned sub, } set_dst(pc, dst, e); - set_src_0(pc, src, e); + + if (sub == 0 || sub == 2) + set_src_0_restricted(pc, src, e); + else + set_src_0(pc, src, e); emit(pc, e); } @@ -810,7 +845,8 @@ emit_precossin(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) #define CVT_F32_U32 0x64 #define CVT_S32_F32 0x8c #define CVT_S32_S32 0x0c -#define CVT_F32_F32_ROP 0xcc +#define CVT_NEG 0x20 +#define CVT_RI 0x08 static void emit_cvt(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src, @@ -890,6 +926,7 @@ emit_set(struct nv50_pc *pc, unsigned ccode, struct nv50_reg *dst, int wp, set_src_1(pc, src1, e); emit(pc, e); + pc->if_cond = pc->p->exec_tail; /* record for OPCODE_IF */ /* cvt.f32.u32/s32 (?) if we didn't only write the predicate */ if (rdst) @@ -917,7 +954,7 @@ map_tgsi_setop_cc(unsigned op) static INLINE void emit_flr(struct nv50_pc *pc, struct nv50_reg *dst, struct nv50_reg *src) { - emit_cvt(pc, dst, src, -1, CVTOP_FLOOR, CVT_F32_F32_ROP); + emit_cvt(pc, dst, src, -1, CVTOP_FLOOR, CVT_F32_F32 | CVT_RI); } static void @@ -1129,10 +1166,11 @@ emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, emit(pc, e); #if 1 - if (mask & 1) emit_mov(pc, dst[0], t[0]); - if (mask & 2) emit_mov(pc, dst[1], t[1]); - if (mask & 4) emit_mov(pc, dst[2], t[2]); - if (mask & 8) emit_mov(pc, dst[3], t[3]); + c = 0; + if (mask & 1) emit_mov(pc, dst[0], t[c++]); + if (mask & 2) emit_mov(pc, dst[1], t[c++]); + if (mask & 4) emit_mov(pc, dst[2], t[c++]); + if (mask & 8) emit_mov(pc, dst[3], t[c]); free_temp4(pc, t); #else @@ -1149,6 +1187,38 @@ emit_tex(struct nv50_pc *pc, struct nv50_reg **dst, unsigned mask, } static void +emit_branch(struct nv50_pc *pc, int pred, unsigned cc, + struct nv50_program_exec **join) +{ + struct nv50_program_exec *e = exec(pc); + + if (join) { + set_long(pc, e); + e->inst[0] |= 0xa0000002; + emit(pc, e); + *join = e; + e = exec(pc); + } + + set_long(pc, e); + e->inst[0] |= 0x10000002; + if (pred >= 0) + set_pred(pc, cc, pred, e); + emit(pc, e); +} + +static void +emit_nop(struct nv50_pc *pc) +{ + struct nv50_program_exec *e = exec(pc); + + e->inst[0] = 0xf0000000; + set_long(pc, e); + e->inst[1] = 0xe0000000; + emit(pc, e); +} + +static void convert_to_long(struct nv50_pc *pc, struct nv50_program_exec *e) { unsigned q = 0, m = ~0; @@ -1444,6 +1514,55 @@ nv50_tgsi_dst_revdep(unsigned op, int s, int c) } } +static INLINE boolean +has_pred(struct nv50_program_exec *e, unsigned cc) +{ + if (!is_long(e) || is_immd(e)) + return FALSE; + return ((e->inst[1] & 0x780) == (cc << 7)); +} + +/* on ENDIF see if we can do "@p0.neu single_op" instead of: + * join_at ENDIF + * @p0.eq bra ENDIF + * single_op + * ENDIF: nop.join + */ +static boolean +nv50_kill_branch(struct nv50_pc *pc) +{ + int lvl = pc->if_lvl; + + if (pc->if_insn[lvl]->next != pc->p->exec_tail) + return FALSE; + + /* if ccode == 'true', the BRA is from an ELSE and the predicate + * reg may no longer be valid, since we currently always use $p0 + */ + if (has_pred(pc->if_insn[lvl], 0xf)) + return FALSE; + assert(pc->if_insn[lvl] && pc->br_join[lvl]); + + /* We'll use the exec allocated for JOIN_AT (as we can't easily + * update prev's next); if exec_tail is BRK, update the pointer. + */ + if (pc->loop_lvl && pc->br_loop[pc->loop_lvl - 1] == pc->p->exec_tail) + pc->br_loop[pc->loop_lvl - 1] = pc->br_join[lvl]; + + pc->p->exec_size -= 4; /* remove JOIN_AT and BRA */ + + *pc->br_join[lvl] = *pc->p->exec_tail; + + FREE(pc->if_insn[lvl]); + FREE(pc->p->exec_tail); + + pc->p->exec_tail = pc->br_join[lvl]; + pc->p->exec_tail->next = NULL; + set_pred(pc, 0xd, 0, pc->p->exec_tail); + + return TRUE; +} + static boolean nv50_program_tx_insn(struct nv50_pc *pc, const struct tgsi_full_instruction *inst) @@ -1513,12 +1632,20 @@ nv50_program_tx_insn(struct nv50_pc *pc, emit_add(pc, dst[c], src[0][c], src[1][c]); } break; + case TGSI_OPCODE_BGNLOOP: + pc->loop_pos[pc->loop_lvl++] = pc->p->exec_size; + break; + case TGSI_OPCODE_BRK: + emit_branch(pc, -1, 0, NULL); + assert(pc->loop_lvl > 0); + pc->br_loop[pc->loop_lvl - 1] = pc->p->exec_tail; + break; case TGSI_OPCODE_CEIL: for (c = 0; c < 4; c++) { if (!(mask & (1 << c))) continue; emit_cvt(pc, dst[c], src[0][c], -1, - CVTOP_CEIL, CVT_F32_F32); + CVTOP_CEIL, CVT_F32_F32 | CVT_RI); } break; case TGSI_OPCODE_COS: @@ -1560,6 +1687,33 @@ nv50_program_tx_insn(struct nv50_pc *pc, if (mask & (1 << 0)) emit_mov_immdval(pc, dst[0], 1.0f); break; + case TGSI_OPCODE_ELSE: + emit_branch(pc, -1, 0, NULL); + pc->if_insn[--pc->if_lvl]->param.index = pc->p->exec_size; + pc->if_insn[pc->if_lvl++] = pc->p->exec_tail; + break; + case TGSI_OPCODE_ENDIF: + pc->if_insn[--pc->if_lvl]->param.index = pc->p->exec_size; + + /* try to replace branch over 1 insn with a predicated insn */ + if (nv50_kill_branch(pc) == TRUE) + break; + + if (pc->br_join[pc->if_lvl]) { + pc->br_join[pc->if_lvl]->param.index = pc->p->exec_size; + pc->br_join[pc->if_lvl] = NULL; + } + /* emit a NOP as join point, we could set it on the next + * one, but would have to make sure it is long and !immd + */ + emit_nop(pc); + pc->p->exec_tail->inst[1] |= 2; + break; + case TGSI_OPCODE_ENDLOOP: + emit_branch(pc, -1, 0, NULL); + pc->p->exec_tail->param.index = pc->loop_pos[--pc->loop_lvl]; + pc->br_loop[pc->loop_lvl]->param.index = pc->p->exec_size; + break; case TGSI_OPCODE_EX2: emit_preex2(pc, temp, src[0][0]); emit_flop(pc, 6, brdc, temp); @@ -1580,6 +1734,13 @@ nv50_program_tx_insn(struct nv50_pc *pc, emit_sub(pc, dst[c], src[0][c], temp); } break; + case TGSI_OPCODE_IF: + /* emitting a join_at may not be necessary */ + assert(pc->if_lvl < MAX_IF_DEPTH); + set_pred_wr(pc, 1, 0, pc->if_cond); + emit_branch(pc, 0, 2, &pc->br_join[pc->if_lvl]); + pc->if_insn[pc->if_lvl++] = pc->p->exec_tail; + break; case TGSI_OPCODE_KIL: emit_kil(pc, src[0][0]); emit_kil(pc, src[0][1]); @@ -1704,7 +1865,7 @@ nv50_program_tx_insn(struct nv50_pc *pc, if (!(mask & (1 << c))) continue; emit_cvt(pc, dst[c], src[0][c], -1, - CVTOP_TRUNC, CVT_F32_F32); + CVTOP_TRUNC, CVT_F32_F32 | CVT_RI); } break; case TGSI_OPCODE_XPD: @@ -2237,6 +2398,8 @@ nv50_program_tx_prep(struct nv50_pc *pc) pc->result[i].rhw = rid++; if (p->info.writes_z) pc->result[2].rhw = rid; + + p->cfg.high_result = rid; } if (pc->immd_nr) { @@ -2362,12 +2525,75 @@ ctor_nv50_pc(struct nv50_pc *pc, struct nv50_program *p) return TRUE; } +static void +nv50_fp_move_results(struct nv50_pc *pc) +{ + struct nv50_reg reg; + unsigned i; + + ctor_reg(®, P_TEMP, -1, -1); + + for (i = 0; i < pc->result_nr * 4; ++i) { + if (pc->result[i].rhw < 0 || pc->result[i].hw < 0) + continue; + if (pc->result[i].rhw != pc->result[i].hw) { + reg.hw = pc->result[i].rhw; + emit_mov(pc, ®, &pc->result[i]); + } + } +} + +static void +nv50_program_fixup_insns(struct nv50_pc *pc) +{ + struct nv50_program_exec *e, *prev = NULL, **bra_list; + unsigned i, n, pos; + + bra_list = CALLOC(pc->p->exec_size, sizeof(struct nv50_program_exec *)); + + /* Collect branch instructions, we need to adjust their offsets + * when converting 32 bit instructions to 64 bit ones + */ + for (n = 0, e = pc->p->exec_head; e; e = e->next) + if (e->param.index >= 0 && !e->param.mask) + bra_list[n++] = e; + + /* Make sure we don't have any single 32 bit instructions. */ + for (e = pc->p->exec_head, pos = 0; e; e = e->next) { + pos += is_long(e) ? 2 : 1; + + if ((pos & 1) && (!e->next || is_long(e->next))) { + for (i = 0; i < n; ++i) + if (bra_list[i]->param.index >= pos) + bra_list[i]->param.index += 1; + convert_to_long(pc, e); + ++pos; + } + if (e->next) + prev = e; + } + + assert(!is_immd(pc->p->exec_head)); + assert(!is_immd(pc->p->exec_tail)); + + /* last instruction must be long so it can have the end bit set */ + if (!is_long(pc->p->exec_tail)) { + convert_to_long(pc, pc->p->exec_tail); + if (prev) + convert_to_long(pc, prev); + } + assert(!(pc->p->exec_tail->inst[1] & 2)); + /* set the end-bit */ + pc->p->exec_tail->inst[1] |= 1; + + FREE(bra_list); +} + static boolean nv50_program_tx(struct nv50_program *p) { struct tgsi_parse_context parse; struct nv50_pc *pc; - unsigned k; boolean ret; pc = CALLOC_STRUCT(nv50_pc); @@ -2405,48 +2631,10 @@ nv50_program_tx(struct nv50_program *p) } } - if (p->type == PIPE_SHADER_FRAGMENT) { - struct nv50_reg out; - ctor_reg(&out, P_TEMP, -1, -1); - - for (k = 0; k < pc->result_nr * 4; k++) { - if (pc->result[k].rhw == -1) - continue; - if (pc->result[k].hw != pc->result[k].rhw) { - out.hw = pc->result[k].rhw; - emit_mov(pc, &out, &pc->result[k]); - } - if (pc->p->cfg.high_result < (pc->result[k].rhw + 1)) - pc->p->cfg.high_result = pc->result[k].rhw + 1; - } - } - - /* look for single half instructions and make them long */ - struct nv50_program_exec *e, *e_prev; + if (pc->p->type == PIPE_SHADER_FRAGMENT) + nv50_fp_move_results(pc); - for (k = 0, e = pc->p->exec_head, e_prev = NULL; e; e = e->next) { - if (!is_long(e)) - k++; - - if (!e->next || is_long(e->next)) { - if (k & 1) - convert_to_long(pc, e); - k = 0; - } - - if (e->next) - e_prev = e; - } - - if (!is_long(pc->p->exec_tail)) { - /* this may occur if moving FP results */ - assert(e_prev && !is_long(e_prev)); - convert_to_long(pc, e_prev); - convert_to_long(pc, pc->p->exec_tail); - } - - assert(is_long(pc->p->exec_tail) && !is_immd(pc->p->exec_head)); - pc->p->exec_tail->inst[1] |= 0x00000001; + nv50_program_fixup_insns(pc); p->param_nr = pc->param_nr * 4; p->immd_nr = pc->immd_nr * 4; @@ -2558,6 +2746,17 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p) if (e->param.index < 0) continue; + + if (e->param.mask == 0) { + assert(!(e->param.index & 1)); + /* seem to be 8 byte steps */ + ei = (e->param.index >> 1) + 0 /* START_ID */; + + e->inst[0] &= 0xf0000fff; + e->inst[0] |= ei << 12; + continue; + } + bs = (e->inst[1] >> 22) & 0x07; assert(bs < 2); ei = e->param.shift >> 5; diff --git a/src/gallium/drivers/nv50/nv50_state_validate.c b/src/gallium/drivers/nv50/nv50_state_validate.c index 5a3559ed181..4ed76973c4b 100644 --- a/src/gallium/drivers/nv50/nv50_state_validate.c +++ b/src/gallium/drivers/nv50/nv50_state_validate.c @@ -312,7 +312,7 @@ scissor_uptodate: goto viewport_uptodate; nv50->state.viewport_bypass = bypass; - so = so_new(12, 0); + so = so_new(14, 0); if (!bypass) { so_method(so, tesla, NV50TCL_VIEWPORT_TRANSLATE(0), 3); so_data (so, fui(nv50->viewport.translate[0])); @@ -325,12 +325,21 @@ scissor_uptodate: so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1); so_data (so, 1); + /* 0x0000 = remove whole primitive only (xyz) + * 0x1018 = remove whole primitive only (xy), clamp z + * 0x1080 = clip primitive (xyz) + * 0x1098 = clip primitive (xy), clamp z + */ + so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1); + so_data (so, 0x1080); /* no idea what 0f90 does */ so_method(so, tesla, 0x0f90, 1); so_data (so, 0); } else { so_method(so, tesla, NV50TCL_VIEWPORT_TRANSFORM_EN, 1); so_data (so, 0); + so_method(so, tesla, NV50TCL_VIEW_VOLUME_CLIP_CTRL, 1); + so_data (so, 0x0000); so_method(so, tesla, 0x0f90, 1); so_data (so, 1); } diff --git a/src/gallium/drivers/nv50/nv50_transfer.c b/src/gallium/drivers/nv50/nv50_transfer.c index bb7731855cd..9c289026bbb 100644 --- a/src/gallium/drivers/nv50/nv50_transfer.c +++ b/src/gallium/drivers/nv50/nv50_transfer.c @@ -161,7 +161,7 @@ nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, return NULL; } - if (usage != PIPE_TRANSFER_WRITE) { + if (usage & PIPE_TRANSFER_READ) { nv50_transfer_rect_m2mf(pscreen, mt->base.bo, tx->level_offset, tx->level_pitch, tx->level_tiling, x, y, @@ -183,7 +183,7 @@ nv50_transfer_del(struct pipe_transfer *ptx) struct nv50_transfer *tx = (struct nv50_transfer *)ptx; struct nv50_miptree *mt = nv50_miptree(ptx->texture); - if (ptx->usage != PIPE_TRANSFER_READ) { + if (ptx->usage & PIPE_TRANSFER_WRITE) { struct pipe_screen *pscreen = ptx->texture->screen; nv50_transfer_rect_m2mf(pscreen, tx->bo, 0, tx->base.stride, tx->bo->tile_mode, 0, 0, diff --git a/src/gallium/drivers/r300/r300_cs.h b/src/gallium/drivers/r300/r300_cs.h index 0a7e4703636..883f0a02dc5 100644 --- a/src/gallium/drivers/r300/r300_cs.h +++ b/src/gallium/drivers/r300/r300_cs.h @@ -68,11 +68,17 @@ } while (0) #define OUT_CS(value) do { \ + if (VERY_VERBOSE_CS || VERY_VERBOSE_REGISTERS) { \ + DBG(cs_context_copy, DBG_CS, "r300: writing %08x\n", value); \ + } \ cs_winsys->write_cs_dword(cs_winsys, (value)); \ cs_count--; \ } while (0) #define OUT_CS_32F(value) do { \ + if (VERY_VERBOSE_CS || VERY_VERBOSE_REGISTERS) { \ + DBG(cs_context_copy, DBG_CS, "r300: writing %f\n", value); \ + } \ cs_winsys->write_cs_dword(cs_winsys, fui(value)); \ cs_count--; \ } while (0) @@ -82,8 +88,9 @@ DBG(cs_context_copy, DBG_CS, "r300: writing 0x%08X to register 0x%04X\n", \ value, register); \ assert(register); \ - OUT_CS(CP_PACKET0(register, 0)); \ - OUT_CS(value); \ + cs_winsys->write_cs_dword(cs_winsys, CP_PACKET0(register, 0)); \ + cs_winsys->write_cs_dword(cs_winsys, value); \ + cs_count -= 2; \ } while (0) /* Note: This expects count to be the number of registers, @@ -93,7 +100,8 @@ DBG(cs_context_copy, DBG_CS, "r300: writing register sequence of %d to 0x%04X\n", \ count, register); \ assert(register); \ - OUT_CS(CP_PACKET0(register, ((count) - 1))); \ + cs_winsys->write_cs_dword(cs_winsys, CP_PACKET0((register), ((count) - 1))); \ + cs_count--; \ } while (0) #define OUT_CS_RELOC(bo, offset, rd, wd, flags) do { \ @@ -101,9 +109,9 @@ "domains (%d, %d, %d)\n", \ bo, offset, rd, wd, flags); \ assert(bo); \ - OUT_CS(offset); \ + cs_winsys->write_cs_dword(cs_winsys, offset); \ cs_winsys->write_cs_reloc(cs_winsys, bo, rd, wd, flags); \ - cs_count -= 2; \ + cs_count -= 3; \ } while (0) #define END_CS do { \ @@ -131,24 +139,26 @@ DBG(cs_context_copy, DBG_CS, "r300: writing data sequence of %d to 0x%04X\n", \ count, register); \ assert(register); \ - OUT_CS(CP_PACKET0(register, ((count) - 1)) | RADEON_ONE_REG_WR); \ + cs_winsys->write_cs_dword(cs_winsys, CP_PACKET0((register), ((count) - 1)) | RADEON_ONE_REG_WR); \ + cs_count--; \ } while (0) #define CP_PACKET3(op, count) \ (RADEON_CP_PACKET3 | (op) | ((count) << 16)) #define OUT_CS_PKT3(op, count) do { \ - OUT_CS(CP_PACKET3(op, count)); \ + cs_winsys->write_cs_dword(cs_winsys, CP_PACKET3(op, count)); \ + cs_count--; \ } while (0) #define OUT_CS_INDEX_RELOC(bo, offset, count, rd, wd, flags) do { \ DBG(cs_context_copy, DBG_CS, "r300: writing relocation for index buffer %p," \ "offset %d\n", bo, offset); \ assert(bo); \ - OUT_CS(offset); \ - OUT_CS(count); \ + cs_winsys->write_cs_dword(cs_winsys, offset); \ + cs_winsys->write_cs_dword(cs_winsys, count); \ cs_winsys->write_cs_reloc(cs_winsys, bo, rd, wd, flags); \ - cs_count -= 2; \ + cs_count -= 4; \ } while (0) #endif /* R300_CS_H */ diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c index a1b36ba2ed1..77ce431cdc3 100644 --- a/src/gallium/drivers/r300/r300_emit.c +++ b/src/gallium/drivers/r300/r300_emit.c @@ -381,6 +381,7 @@ void r300_emit_query_end(struct r300_context* r300, OUT_CS_REG_SEQ(R300_ZB_ZPASS_ADDR, 1); OUT_CS_RELOC(r300->oqbo, query->offset + (sizeof(uint32_t) * 0), 0, RADEON_GEM_DOMAIN_GTT, 0); + break; default: debug_printf("r300: Implementation error: Chipset reports %d" " pixel pipes!\n", caps->num_frag_pipes); diff --git a/src/gallium/drivers/r300/r300_fs.c b/src/gallium/drivers/r300/r300_fs.c index a0e848a59ac..546ad545a53 100644 --- a/src/gallium/drivers/r300/r300_fs.c +++ b/src/gallium/drivers/r300/r300_fs.c @@ -126,9 +126,14 @@ void r300_translate_fragment_shader(struct r300_context* r300, /* Invoke the compiler */ r3xx_compile_fragment_program(&compiler); if (compiler.Base.Error) { - /* Todo: Fail gracefully */ - fprintf(stderr, "r300 FP: Compiler error\n"); - abort(); + /* Todo: Fallback to software rendering gracefully? */ + fprintf(stderr, "r300 FP: Compiler error: %s\n", compiler.Base.ErrorMsg); + + if (compiler.is_r500) { + memcpy(compiler.code, &r5xx_passthrough_fragment_shader, sizeof(r5xx_passthrough_fragment_shader)); + } else { + memcpy(compiler.code, &r3xx_passthrough_fragment_shader, sizeof(r3xx_passthrough_fragment_shader)); + } } /* And, finally... */ diff --git a/src/gallium/drivers/r300/r300_fs.h b/src/gallium/drivers/r300/r300_fs.h index 9fab7894024..967e9f697e9 100644 --- a/src/gallium/drivers/r300/r300_fs.h +++ b/src/gallium/drivers/r300/r300_fs.h @@ -48,4 +48,4 @@ struct r300_fragment_shader { void r300_translate_fragment_shader(struct r300_context* r300, struct r300_fragment_shader* fs); - #endif /* R300_FS_H */ +#endif /* R300_FS_H */ diff --git a/src/gallium/drivers/r300/r300_reg.h b/src/gallium/drivers/r300/r300_reg.h index 03cd219cde9..3abff5db622 100644 --- a/src/gallium/drivers/r300/r300_reg.h +++ b/src/gallium/drivers/r300/r300_reg.h @@ -1478,6 +1478,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. # define R300_TX_PITCH_EN (1 << 31) # define R300_TX_WIDTH(x) ((x) << 0) # define R300_TX_HEIGHT(x) ((x) << 11) +# define R300_TX_DEPTH(x) ((x) << 22) # define R300_TX_NUM_LEVELS(x) ((x) << 26) #define R300_TX_FORMAT1_0 0x44C0 diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c index 737396d8d97..ca44e0f6615 100644 --- a/src/gallium/drivers/r300/r300_render.c +++ b/src/gallium/drivers/r300/r300_render.c @@ -79,12 +79,14 @@ static boolean r300_render_allocate_vertices(struct vbuf_render* render, struct pipe_screen* screen = r300->context.screen; size_t size = (size_t)vertex_size * (size_t)count; - if (size + r300render->vbo_offset > r300render->vbo_size) + if (size + r300render->vbo_offset > r300render->vbo_size) { + pipe_buffer_reference(&r300->vbo, NULL); r300render->vbo = pipe_buffer_create(screen, 64, PIPE_BUFFER_USAGE_VERTEX, R300_MAX_VBO_SIZE); + r300render->vbo_offset = 0; r300render->vbo_size = R300_MAX_VBO_SIZE; } @@ -117,7 +119,7 @@ static void r300_render_unmap_vertices(struct vbuf_render* render, OUT_CS_REG(R300_VAP_VF_MAX_VTX_INDX, max); END_CS; - r300render->vbo_max_used = MAX2(r300render->vbo_max_used, + r300render->vbo_max_used = MAX2(r300render->vbo_max_used, r300render->vertex_size * (max + 1)); pipe_buffer_unmap(screen, r300render->vbo); } @@ -129,7 +131,6 @@ static void r300_render_release_vertices(struct vbuf_render* render) r300render->vbo_offset += r300render->vbo_max_used; r300render->vbo_max_used = 0; - r300->vbo = NULL; } static boolean r300_render_set_primitive(struct vbuf_render* render, @@ -222,13 +223,6 @@ static void r300_render_draw(struct vbuf_render* render, r300_prepare_render(r300render, count); - /* Send our indices into an index buffer. */ - index_buffer = pipe_buffer_create(screen, 64, PIPE_BUFFER_USAGE_VERTEX, - count * 2); - if (!index_buffer) { - return; - } - BEGIN_CS(2 + (count+1)/2); OUT_CS_PKT3(R300_PACKET3_3D_DRAW_INDX_2, (count+1)/2); OUT_CS(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (count << 16) | diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c index 3b5b1bbd37f..f2659ca61f7 100644 --- a/src/gallium/drivers/r300/r300_screen.c +++ b/src/gallium/drivers/r300/r300_screen.c @@ -101,11 +101,9 @@ static int r300_get_param(struct pipe_screen* pscreen, int param) case PIPE_CAP_MAX_RENDER_TARGETS: return 4; case PIPE_CAP_OCCLUSION_QUERY: - /* IN THEORY */ - return 0; + return 1; case PIPE_CAP_TEXTURE_SHADOW_MAP: - /* IN THEORY */ - return 0; + return 1; case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: if (r300screen->caps->is_r500) { /* 13 == 4096x4096 */ @@ -347,16 +345,9 @@ static void* r300_transfer_map(struct pipe_screen* screen, { struct r300_texture* tex = (struct r300_texture*)transfer->texture; char* map; - unsigned flags = 0; - - if (transfer->usage != PIPE_TRANSFER_WRITE) { - flags |= PIPE_BUFFER_USAGE_CPU_READ; - } - if (transfer->usage != PIPE_TRANSFER_READ) { - flags |= PIPE_BUFFER_USAGE_CPU_WRITE; - } - map = pipe_buffer_map(screen, tex->buffer, flags); + map = pipe_buffer_map(screen, tex->buffer, + pipe_transfer_buffer_flags(transfer)); if (!map) { return NULL; diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c index 5f6b225d340..02b7ab91076 100644 --- a/src/gallium/drivers/r300/r300_state_derived.c +++ b/src/gallium/drivers/r300/r300_state_derived.c @@ -53,7 +53,7 @@ static void r300_vs_tab_routes(struct r300_context* r300, if (!r300screen->caps->has_tcl || !r300->rs_state->enable_vte) { for (i = 0; i < info->num_inputs; i++) { - switch (info->input_semantic_name[i]) { + switch (r300->vs->code.inputs[i]) { case TGSI_SEMANTIC_POSITION: pos = TRUE; tab[i] = 0; @@ -63,10 +63,12 @@ static void r300_vs_tab_routes(struct r300_context* r300, cols++; break; case TGSI_SEMANTIC_PSIZE: + assert(psize == FALSE); psize = TRUE; tab[i] = 15; break; case TGSI_SEMANTIC_FOG: + assert(fog == FALSE); fog = TRUE; /* Fall through */ case TGSI_SEMANTIC_GENERIC: @@ -125,7 +127,9 @@ static void r300_vs_tab_routes(struct r300_context* r300, vinfo->hwfmt[0] = 0x5555; /* XXX this is classic Mesa bonghits */ - if (!pos) { + /* We need to add vertex position attribute only for SW TCL case, + * for HW TCL case it could be generated by vertex shader */ + if (!pos && !r300screen->caps->has_tcl) { debug_printf("r300: Forcing vertex position attribute emit...\n"); /* Make room for the position attribute * at the beginning of the tab. */ @@ -164,7 +168,7 @@ static void r300_vs_tab_routes(struct r300_context* r300, vinfo->hwfmt[3] |= (4 << (3 * i)); } - for (i; i < texs; i++) { + for (; i < texs; i++) { draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, draw_find_vs_output(r300->draw, TGSI_SEMANTIC_GENERIC, i)); vinfo->hwfmt[1] |= (R300_INPUT_CNTL_TC0 << i); @@ -334,48 +338,37 @@ static void r300_update_rs_block(struct r300_context* r300) struct r300_rs_block* rs = r300->rs_block; struct tgsi_shader_info* info = &r300->fs->info; int* tab = r300->vertex_info.fs_tab; - int col_count = 0, fp_offset = 0, i, memory_pos, tex_count = 0; - + int col_count = 0, fp_offset = 0, i, tex_count = 0; + int rs_tex_comp = 0; memset(rs, 0, sizeof(struct r300_rs_block)); if (r300_screen(r300->context.screen)->caps->is_r500) { for (i = 0; i < info->num_inputs; i++) { assert(tab[i] != -1); - memory_pos = tab[i] * 4; switch (info->input_semantic_name[i]) { case TGSI_SEMANTIC_COLOR: rs->ip[col_count] |= - R500_RS_COL_PTR(memory_pos) | + R500_RS_COL_PTR(col_count) | R500_RS_COL_FMT(R300_RS_COL_FMT_RGBA); col_count++; break; case TGSI_SEMANTIC_GENERIC: rs->ip[tex_count] |= - R500_RS_SEL_S(memory_pos) | - R500_RS_SEL_T(memory_pos + 1) | - R500_RS_SEL_R(memory_pos + 2) | - R500_RS_SEL_Q(memory_pos + 3); + R500_RS_SEL_S(rs_tex_comp) | + R500_RS_SEL_T(rs_tex_comp + 1) | + R500_RS_SEL_R(rs_tex_comp + 2) | + R500_RS_SEL_Q(rs_tex_comp + 3); tex_count++; + rs_tex_comp += 4; break; default: break; } } - if (col_count == 0) { - rs->ip[0] |= R500_RS_COL_FMT(R300_RS_COL_FMT_0001); - } - - if (tex_count == 0) { - rs->ip[0] |= - R500_RS_SEL_S(R500_RS_IP_PTR_K0) | - R500_RS_SEL_T(R500_RS_IP_PTR_K0) | - R500_RS_SEL_R(R500_RS_IP_PTR_K0) | - R500_RS_SEL_Q(R500_RS_IP_PTR_K1); - } - /* Rasterize at least one color, or bad things happen. */ if ((col_count == 0) && (tex_count == 0)) { + rs->ip[0] |= R500_RS_COL_FMT(R300_RS_COL_FMT_0001); col_count++; } @@ -393,22 +386,22 @@ static void r300_update_rs_block(struct r300_context* r300) } else { for (i = 0; i < info->num_inputs; i++) { assert(tab[i] != -1); - memory_pos = tab[i] * 4; switch (info->input_semantic_name[i]) { case TGSI_SEMANTIC_COLOR: rs->ip[col_count] |= - R300_RS_COL_PTR(memory_pos) | + R300_RS_COL_PTR(col_count) | R300_RS_COL_FMT(R300_RS_COL_FMT_RGBA); col_count++; break; case TGSI_SEMANTIC_GENERIC: rs->ip[tex_count] |= - R300_RS_TEX_PTR(memory_pos) | + R300_RS_TEX_PTR(rs_tex_comp) | R300_RS_SEL_S(R300_RS_SEL_C0) | R300_RS_SEL_T(R300_RS_SEL_C1) | R300_RS_SEL_R(R300_RS_SEL_C2) | R300_RS_SEL_Q(R300_RS_SEL_C3); tex_count++; + rs_tex_comp+=4; break; default: break; @@ -445,7 +438,7 @@ static void r300_update_rs_block(struct r300_context* r300) } } - rs->count = (tex_count * 4) | (col_count << R300_IC_COUNT_SHIFT) | + rs->count = (rs_tex_comp) | (col_count << R300_IC_COUNT_SHIFT) | R300_HIRES_EN; rs->inst_count = MAX2(MAX2(col_count - 1, tex_count - 1), 0); diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c index 6e8c3683200..ce60ded7caf 100644 --- a/src/gallium/drivers/r300/r300_texture.c +++ b/src/gallium/drivers/r300/r300_texture.c @@ -22,34 +22,39 @@ #include "r300_texture.h" -static void r300_setup_texture_state(struct r300_texture* tex, - unsigned width, - unsigned height, - unsigned levels) +static void r300_setup_texture_state(struct r300_texture* tex) { struct r300_texture_state* state = &tex->state; + struct pipe_texture *pt = &tex->tex; - state->format0 = R300_TX_WIDTH((width - 1) & 0x7ff) | - R300_TX_HEIGHT((height - 1) & 0x7ff) | - R300_TX_NUM_LEVELS(levels) | + state->format0 = R300_TX_WIDTH((pt->width[0] - 1) & 0x7ff) | + R300_TX_HEIGHT((pt->height[0] - 1) & 0x7ff) | + R300_TX_DEPTH(util_logbase2(pt->depth[0]) & 0xf) | + R300_TX_NUM_LEVELS(pt->last_level) | R300_TX_PITCH_EN; /* XXX */ - state->format1 = r300_translate_texformat(tex->tex.format); + state->format1 = r300_translate_texformat(pt->format); + if (pt->target == PIPE_TEXTURE_CUBE) { + state->format1 |= R300_TX_FORMAT_CUBIC_MAP; + } + if (pt->target == PIPE_TEXTURE_3D) { + state->format1 |= R300_TX_FORMAT_3D; + } - state->format2 = r300_texture_get_stride(tex, 0); + state->format2 = (r300_texture_get_stride(tex, 0) / pt->block.size) - 1; /* Assume (somewhat foolishly) that oversized textures will * not be permitted by the state tracker. */ - if (width > 2048) { + if (pt->width[0] > 2048) { state->format2 |= R500_TXWIDTH_BIT11; } - if (height > 2048) { + if (pt->height[0] > 2048) { state->format2 |= R500_TXHEIGHT_BIT11; } - debug_printf("r300: Set texture state (%dx%d, pitch %d, %d levels)\n", - width, height, levels); + debug_printf("r300: Set texture state (%dx%d, %d levels)\n", + pt->width[0], pt->height[0], pt->last_level); } /** @@ -62,7 +67,7 @@ unsigned r300_texture_get_stride(struct r300_texture* tex, unsigned level) return tex->stride_override; if (level > tex->tex.last_level) { - debug_printf("%s: level (%u) > last_level (%u)\n", level, tex->tex.last_level); + debug_printf("%s: level (%u) > last_level (%u)\n", __FUNCTION__, level, tex->tex.last_level); return 0; } @@ -120,8 +125,7 @@ static struct pipe_texture* r300_setup_miptree(tex); - r300_setup_texture_state(tex, template->width[0], template->height[0], - template->last_level); + r300_setup_texture_state(tex); tex->buffer = screen->buffer_create(screen, 1024, PIPE_BUFFER_USAGE_PIXEL, @@ -204,7 +208,7 @@ static struct pipe_texture* tex->stride_override = *stride; /* XXX */ - r300_setup_texture_state(tex, tex->tex.width[0], tex->tex.height[0], 0); + r300_setup_texture_state(tex); pipe_buffer_reference(&tex->buffer, buffer); diff --git a/src/gallium/drivers/r300/r300_texture.h b/src/gallium/drivers/r300/r300_texture.h index 3109af5baca..78ee0f1611a 100644 --- a/src/gallium/drivers/r300/r300_texture.h +++ b/src/gallium/drivers/r300/r300_texture.h @@ -72,6 +72,11 @@ static INLINE uint32_t r300_translate_texformat(enum pipe_format format) /* W24_FP */ case PIPE_FORMAT_Z24S8_UNORM: return R300_EASY_TX_FORMAT(X, X, X, X, W24_FP); + /* Z5_Y6_X5 */ + case PIPE_FORMAT_R16_SNORM: + return R300_EASY_TX_FORMAT(X, X, X, X, Z5Y6X5); + case PIPE_FORMAT_Z16_UNORM: + return R300_EASY_TX_FORMAT(X, X, X, X, X16); default: debug_printf("r300: Implementation error: " "Got unsupported texture format %s in %s\n", diff --git a/src/gallium/drivers/softpipe/Makefile b/src/gallium/drivers/softpipe/Makefile index 516e3992fdd..bcb887a0b26 100644 --- a/src/gallium/drivers/softpipe/Makefile +++ b/src/gallium/drivers/softpipe/Makefile @@ -6,26 +6,17 @@ LIBNAME = softpipe C_SOURCES = \ sp_fs_exec.c \ sp_fs_sse.c \ - sp_fs_llvm.c \ sp_clear.c \ sp_flush.c \ sp_query.c \ sp_context.c \ sp_draw_arrays.c \ - sp_prim_setup.c \ sp_prim_vbuf.c \ sp_quad_pipe.c \ - sp_quad_alpha_test.c \ - sp_quad_blend.c \ - sp_quad_colormask.c \ - sp_quad_coverage.c \ + sp_quad_stipple.c \ sp_quad_depth_test.c \ - sp_quad_earlyz.c \ sp_quad_fs.c \ - sp_quad_occlusion.c \ - sp_quad_output.c \ - sp_quad_stencil.c \ - sp_quad_stipple.c \ + sp_quad_blend.c \ sp_screen.c \ sp_setup.c \ sp_state_blend.c \ @@ -38,7 +29,9 @@ C_SOURCES = \ sp_state_vertex.c \ sp_texture.c \ sp_tex_sample.c \ + sp_tex_tile_cache.c \ sp_tile_cache.c \ - sp_surface.c + sp_surface.c \ + sp_video_context.c include ../../Makefile.template diff --git a/src/gallium/drivers/softpipe/SConscript b/src/gallium/drivers/softpipe/SConscript index f8720638a76..aac9edf44e6 100644 --- a/src/gallium/drivers/softpipe/SConscript +++ b/src/gallium/drivers/softpipe/SConscript @@ -7,25 +7,16 @@ softpipe = env.ConvenienceLibrary( source = [ 'sp_fs_exec.c', 'sp_fs_sse.c', - 'sp_fs_llvm.c', 'sp_clear.c', 'sp_context.c', 'sp_draw_arrays.c', 'sp_flush.c', - 'sp_prim_setup.c', 'sp_prim_vbuf.c', 'sp_setup.c', - 'sp_quad_alpha_test.c', 'sp_quad_blend.c', 'sp_quad_pipe.c', - 'sp_quad_colormask.c', - 'sp_quad_coverage.c', 'sp_quad_depth_test.c', - 'sp_quad_earlyz.c', 'sp_quad_fs.c', - 'sp_quad_occlusion.c', - 'sp_quad_output.c', - 'sp_quad_stencil.c', 'sp_quad_stipple.c', 'sp_query.c', 'sp_screen.c', @@ -39,8 +30,10 @@ softpipe = env.ConvenienceLibrary( 'sp_state_vertex.c', 'sp_surface.c', 'sp_tex_sample.c', + 'sp_tex_tile_cache.c', 'sp_texture.c', 'sp_tile_cache.c', + 'sp_video_context.c', ]) -Export('softpipe')
\ No newline at end of file +Export('softpipe') diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c index c699c433e9c..94d000a5acc 100644 --- a/src/gallium/drivers/softpipe/sp_context.c +++ b/src/gallium/drivers/softpipe/sp_context.c @@ -31,17 +31,18 @@ */ #include "draw/draw_context.h" +#include "draw/draw_vbuf.h" #include "pipe/p_defines.h" #include "util/u_math.h" #include "util/u_memory.h" #include "sp_clear.h" #include "sp_context.h" #include "sp_flush.h" -#include "sp_prim_setup.h" #include "sp_prim_vbuf.h" #include "sp_state.h" #include "sp_surface.h" #include "sp_tile_cache.h" +#include "sp_tex_tile_cache.h" #include "sp_texture.h" #include "sp_winsys.h" #include "sp_query.h" @@ -72,13 +73,10 @@ softpipe_unmap_transfers(struct softpipe_context *sp) { uint i; - for (i = 0; i < sp->framebuffer.nr_cbufs; i++) - sp_flush_tile_cache(sp, sp->cbuf_cache[i]); - sp_flush_tile_cache(sp, sp->zsbuf_cache); - for (i = 0; i < sp->framebuffer.nr_cbufs; i++) { sp_tile_cache_unmap_transfers(sp->cbuf_cache[i]); } + sp_tile_cache_unmap_transfers(sp->zsbuf_cache); } @@ -92,26 +90,21 @@ softpipe_destroy( struct pipe_context *pipe ) if (softpipe->draw) draw_destroy( softpipe->draw ); - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - softpipe->quad[i].polygon_stipple->destroy( softpipe->quad[i].polygon_stipple ); - softpipe->quad[i].earlyz->destroy( softpipe->quad[i].earlyz ); - softpipe->quad[i].shade->destroy( softpipe->quad[i].shade ); - softpipe->quad[i].alpha_test->destroy( softpipe->quad[i].alpha_test ); - softpipe->quad[i].depth_test->destroy( softpipe->quad[i].depth_test ); - softpipe->quad[i].stencil_test->destroy( softpipe->quad[i].stencil_test ); - softpipe->quad[i].occlusion->destroy( softpipe->quad[i].occlusion ); - softpipe->quad[i].coverage->destroy( softpipe->quad[i].coverage ); - softpipe->quad[i].blend->destroy( softpipe->quad[i].blend ); - softpipe->quad[i].colormask->destroy( softpipe->quad[i].colormask ); - softpipe->quad[i].output->destroy( softpipe->quad[i].output ); - } + softpipe->quad.shade->destroy( softpipe->quad.shade ); + softpipe->quad.depth_test->destroy( softpipe->quad.depth_test ); + softpipe->quad.blend->destroy( softpipe->quad.blend ); - for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) + for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) { sp_destroy_tile_cache(softpipe->cbuf_cache[i]); + pipe_surface_reference(&softpipe->framebuffer.cbufs[i], NULL); + } sp_destroy_tile_cache(softpipe->zsbuf_cache); + pipe_surface_reference(&softpipe->framebuffer.zsbuf, NULL); - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) - sp_destroy_tile_cache(softpipe->tex_cache[i]); + for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { + sp_destroy_tex_tile_cache(softpipe->tex_cache[i]); + pipe_texture_reference(&softpipe->texture[i], NULL); + } for (i = 0; i < Elements(softpipe->constants); i++) { if (softpipe->constants[i].buffer) { @@ -246,41 +239,14 @@ softpipe_create( struct pipe_screen *screen ) softpipe->zsbuf_cache = sp_create_tile_cache( screen ); for (i = 0; i < PIPE_MAX_SAMPLERS; i++) - softpipe->tex_cache[i] = sp_create_tile_cache( screen ); + softpipe->tex_cache[i] = sp_create_tex_tile_cache( screen ); /* setup quad rendering stages */ - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - softpipe->quad[i].polygon_stipple = sp_quad_polygon_stipple_stage(softpipe); - softpipe->quad[i].earlyz = sp_quad_earlyz_stage(softpipe); - softpipe->quad[i].shade = sp_quad_shade_stage(softpipe); - softpipe->quad[i].alpha_test = sp_quad_alpha_test_stage(softpipe); - softpipe->quad[i].depth_test = sp_quad_depth_test_stage(softpipe); - softpipe->quad[i].stencil_test = sp_quad_stencil_test_stage(softpipe); - softpipe->quad[i].occlusion = sp_quad_occlusion_stage(softpipe); - softpipe->quad[i].coverage = sp_quad_coverage_stage(softpipe); - softpipe->quad[i].blend = sp_quad_blend_stage(softpipe); - softpipe->quad[i].colormask = sp_quad_colormask_stage(softpipe); - softpipe->quad[i].output = sp_quad_output_stage(softpipe); - } + softpipe->quad.shade = sp_quad_shade_stage(softpipe); + softpipe->quad.depth_test = sp_quad_depth_test_stage(softpipe); + softpipe->quad.blend = sp_quad_blend_stage(softpipe); - /* vertex shader samplers */ - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - softpipe->tgsi.vert_samplers[i].base.get_samples = sp_get_samples_vertex; - softpipe->tgsi.vert_samplers[i].unit = i; - softpipe->tgsi.vert_samplers[i].sp = softpipe; - softpipe->tgsi.vert_samplers[i].cache = softpipe->tex_cache[i]; - softpipe->tgsi.vert_samplers_list[i] = &softpipe->tgsi.vert_samplers[i]; - } - - /* fragment shader samplers */ - for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { - softpipe->tgsi.frag_samplers[i].base.get_samples = sp_get_samples_fragment; - softpipe->tgsi.frag_samplers[i].unit = i; - softpipe->tgsi.frag_samplers[i].sp = softpipe; - softpipe->tgsi.frag_samplers[i].cache = softpipe->tex_cache[i]; - softpipe->tgsi.frag_samplers_list[i] = &softpipe->tgsi.frag_samplers[i]; - } /* * Create drawing context and plug our rendering stage into it. @@ -294,30 +260,28 @@ softpipe_create( struct pipe_screen *screen ) (struct tgsi_sampler **) softpipe->tgsi.vert_samplers_list); - softpipe->setup = sp_draw_render_stage(softpipe); - if (!softpipe->setup) - goto fail; - if (debug_get_bool_option( "SP_NO_RAST", FALSE )) softpipe->no_rast = TRUE; - if (debug_get_bool_option( "SP_NO_VBUF", FALSE )) { - /* Deprecated path -- vbuf is the intended interface to the draw module: - */ - draw_set_rasterize_stage(softpipe->draw, softpipe->setup); - } - else { - sp_init_vbuf(softpipe); - } + softpipe->vbuf_backend = sp_create_vbuf_backend(softpipe); + if (!softpipe->vbuf_backend) + goto fail; + + softpipe->vbuf = draw_vbuf_stage(softpipe->draw, softpipe->vbuf_backend); + if (!softpipe->vbuf) + goto fail; + + draw_set_rasterize_stage(softpipe->draw, softpipe->vbuf); + draw_set_render(softpipe->draw, softpipe->vbuf_backend); + + /* plug in AA line/point stages */ draw_install_aaline_stage(softpipe->draw, &softpipe->pipe); draw_install_aapoint_stage(softpipe->draw, &softpipe->pipe); -#if USE_DRAW_STAGE_PSTIPPLE /* Do polygon stipple w/ texture map + frag prog? */ draw_install_pstipple_stage(softpipe->draw, &softpipe->pipe); -#endif sp_init_surface_functions(softpipe); diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h index 7888c2f644b..43a195c8ef5 100644 --- a/src/gallium/drivers/softpipe/sp_context.h +++ b/src/gallium/drivers/softpipe/sp_context.h @@ -36,24 +36,13 @@ #include "draw/draw_vertex.h" #include "sp_quad_pipe.h" -#include "sp_tex_sample.h" -/** - * This is a temporary variable for testing draw-stage polygon stipple. - * If zero, do stipple in sp_quad_stipple.c - */ -#define USE_DRAW_STAGE_PSTIPPLE 1 - -/* Number of threads working on individual quads. - * Setting to 1 disables this feature. - */ -#define SP_NUM_QUAD_THREADS 1 - struct softpipe_vbuf_render; struct draw_context; struct draw_stage; struct softpipe_tile_cache; +struct softpipe_tex_tile_cache; struct sp_fragment_shader; struct sp_vertex_shader; @@ -62,12 +51,12 @@ struct softpipe_context { struct pipe_context pipe; /**< base class */ /** Constant state objects */ - const struct pipe_blend_state *blend; - const struct pipe_sampler_state *sampler[PIPE_MAX_SAMPLERS]; - const struct pipe_depth_stencil_alpha_state *depth_stencil; - const struct pipe_rasterizer_state *rasterizer; - const struct sp_fragment_shader *fs; - const struct sp_vertex_shader *vs; + struct pipe_blend_state *blend; + struct pipe_sampler_state *sampler[PIPE_MAX_SAMPLERS]; + struct pipe_depth_stencil_alpha_state *depth_stencil; + struct pipe_rasterizer_state *rasterizer; + struct sp_fragment_shader *fs; + struct sp_vertex_shader *vs; /** Other rendering state */ struct pipe_blend_color blend_color; @@ -107,7 +96,16 @@ struct softpipe_context { /** Which vertex shader output slot contains point size */ int psize_slot; - unsigned reduced_api_prim; /**< PIPE_PRIM_POINTS, _LINES or _TRIANGLES */ + /* The reduced version of the primitive supplied by the state + * tracker. + */ + unsigned reduced_api_prim; + + /* The reduced primitive after unfilled triangles, wide-line + * decomposition, etc, are taken into account. This is the + * primitive actually rasterized. + */ + unsigned reduced_prim; /** Derived from scissor and surface bounds: */ struct pipe_scissor_state cliprect; @@ -116,41 +114,33 @@ struct softpipe_context { /** Software quad rendering pipeline */ struct { - struct quad_stage *polygon_stipple; - struct quad_stage *earlyz; struct quad_stage *shade; - struct quad_stage *alpha_test; - struct quad_stage *stencil_test; struct quad_stage *depth_test; - struct quad_stage *occlusion; - struct quad_stage *coverage; struct quad_stage *blend; - struct quad_stage *colormask; - struct quad_stage *output; struct quad_stage *first; /**< points to one of the above stages */ - } quad[SP_NUM_QUAD_THREADS]; + } quad; /** TGSI exec things */ struct { - struct sp_shader_sampler vert_samplers[PIPE_MAX_SAMPLERS]; - struct sp_shader_sampler *vert_samplers_list[PIPE_MAX_SAMPLERS]; - struct sp_shader_sampler frag_samplers[PIPE_MAX_SAMPLERS]; - struct sp_shader_sampler *frag_samplers_list[PIPE_MAX_SAMPLERS]; + struct sp_sampler_varient *vert_samplers_list[PIPE_MAX_SAMPLERS]; + struct sp_sampler_varient *frag_samplers_list[PIPE_MAX_SAMPLERS]; } tgsi; /** The primitive drawing context */ struct draw_context *draw; - struct draw_stage *setup; + + /** Draw module backend */ + struct vbuf_render *vbuf_backend; struct draw_stage *vbuf; - struct softpipe_vbuf_render *vbuf_render; boolean dirty_render_cache; struct softpipe_tile_cache *cbuf_cache[PIPE_MAX_COLOR_BUFS]; struct softpipe_tile_cache *zsbuf_cache; - - struct softpipe_tile_cache *tex_cache[PIPE_MAX_SAMPLERS]; + + unsigned tex_timestamp; + struct softpipe_tex_tile_cache *tex_cache[PIPE_MAX_SAMPLERS]; unsigned use_sse : 1; unsigned dump_fs : 1; @@ -164,5 +154,9 @@ softpipe_context( struct pipe_context *pipe ) return (struct softpipe_context *)pipe; } +void +softpipe_reset_sampler_varients(struct softpipe_context *softpipe); + + #endif /* SP_CONTEXT_H */ diff --git a/src/gallium/drivers/softpipe/sp_flush.c b/src/gallium/drivers/softpipe/sp_flush.c index 4a14d49686e..e38b767cf2c 100644 --- a/src/gallium/drivers/softpipe/sp_flush.c +++ b/src/gallium/drivers/softpipe/sp_flush.c @@ -37,6 +37,7 @@ #include "sp_surface.h" #include "sp_state.h" #include "sp_tile_cache.h" +#include "sp_tex_tile_cache.h" #include "sp_winsys.h" @@ -52,17 +53,19 @@ softpipe_flush( struct pipe_context *pipe, if (flags & PIPE_FLUSH_TEXTURE_CACHE) { for (i = 0; i < softpipe->num_textures; i++) { - sp_flush_tile_cache(softpipe, softpipe->tex_cache[i]); + sp_flush_tex_tile_cache(softpipe->tex_cache[i]); } } - if (flags & PIPE_FLUSH_RENDER_CACHE) { + if (flags & PIPE_FLUSH_SWAPBUFFERS) { + /* If this is a swapbuffers, just flush color buffers. + * + * The zbuffer changes are not discarded, but held in the cache + * in the hope that a later clear will wipe them out. + */ for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) if (softpipe->cbuf_cache[i]) - sp_flush_tile_cache(softpipe, softpipe->cbuf_cache[i]); - - if (softpipe->zsbuf_cache) - sp_flush_tile_cache(softpipe, softpipe->zsbuf_cache); + sp_flush_tile_cache(softpipe->cbuf_cache[i]); /* Need this call for hardware buffers before swapbuffers. * @@ -71,7 +74,15 @@ softpipe_flush( struct pipe_context *pipe, * to unmap surfaces when flushing. */ softpipe_unmap_transfers(softpipe); - + } + else if (flags & PIPE_FLUSH_RENDER_CACHE) { + for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) + if (softpipe->cbuf_cache[i]) + sp_flush_tile_cache(softpipe->cbuf_cache[i]); + + if (softpipe->zsbuf_cache) + sp_flush_tile_cache(softpipe->zsbuf_cache); + softpipe->dirty_render_cache = FALSE; } diff --git a/src/gallium/drivers/softpipe/sp_fs_exec.c b/src/gallium/drivers/softpipe/sp_fs_exec.c index 9ee86fe7878..4076114d392 100644 --- a/src/gallium/drivers/softpipe/sp_fs_exec.c +++ b/src/gallium/drivers/softpipe/sp_fs_exec.c @@ -59,15 +59,34 @@ sp_exec_fragment_shader(const struct sp_fragment_shader *base) } +static void +exec_prepare( const struct sp_fragment_shader *base, + struct tgsi_exec_machine *machine, + struct tgsi_sampler **samplers ) +{ + /* + * Bind tokens/shader to the interpreter's machine state. + * Avoid redundant binding. + */ + if (machine->Tokens != base->shader.tokens) { + tgsi_exec_machine_bind_shader( machine, + base->shader.tokens, + PIPE_MAX_SAMPLERS, + samplers ); + } +} + + + /** * Compute quad X,Y,Z,W for the four fragments in a quad. * * This should really be part of the compiled shader. */ -void -sp_setup_pos_vector(const struct tgsi_interp_coef *coef, - float x, float y, - struct tgsi_exec_vector *quadpos) +static void +setup_pos_vector(const struct tgsi_interp_coef *coef, + float x, float y, + struct tgsi_exec_vector *quadpos) { uint chan; /* do X */ @@ -95,24 +114,6 @@ sp_setup_pos_vector(const struct tgsi_interp_coef *coef, } -static void -exec_prepare( const struct sp_fragment_shader *base, - struct tgsi_exec_machine *machine, - struct tgsi_sampler **samplers ) -{ - /* - * Bind tokens/shader to the interpreter's machine state. - * Avoid redundant binding. - */ - if (machine->Tokens != base->shader.tokens) { - tgsi_exec_machine_bind_shader( machine, - base->shader.tokens, - PIPE_MAX_SAMPLERS, - samplers ); - } -} - - /* TODO: hide the machine struct in here somewhere, remove from this * interface: */ @@ -122,11 +123,43 @@ exec_run( const struct sp_fragment_shader *base, struct quad_header *quad ) { /* Compute X, Y, Z, W vals for this quad */ - sp_setup_pos_vector(quad->posCoef, - (float)quad->input.x0, (float)quad->input.y0, - &machine->QuadPos); + setup_pos_vector(quad->posCoef, + (float)quad->input.x0, (float)quad->input.y0, + &machine->QuadPos); - return tgsi_exec_machine_run( machine ); + quad->inout.mask &= tgsi_exec_machine_run( machine ); + if (quad->inout.mask == 0) + return FALSE; + + /* store outputs */ + { + const ubyte *sem_name = base->info.output_semantic_name; + const ubyte *sem_index = base->info.output_semantic_index; + const uint n = base->info.num_outputs; + uint i; + for (i = 0; i < n; i++) { + switch (sem_name[i]) { + case TGSI_SEMANTIC_COLOR: + { + uint cbuf = sem_index[i]; + memcpy(quad->output.color[cbuf], + &machine->Outputs[i].xyzw[0].f[0], + sizeof(quad->output.color[0]) ); + } + break; + case TGSI_SEMANTIC_POSITION: + { + uint j; + for (j = 0; j < 4; j++) { + quad->output.depth[j] = machine->Outputs[i].xyzw[2].f[j]; + } + } + break; + } + } + } + + return TRUE; } diff --git a/src/gallium/drivers/softpipe/sp_fs_llvm.c b/src/gallium/drivers/softpipe/sp_fs_llvm.c deleted file mode 100644 index 95c0d982d12..00000000000 --- a/src/gallium/drivers/softpipe/sp_fs_llvm.c +++ /dev/null @@ -1,205 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/** - * Execute fragment shader using LLVM code generation. - * Authors: - * Zack Rusin - */ - -#include "sp_context.h" -#include "sp_state.h" -#include "sp_fs.h" - -#include "pipe/p_state.h" -#include "pipe/p_defines.h" -#include "util/u_memory.h" -#include "tgsi/tgsi_sse2.h" - -#if 0 - -/** - * Subclass of sp_fragment_shader - */ -struct sp_llvm_fragment_shader -{ - struct sp_fragment_shader base; - struct gallivm_prog *llvm_prog; -}; - - -static void -shade_quad_llvm(struct quad_stage *qs, - struct quad_header *quad) -{ - struct quad_shade_stage *qss = quad_shade_stage(qs); - struct softpipe_context *softpipe = qs->softpipe; - float dests[4][16][4] ALIGN16_ATTRIB; - float inputs[4][16][4] ALIGN16_ATTRIB; - const float fx = (float) quad->x0; - const float fy = (float) quad->y0; - struct gallivm_prog *llvm = qss->llvm_prog; - - inputs[0][0][0] = fx; - inputs[1][0][0] = fx + 1.0f; - inputs[2][0][0] = fx; - inputs[3][0][0] = fx + 1.0f; - - inputs[0][0][1] = fy; - inputs[1][0][1] = fy; - inputs[2][0][1] = fy + 1.0f; - inputs[3][0][1] = fy + 1.0f; - - - gallivm_prog_inputs_interpolate(llvm, inputs, quad->coef); - -#if DLLVM - debug_printf("MASK = %d\n", quad->mask); - for (int i = 0; i < 4; ++i) { - for (int j = 0; j < 2; ++j) { - debug_printf("IN(%d,%d) [%f %f %f %f]\n", i, j, - inputs[i][j][0], inputs[i][j][1], inputs[i][j][2], inputs[i][j][3]); - } - } -#endif - - quad->mask &= - gallivm_fragment_shader_exec(llvm, fx, fy, dests, inputs, - softpipe->mapped_constants[PIPE_SHADER_FRAGMENT], - qss->samplers); -#if DLLVM - debug_printf("OUT LLVM = 1[%f %f %f %f], 2[%f %f %f %f]\n", - dests[0][0][0], dests[0][0][1], dests[0][0][2], dests[0][0][3], - dests[0][1][0], dests[0][1][1], dests[0][1][2], dests[0][1][3]); -#endif - - /* store result color */ - if (qss->colorOutSlot >= 0) { - unsigned i; - /* XXX need to handle multiple color outputs someday */ - allvmrt(qss->stage.softpipe->fs->info.output_semantic_name[qss->colorOutSlot] - == TGSI_SEMANTIC_COLOR); - for (i = 0; i < QUAD_SIZE; ++i) { - quad->outputs.color[0][0][i] = dests[i][qss->colorOutSlot][0]; - quad->outputs.color[0][1][i] = dests[i][qss->colorOutSlot][1]; - quad->outputs.color[0][2][i] = dests[i][qss->colorOutSlot][2]; - quad->outputs.color[0][3][i] = dests[i][qss->colorOutSlot][3]; - } - } -#if DLLVM - for (int i = 0; i < QUAD_SIZE; ++i) { - debug_printf("QLLVM%d(%d) [%f, %f, %f, %f]\n", i, qss->colorOutSlot, - quad->outputs.color[0][0][i], - quad->outputs.color[0][1][i], - quad->outputs.color[0][2][i], - quad->outputs.color[0][3][i]); - } -#endif - - /* store result Z */ - if (qss->depthOutSlot >= 0) { - /* output[slot] is new Z */ - uint i; - for (i = 0; i < 4; i++) { - quad->outputs.depth[i] = dests[i][0][2]; - } - } - else { - /* copy input Z (which was interpolated by the executor) to output Z */ - uint i; - for (i = 0; i < 4; i++) { - quad->outputs.depth[i] = inputs[i][0][2]; - } - } -#if DLLVM - debug_printf("D [%f, %f, %f, %f] mask = %d\n", - quad->outputs.depth[0], - quad->outputs.depth[1], - quad->outputs.depth[2], - quad->outputs.depth[3], quad->mask); -#endif - - /* shader may cull fragments */ - if( quad->mask ) { - qs->next->run( qs->next, quad ); - } -} - - -unsigned -run_llvm_fs( const struct sp_fragment_shader *base, - struct foo *machine ) -{ -} - - -void -delete_llvm_fs( struct sp_fragment_shader *base ) -{ - FREE(base); -} - - -struct sp_fragment_shader * -softpipe_create_fs_llvm(struct softpipe_context *softpipe, - const struct pipe_shader_state *templ) -{ - struct sp_llvm_fragment_shader *shader = NULL; - - /* LLVM fragment shaders currently disabled: - */ - state = CALLOC_STRUCT(sp_llvm_shader_state); - if (!state) - return NULL; - - state->llvm_prog = 0; - - if (!gallivm_global_cpu_engine()) { - gallivm_cpu_engine_create(state->llvm_prog); - } - else - gallivm_cpu_jit_compile(gallivm_global_cpu_engine(), state->llvm_prog); - - if (shader) { - shader->base.run = run_llvm_fs; - shader->base.delete = delete_llvm_fs; - } - - return shader; -} - - -#else - -struct sp_fragment_shader * -softpipe_create_fs_llvm(struct softpipe_context *softpipe, - const struct pipe_shader_state *templ) -{ - return NULL; -} - -#endif diff --git a/src/gallium/drivers/softpipe/sp_fs_sse.c b/src/gallium/drivers/softpipe/sp_fs_sse.c index f4fa0905d74..9d3e4670eef 100644 --- a/src/gallium/drivers/softpipe/sp_fs_sse.c +++ b/src/gallium/drivers/softpipe/sp_fs_sse.c @@ -76,6 +76,43 @@ fs_sse_prepare( const struct sp_fragment_shader *base, } + +/** + * Compute quad X,Y,Z,W for the four fragments in a quad. + * + * This should really be part of the compiled shader. + */ +static void +setup_pos_vector(const struct tgsi_interp_coef *coef, + float x, float y, + struct tgsi_exec_vector *quadpos) +{ + uint chan; + /* do X */ + quadpos->xyzw[0].f[0] = x; + quadpos->xyzw[0].f[1] = x + 1; + quadpos->xyzw[0].f[2] = x; + quadpos->xyzw[0].f[3] = x + 1; + + /* do Y */ + quadpos->xyzw[1].f[0] = y; + quadpos->xyzw[1].f[1] = y; + quadpos->xyzw[1].f[2] = y + 1; + quadpos->xyzw[1].f[3] = y + 1; + + /* do Z and W for all fragments in the quad */ + for (chan = 2; chan < 4; chan++) { + const float dadx = coef->dadx[chan]; + const float dady = coef->dady[chan]; + const float a0 = coef->a0[chan] + dadx * x + dady * y; + quadpos->xyzw[chan].f[0] = a0; + quadpos->xyzw[chan].f[1] = a0 + dadx; + quadpos->xyzw[chan].f[2] = a0 + dady; + quadpos->xyzw[chan].f[3] = a0 + dadx + dady; + } +} + + /* TODO: codegenerate the whole run function, skip this wrapper. * TODO: break dependency on tgsi_exec_machine struct * TODO: push Position calculation into the generated shader @@ -89,9 +126,9 @@ fs_sse_run( const struct sp_fragment_shader *base, struct sp_sse_fragment_shader *shader = sp_sse_fragment_shader(base); /* Compute X, Y, Z, W vals for this quad -- place in temp[0] for now */ - sp_setup_pos_vector(quad->posCoef, - (float)quad->input.x0, (float)quad->input.y0, - machine->Temps); + setup_pos_vector(quad->posCoef, + (float)quad->input.x0, (float)quad->input.y0, + machine->Temps); /* init kill mask */ tgsi_set_kill_mask(machine, 0x0); @@ -104,7 +141,39 @@ fs_sse_run( const struct sp_fragment_shader *base, // , &machine->QuadPos ); - return ~(machine->Temps[TGSI_EXEC_TEMP_KILMASK_I].xyzw[TGSI_EXEC_TEMP_KILMASK_C].u[0]); + quad->inout.mask &= ~(machine->Temps[TGSI_EXEC_TEMP_KILMASK_I].xyzw[TGSI_EXEC_TEMP_KILMASK_C].u[0]); + if (quad->inout.mask == 0) + return FALSE; + + /* store outputs */ + { + const ubyte *sem_name = base->info.output_semantic_name; + const ubyte *sem_index = base->info.output_semantic_index; + const uint n = base->info.num_outputs; + uint i; + for (i = 0; i < n; i++) { + switch (sem_name[i]) { + case TGSI_SEMANTIC_COLOR: + { + uint cbuf = sem_index[i]; + memcpy(quad->output.color[cbuf], + &machine->Outputs[i].xyzw[0].f[0], + sizeof(quad->output.color[0]) ); + } + break; + case TGSI_SEMANTIC_POSITION: + { + uint j; + for (j = 0; j < 4; j++) { + quad->output.depth[j] = machine->Outputs[0].xyzw[2].f[j]; + } + } + break; + } + } + } + + return TRUE; } diff --git a/src/gallium/drivers/softpipe/sp_prim_setup.c b/src/gallium/drivers/softpipe/sp_prim_setup.c deleted file mode 100644 index 038ff04d4f1..00000000000 --- a/src/gallium/drivers/softpipe/sp_prim_setup.c +++ /dev/null @@ -1,190 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/** - * \brief A draw stage that drives our triangle setup routines from - * within the draw pipeline. One of two ways to drive setup, the - * other being in sp_prim_vbuf.c. - * - * \author Keith Whitwell <[email protected]> - * \author Brian Paul - */ - - -#include "sp_context.h" -#include "sp_setup.h" -#include "sp_state.h" -#include "sp_prim_setup.h" -#include "draw/draw_pipe.h" -#include "draw/draw_vertex.h" -#include "util/u_memory.h" - -/** - * Triangle setup info (derived from draw_stage). - * Also used for line drawing (taking some liberties). - */ -struct setup_stage { - struct draw_stage stage; /**< This must be first (base class) */ - - struct setup_context *setup; -}; - - - -/** - * Basically a cast wrapper. - */ -static INLINE struct setup_stage *setup_stage( struct draw_stage *stage ) -{ - return (struct setup_stage *)stage; -} - - -typedef const float (*cptrf4)[4]; - -static void -do_tri(struct draw_stage *stage, struct prim_header *prim) -{ - struct setup_stage *setup = setup_stage( stage ); - - setup_tri( setup->setup, - (cptrf4)prim->v[0]->data, - (cptrf4)prim->v[1]->data, - (cptrf4)prim->v[2]->data ); -} - -static void -do_line(struct draw_stage *stage, struct prim_header *prim) -{ - struct setup_stage *setup = setup_stage( stage ); - - setup_line( setup->setup, - (cptrf4)prim->v[0]->data, - (cptrf4)prim->v[1]->data ); -} - -static void -do_point(struct draw_stage *stage, struct prim_header *prim) -{ - struct setup_stage *setup = setup_stage( stage ); - - setup_point( setup->setup, - (cptrf4)prim->v[0]->data ); -} - - - - -static void setup_begin( struct draw_stage *stage ) -{ - struct setup_stage *setup = setup_stage(stage); - - setup_prepare( setup->setup ); - - stage->point = do_point; - stage->line = do_line; - stage->tri = do_tri; -} - - -static void setup_first_point( struct draw_stage *stage, - struct prim_header *header ) -{ - setup_begin(stage); - stage->point( stage, header ); -} - -static void setup_first_line( struct draw_stage *stage, - struct prim_header *header ) -{ - setup_begin(stage); - stage->line( stage, header ); -} - - -static void setup_first_tri( struct draw_stage *stage, - struct prim_header *header ) -{ - setup_begin(stage); - stage->tri( stage, header ); -} - - - -static void setup_flush( struct draw_stage *stage, - unsigned flags ) -{ - stage->point = setup_first_point; - stage->line = setup_first_line; - stage->tri = setup_first_tri; -} - - -static void reset_stipple_counter( struct draw_stage *stage ) -{ -} - - -static void render_destroy( struct draw_stage *stage ) -{ - struct setup_stage *ssetup = setup_stage(stage); - setup_destroy_context(ssetup->setup); - FREE( stage ); -} - - -/** - * Create a new primitive setup/render stage. - */ -struct draw_stage *sp_draw_render_stage( struct softpipe_context *softpipe ) -{ - struct setup_stage *sstage = CALLOC_STRUCT(setup_stage); - - sstage->setup = setup_create_context(softpipe); - sstage->stage.draw = softpipe->draw; - sstage->stage.point = setup_first_point; - sstage->stage.line = setup_first_line; - sstage->stage.tri = setup_first_tri; - sstage->stage.flush = setup_flush; - sstage->stage.reset_stipple_counter = reset_stipple_counter; - sstage->stage.destroy = render_destroy; - - return (struct draw_stage *)sstage; -} - -struct setup_context * -sp_draw_setup_context( struct draw_stage *stage ) -{ - struct setup_stage *ssetup = setup_stage(stage); - return ssetup->setup; -} - -void -sp_draw_flush( struct draw_stage *stage ) -{ - stage->flush( stage, 0 ); -} diff --git a/src/gallium/drivers/softpipe/sp_prim_setup.h b/src/gallium/drivers/softpipe/sp_prim_setup.h deleted file mode 100644 index 49bdd98ed87..00000000000 --- a/src/gallium/drivers/softpipe/sp_prim_setup.h +++ /dev/null @@ -1,85 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - -#ifndef SP_PRIM_SETUP_H -#define SP_PRIM_SETUP_H - - -/** - * vbuf is a special stage to gather the stream of triangles, lines, points - * together and reconstruct vertex buffers for hardware upload. - * - * First attempt, work in progress. - * - * TODO: - * - separate out vertex buffer building and primitive emit, ie >1 draw per vb. - * - tell vbuf stage how to build hw vertices directly - * - pass vbuf stage a buffer pointer for direct emit to agp/vram. - * - * - * - * Vertices are just an array of floats, with all the attributes - * packed. We currently assume a layout like: - * - * attr[0][0..3] - window position - * attr[1..n][0..3] - remaining attributes. - * - * Attributes are assumed to be 4 floats wide but are packed so that - * all the enabled attributes run contiguously. - */ - - -struct draw_stage; -struct softpipe_context; - - -typedef void (*vbuf_draw_func)( struct pipe_context *pipe, - unsigned prim, - const ushort *elements, - unsigned nr_elements, - const void *vertex_buffer, - unsigned nr_vertices ); - - -extern struct draw_stage * -sp_draw_render_stage( struct softpipe_context *softpipe ); - -extern struct setup_context * -sp_draw_setup_context( struct draw_stage * ); - -extern void -sp_draw_flush( struct draw_stage * ); - - -extern struct draw_stage * -sp_draw_vbuf_stage( struct draw_context *draw_context, - struct pipe_context *pipe, - vbuf_draw_func draw ); - - -#endif /* SP_PRIM_SETUP_H */ diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.c b/src/gallium/drivers/softpipe/sp_prim_vbuf.c index 42021789ea8..e603c20fc4c 100644 --- a/src/gallium/drivers/softpipe/sp_prim_vbuf.c +++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.c @@ -37,13 +37,13 @@ #include "sp_context.h" +#include "sp_setup.h" #include "sp_state.h" #include "sp_prim_vbuf.h" -#include "sp_prim_setup.h" -#include "sp_setup.h" #include "draw/draw_context.h" #include "draw/draw_vbuf.h" #include "util/u_memory.h" +#include "util/u_prim.h" #define SP_MAX_VBUF_INDEXES 1024 @@ -58,6 +58,8 @@ struct softpipe_vbuf_render { struct vbuf_render base; struct softpipe_context *softpipe; + struct setup_context *setup; + uint prim; uint vertex_size; uint nr_vertices; @@ -74,6 +76,11 @@ softpipe_vbuf_render(struct vbuf_render *vbr) } + + + + + static const struct vertex_info * sp_vbuf_get_vertex_info(struct vbuf_render *vbr) { @@ -104,36 +111,6 @@ sp_vbuf_allocate_vertices(struct vbuf_render *vbr, static void sp_vbuf_release_vertices(struct vbuf_render *vbr) { -#if 0 - { - struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); - const struct vertex_info *info = - softpipe_get_vbuf_vertex_info(cvbr->softpipe); - const float *vtx = (const float *) cvbr->vertex_buffer; - uint i, j; - debug_printf("%s (vtx_size = %u, vtx_used = %u)\n", - __FUNCTION__, cvbr->vertex_size, cvbr->nr_vertices); - for (i = 0; i < cvbr->nr_vertices; i++) { - for (j = 0; j < info->num_attribs; j++) { - uint k; - switch (info->attrib[j].emit) { - case EMIT_4F: k = 4; break; - case EMIT_3F: k = 3; break; - case EMIT_2F: k = 2; break; - case EMIT_1F: k = 1; break; - default: assert(0); - } - debug_printf("Vert %u attr %u: ", i, j); - while (k-- > 0) { - debug_printf("%g ", vtx[0]); - vtx++; - } - debug_printf("\n"); - } - } - } -#endif - /* keep the old allocation for next time */ } @@ -159,14 +136,11 @@ static boolean sp_vbuf_set_primitive(struct vbuf_render *vbr, unsigned prim) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); - - /* XXX: break this dependency - make setup_context live under - * softpipe, rename the old "setup" draw stage to something else. - */ - struct setup_context *setup_ctx = sp_draw_setup_context(cvbr->softpipe->setup); + struct setup_context *setup_ctx = cvbr->setup; setup_prepare( setup_ctx ); + cvbr->softpipe->reduced_prim = u_reduced_prim(prim); cvbr->prim = prim; return TRUE; @@ -191,14 +165,9 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) struct softpipe_context *softpipe = cvbr->softpipe; const unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float); const void *vertex_buffer = cvbr->vertex_buffer; + struct setup_context *setup_ctx = cvbr->setup; unsigned i; - /* XXX: break this dependency - make setup_context live under - * softpipe, rename the old "setup" draw stage to something else. - */ - struct draw_stage *setup = softpipe->setup; - struct setup_context *setup_ctx = sp_draw_setup_context(setup); - switch (cvbr->prim) { case PIPE_PRIM_POINTS: for (i = 0; i < nr; i++) { @@ -237,14 +206,16 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_TRIANGLES: - for (i = 2; i < nr; i += 3) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 3) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-1], stride), get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[i-2], stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 3) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-2], stride), get_vert(vertex_buffer, indices[i-1], stride), @@ -254,14 +225,16 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_TRIANGLE_STRIP: - for (i = 2; i < nr; i += 1) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i+(i&1)-1], stride), get_vert(vertex_buffer, indices[i-(i&1)], stride), get_vert(vertex_buffer, indices[i-2], stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i+(i&1)-2], stride), get_vert(vertex_buffer, indices[i-(i&1)-1], stride), @@ -271,14 +244,16 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_TRIANGLE_FAN: - for (i = 2; i < nr; i += 1) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[0], stride), get_vert(vertex_buffer, indices[i-1], stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[0], stride), get_vert(vertex_buffer, indices[i-1], stride), @@ -288,8 +263,8 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_QUADS: - for (i = 3; i < nr; i += 4) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 3; i < nr; i += 4) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-2], stride), get_vert(vertex_buffer, indices[i-1], stride), @@ -299,7 +274,9 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[i-3], stride) ); } - else { + } + else { + for (i = 3; i < nr; i += 4) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-3], stride), get_vert(vertex_buffer, indices[i-2], stride), @@ -314,8 +291,8 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) break; case PIPE_PRIM_QUAD_STRIP: - for (i = 3; i < nr; i += 2) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 3; i < nr; i += 2) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[i-1], stride), @@ -325,7 +302,9 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) get_vert(vertex_buffer, indices[i-0], stride), get_vert(vertex_buffer, indices[i-3], stride) ); } - else { + } + else { + for (i = 3; i < nr; i += 2) { setup_tri( setup_ctx, get_vert(vertex_buffer, indices[i-3], stride), get_vert(vertex_buffer, indices[i-2], stride), @@ -355,11 +334,6 @@ sp_vbuf_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) default: assert(0); } - - /* XXX: why are we calling this??? If we had to call something, it - * would be a function in sp_setup.c: - */ - sp_draw_flush( setup ); } @@ -372,17 +346,12 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); struct softpipe_context *softpipe = cvbr->softpipe; + struct setup_context *setup_ctx = cvbr->setup; const unsigned stride = softpipe->vertex_info_vbuf.size * sizeof(float); const void *vertex_buffer = (void *) get_vert(cvbr->vertex_buffer, start, stride); unsigned i; - /* XXX: break this dependency - make setup_context live under - * softpipe, rename the old "setup" draw stage to something else. - */ - struct draw_stage *setup = softpipe->setup; - struct setup_context *setup_ctx = sp_draw_setup_context(setup); - switch (cvbr->prim) { case PIPE_PRIM_POINTS: for (i = 0; i < nr; i++) { @@ -421,14 +390,16 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_TRIANGLES: - for (i = 2; i < nr; i += 3) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 3) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-1, stride), get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, i-2, stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 3) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-2, stride), get_vert(vertex_buffer, i-1, stride), @@ -438,14 +409,16 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_TRIANGLE_STRIP: - for (i = 2; i < nr; i++) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i++) { setup_tri( setup_ctx, get_vert(vertex_buffer, i+(i&1)-1, stride), get_vert(vertex_buffer, i-(i&1), stride), get_vert(vertex_buffer, i-2, stride) ); } - else { + } + else { + for (i = 2; i < nr; i++) { setup_tri( setup_ctx, get_vert(vertex_buffer, i+(i&1)-2, stride), get_vert(vertex_buffer, i-(i&1)-1, stride), @@ -455,14 +428,16 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_TRIANGLE_FAN: - for (i = 2; i < nr; i += 1) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, 0, stride), get_vert(vertex_buffer, i-1, stride) ); } - else { + } + else { + for (i = 2; i < nr; i += 1) { setup_tri( setup_ctx, get_vert(vertex_buffer, 0, stride), get_vert(vertex_buffer, i-1, stride), @@ -472,8 +447,8 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_QUADS: - for (i = 3; i < nr; i += 4) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 3; i < nr; i += 4) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-2, stride), get_vert(vertex_buffer, i-1, stride), @@ -483,7 +458,9 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, i-3, stride) ); } - else { + } + else { + for (i = 3; i < nr; i += 4) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-3, stride), get_vert(vertex_buffer, i-2, stride), @@ -497,8 +474,8 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) break; case PIPE_PRIM_QUAD_STRIP: - for (i = 3; i < nr; i += 2) { - if (softpipe->rasterizer->flatshade_first) { + if (softpipe->rasterizer->flatshade_first) { + for (i = 3; i < nr; i += 2) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, i-1, stride), @@ -508,7 +485,9 @@ sp_vbuf_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) get_vert(vertex_buffer, i-0, stride), get_vert(vertex_buffer, i-3, stride) ); } - else { + } + else { + for (i = 3; i < nr; i += 2) { setup_tri( setup_ctx, get_vert(vertex_buffer, i-3, stride), get_vert(vertex_buffer, i-2, stride), @@ -546,40 +525,38 @@ static void sp_vbuf_destroy(struct vbuf_render *vbr) { struct softpipe_vbuf_render *cvbr = softpipe_vbuf_render(vbr); - cvbr->softpipe->vbuf_render = NULL; + setup_destroy_context(cvbr->setup); FREE(cvbr); } /** - * Initialize the post-transform vertex buffer information for the given - * context. + * Create the post-transform vertex handler for the given context. */ -void -sp_init_vbuf(struct softpipe_context *sp) +struct vbuf_render * +sp_create_vbuf_backend(struct softpipe_context *sp) { - assert(sp->draw); + struct softpipe_vbuf_render *cvbr = CALLOC_STRUCT(softpipe_vbuf_render); - sp->vbuf_render = CALLOC_STRUCT(softpipe_vbuf_render); + assert(sp->draw); - sp->vbuf_render->base.max_indices = SP_MAX_VBUF_INDEXES; - sp->vbuf_render->base.max_vertex_buffer_bytes = SP_MAX_VBUF_SIZE; - sp->vbuf_render->base.get_vertex_info = sp_vbuf_get_vertex_info; - sp->vbuf_render->base.allocate_vertices = sp_vbuf_allocate_vertices; - sp->vbuf_render->base.map_vertices = sp_vbuf_map_vertices; - sp->vbuf_render->base.unmap_vertices = sp_vbuf_unmap_vertices; - sp->vbuf_render->base.set_primitive = sp_vbuf_set_primitive; - sp->vbuf_render->base.draw = sp_vbuf_draw; - sp->vbuf_render->base.draw_arrays = sp_vbuf_draw_arrays; - sp->vbuf_render->base.release_vertices = sp_vbuf_release_vertices; - sp->vbuf_render->base.destroy = sp_vbuf_destroy; + cvbr->base.max_indices = SP_MAX_VBUF_INDEXES; + cvbr->base.max_vertex_buffer_bytes = SP_MAX_VBUF_SIZE; - sp->vbuf_render->softpipe = sp; + cvbr->base.get_vertex_info = sp_vbuf_get_vertex_info; + cvbr->base.allocate_vertices = sp_vbuf_allocate_vertices; + cvbr->base.map_vertices = sp_vbuf_map_vertices; + cvbr->base.unmap_vertices = sp_vbuf_unmap_vertices; + cvbr->base.set_primitive = sp_vbuf_set_primitive; + cvbr->base.draw = sp_vbuf_draw; + cvbr->base.draw_arrays = sp_vbuf_draw_arrays; + cvbr->base.release_vertices = sp_vbuf_release_vertices; + cvbr->base.destroy = sp_vbuf_destroy; - sp->vbuf = draw_vbuf_stage(sp->draw, &sp->vbuf_render->base); + cvbr->softpipe = sp; - draw_set_rasterize_stage(sp->draw, sp->vbuf); + cvbr->setup = setup_create_context(cvbr->softpipe); - draw_set_render(sp->draw, &sp->vbuf_render->base); + return &cvbr->base; } diff --git a/src/gallium/drivers/softpipe/sp_prim_vbuf.h b/src/gallium/drivers/softpipe/sp_prim_vbuf.h index 1de9cc2a894..ad01cc2f289 100644 --- a/src/gallium/drivers/softpipe/sp_prim_vbuf.h +++ b/src/gallium/drivers/softpipe/sp_prim_vbuf.h @@ -31,8 +31,8 @@ struct softpipe_context; -extern void -sp_init_vbuf(struct softpipe_context *softpipe); +extern struct vbuf_render * +sp_create_vbuf_backend(struct softpipe_context *softpipe); #endif /* SP_VBUF_H */ diff --git a/src/gallium/drivers/softpipe/sp_quad.h b/src/gallium/drivers/softpipe/sp_quad.h index bd6c6cb9123..a3236bd1169 100644 --- a/src/gallium/drivers/softpipe/sp_quad.h +++ b/src/gallium/drivers/softpipe/sp_quad.h @@ -97,10 +97,10 @@ struct quad_header { struct quad_header_inout inout; struct quad_header_output output; - const struct tgsi_interp_coef *coef; + /* Redundant/duplicated: + */ const struct tgsi_interp_coef *posCoef; - - unsigned nr_attrs; + const struct tgsi_interp_coef *coef; }; #endif /* SP_QUAD_H */ diff --git a/src/gallium/drivers/softpipe/sp_quad_alpha_test.c b/src/gallium/drivers/softpipe/sp_quad_alpha_test.c deleted file mode 100644 index 0845bae0e68..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_alpha_test.c +++ /dev/null @@ -1,108 +0,0 @@ - -/** - * quad alpha test - */ - -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_quad_pipe.h" -#include "pipe/p_defines.h" -#include "util/u_memory.h" - - -static void -alpha_test_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - const float ref = softpipe->depth_stencil->alpha.ref_value; - unsigned passMask = 0x0, j; - const uint cbuf = 0; /* only output[0].alpha is tested */ - const float *aaaa = quad->output.color[cbuf][3]; - - switch (softpipe->depth_stencil->alpha.func) { - case PIPE_FUNC_NEVER: - break; - case PIPE_FUNC_LESS: - /* - * If mask were an array [4] we could do this SIMD-style: - * passMask = (quad->outputs.color[0][3] <= vec4(ref)); - */ - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] < ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_EQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] == ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_LEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] <= ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_GREATER: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] > ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_NOTEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] != ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_GEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (aaaa[j] >= ref) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_ALWAYS: - passMask = MASK_ALL; - break; - default: - assert(0); - } - - quad->inout.mask &= passMask; - - if (quad->inout.mask) - qs->next->run(qs->next, quad); -} - - -static void alpha_test_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void alpha_test_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage * -sp_quad_alpha_test_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = alpha_test_begin; - stage->run = alpha_test_quad; - stage->destroy = alpha_test_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_blend.c b/src/gallium/drivers/softpipe/sp_quad_blend.c index b1e18805c70..e243c63fa23 100644 --- a/src/gallium/drivers/softpipe/sp_quad_blend.c +++ b/src/gallium/drivers/softpipe/sp_quad_blend.c @@ -117,644 +117,865 @@ do { \ static void -logicop_quad(struct quad_stage *qs, struct quad_header *quad) +logicop_quad(struct quad_stage *qs, + float (*quadColor)[4], + float (*dest)[4]) { struct softpipe_context *softpipe = qs->softpipe; - uint cbuf; + ubyte src[4][4], dst[4][4], res[4][4]; + uint *src4 = (uint *) src; + uint *dst4 = (uint *) dst; + uint *res4 = (uint *) res; + uint j; + + + /* convert to ubyte */ + for (j = 0; j < 4; j++) { /* loop over R,G,B,A channels */ + dst[j][0] = float_to_ubyte(dest[j][0]); /* P0 */ + dst[j][1] = float_to_ubyte(dest[j][1]); /* P1 */ + dst[j][2] = float_to_ubyte(dest[j][2]); /* P2 */ + dst[j][3] = float_to_ubyte(dest[j][3]); /* P3 */ + + src[j][0] = float_to_ubyte(quadColor[j][0]); /* P0 */ + src[j][1] = float_to_ubyte(quadColor[j][1]); /* P1 */ + src[j][2] = float_to_ubyte(quadColor[j][2]); /* P2 */ + src[j][3] = float_to_ubyte(quadColor[j][3]); /* P3 */ + } - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - float dest[4][QUAD_SIZE]; - ubyte src[4][4], dst[4][4], res[4][4]; - uint *src4 = (uint *) src; - uint *dst4 = (uint *) dst; - uint *res4 = (uint *) res; - struct softpipe_cached_tile * - tile = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[cbuf], - quad->input.x0, quad->input.y0); - float (*quadColor)[4] = quad->output.color[cbuf]; - uint i, j; + switch (softpipe->blend->logicop_func) { + case PIPE_LOGICOP_CLEAR: + for (j = 0; j < 4; j++) + res4[j] = 0; + break; + case PIPE_LOGICOP_NOR: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] | dst4[j]); + break; + case PIPE_LOGICOP_AND_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j] & dst4[j]; + break; + case PIPE_LOGICOP_COPY_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j]; + break; + case PIPE_LOGICOP_AND_REVERSE: + for (j = 0; j < 4; j++) + res4[j] = src4[j] & ~dst4[j]; + break; + case PIPE_LOGICOP_INVERT: + for (j = 0; j < 4; j++) + res4[j] = ~dst4[j]; + break; + case PIPE_LOGICOP_XOR: + for (j = 0; j < 4; j++) + res4[j] = dst4[j] ^ src4[j]; + break; + case PIPE_LOGICOP_NAND: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] & dst4[j]); + break; + case PIPE_LOGICOP_AND: + for (j = 0; j < 4; j++) + res4[j] = src4[j] & dst4[j]; + break; + case PIPE_LOGICOP_EQUIV: + for (j = 0; j < 4; j++) + res4[j] = ~(src4[j] ^ dst4[j]); + break; + case PIPE_LOGICOP_NOOP: + for (j = 0; j < 4; j++) + res4[j] = dst4[j]; + break; + case PIPE_LOGICOP_OR_INVERTED: + for (j = 0; j < 4; j++) + res4[j] = ~src4[j] | dst4[j]; + break; + case PIPE_LOGICOP_COPY: + for (j = 0; j < 4; j++) + res4[j] = src4[j]; + break; + case PIPE_LOGICOP_OR_REVERSE: + for (j = 0; j < 4; j++) + res4[j] = src4[j] | ~dst4[j]; + break; + case PIPE_LOGICOP_OR: + for (j = 0; j < 4; j++) + res4[j] = src4[j] | dst4[j]; + break; + case PIPE_LOGICOP_SET: + for (j = 0; j < 4; j++) + res4[j] = ~0; + break; + default: + assert(0); + } - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1); - for (i = 0; i < 4; i++) { - dest[i][j] = tile->data.color[y][x][i]; - } - } + for (j = 0; j < 4; j++) { + quadColor[j][0] = ubyte_to_float(res[j][0]); + quadColor[j][1] = ubyte_to_float(res[j][1]); + quadColor[j][2] = ubyte_to_float(res[j][2]); + quadColor[j][3] = ubyte_to_float(res[j][3]); + } +} - /* convert to ubyte */ - for (j = 0; j < 4; j++) { /* loop over R,G,B,A channels */ - dst[j][0] = float_to_ubyte(dest[j][0]); /* P0 */ - dst[j][1] = float_to_ubyte(dest[j][1]); /* P1 */ - dst[j][2] = float_to_ubyte(dest[j][2]); /* P2 */ - dst[j][3] = float_to_ubyte(dest[j][3]); /* P3 */ - - src[j][0] = float_to_ubyte(quadColor[j][0]); /* P0 */ - src[j][1] = float_to_ubyte(quadColor[j][1]); /* P1 */ - src[j][2] = float_to_ubyte(quadColor[j][2]); /* P2 */ - src[j][3] = float_to_ubyte(quadColor[j][3]); /* P3 */ - } - switch (softpipe->blend->logicop_func) { - case PIPE_LOGICOP_CLEAR: - for (j = 0; j < 4; j++) - res4[j] = 0; - break; - case PIPE_LOGICOP_NOR: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] | dst4[j]); - break; - case PIPE_LOGICOP_AND_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j] & dst4[j]; - break; - case PIPE_LOGICOP_COPY_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j]; - break; - case PIPE_LOGICOP_AND_REVERSE: - for (j = 0; j < 4; j++) - res4[j] = src4[j] & ~dst4[j]; - break; - case PIPE_LOGICOP_INVERT: - for (j = 0; j < 4; j++) - res4[j] = ~dst4[j]; - break; - case PIPE_LOGICOP_XOR: - for (j = 0; j < 4; j++) - res4[j] = dst4[j] ^ src4[j]; - break; - case PIPE_LOGICOP_NAND: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] & dst4[j]); - break; - case PIPE_LOGICOP_AND: - for (j = 0; j < 4; j++) - res4[j] = src4[j] & dst4[j]; - break; - case PIPE_LOGICOP_EQUIV: - for (j = 0; j < 4; j++) - res4[j] = ~(src4[j] ^ dst4[j]); - break; - case PIPE_LOGICOP_NOOP: - for (j = 0; j < 4; j++) - res4[j] = dst4[j]; - break; - case PIPE_LOGICOP_OR_INVERTED: - for (j = 0; j < 4; j++) - res4[j] = ~src4[j] | dst4[j]; - break; - case PIPE_LOGICOP_COPY: - for (j = 0; j < 4; j++) - res4[j] = src4[j]; - break; - case PIPE_LOGICOP_OR_REVERSE: - for (j = 0; j < 4; j++) - res4[j] = src4[j] | ~dst4[j]; - break; - case PIPE_LOGICOP_OR: - for (j = 0; j < 4; j++) - res4[j] = src4[j] | dst4[j]; - break; - case PIPE_LOGICOP_SET: - for (j = 0; j < 4; j++) - res4[j] = ~0; - break; - default: - assert(0); - } - for (j = 0; j < 4; j++) { - quadColor[j][0] = ubyte_to_float(res[j][0]); - quadColor[j][1] = ubyte_to_float(res[j][1]); - quadColor[j][2] = ubyte_to_float(res[j][2]); - quadColor[j][3] = ubyte_to_float(res[j][3]); - } +static void +blend_quad(struct quad_stage *qs, + float (*quadColor)[4], + float (*dest)[4]) +{ + static const float zero[4] = { 0, 0, 0, 0 }; + static const float one[4] = { 1, 1, 1, 1 }; + struct softpipe_context *softpipe = qs->softpipe; + float source[4][QUAD_SIZE]; + + /* + * Compute src/first term RGB + */ + switch (softpipe->blend->rgb_src_factor) { + case PIPE_BLENDFACTOR_ONE: + VEC4_COPY(source[0], quadColor[0]); /* R */ + VEC4_COPY(source[1], quadColor[1]); /* G */ + VEC4_COPY(source[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + VEC4_MUL(source[0], quadColor[0], quadColor[0]); /* R */ + VEC4_MUL(source[1], quadColor[1], quadColor[1]); /* G */ + VEC4_MUL(source[2], quadColor[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA: + { + const float *alpha = quadColor[3]; + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_DST_COLOR: + VEC4_MUL(source[0], quadColor[0], dest[0]); /* R */ + VEC4_MUL(source[1], quadColor[1], dest[1]); /* G */ + VEC4_MUL(source[2], quadColor[2], dest[2]); /* B */ + break; + case PIPE_BLENDFACTOR_DST_ALPHA: + { + const float *alpha = dest[3]; + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + { + const float *alpha = quadColor[3]; + float diff[4], temp[4]; + VEC4_SUB(diff, one, dest[3]); + VEC4_MIN(temp, alpha, diff); + VEC4_MUL(source[0], quadColor[0], temp); /* R */ + VEC4_MUL(source[1], quadColor[1], temp); /* G */ + VEC4_MUL(source[2], quadColor[2], temp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], comp); /* R */ + VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], comp); /* G */ + VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float alpha[4]; + VEC4_SCALAR(alpha, softpipe->blend_color.color[3]); + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_SRC1_COLOR: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_SRC1_ALPHA: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(source[0], zero); /* R */ + VEC4_COPY(source[1], zero); /* G */ + VEC4_COPY(source[2], zero); /* B */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, quadColor[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, dest[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[0]); /* R */ + VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, dest[1]); /* G */ + VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, dest[2]); /* B */ + VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + { + float inv_comp[4]; + /* R */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); + VEC4_MUL(source[0], quadColor[0], inv_comp); + /* G */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); + VEC4_MUL(source[1], quadColor[1], inv_comp); + /* B */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); + VEC4_MUL(source[2], quadColor[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_alpha[4]; + VEC4_SCALAR(inv_alpha, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC1_COLOR: + assert(0); /* to do */ + break; + case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: + assert(0); /* to do */ + break; + default: + assert(0); + } + + /* + * Compute src/first term A + */ + switch (softpipe->blend->alpha_src_factor) { + case PIPE_BLENDFACTOR_ONE: + VEC4_COPY(source[3], quadColor[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_SRC_ALPHA: + { + const float *alpha = quadColor[3]; + VEC4_MUL(source[3], quadColor[3], alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(source[3], quadColor[3], dest[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + /* multiply alpha by 1.0 */ + VEC4_COPY(source[3], quadColor[3]); /* A */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(source[3], quadColor[3], comp); /* A */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(source[3], zero); /* A */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, quadColor[3]); + VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_alpha[4]; + VEC4_SUB(inv_alpha, one, dest[3]); + VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + /* A */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(source[3], quadColor[3], inv_comp); + } + break; + default: + assert(0); } - /* pass quad to next stage */ - qs->next->run(qs->next, quad); -} + /* + * Compute dest/second term RGB + */ + switch (softpipe->blend->rgb_dst_factor) { + case PIPE_BLENDFACTOR_ONE: + /* dest = dest * 1 NO-OP, leave dest as-is */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + VEC4_MUL(dest[0], dest[0], quadColor[0]); /* R */ + VEC4_MUL(dest[1], dest[1], quadColor[1]); /* G */ + VEC4_MUL(dest[2], dest[2], quadColor[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA: + VEC4_MUL(dest[0], dest[0], quadColor[3]); /* R * A */ + VEC4_MUL(dest[1], dest[1], quadColor[3]); /* G * A */ + VEC4_MUL(dest[2], dest[2], quadColor[3]); /* B * A */ + break; + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(dest[0], dest[0], dest[3]); /* R * A */ + VEC4_MUL(dest[1], dest[1], dest[3]); /* G * A */ + VEC4_MUL(dest[2], dest[2], dest[3]); /* B * A */ + break; + case PIPE_BLENDFACTOR_DST_COLOR: + VEC4_MUL(dest[0], dest[0], dest[0]); /* R */ + VEC4_MUL(dest[1], dest[1], dest[1]); /* G */ + VEC4_MUL(dest[2], dest[2], dest[2]); /* B */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + assert(0); /* illegal */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ + VEC4_MUL(dest[0], dest[0], comp); /* R */ + VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ + VEC4_MUL(dest[1], dest[1], comp); /* G */ + VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ + VEC4_MUL(dest[2], dest[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(dest[0], dest[0], comp); /* R */ + VEC4_MUL(dest[1], dest[1], comp); /* G */ + VEC4_MUL(dest[2], dest[2], comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(dest[0], zero); /* R */ + VEC4_COPY(dest[1], zero); /* G */ + VEC4_COPY(dest[2], zero); /* B */ + break; + case PIPE_BLENDFACTOR_SRC1_COLOR: + case PIPE_BLENDFACTOR_SRC1_ALPHA: + /* XXX what are these? */ + assert(0); + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ + VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ + VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ + VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ + VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ + VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float one_minus_alpha[QUAD_SIZE]; + VEC4_SUB(one_minus_alpha, one, quadColor[3]); + VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */ + VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */ + VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[3]); /* A */ + VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ + VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ + VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[0]); /* R */ + VEC4_MUL(dest[0], dest[0], inv_comp); /* R */ + VEC4_SUB(inv_comp, one, dest[1]); /* G */ + VEC4_MUL(dest[1], dest[1], inv_comp); /* G */ + VEC4_SUB(inv_comp, one, dest[2]); /* B */ + VEC4_MUL(dest[2], dest[2], inv_comp); /* B */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + { + float inv_comp[4]; + /* R */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); + VEC4_MUL(dest[0], dest[0], inv_comp); + /* G */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); + VEC4_MUL(dest[1], dest[1], inv_comp); + /* B */ + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); + VEC4_MUL(dest[2], dest[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(dest[0], dest[0], inv_comp); + VEC4_MUL(dest[1], dest[1], inv_comp); + VEC4_MUL(dest[2], dest[2], inv_comp); + } + break; + case PIPE_BLENDFACTOR_INV_SRC1_COLOR: + case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: + /* XXX what are these? */ + assert(0); + break; + default: + assert(0); + } + /* + * Compute dest/second term A + */ + switch (softpipe->blend->alpha_dst_factor) { + case PIPE_BLENDFACTOR_ONE: + /* dest = dest * 1 NO-OP, leave dest as-is */ + break; + case PIPE_BLENDFACTOR_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_SRC_ALPHA: + VEC4_MUL(dest[3], dest[3], quadColor[3]); /* A * A */ + break; + case PIPE_BLENDFACTOR_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_DST_ALPHA: + VEC4_MUL(dest[3], dest[3], dest[3]); /* A */ + break; + case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: + assert(0); /* illegal */ + break; + case PIPE_BLENDFACTOR_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_CONST_ALPHA: + { + float comp[4]; + VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ + VEC4_MUL(dest[3], dest[3], comp); /* A */ + } + break; + case PIPE_BLENDFACTOR_ZERO: + VEC4_COPY(dest[3], zero); /* A */ + break; + case PIPE_BLENDFACTOR_INV_SRC_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_SRC_ALPHA: + { + float one_minus_alpha[QUAD_SIZE]; + VEC4_SUB(one_minus_alpha, one, quadColor[3]); + VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_DST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_DST_ALPHA: + { + float inv_comp[4]; + VEC4_SUB(inv_comp, one, dest[3]); /* A */ + VEC4_MUL(dest[3], inv_comp, dest[3]); /* A */ + } + break; + case PIPE_BLENDFACTOR_INV_CONST_COLOR: + /* fall-through */ + case PIPE_BLENDFACTOR_INV_CONST_ALPHA: + { + float inv_comp[4]; + VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); + VEC4_MUL(dest[3], dest[3], inv_comp); + } + break; + default: + assert(0); + } + /* + * Combine RGB terms + */ + switch (softpipe->blend->rgb_func) { + case PIPE_BLEND_ADD: + VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */ + VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */ + VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_SUBTRACT: + VEC4_SUB_SAT(quadColor[0], source[0], dest[0]); /* R */ + VEC4_SUB_SAT(quadColor[1], source[1], dest[1]); /* G */ + VEC4_SUB_SAT(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_REVERSE_SUBTRACT: + VEC4_SUB_SAT(quadColor[0], dest[0], source[0]); /* R */ + VEC4_SUB_SAT(quadColor[1], dest[1], source[1]); /* G */ + VEC4_SUB_SAT(quadColor[2], dest[2], source[2]); /* B */ + break; + case PIPE_BLEND_MIN: + VEC4_MIN(quadColor[0], source[0], dest[0]); /* R */ + VEC4_MIN(quadColor[1], source[1], dest[1]); /* G */ + VEC4_MIN(quadColor[2], source[2], dest[2]); /* B */ + break; + case PIPE_BLEND_MAX: + VEC4_MAX(quadColor[0], source[0], dest[0]); /* R */ + VEC4_MAX(quadColor[1], source[1], dest[1]); /* G */ + VEC4_MAX(quadColor[2], source[2], dest[2]); /* B */ + break; + default: + assert(0); + } + + /* + * Combine A terms + */ + switch (softpipe->blend->alpha_func) { + case PIPE_BLEND_ADD: + VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_SUBTRACT: + VEC4_SUB_SAT(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_REVERSE_SUBTRACT: + VEC4_SUB_SAT(quadColor[3], dest[3], source[3]); /* A */ + break; + case PIPE_BLEND_MIN: + VEC4_MIN(quadColor[3], source[3], dest[3]); /* A */ + break; + case PIPE_BLEND_MAX: + VEC4_MAX(quadColor[3], source[3], dest[3]); /* A */ + break; + default: + assert(0); + } +} static void -blend_quad(struct quad_stage *qs, struct quad_header *quad) +colormask_quad(struct quad_stage *qs, + float (*quadColor)[4], + float (*dest)[4]) { - static const float zero[4] = { 0, 0, 0, 0 }; - static const float one[4] = { 1, 1, 1, 1 }; + struct softpipe_context *softpipe = qs->softpipe; + /* R */ + if (!(softpipe->blend->colormask & PIPE_MASK_R)) + COPY_4V(quadColor[0], dest[0]); + + /* G */ + if (!(softpipe->blend->colormask & PIPE_MASK_G)) + COPY_4V(quadColor[1], dest[1]); + + /* B */ + if (!(softpipe->blend->colormask & PIPE_MASK_B)) + COPY_4V(quadColor[2], dest[2]); + + /* A */ + if (!(softpipe->blend->colormask & PIPE_MASK_A)) + COPY_4V(quadColor[3], dest[3]); +} + + +static void +blend_fallback(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ struct softpipe_context *softpipe = qs->softpipe; - uint cbuf; + const struct pipe_blend_state *blend = softpipe->blend; + unsigned cbuf; + + for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) + { + float dest[4][QUAD_SIZE]; + struct softpipe_cached_tile *tile + = sp_get_cached_tile(softpipe->cbuf_cache[cbuf], + quads[0]->input.x0, + quads[0]->input.y0); + uint q, i, j; + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; + float (*quadColor)[4] = quad->output.color[cbuf]; + const int itx = (quad->input.x0 & (TILE_SIZE-1)); + const int ity = (quad->input.y0 & (TILE_SIZE-1)); + + /* get/swizzle dest colors + */ + for (j = 0; j < QUAD_SIZE; j++) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { + dest[i][j] = tile->data.color[y][x][i]; + } + } - if (softpipe->blend->logicop_enable) { - logicop_quad(qs, quad); - return; + + if (blend->logicop_enable) { + logicop_quad( qs, quadColor, dest ); + } + else if (blend->blend_enable) { + blend_quad( qs, quadColor, dest ); + } + + if (blend->colormask != 0xf) + colormask_quad( qs, quadColor, dest ); + + /* Output color values + */ + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->inout.mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } + } + } + } } +} - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - float source[4][QUAD_SIZE], dest[4][QUAD_SIZE]; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[cbuf], - quad->input.x0, quad->input.y0); - float (*quadColor)[4] = quad->output.color[cbuf]; - uint i, j; +static void +blend_single_add_src_alpha_inv_src_alpha(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + static const float one[4] = { 1, 1, 1, 1 }; + float one_minus_alpha[QUAD_SIZE]; + float dest[4][QUAD_SIZE]; + float source[4][QUAD_SIZE]; + uint i, j, q; + + struct softpipe_cached_tile *tile + = sp_get_cached_tile(qs->softpipe->cbuf_cache[0], + quads[0]->input.x0, + quads[0]->input.y0); + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; + float (*quadColor)[4] = quad->output.color[0]; + const float *alpha = quadColor[3]; + const int itx = (quad->input.x0 & (TILE_SIZE-1)); + const int ity = (quad->input.y0 & (TILE_SIZE-1)); + /* get/swizzle dest colors */ for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1); + int x = itx + (j & 1); + int y = ity + (j >> 1); for (i = 0; i < 4; i++) { dest[i][j] = tile->data.color[y][x][i]; } } - /* - * Compute src/first term RGB - */ - switch (softpipe->blend->rgb_src_factor) { - case PIPE_BLENDFACTOR_ONE: - VEC4_COPY(source[0], quadColor[0]); /* R */ - VEC4_COPY(source[1], quadColor[1]); /* G */ - VEC4_COPY(source[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - VEC4_MUL(source[0], quadColor[0], quadColor[0]); /* R */ - VEC4_MUL(source[1], quadColor[1], quadColor[1]); /* G */ - VEC4_MUL(source[2], quadColor[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA: - { - const float *alpha = quadColor[3]; - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_DST_COLOR: - VEC4_MUL(source[0], quadColor[0], dest[0]); /* R */ - VEC4_MUL(source[1], quadColor[1], dest[1]); /* G */ - VEC4_MUL(source[2], quadColor[2], dest[2]); /* B */ - break; - case PIPE_BLENDFACTOR_DST_ALPHA: - { - const float *alpha = dest[3]; - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - { - const float *alpha = quadColor[3]; - float diff[4], temp[4]; - VEC4_SUB(diff, one, dest[3]); - VEC4_MIN(temp, alpha, diff); - VEC4_MUL(source[0], quadColor[0], temp); /* R */ - VEC4_MUL(source[1], quadColor[1], temp); /* G */ - VEC4_MUL(source[2], quadColor[2], temp); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], comp); /* R */ - VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], comp); /* G */ - VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float alpha[4]; - VEC4_SCALAR(alpha, softpipe->blend_color.color[3]); - VEC4_MUL(source[0], quadColor[0], alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_SRC1_COLOR: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_SRC1_ALPHA: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(source[0], zero); /* R */ - VEC4_COPY(source[1], zero); /* G */ - VEC4_COPY(source[2], zero); /* B */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, quadColor[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, dest[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[0]); /* R */ - VEC4_MUL(source[0], quadColor[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, dest[1]); /* G */ - VEC4_MUL(source[1], quadColor[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, dest[2]); /* B */ - VEC4_MUL(source[2], quadColor[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - { - float inv_comp[4]; - /* R */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); - VEC4_MUL(source[0], quadColor[0], inv_comp); - /* G */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); - VEC4_MUL(source[1], quadColor[1], inv_comp); - /* B */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); - VEC4_MUL(source[2], quadColor[2], inv_comp); - } - break; - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_alpha[4]; - VEC4_SCALAR(inv_alpha, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(source[0], quadColor[0], inv_alpha); /* R */ - VEC4_MUL(source[1], quadColor[1], inv_alpha); /* G */ - VEC4_MUL(source[2], quadColor[2], inv_alpha); /* B */ + VEC4_MUL(source[0], quadColor[0], alpha); /* R */ + VEC4_MUL(source[1], quadColor[1], alpha); /* G */ + VEC4_MUL(source[2], quadColor[2], alpha); /* B */ + VEC4_MUL(source[3], quadColor[3], alpha); /* A */ + + VEC4_SUB(one_minus_alpha, one, alpha); + VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */ + VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */ + VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */ + VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* B */ + + VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */ + VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */ + VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */ + VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */ + + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->inout.mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } } - break; - case PIPE_BLENDFACTOR_INV_SRC1_COLOR: - assert(0); /* to do */ - break; - case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: - assert(0); /* to do */ - break; - default: - assert(0); } + } +} - /* - * Compute src/first term A - */ - switch (softpipe->blend->alpha_src_factor) { - case PIPE_BLENDFACTOR_ONE: - VEC4_COPY(source[3], quadColor[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_SRC_ALPHA: - { - const float *alpha = quadColor[3]; - VEC4_MUL(source[3], quadColor[3], alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(source[3], quadColor[3], dest[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - /* multiply alpha by 1.0 */ - VEC4_COPY(source[3], quadColor[3]); /* A */ - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(source[3], quadColor[3], comp); /* A */ - } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(source[3], zero); /* A */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, quadColor[3]); - VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_alpha[4]; - VEC4_SUB(inv_alpha, one, dest[3]); - VEC4_MUL(source[3], quadColor[3], inv_alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - /* A */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(source[3], quadColor[3], inv_comp); +static void +blend_single_add_one_one(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + float dest[4][QUAD_SIZE]; + uint i, j, q; + + struct softpipe_cached_tile *tile + = sp_get_cached_tile(qs->softpipe->cbuf_cache[0], + quads[0]->input.x0, + quads[0]->input.y0); + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; + float (*quadColor)[4] = quad->output.color[0]; + const int itx = (quad->input.x0 & (TILE_SIZE-1)); + const int ity = (quad->input.y0 & (TILE_SIZE-1)); + + /* get/swizzle dest colors */ + for (j = 0; j < QUAD_SIZE; j++) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { + dest[i][j] = tile->data.color[y][x][i]; } - break; - default: - assert(0); } + + VEC4_ADD_SAT(quadColor[0], quadColor[0], dest[0]); /* R */ + VEC4_ADD_SAT(quadColor[1], quadColor[1], dest[1]); /* G */ + VEC4_ADD_SAT(quadColor[2], quadColor[2], dest[2]); /* B */ + VEC4_ADD_SAT(quadColor[3], quadColor[3], dest[3]); /* A */ - - /* - * Compute dest/second term RGB - */ - switch (softpipe->blend->rgb_dst_factor) { - case PIPE_BLENDFACTOR_ONE: - /* dest = dest * 1 NO-OP, leave dest as-is */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - VEC4_MUL(dest[0], dest[0], quadColor[0]); /* R */ - VEC4_MUL(dest[1], dest[1], quadColor[1]); /* G */ - VEC4_MUL(dest[2], dest[2], quadColor[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA: - VEC4_MUL(dest[0], dest[0], quadColor[3]); /* R * A */ - VEC4_MUL(dest[1], dest[1], quadColor[3]); /* G * A */ - VEC4_MUL(dest[2], dest[2], quadColor[3]); /* B * A */ - break; - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(dest[0], dest[0], dest[3]); /* R * A */ - VEC4_MUL(dest[1], dest[1], dest[3]); /* G * A */ - VEC4_MUL(dest[2], dest[2], dest[3]); /* B * A */ - break; - case PIPE_BLENDFACTOR_DST_COLOR: - VEC4_MUL(dest[0], dest[0], dest[0]); /* R */ - VEC4_MUL(dest[1], dest[1], dest[1]); /* G */ - VEC4_MUL(dest[2], dest[2], dest[2]); /* B */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - assert(0); /* illegal */ - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[0]); /* R */ - VEC4_MUL(dest[0], dest[0], comp); /* R */ - VEC4_SCALAR(comp, softpipe->blend_color.color[1]); /* G */ - VEC4_MUL(dest[1], dest[1], comp); /* G */ - VEC4_SCALAR(comp, softpipe->blend_color.color[2]); /* B */ - VEC4_MUL(dest[2], dest[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(dest[0], dest[0], comp); /* R */ - VEC4_MUL(dest[1], dest[1], comp); /* G */ - VEC4_MUL(dest[2], dest[2], comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(dest[0], zero); /* R */ - VEC4_COPY(dest[1], zero); /* G */ - VEC4_COPY(dest[2], zero); /* B */ - break; - case PIPE_BLENDFACTOR_SRC1_COLOR: - case PIPE_BLENDFACTOR_SRC1_ALPHA: - /* XXX what are these? */ - assert(0); - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, quadColor[0]); /* R */ - VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ - VEC4_SUB(inv_comp, one, quadColor[1]); /* G */ - VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ - VEC4_SUB(inv_comp, one, quadColor[2]); /* B */ - VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float one_minus_alpha[QUAD_SIZE]; - VEC4_SUB(one_minus_alpha, one, quadColor[3]); - VEC4_MUL(dest[0], dest[0], one_minus_alpha); /* R */ - VEC4_MUL(dest[1], dest[1], one_minus_alpha); /* G */ - VEC4_MUL(dest[2], dest[2], one_minus_alpha); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[3]); /* A */ - VEC4_MUL(dest[0], inv_comp, dest[0]); /* R */ - VEC4_MUL(dest[1], inv_comp, dest[1]); /* G */ - VEC4_MUL(dest[2], inv_comp, dest[2]); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[0]); /* R */ - VEC4_MUL(dest[0], dest[0], inv_comp); /* R */ - VEC4_SUB(inv_comp, one, dest[1]); /* G */ - VEC4_MUL(dest[1], dest[1], inv_comp); /* G */ - VEC4_SUB(inv_comp, one, dest[2]); /* B */ - VEC4_MUL(dest[2], dest[2], inv_comp); /* B */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - { - float inv_comp[4]; - /* R */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[0]); - VEC4_MUL(dest[0], dest[0], inv_comp); - /* G */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[1]); - VEC4_MUL(dest[1], dest[1], inv_comp); - /* B */ - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[2]); - VEC4_MUL(dest[2], dest[2], inv_comp); - } - break; - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(dest[0], dest[0], inv_comp); - VEC4_MUL(dest[1], dest[1], inv_comp); - VEC4_MUL(dest[2], dest[2], inv_comp); + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->inout.mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } } - break; - case PIPE_BLENDFACTOR_INV_SRC1_COLOR: - case PIPE_BLENDFACTOR_INV_SRC1_ALPHA: - /* XXX what are these? */ - assert(0); - break; - default: - assert(0); } + } +} - /* - * Compute dest/second term A - */ - switch (softpipe->blend->alpha_dst_factor) { - case PIPE_BLENDFACTOR_ONE: - /* dest = dest * 1 NO-OP, leave dest as-is */ - break; - case PIPE_BLENDFACTOR_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_SRC_ALPHA: - VEC4_MUL(dest[3], dest[3], quadColor[3]); /* A * A */ - break; - case PIPE_BLENDFACTOR_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_DST_ALPHA: - VEC4_MUL(dest[3], dest[3], dest[3]); /* A */ - break; - case PIPE_BLENDFACTOR_SRC_ALPHA_SATURATE: - assert(0); /* illegal */ - break; - case PIPE_BLENDFACTOR_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_CONST_ALPHA: - { - float comp[4]; - VEC4_SCALAR(comp, softpipe->blend_color.color[3]); /* A */ - VEC4_MUL(dest[3], dest[3], comp); /* A */ - } - break; - case PIPE_BLENDFACTOR_ZERO: - VEC4_COPY(dest[3], zero); /* A */ - break; - case PIPE_BLENDFACTOR_INV_SRC_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_SRC_ALPHA: - { - float one_minus_alpha[QUAD_SIZE]; - VEC4_SUB(one_minus_alpha, one, quadColor[3]); - VEC4_MUL(dest[3], dest[3], one_minus_alpha); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_DST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_DST_ALPHA: - { - float inv_comp[4]; - VEC4_SUB(inv_comp, one, dest[3]); /* A */ - VEC4_MUL(dest[3], inv_comp, dest[3]); /* A */ - } - break; - case PIPE_BLENDFACTOR_INV_CONST_COLOR: - /* fall-through */ - case PIPE_BLENDFACTOR_INV_CONST_ALPHA: - { - float inv_comp[4]; - VEC4_SCALAR(inv_comp, 1.0f - softpipe->blend_color.color[3]); - VEC4_MUL(dest[3], dest[3], inv_comp); + +static void +single_output_color(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + uint i, j, q; + + struct softpipe_cached_tile *tile + = sp_get_cached_tile(qs->softpipe->cbuf_cache[0], + quads[0]->input.x0, + quads[0]->input.y0); + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; + float (*quadColor)[4] = quad->output.color[0]; + const int itx = (quad->input.x0 & (TILE_SIZE-1)); + const int ity = (quad->input.y0 & (TILE_SIZE-1)); + + for (j = 0; j < QUAD_SIZE; j++) { + if (quad->inout.mask & (1 << j)) { + int x = itx + (j & 1); + int y = ity + (j >> 1); + for (i = 0; i < 4; i++) { /* loop over color chans */ + tile->data.color[y][x][i] = quadColor[i][j]; + } } - break; - default: - assert(0); } + } +} + +static void +blend_noop(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ +} - /* - * Combine RGB terms - */ - switch (softpipe->blend->rgb_func) { - case PIPE_BLEND_ADD: - VEC4_ADD_SAT(quadColor[0], source[0], dest[0]); /* R */ - VEC4_ADD_SAT(quadColor[1], source[1], dest[1]); /* G */ - VEC4_ADD_SAT(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_SUBTRACT: - VEC4_SUB_SAT(quadColor[0], source[0], dest[0]); /* R */ - VEC4_SUB_SAT(quadColor[1], source[1], dest[1]); /* G */ - VEC4_SUB_SAT(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_REVERSE_SUBTRACT: - VEC4_SUB_SAT(quadColor[0], dest[0], source[0]); /* R */ - VEC4_SUB_SAT(quadColor[1], dest[1], source[1]); /* G */ - VEC4_SUB_SAT(quadColor[2], dest[2], source[2]); /* B */ - break; - case PIPE_BLEND_MIN: - VEC4_MIN(quadColor[0], source[0], dest[0]); /* R */ - VEC4_MIN(quadColor[1], source[1], dest[1]); /* G */ - VEC4_MIN(quadColor[2], source[2], dest[2]); /* B */ - break; - case PIPE_BLEND_MAX: - VEC4_MAX(quadColor[0], source[0], dest[0]); /* R */ - VEC4_MAX(quadColor[1], source[1], dest[1]); /* G */ - VEC4_MAX(quadColor[2], source[2], dest[2]); /* B */ - break; - default: - assert(0); - } - /* - * Combine A terms - */ - switch (softpipe->blend->alpha_func) { - case PIPE_BLEND_ADD: - VEC4_ADD_SAT(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_SUBTRACT: - VEC4_SUB_SAT(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_REVERSE_SUBTRACT: - VEC4_SUB_SAT(quadColor[3], dest[3], source[3]); /* A */ - break; - case PIPE_BLEND_MIN: - VEC4_MIN(quadColor[3], source[3], dest[3]); /* A */ - break; - case PIPE_BLEND_MAX: - VEC4_MAX(quadColor[3], source[3], dest[3]); /* A */ - break; - default: - assert(0); +static void +choose_blend_quad(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + struct softpipe_context *softpipe = qs->softpipe; + const struct pipe_blend_state *blend = softpipe->blend; + + qs->run = blend_fallback; + + if (softpipe->framebuffer.nr_cbufs == 0) { + qs->run = blend_noop; + } + else if (!softpipe->blend->logicop_enable && + softpipe->blend->colormask == 0xf) + { + if (!blend->blend_enable) { + qs->run = single_output_color; } + else if (blend->rgb_src_factor == blend->alpha_src_factor && + blend->rgb_dst_factor == blend->alpha_dst_factor && + blend->rgb_func == blend->alpha_func && + softpipe->framebuffer.nr_cbufs == 1) + { + if (blend->alpha_func == PIPE_BLEND_ADD) { + if (blend->rgb_src_factor == PIPE_BLENDFACTOR_ONE && + blend->rgb_dst_factor == PIPE_BLENDFACTOR_ONE) { + qs->run = blend_single_add_one_one; + } + else if (blend->rgb_src_factor == PIPE_BLENDFACTOR_SRC_ALPHA && + blend->rgb_dst_factor == PIPE_BLENDFACTOR_INV_SRC_ALPHA) + qs->run = blend_single_add_src_alpha_inv_src_alpha; - } /* cbuf loop */ + } + } + } - /* pass blended quad to next stage */ - qs->next->run(qs->next, quad); + qs->run(qs, quads, nr); } static void blend_begin(struct quad_stage *qs) { - qs->next->begin(qs->next); + qs->run = choose_blend_quad; } @@ -770,7 +991,7 @@ struct quad_stage *sp_quad_blend_stage( struct softpipe_context *softpipe ) stage->softpipe = softpipe; stage->begin = blend_begin; - stage->run = blend_quad; + stage->run = choose_blend_quad; stage->destroy = blend_destroy; return stage; diff --git a/src/gallium/drivers/softpipe/sp_quad_bufloop.c b/src/gallium/drivers/softpipe/sp_quad_bufloop.c deleted file mode 100644 index 953d8516b90..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_bufloop.c +++ /dev/null @@ -1,74 +0,0 @@ - -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_quad_pipe.h" - - -/** - * Loop over colorbuffers, passing quad to next stage each time. - */ -static void -cbuf_loop_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - float tmp[PIPE_MAX_COLOR_BUFS][4][QUAD_SIZE]; - unsigned i; - - assert(sizeof(quad->outputs.color) == sizeof(tmp)); - assert(softpipe->framebuffer.nr_cbufs <= PIPE_MAX_COLOR_BUFS); - - /* make copy of original colors since they can get modified - * by blending and masking. - * XXX we won't have to do this if the fragment program actually emits - * N separate colors and we're drawing to N color buffers (MRT). - * But if we emitted one color and glDrawBuffer(GL_FRONT_AND_BACK) is - * in effect, we need to save/restore colors like this. - */ - memcpy(tmp, quad->outputs.color, sizeof(tmp)); - - for (i = 0; i < softpipe->framebuffer.nr_cbufs; i++) { - /* set current cbuffer */ -#if 0 /* obsolete & going away */ - softpipe->current_cbuf = i; -#endif - - /* pass blended quad to next stage */ - qs->next->run(qs->next, quad); - - /* restore quad's colors for next buffer */ - memcpy(quad->outputs.color, tmp, sizeof(tmp)); - } -} - - -static void cbuf_loop_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void cbuf_loop_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -/** - * Create the colorbuffer loop stage. - * This is used to implement multiple render targets and GL_FRONT_AND_BACK - * rendering. - */ -struct quad_stage *sp_quad_bufloop_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = cbuf_loop_begin; - stage->run = cbuf_loop_quad; - stage->destroy = cbuf_loop_destroy; - - return stage; -} - diff --git a/src/gallium/drivers/softpipe/sp_quad_colormask.c b/src/gallium/drivers/softpipe/sp_quad_colormask.c deleted file mode 100644 index dc90e5d5e99..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_colormask.c +++ /dev/null @@ -1,116 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/** - * \brief quad colormask stage - * \author Brian Paul - */ - -#include "pipe/p_defines.h" -#include "util/u_math.h" -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_quad_pipe.h" -#include "sp_tile_cache.h" - - - -/** - * XXX colormask could be rolled into blending... - */ -static void -colormask_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - uint cbuf; - - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - float dest[4][QUAD_SIZE]; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[cbuf], - quad->input.x0, quad->input.y0); - float (*quadColor)[4] = quad->output.color[cbuf]; - uint i, j; - - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - int x = (quad->input.x0 & (TILE_SIZE-1)) + (j & 1); - int y = (quad->input.y0 & (TILE_SIZE-1)) + (j >> 1); - for (i = 0; i < 4; i++) { - dest[i][j] = tile->data.color[y][x][i]; - } - } - - /* R */ - if (!(softpipe->blend->colormask & PIPE_MASK_R)) - COPY_4V(quadColor[0], dest[0]); - - /* G */ - if (!(softpipe->blend->colormask & PIPE_MASK_G)) - COPY_4V(quadColor[1], dest[1]); - - /* B */ - if (!(softpipe->blend->colormask & PIPE_MASK_B)) - COPY_4V(quadColor[2], dest[2]); - - /* A */ - if (!(softpipe->blend->colormask & PIPE_MASK_A)) - COPY_4V(quadColor[3], dest[3]); - } - - /* pass quad to next stage */ - qs->next->run(qs->next, quad); -} - - -static void colormask_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void colormask_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_colormask_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = colormask_begin; - stage->run = colormask_quad; - stage->destroy = colormask_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_coverage.c b/src/gallium/drivers/softpipe/sp_quad_coverage.c deleted file mode 100644 index 4aeee858705..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_coverage.c +++ /dev/null @@ -1,94 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - - -/** - * \brief Apply AA coverage to quad alpha valus - * \author Brian Paul - */ - - -#include "pipe/p_defines.h" -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_quad_pipe.h" - - -/** - * Multiply quad's alpha values by the fragment coverage. - */ -static void -coverage_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - const uint prim = quad->input.prim; - - if ((softpipe->rasterizer->poly_smooth && prim == QUAD_PRIM_TRI) || - (softpipe->rasterizer->line_smooth && prim == QUAD_PRIM_LINE) || - (softpipe->rasterizer->point_smooth && prim == QUAD_PRIM_POINT)) { - uint cbuf; - - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - float (*quadColor)[4] = quad->output.color[cbuf]; - unsigned j; - for (j = 0; j < QUAD_SIZE; j++) { - assert(quad->input.coverage[j] >= 0.0); - assert(quad->input.coverage[j] <= 1.0); - quadColor[3][j] *= quad->input.coverage[j]; - } - } - } - - qs->next->run(qs->next, quad); -} - - -static void coverage_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void coverage_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_coverage_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = coverage_begin; - stage->run = coverage_quad; - stage->destroy = coverage_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_depth_test.c b/src/gallium/drivers/softpipe/sp_quad_depth_test.c index d463930bae1..0ca86c4e1cb 100644 --- a/src/gallium/drivers/softpipe/sp_quad_depth_test.c +++ b/src/gallium/drivers/softpipe/sp_quad_depth_test.c @@ -31,61 +31,109 @@ #include "pipe/p_defines.h" #include "util/u_memory.h" +#include "tgsi/tgsi_scan.h" #include "sp_context.h" #include "sp_quad.h" #include "sp_surface.h" #include "sp_quad_pipe.h" #include "sp_tile_cache.h" +#include "sp_state.h" /* for sp_fragment_shader */ -/** - * Do depth testing for a quad. - * Not static since it's used by the stencil code. - */ +struct depth_data { + struct pipe_surface *ps; + enum pipe_format format; + unsigned bzzzz[QUAD_SIZE]; /**< Z values fetched from depth buffer */ + unsigned qzzzz[QUAD_SIZE]; /**< Z values from the quad */ + ubyte stencilVals[QUAD_SIZE]; + struct softpipe_cached_tile *tile; +}; -/* - * To increase efficiency, we should probably have multiple versions - * of this function that are specifically for Z16, Z32 and FP Z buffers. - * Try to effectively do that with codegen... - */ -void -sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) + +static void +get_depth_stencil_values( struct depth_data *data, + const struct quad_header *quad ) { - struct softpipe_context *softpipe = qs->softpipe; - struct pipe_surface *ps = softpipe->framebuffer.zsbuf; - const enum pipe_format format = ps->format; - unsigned bzzzz[QUAD_SIZE]; /**< Z values fetched from depth buffer */ - unsigned qzzzz[QUAD_SIZE]; /**< Z values from the quad */ - unsigned zmask = 0; unsigned j; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, softpipe->zsbuf_cache, quad->input.x0, quad->input.y0); + const struct softpipe_cached_tile *tile = data->tile; + + switch (data->format) { + case PIPE_FORMAT_Z16_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + data->bzzzz[j] = tile->data.depth16[y][x]; + } + break; + case PIPE_FORMAT_Z32_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + data->bzzzz[j] = tile->data.depth32[y][x]; + } + break; + case PIPE_FORMAT_X8Z24_UNORM: + case PIPE_FORMAT_S8Z24_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + data->bzzzz[j] = tile->data.depth32[y][x] & 0xffffff; + data->stencilVals[j] = tile->data.depth32[y][x] >> 24; + } + break; + case PIPE_FORMAT_Z24X8_UNORM: + case PIPE_FORMAT_Z24S8_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + data->bzzzz[j] = tile->data.depth32[y][x] >> 8; + data->stencilVals[j] = tile->data.depth32[y][x] & 0xff; + } + break; + default: + assert(0); + } +} - assert(ps); /* shouldn't get here if there's no zbuffer */ +/* If the shader has not been run, interpolate the depth values + * ourselves. + */ +static void +interpolate_quad_depth( struct quad_header *quad ) +{ + const float fx = (float) quad->input.x0; + const float fy = (float) quad->input.y0; + const float dzdx = quad->posCoef->dadx[2]; + const float dzdy = quad->posCoef->dady[2]; + const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy; - /* - * Convert quad's float depth values to int depth values (qzzzz). + quad->output.depth[0] = z0; + quad->output.depth[1] = z0 + dzdx; + quad->output.depth[2] = z0 + dzdy; + quad->output.depth[3] = z0 + dzdx + dzdy; +} + + +static void +convert_quad_depth( struct depth_data *data, + const struct quad_header *quad ) +{ + unsigned j; + + /* Convert quad's float depth values to int depth values (qzzzz). * If the Z buffer stores integer values, we _have_ to do the depth * compares with integers (not floats). Otherwise, the float->int->float * conversion of Z values (which isn't an identity function) will cause * Z-fighting errors. - * - * Also, get the zbuffer values (bzzzz) from the cached tile. */ - switch (format) { + switch (data->format) { case PIPE_FORMAT_Z16_UNORM: { float scale = 65535.0; for (j = 0; j < QUAD_SIZE; j++) { - qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); - } - - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - bzzzz[j] = tile->data.depth16[y][x]; + data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); } } break; @@ -94,47 +142,247 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) double scale = (double) (uint) ~0UL; for (j = 0; j < QUAD_SIZE; j++) { - qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); - } - - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - bzzzz[j] = tile->data.depth32[y][x]; + data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); } } break; case PIPE_FORMAT_X8Z24_UNORM: - /* fall-through */ case PIPE_FORMAT_S8Z24_UNORM: { float scale = (float) ((1 << 24) - 1); for (j = 0; j < QUAD_SIZE; j++) { - qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); - } - - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - bzzzz[j] = tile->data.depth32[y][x] & 0xffffff; + data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); } } break; case PIPE_FORMAT_Z24X8_UNORM: - /* fall-through */ case PIPE_FORMAT_Z24S8_UNORM: { float scale = (float) ((1 << 24) - 1); for (j = 0; j < QUAD_SIZE; j++) { - qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); + data->qzzzz[j] = (unsigned) (quad->output.depth[j] * scale); } + } + break; + default: + assert(0); + } +} - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - bzzzz[j] = tile->data.depth32[y][x] >> 8; + + +static void +write_depth_stencil_values( struct depth_data *data, + struct quad_header *quad ) +{ + struct softpipe_cached_tile *tile = data->tile; + unsigned j; + + /* put updated Z values back into cached tile */ + switch (data->format) { + case PIPE_FORMAT_Z16_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth16[y][x] = (ushort) data->bzzzz[j]; + } + break; + case PIPE_FORMAT_X8Z24_UNORM: + case PIPE_FORMAT_Z32_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth32[y][x] = data->bzzzz[j]; + } + break; + case PIPE_FORMAT_S8Z24_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth32[y][x] = (data->stencilVals[j] << 24) | data->bzzzz[j]; + } + break; + case PIPE_FORMAT_Z24S8_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth32[y][x] = (data->bzzzz[j] << 8) | data->stencilVals[j]; + } + break; + case PIPE_FORMAT_Z24X8_UNORM: + for (j = 0; j < QUAD_SIZE; j++) { + int x = quad->input.x0 % TILE_SIZE + (j & 1); + int y = quad->input.y0 % TILE_SIZE + (j >> 1); + tile->data.depth32[y][x] = data->bzzzz[j] << 8; + } + break; + default: + assert(0); + } +} + + + + +/** Only 8-bit stencil supported */ +#define STENCIL_MAX 0xff + + +/** + * Do the basic stencil test (compare stencil buffer values against the + * reference value. + * + * \param data->stencilVals the stencil values from the stencil buffer + * \param func the stencil func (PIPE_FUNC_x) + * \param ref the stencil reference value + * \param valMask the stencil value mask indicating which bits of the stencil + * values and ref value are to be used. + * \return mask indicating which pixels passed the stencil test + */ +static unsigned +do_stencil_test(struct depth_data *data, + unsigned func, + unsigned ref, unsigned valMask) +{ + unsigned passMask = 0x0; + unsigned j; + + ref &= valMask; + + switch (func) { + case PIPE_FUNC_NEVER: + /* passMask = 0x0 */ + break; + case PIPE_FUNC_LESS: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref < (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_EQUAL: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref == (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_LEQUAL: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref <= (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_GREATER: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref > (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_NOTEQUAL: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref != (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_GEQUAL: + for (j = 0; j < QUAD_SIZE; j++) { + if (ref >= (data->stencilVals[j] & valMask)) { + passMask |= (1 << j); + } + } + break; + case PIPE_FUNC_ALWAYS: + passMask = MASK_ALL; + break; + default: + assert(0); + } + + return passMask; +} + + +/** + * Apply the stencil operator to stencil values. + * + * \param data->stencilVals the stencil buffer values (read and written) + * \param mask indicates which pixels to update + * \param op the stencil operator (PIPE_STENCIL_OP_x) + * \param ref the stencil reference value + * \param wrtMask writemask controlling which bits are changed in the + * stencil values + */ +static void +apply_stencil_op(struct depth_data *data, + unsigned mask, unsigned op, ubyte ref, ubyte wrtMask) +{ + unsigned j; + ubyte newstencil[QUAD_SIZE]; + + for (j = 0; j < QUAD_SIZE; j++) { + newstencil[j] = data->stencilVals[j]; + } + + switch (op) { + case PIPE_STENCIL_OP_KEEP: + /* no-op */ + break; + case PIPE_STENCIL_OP_ZERO: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = 0; + } + } + break; + case PIPE_STENCIL_OP_REPLACE: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = ref; + } + } + break; + case PIPE_STENCIL_OP_INCR: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + if (data->stencilVals[j] < STENCIL_MAX) { + newstencil[j] = data->stencilVals[j] + 1; + } + } + } + break; + case PIPE_STENCIL_OP_DECR: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + if (data->stencilVals[j] > 0) { + newstencil[j] = data->stencilVals[j] - 1; + } + } + } + break; + case PIPE_STENCIL_OP_INCR_WRAP: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = data->stencilVals[j] + 1; + } + } + break; + case PIPE_STENCIL_OP_DECR_WRAP: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = data->stencilVals[j] - 1; + } + } + break; + case PIPE_STENCIL_OP_INVERT: + for (j = 0; j < QUAD_SIZE; j++) { + if (mask & (1 << j)) { + newstencil[j] = ~data->stencilVals[j]; } } break; @@ -142,6 +390,39 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) assert(0); } + /* + * update the stencil values + */ + if (wrtMask != STENCIL_MAX) { + /* apply bit-wise stencil buffer writemask */ + for (j = 0; j < QUAD_SIZE; j++) { + data->stencilVals[j] = (wrtMask & newstencil[j]) | (~wrtMask & data->stencilVals[j]); + } + } + else { + for (j = 0; j < QUAD_SIZE; j++) { + data->stencilVals[j] = newstencil[j]; + } + } +} + + + +/* + * To increase efficiency, we should probably have multiple versions + * of this function that are specifically for Z16, Z32 and FP Z buffers. + * Try to effectively do that with codegen... + */ + +static boolean +depth_test_quad(struct quad_stage *qs, + struct depth_data *data, + struct quad_header *quad) +{ + struct softpipe_context *softpipe = qs->softpipe; + unsigned zmask = 0; + unsigned j; + switch (softpipe->depth_stencil->depth.func) { case PIPE_FUNC_NEVER: /* zmask = 0 */ @@ -151,37 +432,37 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) * Like this: quad->mask &= (quad->outputs.depth < zzzz); */ for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] < bzzzz[j]) + if (data->qzzzz[j] < data->bzzzz[j]) zmask |= 1 << j; } break; case PIPE_FUNC_EQUAL: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] == bzzzz[j]) + if (data->qzzzz[j] == data->bzzzz[j]) zmask |= 1 << j; } break; case PIPE_FUNC_LEQUAL: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] <= bzzzz[j]) + if (data->qzzzz[j] <= data->bzzzz[j]) zmask |= (1 << j); } break; case PIPE_FUNC_GREATER: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] > bzzzz[j]) + if (data->qzzzz[j] > data->bzzzz[j]) zmask |= (1 << j); } break; case PIPE_FUNC_NOTEQUAL: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] != bzzzz[j]) + if (data->qzzzz[j] != data->bzzzz[j]) zmask |= (1 << j); } break; case PIPE_FUNC_GEQUAL: for (j = 0; j < QUAD_SIZE; j++) { - if (qzzzz[j] >= bzzzz[j]) + if (data->qzzzz[j] >= data->bzzzz[j]) zmask |= (1 << j); } break; @@ -193,80 +474,480 @@ sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad) } quad->inout.mask &= zmask; + if (quad->inout.mask == 0) + return FALSE; + /* Update our internal copy only if writemask set. Even if + * depth.writemask is FALSE, may still need to write out buffer + * data due to stencil changes. + */ if (softpipe->depth_stencil->depth.writemask) { - - /* This is also efficient with sse / spe instructions: - */ for (j = 0; j < QUAD_SIZE; j++) { - if (quad->inout.mask & (1 << j)) { - bzzzz[j] = qzzzz[j]; - } + if (quad->inout.mask & (1 << j)) { + data->bzzzz[j] = data->qzzzz[j]; + } } + } - /* put updated Z values back into cached tile */ - switch (format) { - case PIPE_FORMAT_Z16_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - tile->data.depth16[y][x] = (ushort) bzzzz[j]; + return TRUE; +} + + + +/** + * Do stencil (and depth) testing. Stenciling depends on the outcome of + * depth testing. + */ +static void +depth_stencil_test_quad(struct quad_stage *qs, + struct depth_data *data, + struct quad_header *quad) +{ + struct softpipe_context *softpipe = qs->softpipe; + unsigned func, zFailOp, zPassOp, failOp; + ubyte ref, wrtMask, valMask; + uint face = quad->input.facing; + + if (!softpipe->depth_stencil->stencil[1].enabled) { + /* single-sided stencil test, use front (face=0) state */ + face = 0; + } + + /* choose front or back face function, operator, etc */ + /* XXX we could do these initializations once per primitive */ + func = softpipe->depth_stencil->stencil[face].func; + failOp = softpipe->depth_stencil->stencil[face].fail_op; + zFailOp = softpipe->depth_stencil->stencil[face].zfail_op; + zPassOp = softpipe->depth_stencil->stencil[face].zpass_op; + ref = softpipe->depth_stencil->stencil[face].ref_value; + wrtMask = softpipe->depth_stencil->stencil[face].writemask; + valMask = softpipe->depth_stencil->stencil[face].valuemask; + + + /* do the stencil test first */ + { + unsigned passMask, failMask; + passMask = do_stencil_test(data, func, ref, valMask); + failMask = quad->inout.mask & ~passMask; + quad->inout.mask &= passMask; + + if (failOp != PIPE_STENCIL_OP_KEEP) { + apply_stencil_op(data, failMask, failOp, ref, wrtMask); + } + } + + if (quad->inout.mask) { + /* now the pixels that passed the stencil test are depth tested */ + if (softpipe->depth_stencil->depth.enabled) { + const unsigned origMask = quad->inout.mask; + + depth_test_quad(qs, data, quad); /* quad->mask is updated */ + + /* update stencil buffer values according to z pass/fail result */ + if (zFailOp != PIPE_STENCIL_OP_KEEP) { + const unsigned zFailMask = origMask & ~quad->inout.mask; + apply_stencil_op(data, zFailMask, zFailOp, ref, wrtMask); } - break; - case PIPE_FORMAT_X8Z24_UNORM: - /* fall-through */ - /* (yes, this falls through to a different case than above) */ - case PIPE_FORMAT_Z32_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - tile->data.depth32[y][x] = bzzzz[j]; + + if (zPassOp != PIPE_STENCIL_OP_KEEP) { + const unsigned zPassMask = origMask & quad->inout.mask; + apply_stencil_op(data, zPassMask, zPassOp, ref, wrtMask); } - break; - case PIPE_FORMAT_S8Z24_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - uint s8z24 = tile->data.depth32[y][x]; - s8z24 = (s8z24 & 0xff000000) | bzzzz[j]; - tile->data.depth32[y][x] = s8z24; + } + else { + /* no depth test, apply Zpass operator to stencil buffer values */ + apply_stencil_op(data, quad->inout.mask, zPassOp, ref, wrtMask); + } + } +} + + +#define ALPHATEST( FUNC, COMP ) \ + static int \ + alpha_test_quads_##FUNC( struct quad_stage *qs, \ + struct quad_header *quads[], \ + unsigned nr ) \ + { \ + const float ref = qs->softpipe->depth_stencil->alpha.ref_value; \ + const uint cbuf = 0; /* only output[0].alpha is tested */ \ + unsigned pass_nr = 0; \ + unsigned i; \ + \ + for (i = 0; i < nr; i++) { \ + const float *aaaa = quads[i]->output.color[cbuf][3]; \ + unsigned passMask = 0; \ + \ + if (aaaa[0] COMP ref) passMask |= (1 << 0); \ + if (aaaa[1] COMP ref) passMask |= (1 << 1); \ + if (aaaa[2] COMP ref) passMask |= (1 << 2); \ + if (aaaa[3] COMP ref) passMask |= (1 << 3); \ + \ + quads[i]->inout.mask &= passMask; \ + \ + if (quads[i]->inout.mask) \ + quads[pass_nr++] = quads[i]; \ + } \ + \ + return pass_nr; \ + } + + +ALPHATEST( LESS, < ) +ALPHATEST( EQUAL, == ) +ALPHATEST( LEQUAL, <= ) +ALPHATEST( GREATER, > ) +ALPHATEST( NOTEQUAL, != ) +ALPHATEST( GEQUAL, >= ) + + +/* XXX: Incorporate into shader using KILP. + */ +static int +alpha_test_quads(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + switch (qs->softpipe->depth_stencil->alpha.func) { + case PIPE_FUNC_LESS: + return alpha_test_quads_LESS( qs, quads, nr ); + case PIPE_FUNC_EQUAL: + return alpha_test_quads_EQUAL( qs, quads, nr ); + break; + case PIPE_FUNC_LEQUAL: + return alpha_test_quads_LEQUAL( qs, quads, nr ); + case PIPE_FUNC_GREATER: + return alpha_test_quads_GREATER( qs, quads, nr ); + case PIPE_FUNC_NOTEQUAL: + return alpha_test_quads_NOTEQUAL( qs, quads, nr ); + case PIPE_FUNC_GEQUAL: + return alpha_test_quads_GEQUAL( qs, quads, nr ); + case PIPE_FUNC_ALWAYS: + return nr; + case PIPE_FUNC_NEVER: + default: + return 0; + } +} + +static unsigned mask_count[16] = +{ + 0, /* 0x0 */ + 1, /* 0x1 */ + 1, /* 0x2 */ + 2, /* 0x3 */ + 1, /* 0x4 */ + 2, /* 0x5 */ + 2, /* 0x6 */ + 3, /* 0x7 */ + 1, /* 0x8 */ + 2, /* 0x9 */ + 2, /* 0xa */ + 3, /* 0xb */ + 2, /* 0xc */ + 3, /* 0xd */ + 3, /* 0xe */ + 4, /* 0xf */ +}; + + + +static void +depth_test_quads_fallback(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + unsigned i, pass = 0; + const struct sp_fragment_shader *fs = qs->softpipe->fs; + boolean interp_depth = !fs->info.writes_z; + struct depth_data data; + + + if (qs->softpipe->depth_stencil->alpha.enabled) { + nr = alpha_test_quads(qs, quads, nr); + } + + if (qs->softpipe->framebuffer.zsbuf && + (qs->softpipe->depth_stencil->depth.enabled || + qs->softpipe->depth_stencil->stencil[0].enabled)) { + + data.ps = qs->softpipe->framebuffer.zsbuf; + data.format = data.ps->format; + data.tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, + quads[0]->input.x0, + quads[0]->input.y0); + + for (i = 0; i < nr; i++) { + get_depth_stencil_values(&data, quads[i]); + + if (qs->softpipe->depth_stencil->depth.enabled) { + if (interp_depth) + interpolate_quad_depth(quads[i]); + + convert_quad_depth(&data, quads[i]); } - break; - case PIPE_FORMAT_Z24S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - uint z24s8 = tile->data.depth32[y][x]; - z24s8 = (z24s8 & 0xff) | (bzzzz[j] << 8); - tile->data.depth32[y][x] = z24s8; + + if (qs->softpipe->depth_stencil->stencil[0].enabled) { + depth_stencil_test_quad(qs, &data, quads[i]); + write_depth_stencil_values(&data, quads[i]); } - break; - case PIPE_FORMAT_Z24X8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - tile->data.depth32[y][x] = bzzzz[j] << 8; + else { + if (!depth_test_quad(qs, &data, quads[i])) + continue; + + if (qs->softpipe->depth_stencil->depth.writemask) + write_depth_stencil_values(&data, quads[i]); } - break; - default: - assert(0); + + + quads[pass++] = quads[i]; + } + + nr = pass; + } + + if (qs->softpipe->active_query_count) { + for (i = 0; i < nr; i++) + qs->softpipe->occlusion_count += mask_count[quads[i]->inout.mask]; + } + + if (nr) + qs->next->run(qs->next, quads, nr); +} + +/* XXX: this function assumes setup function actually emits linear + * spans of quads. It seems a lot more natural to do (early) + * depth-testing on spans rather than quads. + */ +static void +depth_interp_z16_less_write(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + unsigned i, pass = 0; + const unsigned ix = quads[0]->input.x0; + const unsigned iy = quads[0]->input.y0; + const float fx = (float) ix; + const float fy = (float) iy; + const float dzdx = quads[0]->posCoef->dadx[2]; + const float dzdy = quads[0]->posCoef->dady[2]; + const float z0 = quads[0]->posCoef->a0[2] + dzdx * fx + dzdy * fy; + struct softpipe_cached_tile *tile; + ushort (*depth16)[TILE_SIZE]; + ushort idepth[4], depth_step; + const float scale = 65535.0; + + idepth[0] = (ushort)((z0) * scale); + idepth[1] = (ushort)((z0 + dzdx) * scale); + idepth[2] = (ushort)((z0 + dzdy) * scale); + idepth[3] = (ushort)((z0 + dzdx + dzdy) * scale); + + depth_step = (ushort)(dzdx * 2 * scale); + + tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, ix, iy); + + depth16 = (ushort (*)[TILE_SIZE]) + &tile->data.depth16[iy % TILE_SIZE][ix % TILE_SIZE]; + + for (i = 0; i < nr; i++) { + unsigned outmask = quads[i]->inout.mask; + unsigned mask = 0; + + if ((outmask & 1) && idepth[0] < depth16[0][0]) { + depth16[0][0] = idepth[0]; + mask |= (1 << 0); + } + + if ((outmask & 2) && idepth[1] < depth16[0][1]) { + depth16[0][1] = idepth[1]; + mask |= (1 << 1); + } + + if ((outmask & 4) && idepth[2] < depth16[1][0]) { + depth16[1][0] = idepth[2]; + mask |= (1 << 2); + } + + if ((outmask & 8) && idepth[3] < depth16[1][1]) { + depth16[1][1] = idepth[3]; + mask |= (1 << 3); } + + idepth[0] += depth_step; + idepth[1] += depth_step; + idepth[2] += depth_step; + idepth[3] += depth_step; + + depth16 = (ushort (*)[TILE_SIZE]) &depth16[0][2]; + + quads[i]->inout.mask = mask; + if (quads[i]->inout.mask) + quads[pass++] = quads[i]; } + + if (pass) + qs->next->run(qs->next, quads, pass); + } static void -depth_test_quad(struct quad_stage *qs, struct quad_header *quad) +depth_interp_z16_lequal_write(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) { - sp_depth_test_quad(qs, quad); + unsigned i, pass = 0; + const unsigned ix = quads[0]->input.x0; + const unsigned iy = quads[0]->input.y0; + const float fx = (float) ix; + const float fy = (float) iy; + const float dzdx = quads[0]->posCoef->dadx[2]; + const float dzdy = quads[0]->posCoef->dady[2]; + const float z0 = quads[0]->posCoef->a0[2] + dzdx * fx + dzdy * fy; + struct softpipe_cached_tile *tile; + ushort (*depth16)[TILE_SIZE]; + ushort idepth[4], depth_step; + const float scale = 65535.0; + + idepth[0] = (ushort)((z0) * scale); + idepth[1] = (ushort)((z0 + dzdx) * scale); + idepth[2] = (ushort)((z0 + dzdy) * scale); + idepth[3] = (ushort)((z0 + dzdx + dzdy) * scale); + + depth_step = (ushort)(dzdx * 2 * scale); + + tile = sp_get_cached_tile(qs->softpipe->zsbuf_cache, ix, iy); + + depth16 = (ushort (*)[TILE_SIZE]) + &tile->data.depth16[iy % TILE_SIZE][ix % TILE_SIZE]; + + for (i = 0; i < nr; i++) { + unsigned outmask = quads[i]->inout.mask; + unsigned mask = 0; + + if ((outmask & 1) && idepth[0] <= depth16[0][0]) { + depth16[0][0] = idepth[0]; + mask |= (1 << 0); + } + + if ((outmask & 2) && idepth[1] <= depth16[0][1]) { + depth16[0][1] = idepth[1]; + mask |= (1 << 1); + } + + if ((outmask & 4) && idepth[2] <= depth16[1][0]) { + depth16[1][0] = idepth[2]; + mask |= (1 << 2); + } + + if ((outmask & 8) && idepth[3] <= depth16[1][1]) { + depth16[1][1] = idepth[3]; + mask |= (1 << 3); + } + + idepth[0] += depth_step; + idepth[1] += depth_step; + idepth[2] += depth_step; + idepth[3] += depth_step; + + depth16 = (ushort (*)[TILE_SIZE]) &depth16[0][2]; + + quads[i]->inout.mask = mask; + if (quads[i]->inout.mask) + quads[pass++] = quads[i]; + } + + if (pass) + qs->next->run(qs->next, quads, pass); - if (quad->inout.mask) - qs->next->run(qs->next, quad); } + + + +static void +depth_noop(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + qs->next->run(qs->next, quads, nr); +} + + + +static void +choose_depth_test(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + boolean interp_depth = !qs->softpipe->fs->info.writes_z; + + boolean alpha = qs->softpipe->depth_stencil->alpha.enabled; + + boolean depth = (qs->softpipe->framebuffer.zsbuf && + qs->softpipe->depth_stencil->depth.enabled); + + unsigned depthfunc = qs->softpipe->depth_stencil->depth.func; + + boolean stencil = qs->softpipe->depth_stencil->stencil[0].enabled; + + boolean depthwrite = qs->softpipe->depth_stencil->depth.writemask; + + boolean occlusion = qs->softpipe->active_query_count; + + + if (!alpha && + !depth && + !stencil) { + qs->run = depth_noop; + } + else if (!alpha && + interp_depth && + depth && + depthwrite && + !occlusion && + !stencil) + { + switch (depthfunc) { + case PIPE_FUNC_LESS: + switch (qs->softpipe->framebuffer.zsbuf->format) { + case PIPE_FORMAT_Z16_UNORM: + qs->run = depth_interp_z16_less_write; + break; + default: + qs->run = depth_test_quads_fallback; + break; + } + break; + case PIPE_FUNC_LEQUAL: + switch (qs->softpipe->framebuffer.zsbuf->format) { + case PIPE_FORMAT_Z16_UNORM: + qs->run = depth_interp_z16_lequal_write; + break; + default: + qs->run = depth_test_quads_fallback; + break; + } + break; + default: + qs->run = depth_test_quads_fallback; + } + } + else { + qs->run = depth_test_quads_fallback; + } + + + qs->run( qs, quads, nr ); +} + + + + + static void depth_test_begin(struct quad_stage *qs) { + qs->run = choose_depth_test; qs->next->begin(qs->next); } @@ -283,7 +964,7 @@ struct quad_stage *sp_quad_depth_test_stage( struct softpipe_context *softpipe ) stage->softpipe = softpipe; stage->begin = depth_test_begin; - stage->run = depth_test_quad; + stage->run = choose_depth_test; stage->destroy = depth_test_destroy; return stage; diff --git a/src/gallium/drivers/softpipe/sp_quad_earlyz.c b/src/gallium/drivers/softpipe/sp_quad_earlyz.c deleted file mode 100644 index 496fd39ed1a..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_earlyz.c +++ /dev/null @@ -1,88 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -/** - * \brief Quad early-z testing - */ - -#include "pipe/p_defines.h" -#include "util/u_memory.h" -#include "sp_quad.h" -#include "sp_quad_pipe.h" - - -/** - * All this stage does is compute the quad's Z values (which is normally - * done by the shading stage). - * The next stage will do the actual depth test. - */ -static void -earlyz_quad( - struct quad_stage *qs, - struct quad_header *quad ) -{ - const float fx = (float) quad->input.x0; - const float fy = (float) quad->input.y0; - const float dzdx = quad->posCoef->dadx[2]; - const float dzdy = quad->posCoef->dady[2]; - const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy; - - quad->output.depth[0] = z0; - quad->output.depth[1] = z0 + dzdx; - quad->output.depth[2] = z0 + dzdy; - quad->output.depth[3] = z0 + dzdx + dzdy; - - qs->next->run( qs->next, quad ); -} - -static void -earlyz_begin( - struct quad_stage *qs ) -{ - qs->next->begin( qs->next ); -} - -static void -earlyz_destroy( - struct quad_stage *qs ) -{ - FREE( qs ); -} - -struct quad_stage * -sp_quad_earlyz_stage( - struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT( quad_stage ); - - stage->softpipe = softpipe; - stage->begin = earlyz_begin; - stage->run = earlyz_quad; - stage->destroy = earlyz_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c index 28f8d1a60ea..1e7533d0f9e 100644 --- a/src/gallium/drivers/softpipe/sp_quad_fs.c +++ b/src/gallium/drivers/softpipe/sp_quad_fs.c @@ -68,72 +68,69 @@ quad_shade_stage(struct quad_stage *qs) /** * Execute fragment shader for the four fragments in the quad. */ -static void +static INLINE boolean shade_quad(struct quad_stage *qs, struct quad_header *quad) { struct quad_shade_stage *qss = quad_shade_stage( qs ); struct softpipe_context *softpipe = qs->softpipe; struct tgsi_exec_machine *machine = qss->machine; - boolean z_written; - - /* Consts do not require 16 byte alignment. */ - machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT]; - - machine->InterpCoefs = quad->coef; /* run shader */ - quad->inout.mask &= softpipe->fs->run( softpipe->fs, machine, quad ); - - /* store outputs */ - z_written = FALSE; - { - const ubyte *sem_name = softpipe->fs->info.output_semantic_name; - const ubyte *sem_index = softpipe->fs->info.output_semantic_index; - const uint n = qss->stage.softpipe->fs->info.num_outputs; - uint i; - for (i = 0; i < n; i++) { - switch (sem_name[i]) { - case TGSI_SEMANTIC_COLOR: - { - uint cbuf = sem_index[i]; - memcpy(quad->output.color[cbuf], - &machine->Outputs[i].xyzw[0].f[0], - sizeof(quad->output.color[0]) ); - } - break; - case TGSI_SEMANTIC_POSITION: - { - uint j; - for (j = 0; j < 4; j++) { - quad->output.depth[j] = machine->Outputs[0].xyzw[2].f[j]; - } - z_written = TRUE; - } - break; - } + return softpipe->fs->run( softpipe->fs, machine, quad ); +} + + + +static void +coverage_quad(struct quad_stage *qs, struct quad_header *quad) +{ + struct softpipe_context *softpipe = qs->softpipe; + uint cbuf; + + /* loop over colorbuffer outputs */ + for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { + float (*quadColor)[4] = quad->output.color[cbuf]; + unsigned j; + for (j = 0; j < QUAD_SIZE; j++) { + assert(quad->input.coverage[j] >= 0.0); + assert(quad->input.coverage[j] <= 1.0); + quadColor[3][j] *= quad->input.coverage[j]; } } +} + - if (!z_written) { - /* compute Z values now, as in the quad earlyz stage */ - /* XXX we should really only do this if the earlyz stage is not used */ - const float fx = (float) quad->input.x0; - const float fy = (float) quad->input.y0; - const float dzdx = quad->posCoef->dadx[2]; - const float dzdy = quad->posCoef->dady[2]; - const float z0 = quad->posCoef->a0[2] + dzdx * fx + dzdy * fy; - - quad->output.depth[0] = z0; - quad->output.depth[1] = z0 + dzdx; - quad->output.depth[2] = z0 + dzdy; - quad->output.depth[3] = z0 + dzdx + dzdy; - } - /* shader may cull fragments */ - if (quad->inout.mask) { - qs->next->run( qs->next, quad ); +static void +shade_quads(struct quad_stage *qs, + struct quad_header *quads[], + unsigned nr) +{ + struct quad_shade_stage *qss = quad_shade_stage( qs ); + struct softpipe_context *softpipe = qs->softpipe; + struct tgsi_exec_machine *machine = qss->machine; + + unsigned i, pass = 0; + + machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT]; + machine->InterpCoefs = quads[0]->coef; + + for (i = 0; i < nr; i++) { + if (!shade_quad(qs, quads[i])) + continue; + + if (/*do_coverage*/ 0) + coverage_quad( qs, quads[i] ); + + quads[pass++] = quads[i]; } + + if (pass) + qs->next->run(qs->next, quads, pass); } + + + /** @@ -174,7 +171,7 @@ sp_quad_shade_stage( struct softpipe_context *softpipe ) qss->stage.softpipe = softpipe; qss->stage.begin = shade_begin; - qss->stage.run = shade_quad; + qss->stage.run = shade_quads; qss->stage.destroy = shade_destroy; qss->machine = tgsi_exec_machine_create(); diff --git a/src/gallium/drivers/softpipe/sp_quad_output.c b/src/gallium/drivers/softpipe/sp_quad_output.c deleted file mode 100644 index 92d5f9f3c1a..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_output.c +++ /dev/null @@ -1,103 +0,0 @@ -/************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_quad_pipe.h" -#include "sp_tile_cache.h" - - -/** - * Last step of quad processing: write quad colors to the framebuffer, - * taking mask into account. - */ -static void -output_quad(struct quad_stage *qs, struct quad_header *quad) -{ - /* in-tile pos: */ - const int itx = quad->input.x0 % TILE_SIZE; - const int ity = quad->input.y0 % TILE_SIZE; - - struct softpipe_context *softpipe = qs->softpipe; - uint cbuf; - - /* loop over colorbuffer outputs */ - for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) { - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, - softpipe->cbuf_cache[cbuf], - quad->input.x0, quad->input.y0); - float (*quadColor)[4] = quad->output.color[cbuf]; - int i, j; - - /* get/swizzle dest colors */ - for (j = 0; j < QUAD_SIZE; j++) { - if (quad->inout.mask & (1 << j)) { - int x = itx + (j & 1); - int y = ity + (j >> 1); - for (i = 0; i < 4; i++) { /* loop over color chans */ - tile->data.color[y][x][i] = quadColor[i][j]; - } - if (0) { - debug_printf("sp write pixel %d,%d: %g, %g, %g\n", - quad->input.x0 + x, - quad->input.y0 + y, - quadColor[0][j], - quadColor[1][j], - quadColor[2][j]); - } - } - } - } -} - - -static void output_begin(struct quad_stage *qs) -{ - assert(qs->next == NULL); -} - - -static void output_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_output_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = output_begin; - stage->run = output_quad; - stage->destroy = output_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_pipe.c b/src/gallium/drivers/softpipe/sp_quad_pipe.c index b5f69b74264..1b5bab4eca6 100644 --- a/src/gallium/drivers/softpipe/sp_quad_pipe.c +++ b/src/gallium/drivers/softpipe/sp_quad_pipe.c @@ -31,88 +31,33 @@ #include "pipe/p_shader_tokens.h" static void -sp_push_quad_first( - struct softpipe_context *sp, - struct quad_stage *quad, - uint i ) +sp_push_quad_first( struct softpipe_context *sp, + struct quad_stage *quad ) { - quad->next = sp->quad[i].first; - sp->quad[i].first = quad; + quad->next = sp->quad.first; + sp->quad.first = quad; } -static void -sp_build_depth_stencil( - struct softpipe_context *sp, - uint i ) -{ - if (sp->depth_stencil->stencil[0].enabled || - sp->depth_stencil->stencil[1].enabled) { - sp_push_quad_first( sp, sp->quad[i].stencil_test, i ); - } - else if (sp->depth_stencil->depth.enabled && - sp->framebuffer.zsbuf) { - sp_push_quad_first( sp, sp->quad[i].depth_test, i ); - } -} void sp_build_quad_pipeline(struct softpipe_context *sp) { - uint i; - boolean early_depth_test = - sp->depth_stencil->depth.enabled && - sp->framebuffer.zsbuf && - !sp->depth_stencil->alpha.enabled && - !sp->fs->info.uses_kill && - !sp->fs->info.writes_z; - - /* build up the pipeline in reverse order... */ - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - sp->quad[i].first = sp->quad[i].output; - - if (sp->blend->colormask != 0xf) { - sp_push_quad_first( sp, sp->quad[i].colormask, i ); - } + sp->depth_stencil->depth.enabled && + sp->framebuffer.zsbuf && + !sp->depth_stencil->alpha.enabled && + !sp->fs->info.uses_kill && + !sp->fs->info.writes_z; - if (sp->blend->blend_enable || - sp->blend->logicop_enable) { - sp_push_quad_first( sp, sp->quad[i].blend, i ); - } + sp->quad.first = sp->quad.blend; - if (sp->active_query_count) { - sp_push_quad_first( sp, sp->quad[i].occlusion, i ); - } - - if (sp->rasterizer->poly_smooth || - sp->rasterizer->line_smooth || - sp->rasterizer->point_smooth) { - sp_push_quad_first( sp, sp->quad[i].coverage, i ); - } - - if (!early_depth_test) { - sp_build_depth_stencil( sp, i ); - } - - if (sp->depth_stencil->alpha.enabled) { - sp_push_quad_first( sp, sp->quad[i].alpha_test, i ); - } - - /* XXX always enable shader? */ - if (1) { - sp_push_quad_first( sp, sp->quad[i].shade, i ); - } - - if (early_depth_test) { - sp_build_depth_stencil( sp, i ); - sp_push_quad_first( sp, sp->quad[i].earlyz, i ); - } - -#if !USE_DRAW_STAGE_PSTIPPLE - if (sp->rasterizer->poly_stipple_enable) { - sp_push_quad_first( sp, sp->quad[i].polygon_stipple, i ); - } -#endif + if (early_depth_test) { + sp_push_quad_first( sp, sp->quad.shade ); + sp_push_quad_first( sp, sp->quad.depth_test ); + } + else { + sp_push_quad_first( sp, sp->quad.depth_test ); + sp_push_quad_first( sp, sp->quad.shade ); } } diff --git a/src/gallium/drivers/softpipe/sp_quad_pipe.h b/src/gallium/drivers/softpipe/sp_quad_pipe.h index 0e40586ffc8..c0aa1348319 100644 --- a/src/gallium/drivers/softpipe/sp_quad_pipe.h +++ b/src/gallium/drivers/softpipe/sp_quad_pipe.h @@ -49,7 +49,7 @@ struct quad_stage { void (*begin)(struct quad_stage *qs); /** the stage action */ - void (*run)(struct quad_stage *qs, struct quad_header *quad); + void (*run)(struct quad_stage *qs, struct quad_header *quad[], unsigned nr); void (*destroy)(struct quad_stage *qs); }; @@ -69,6 +69,4 @@ struct quad_stage *sp_quad_output_stage( struct softpipe_context *softpipe ); void sp_build_quad_pipeline(struct softpipe_context *sp); -void sp_depth_test_quad(struct quad_stage *qs, struct quad_header *quad); - #endif /* SP_QUAD_PIPE_H */ diff --git a/src/gallium/drivers/softpipe/sp_quad_stencil.c b/src/gallium/drivers/softpipe/sp_quad_stencil.c deleted file mode 100644 index 5e9d447737d..00000000000 --- a/src/gallium/drivers/softpipe/sp_quad_stencil.c +++ /dev/null @@ -1,352 +0,0 @@ - -/** - * \brief Quad stencil testing - */ - - -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_tile_cache.h" -#include "sp_quad_pipe.h" -#include "pipe/p_defines.h" -#include "util/u_memory.h" - - -/** Only 8-bit stencil supported */ -#define STENCIL_MAX 0xff - - -/** - * Do the basic stencil test (compare stencil buffer values against the - * reference value. - * - * \param stencilVals the stencil values from the stencil buffer - * \param func the stencil func (PIPE_FUNC_x) - * \param ref the stencil reference value - * \param valMask the stencil value mask indicating which bits of the stencil - * values and ref value are to be used. - * \return mask indicating which pixels passed the stencil test - */ -static unsigned -do_stencil_test(const ubyte stencilVals[QUAD_SIZE], unsigned func, - unsigned ref, unsigned valMask) -{ - unsigned passMask = 0x0; - unsigned j; - - ref &= valMask; - - switch (func) { - case PIPE_FUNC_NEVER: - /* passMask = 0x0 */ - break; - case PIPE_FUNC_LESS: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref < (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_EQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref == (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_LEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref <= (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_GREATER: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref > (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_NOTEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref != (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_GEQUAL: - for (j = 0; j < QUAD_SIZE; j++) { - if (ref >= (stencilVals[j] & valMask)) { - passMask |= (1 << j); - } - } - break; - case PIPE_FUNC_ALWAYS: - passMask = MASK_ALL; - break; - default: - assert(0); - } - - return passMask; -} - - -/** - * Apply the stencil operator to stencil values. - * - * \param stencilVals the stencil buffer values (read and written) - * \param mask indicates which pixels to update - * \param op the stencil operator (PIPE_STENCIL_OP_x) - * \param ref the stencil reference value - * \param wrtMask writemask controlling which bits are changed in the - * stencil values - */ -static void -apply_stencil_op(ubyte stencilVals[QUAD_SIZE], - unsigned mask, unsigned op, ubyte ref, ubyte wrtMask) -{ - unsigned j; - ubyte newstencil[QUAD_SIZE]; - - for (j = 0; j < QUAD_SIZE; j++) { - newstencil[j] = stencilVals[j]; - } - - switch (op) { - case PIPE_STENCIL_OP_KEEP: - /* no-op */ - break; - case PIPE_STENCIL_OP_ZERO: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = 0; - } - } - break; - case PIPE_STENCIL_OP_REPLACE: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = ref; - } - } - break; - case PIPE_STENCIL_OP_INCR: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - if (stencilVals[j] < STENCIL_MAX) { - newstencil[j] = stencilVals[j] + 1; - } - } - } - break; - case PIPE_STENCIL_OP_DECR: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - if (stencilVals[j] > 0) { - newstencil[j] = stencilVals[j] - 1; - } - } - } - break; - case PIPE_STENCIL_OP_INCR_WRAP: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = stencilVals[j] + 1; - } - } - break; - case PIPE_STENCIL_OP_DECR_WRAP: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = stencilVals[j] - 1; - } - } - break; - case PIPE_STENCIL_OP_INVERT: - for (j = 0; j < QUAD_SIZE; j++) { - if (mask & (1 << j)) { - newstencil[j] = ~stencilVals[j]; - } - } - break; - default: - assert(0); - } - - /* - * update the stencil values - */ - if (wrtMask != STENCIL_MAX) { - /* apply bit-wise stencil buffer writemask */ - for (j = 0; j < QUAD_SIZE; j++) { - stencilVals[j] = (wrtMask & newstencil[j]) | (~wrtMask & stencilVals[j]); - } - } - else { - for (j = 0; j < QUAD_SIZE; j++) { - stencilVals[j] = newstencil[j]; - } - } -} - - -/** - * Do stencil (and depth) testing. Stenciling depends on the outcome of - * depth testing. - */ -static void -stencil_test_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - struct pipe_surface *ps = softpipe->framebuffer.zsbuf; - unsigned func, zFailOp, zPassOp, failOp; - ubyte ref, wrtMask, valMask; - ubyte stencilVals[QUAD_SIZE]; - struct softpipe_cached_tile *tile - = sp_get_cached_tile(softpipe, softpipe->zsbuf_cache, quad->input.x0, quad->input.y0); - uint j; - uint face = quad->input.facing; - - if (!softpipe->depth_stencil->stencil[1].enabled) { - /* single-sided stencil test, use front (face=0) state */ - face = 0; - } - - /* choose front or back face function, operator, etc */ - /* XXX we could do these initializations once per primitive */ - func = softpipe->depth_stencil->stencil[face].func; - failOp = softpipe->depth_stencil->stencil[face].fail_op; - zFailOp = softpipe->depth_stencil->stencil[face].zfail_op; - zPassOp = softpipe->depth_stencil->stencil[face].zpass_op; - ref = softpipe->depth_stencil->stencil[face].ref_value; - wrtMask = softpipe->depth_stencil->stencil[face].writemask; - valMask = softpipe->depth_stencil->stencil[face].valuemask; - - assert(ps); /* shouldn't get here if there's no stencil buffer */ - - /* get stencil values from cached tile */ - switch (ps->format) { - case PIPE_FORMAT_S8Z24_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - stencilVals[j] = tile->data.depth32[y][x] >> 24; - } - break; - case PIPE_FORMAT_Z24S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - stencilVals[j] = tile->data.depth32[y][x] & 0xff; - } - break; - case PIPE_FORMAT_S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - stencilVals[j] = tile->data.stencil8[y][x]; - } - break; - default: - assert(0); - } - - /* do the stencil test first */ - { - unsigned passMask, failMask; - passMask = do_stencil_test(stencilVals, func, ref, valMask); - failMask = quad->inout.mask & ~passMask; - quad->inout.mask &= passMask; - - if (failOp != PIPE_STENCIL_OP_KEEP) { - apply_stencil_op(stencilVals, failMask, failOp, ref, wrtMask); - } - } - - if (quad->inout.mask) { - /* now the pixels that passed the stencil test are depth tested */ - if (softpipe->depth_stencil->depth.enabled) { - const unsigned origMask = quad->inout.mask; - - sp_depth_test_quad(qs, quad); /* quad->mask is updated */ - - /* update stencil buffer values according to z pass/fail result */ - if (zFailOp != PIPE_STENCIL_OP_KEEP) { - const unsigned failMask = origMask & ~quad->inout.mask; - apply_stencil_op(stencilVals, failMask, zFailOp, ref, wrtMask); - } - - if (zPassOp != PIPE_STENCIL_OP_KEEP) { - const unsigned passMask = origMask & quad->inout.mask; - apply_stencil_op(stencilVals, passMask, zPassOp, ref, wrtMask); - } - } - else { - /* no depth test, apply Zpass operator to stencil buffer values */ - apply_stencil_op(stencilVals, quad->inout.mask, zPassOp, ref, wrtMask); - } - - } - - /* put new stencil values into cached tile */ - switch (ps->format) { - case PIPE_FORMAT_S8Z24_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - uint s8z24 = tile->data.depth32[y][x]; - s8z24 = (stencilVals[j] << 24) | (s8z24 & 0xffffff); - tile->data.depth32[y][x] = s8z24; - } - break; - case PIPE_FORMAT_Z24S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - uint z24s8 = tile->data.depth32[y][x]; - z24s8 = (z24s8 & 0xffffff00) | stencilVals[j]; - tile->data.depth32[y][x] = z24s8; - } - break; - case PIPE_FORMAT_S8_UNORM: - for (j = 0; j < QUAD_SIZE; j++) { - int x = quad->input.x0 % TILE_SIZE + (j & 1); - int y = quad->input.y0 % TILE_SIZE + (j >> 1); - tile->data.stencil8[y][x] = stencilVals[j]; - } - break; - default: - assert(0); - } - - if (quad->inout.mask) - qs->next->run(qs->next, quad); -} - - -static void stencil_begin(struct quad_stage *qs) -{ - qs->next->begin(qs->next); -} - - -static void stencil_destroy(struct quad_stage *qs) -{ - FREE( qs ); -} - - -struct quad_stage *sp_quad_stencil_test_stage( struct softpipe_context *softpipe ) -{ - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = stencil_begin; - stage->run = stencil_test_quad; - stage->destroy = stencil_destroy; - - return stage; -} diff --git a/src/gallium/drivers/softpipe/sp_quad_stipple.c b/src/gallium/drivers/softpipe/sp_quad_stipple.c index 07162db7b6e..a0527a596a6 100644 --- a/src/gallium/drivers/softpipe/sp_quad_stipple.c +++ b/src/gallium/drivers/softpipe/sp_quad_stipple.c @@ -14,14 +14,20 @@ * Apply polygon stipple to quads produced by triangle rasterization */ static void -stipple_quad(struct quad_stage *qs, struct quad_header *quad) +stipple_quad(struct quad_stage *qs, struct quad_header *quads[], unsigned nr) { static const uint bit31 = 1 << 31; static const uint bit30 = 1 << 30; + unsigned pass = nr; + + struct softpipe_context *softpipe = qs->softpipe; + unsigned q; + + pass = 0; + + for (q = 0; q < nr; q++) { + struct quad_header *quad = quads[q]; - if (quad->input.prim == QUAD_PRIM_TRI) { - struct softpipe_context *softpipe = qs->softpipe; - /* need to invert Y to index into OpenGL's stipple pattern */ const int col0 = quad->input.x0 % 32; const int y0 = quad->input.y0; const int y1 = y0 + 1; @@ -41,13 +47,11 @@ stipple_quad(struct quad_stage *qs, struct quad_header *quad) if ((stipple1 & (bit30 >> col0)) == 0) quad->inout.mask &= ~MASK_BOTTOM_RIGHT; - if (!quad->inout.mask) { - /* all fragments failed stipple test, end of quad pipeline */ - return; - } + if (quad->inout.mask) + quads[pass++] = quad; } - qs->next->run(qs->next, quad); + qs->next->run(qs->next, quads, pass); } diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c index ce770184158..81fb7aa20c6 100644 --- a/src/gallium/drivers/softpipe/sp_screen.c +++ b/src/gallium/drivers/softpipe/sp_screen.c @@ -40,7 +40,7 @@ static const char * softpipe_get_vendor(struct pipe_screen *screen) { - return "Tungsten Graphics, Inc."; + return "VMware, Inc."; } @@ -82,7 +82,7 @@ softpipe_get_param(struct pipe_screen *screen, int param) case PIPE_CAP_MAX_TEXTURE_2D_LEVELS: return 13; /* max 4Kx4K */ case PIPE_CAP_MAX_TEXTURE_3D_LEVELS: - return 8; /* max 128x128x128 */ + return 9; /* max 256x256x256 */ case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS: return 13; /* max 4Kx4K */ case PIPE_CAP_TGSI_CONT_SUPPORTED: @@ -135,6 +135,9 @@ softpipe_is_format_supported( struct pipe_screen *screen, target == PIPE_TEXTURE_CUBE); switch(format) { + case PIPE_FORMAT_L16_UNORM: + case PIPE_FORMAT_YCBCR_REV: + case PIPE_FORMAT_YCBCR: case PIPE_FORMAT_DXT1_RGB: case PIPE_FORMAT_DXT1_RGBA: case PIPE_FORMAT_DXT3_RGBA: diff --git a/src/gallium/drivers/softpipe/sp_setup.c b/src/gallium/drivers/softpipe/sp_setup.c index de3ae3c3696..ade125662ad 100644 --- a/src/gallium/drivers/softpipe/sp_setup.c +++ b/src/gallium/drivers/softpipe/sp_setup.c @@ -33,7 +33,6 @@ */ #include "sp_context.h" -#include "sp_prim_setup.h" #include "sp_quad.h" #include "sp_quad_pipe.h" #include "sp_setup.h" @@ -61,87 +60,9 @@ struct edge { int lines; /**< number of lines on this edge */ }; -#if SP_NUM_QUAD_THREADS > 1 -/* Set to 1 if you want other threads to be instantly - * notified of pending jobs. - */ -#define INSTANT_NOTEMPTY_NOTIFY 0 - -struct thread_info -{ - struct setup_context *setup; - uint id; - pipe_thread handle; -}; - -struct quad_job; - -typedef void (* quad_job_routine)( struct setup_context *setup, uint thread, struct quad_job *job ); - -struct quad_job -{ - struct quad_header_input input; - struct quad_header_inout inout; - quad_job_routine routine; -}; - -#define NUM_QUAD_JOBS 64 - -struct quad_job_que -{ - struct quad_job jobs[NUM_QUAD_JOBS]; - uint first; - uint last; - pipe_mutex que_mutex; - pipe_condvar que_notfull_condvar; - pipe_condvar que_notempty_condvar; - uint jobs_added; - uint jobs_done; - pipe_condvar que_done_condvar; -}; - -static void -add_quad_job( struct quad_job_que *que, struct quad_header *quad, quad_job_routine routine ) -{ -#if INSTANT_NOTEMPTY_NOTIFY - boolean empty; -#endif - - /* Wait for empty slot, see if the que is empty. - */ - pipe_mutex_lock( que->que_mutex ); - while ((que->last + 1) % NUM_QUAD_JOBS == que->first) { -#if !INSTANT_NOTEMPTY_NOTIFY - pipe_condvar_broadcast( que->que_notempty_condvar ); -#endif - pipe_condvar_wait( que->que_notfull_condvar, que->que_mutex ); - } -#if INSTANT_NOTEMPTY_NOTIFY - empty = que->last == que->first; -#endif - que->jobs_added++; - pipe_mutex_unlock( que->que_mutex ); +#define MAX_QUADS 16 - /* Submit new job. - */ - que->jobs[que->last].input = quad->input; - que->jobs[que->last].inout = quad->inout; - que->jobs[que->last].routine = routine; - que->last = (que->last + 1) % NUM_QUAD_JOBS; - -#if INSTANT_NOTEMPTY_NOTIFY - /* If the que was empty, notify consumers there's a job to be done. - */ - if (empty) { - pipe_mutex_lock( que->que_mutex ); - pipe_condvar_broadcast( que->que_notempty_condvar ); - pipe_mutex_unlock( que->que_mutex ); - } -#endif -} - -#endif /** * Triangle setup info (derived from draw_stage). @@ -164,22 +85,19 @@ struct setup_context { struct edge emaj; float oneoverarea; + int facing; + + struct quad_header quad[MAX_QUADS]; + struct quad_header *quad_ptrs[MAX_QUADS]; + unsigned count; struct tgsi_interp_coef coef[PIPE_MAX_SHADER_INPUTS]; struct tgsi_interp_coef posCoef; /* For Z, W */ - struct quad_header quad; - -#if SP_NUM_QUAD_THREADS > 1 - struct quad_job_que que; - struct thread_info threads[SP_NUM_QUAD_THREADS]; -#endif struct { int left[2]; /**< [0] = row0, [1] = row1 */ int right[2]; int y; - unsigned y_flags; - unsigned mask; /**< mask of MASK_BOTTOM/TOP_LEFT/RIGHT bits */ } span; #if DEBUG_FRAGS @@ -190,67 +108,6 @@ struct setup_context { unsigned winding; /* which winding to cull */ }; -#if SP_NUM_QUAD_THREADS > 1 - -static PIPE_THREAD_ROUTINE( quad_thread, param ) -{ - struct thread_info *info = (struct thread_info *) param; - struct quad_job_que *que = &info->setup->que; - - for (;;) { - struct quad_job job; - boolean full; - - /* Wait for an available job. - */ - pipe_mutex_lock( que->que_mutex ); - while (que->last == que->first) - pipe_condvar_wait( que->que_notempty_condvar, que->que_mutex ); - - /* See if the que is full. - */ - full = (que->last + 1) % NUM_QUAD_JOBS == que->first; - - /* Take a job and remove it from que. - */ - job = que->jobs[que->first]; - que->first = (que->first + 1) % NUM_QUAD_JOBS; - - /* Notify the producer if the que is not full. - */ - if (full) - pipe_condvar_signal( que->que_notfull_condvar ); - pipe_mutex_unlock( que->que_mutex ); - - job.routine( info->setup, info->id, &job ); - - /* Notify the producer if that's the last finished job. - */ - pipe_mutex_lock( que->que_mutex ); - que->jobs_done++; - if (que->jobs_added == que->jobs_done) - pipe_condvar_signal( que->que_done_condvar ); - pipe_mutex_unlock( que->que_mutex ); - } - - return NULL; -} - -#define WAIT_FOR_COMPLETION(setup) \ - do {\ - pipe_mutex_lock( setup->que.que_mutex );\ - if (!INSTANT_NOTEMPTY_NOTIFY)\ - pipe_condvar_broadcast( setup->que.que_notempty_condvar );\ - while (setup->que.jobs_added != setup->que.jobs_done)\ - pipe_condvar_wait( setup->que.que_done_condvar, setup->que.que_mutex );\ - pipe_mutex_unlock( setup->que.que_mutex );\ - } while (0) - -#else - -#define WAIT_FOR_COMPLETION(setup) ((void) 0) - -#endif @@ -313,98 +170,18 @@ quad_clip( struct setup_context *setup, struct quad_header *quad ) * Emit a quad (pass to next stage) with clipping. */ static INLINE void -clip_emit_quad( struct setup_context *setup, struct quad_header *quad, uint thread ) +clip_emit_quad( struct setup_context *setup, struct quad_header *quad ) { quad_clip( setup, quad ); + if (quad->inout.mask) { struct softpipe_context *sp = setup->softpipe; - sp->quad[thread].first->run( sp->quad[thread].first, quad ); + sp->quad.first->run( sp->quad.first, &quad, 1 ); } } -#if SP_NUM_QUAD_THREADS > 1 - -static void -clip_emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job ) -{ - struct quad_header quad; - - quad.input = job->input; - quad.inout = job->inout; - quad.coef = setup->quad.coef; - quad.posCoef = setup->quad.posCoef; - quad.nr_attrs = setup->quad.nr_attrs; - clip_emit_quad( setup, &quad, thread ); -} - -#define CLIP_EMIT_QUAD(setup) add_quad_job( &setup->que, &setup->quad, clip_emit_quad_job ) - -#else - -#define CLIP_EMIT_QUAD(setup) clip_emit_quad( setup, &setup->quad, 0 ) - -#endif - -/** - * Emit a quad (pass to next stage). No clipping is done. - */ -static INLINE void -emit_quad( struct setup_context *setup, struct quad_header *quad, uint thread ) -{ - struct softpipe_context *sp = setup->softpipe; -#if DEBUG_FRAGS - uint mask = quad->inout.mask; -#endif - -#if DEBUG_FRAGS - if (mask & 1) setup->numFragsEmitted++; - if (mask & 2) setup->numFragsEmitted++; - if (mask & 4) setup->numFragsEmitted++; - if (mask & 8) setup->numFragsEmitted++; -#endif - sp->quad[thread].first->run( sp->quad[thread].first, quad ); -#if DEBUG_FRAGS - mask = quad->inout.mask; - if (mask & 1) setup->numFragsWritten++; - if (mask & 2) setup->numFragsWritten++; - if (mask & 4) setup->numFragsWritten++; - if (mask & 8) setup->numFragsWritten++; -#endif -} - -#if SP_NUM_QUAD_THREADS > 1 -static void -emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job ) -{ - struct quad_header quad; - - quad.input = job->input; - quad.inout = job->inout; - quad.coef = setup->quad.coef; - quad.posCoef = setup->quad.posCoef; - quad.nr_attrs = setup->quad.nr_attrs; - emit_quad( setup, &quad, thread ); -} - -#define EMIT_QUAD(setup,x,y,mask) do {\ - setup->quad.input.x0 = x;\ - setup->quad.input.y0 = y;\ - setup->quad.inout.mask = mask;\ - add_quad_job( &setup->que, &setup->quad, emit_quad_job );\ - } while (0) - -#else - -#define EMIT_QUAD(setup,x,y,mask) do {\ - setup->quad.input.x0 = x;\ - setup->quad.input.y0 = y;\ - setup->quad.inout.mask = mask;\ - emit_quad( setup, &setup->quad, 0 );\ - } while (0) - -#endif /** * Given an X or Y coordinate, return the block/quad coordinate that it @@ -412,7 +189,12 @@ emit_quad_job( struct setup_context *setup, uint thread, struct quad_job *job ) */ static INLINE int block( int x ) { - return x & ~1; + return x & ~(2-1); +} + +static INLINE int block_x( int x ) +{ + return x & ~(16-1); } @@ -421,72 +203,63 @@ static INLINE int block( int x ) */ static void flush_spans( struct setup_context *setup ) { + const int step = 16; const int xleft0 = setup->span.left[0]; const int xleft1 = setup->span.left[1]; const int xright0 = setup->span.right[0]; const int xright1 = setup->span.right[1]; - int minleft, maxright; + struct quad_stage *pipe = setup->softpipe->quad.first; + + + int minleft = block_x(MIN2(xleft0, xleft1)); + int maxright = MAX2(xright0, xright1); int x; - switch (setup->span.y_flags) { - case 0x3: - /* both odd and even lines written (both quad rows) */ - minleft = block(MIN2(xleft0, xleft1)); - maxright = block(MAX2(xright0, xright1)); - for (x = minleft; x <= maxright; x += 2) { - /* determine which of the four pixels is inside the span bounds */ - uint mask = 0x0; - if (x >= xleft0 && x < xright0) - mask |= MASK_TOP_LEFT; - if (x >= xleft1 && x < xright1) - mask |= MASK_BOTTOM_LEFT; - if (x+1 >= xleft0 && x+1 < xright0) - mask |= MASK_TOP_RIGHT; - if (x+1 >= xleft1 && x+1 < xright1) - mask |= MASK_BOTTOM_RIGHT; - if (mask) - EMIT_QUAD( setup, x, setup->span.y, mask ); - } - break; - - case 0x1: - /* only even line written (quad top row) */ - minleft = block(xleft0); - maxright = block(xright0); - for (x = minleft; x <= maxright; x += 2) { - uint mask = 0x0; - if (x >= xleft0 && x < xright0) - mask |= MASK_TOP_LEFT; - if (x+1 >= xleft0 && x+1 < xright0) - mask |= MASK_TOP_RIGHT; - if (mask) - EMIT_QUAD( setup, x, setup->span.y, mask ); - } - break; - - case 0x2: - /* only odd line written (quad bottom row) */ - minleft = block(xleft1); - maxright = block(xright1); - for (x = minleft; x <= maxright; x += 2) { - uint mask = 0x0; - if (x >= xleft1 && x < xright1) - mask |= MASK_BOTTOM_LEFT; - if (x+1 >= xleft1 && x+1 < xright1) - mask |= MASK_BOTTOM_RIGHT; - if (mask) - EMIT_QUAD( setup, x, setup->span.y, mask ); - } - break; + for (x = minleft; x < maxright; x += step) { + unsigned skip_left0 = CLAMP(xleft0 - x, 0, step); + unsigned skip_left1 = CLAMP(xleft1 - x, 0, step); + unsigned skip_right0 = CLAMP(x + step - xright0, 0, step); + unsigned skip_right1 = CLAMP(x + step - xright1, 0, step); + unsigned lx = x; + unsigned q = 0; - default: - return; + unsigned skipmask_left0 = (1U << skip_left0) - 1U; + unsigned skipmask_left1 = (1U << skip_left1) - 1U; + + /* These calculations fail when step == 32 and skip_right == 0. + */ + unsigned skipmask_right0 = ~0U << (unsigned)(step - skip_right0); + unsigned skipmask_right1 = ~0U << (unsigned)(step - skip_right1); + + unsigned mask0 = ~skipmask_left0 & ~skipmask_right0; + unsigned mask1 = ~skipmask_left1 & ~skipmask_right1; + + if (mask0 | mask1) { + do { + unsigned quadmask = (mask0 & 3) | ((mask1 & 3) << 2); + if (quadmask) { + setup->quad[q].input.x0 = lx; + setup->quad[q].input.y0 = setup->span.y; + setup->quad[q].input.facing = setup->facing; + setup->quad[q].inout.mask = quadmask; + setup->quad_ptrs[q] = &setup->quad[q]; + q++; + } + mask0 >>= 2; + mask1 >>= 2; + lx += 2; + } while (mask0 | mask1); + + pipe->run( pipe, setup->quad_ptrs, q ); + } } + setup->span.y = 0; - setup->span.y_flags = 0; setup->span.right[0] = 0; setup->span.right[1] = 0; + setup->span.left[0] = 1000000; /* greater than right[0] */ + setup->span.left[1] = 1000000; /* greater than right[1] */ } @@ -496,7 +269,7 @@ static void print_vertex(const struct setup_context *setup, { int i; debug_printf(" Vertex: (%p)\n", v); - for (i = 0; i < setup->quad.nr_attrs; i++) { + for (i = 0; i < setup->quad[0].nr_attrs; i++) { debug_printf(" %d: %f %f %f %f\n", i, v[i][0], v[i][1], v[i][2], v[i][3]); if (util_is_inf_or_nan(v[i][0])) { @@ -601,7 +374,9 @@ static boolean setup_sort_vertices( struct setup_context *setup, * - the GLSL gl_FrontFacing fragment attribute (bool) * - two-sided stencil test */ - setup->quad.input.facing = (det > 0.0) ^ (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW); + setup->facing = + ((det > 0.0) ^ + (setup->softpipe->rasterizer->front_winding == PIPE_WINDING_CW)); return TRUE; } @@ -788,7 +563,7 @@ static void setup_tri_coefficients( struct setup_context *setup ) } if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) { - setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing; + setup->coef[fragSlot].a0[0] = 1.0f - setup->facing; setup->coef[fragSlot].dadx[0] = 0.0; setup->coef[fragSlot].dady[0] = 0.0; } @@ -844,11 +619,10 @@ static void subtriangle( struct setup_context *setup, /* clip top/bottom */ start_y = sy; - finish_y = sy + lines; - if (start_y < miny) start_y = miny; + finish_y = sy + lines; if (finish_y > maxy) finish_y = maxy; @@ -885,7 +659,6 @@ static void subtriangle( struct setup_context *setup, setup->span.left[_y&1] = left; setup->span.right[_y&1] = right; - setup->span.y_flags |= 1<<(_y&1); } } @@ -958,10 +731,9 @@ void setup_tri( struct setup_context *setup, setup_tri_coefficients( setup ); setup_tri_edges( setup ); - setup->quad.input.prim = QUAD_PRIM_TRI; + assert(setup->softpipe->reduced_prim == PIPE_PRIM_TRIANGLES); setup->span.y = 0; - setup->span.y_flags = 0; setup->span.right[0] = 0; setup->span.right[1] = 0; /* setup->span.z_mode = tri_z_mode( setup->ctx ); */ @@ -983,8 +755,6 @@ void setup_tri( struct setup_context *setup, flush_spans( setup ); - WAIT_FOR_COMPLETION(setup); - #if DEBUG_FRAGS printf("Tri: %u frags emitted, %u written\n", setup->numFragsEmitted, @@ -1101,7 +871,7 @@ setup_line_coefficients(struct setup_context *setup, } if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) { - setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing; + setup->coef[fragSlot].a0[0] = 1.0f - setup->facing; setup->coef[fragSlot].dadx[0] = 0.0; setup->coef[fragSlot].dady[0] = 0.0; } @@ -1122,20 +892,20 @@ plot(struct setup_context *setup, int x, int y) const int quadY = y - iy; const int mask = (1 << ix) << (2 * iy); - if (quadX != setup->quad.input.x0 || - quadY != setup->quad.input.y0) + if (quadX != setup->quad[0].input.x0 || + quadY != setup->quad[0].input.y0) { /* flush prev quad, start new quad */ - if (setup->quad.input.x0 != -1) - CLIP_EMIT_QUAD(setup); + if (setup->quad[0].input.x0 != -1) + clip_emit_quad( setup, &setup->quad[0] ); - setup->quad.input.x0 = quadX; - setup->quad.input.y0 = quadY; - setup->quad.inout.mask = 0x0; + setup->quad[0].input.x0 = quadX; + setup->quad[0].input.y0 = quadY; + setup->quad[0].inout.mask = 0x0; } - setup->quad.inout.mask |= mask; + setup->quad[0].inout.mask |= mask; } @@ -1195,17 +965,18 @@ setup_line(struct setup_context *setup, assert(dx >= 0); assert(dy >= 0); + assert(setup->softpipe->reduced_prim == PIPE_PRIM_LINES); + + setup->quad[0].input.x0 = setup->quad[0].input.y0 = -1; + setup->quad[0].inout.mask = 0x0; - setup->quad.input.x0 = setup->quad.input.y0 = -1; - setup->quad.inout.mask = 0x0; - setup->quad.input.prim = QUAD_PRIM_LINE; /* XXX temporary: set coverage to 1.0 so the line appears * if AA mode happens to be enabled. */ - setup->quad.input.coverage[0] = - setup->quad.input.coverage[1] = - setup->quad.input.coverage[2] = - setup->quad.input.coverage[3] = 1.0; + setup->quad[0].input.coverage[0] = + setup->quad[0].input.coverage[1] = + setup->quad[0].input.coverage[2] = + setup->quad[0].input.coverage[3] = 1.0; if (dx > dy) { /*** X-major line ***/ @@ -1249,11 +1020,9 @@ setup_line(struct setup_context *setup, } /* draw final quad */ - if (setup->quad.inout.mask) { - CLIP_EMIT_QUAD(setup); + if (setup->quad[0].inout.mask) { + clip_emit_quad( setup, &setup->quad[0] ); } - - WAIT_FOR_COMPLETION(setup); } @@ -1300,6 +1069,8 @@ setup_point( struct setup_context *setup, if (softpipe->no_rast) return; + assert(setup->softpipe->reduced_prim == PIPE_PRIM_POINTS); + /* For points, all interpolants are constant-valued. * However, for point sprites, we'll need to setup texcoords appropriately. * XXX: which coefficients are the texcoords??? @@ -1346,22 +1117,21 @@ setup_point( struct setup_context *setup, } if (spfs->info.input_semantic_name[fragSlot] == TGSI_SEMANTIC_FACE) { - setup->coef[fragSlot].a0[0] = 1.0f - setup->quad.input.facing; + setup->coef[fragSlot].a0[0] = 1.0f - setup->facing; setup->coef[fragSlot].dadx[0] = 0.0; setup->coef[fragSlot].dady[0] = 0.0; } } - setup->quad.input.prim = QUAD_PRIM_POINT; if (halfSize <= 0.5 && !round) { /* special case for 1-pixel points */ const int ix = ((int) x) & 1; const int iy = ((int) y) & 1; - setup->quad.input.x0 = (int) x - ix; - setup->quad.input.y0 = (int) y - iy; - setup->quad.inout.mask = (1 << ix) << (2 * iy); - CLIP_EMIT_QUAD(setup); + setup->quad[0].input.x0 = (int) x - ix; + setup->quad[0].input.y0 = (int) y - iy; + setup->quad[0].inout.mask = (1 << ix) << (2 * iy); + clip_emit_quad( setup, &setup->quad[0] ); } else { if (round) { @@ -1381,15 +1151,15 @@ setup_point( struct setup_context *setup, for (ix = ixmin; ix <= ixmax; ix += 2) { float dx, dy, dist2, cover; - setup->quad.inout.mask = 0x0; + setup->quad[0].inout.mask = 0x0; dx = (ix + 0.5f) - x; dy = (iy + 0.5f) - y; dist2 = dx * dx + dy * dy; if (dist2 <= rmax2) { cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.input.coverage[QUAD_TOP_LEFT] = MIN2(cover, 1.0f); - setup->quad.inout.mask |= MASK_TOP_LEFT; + setup->quad[0].input.coverage[QUAD_TOP_LEFT] = MIN2(cover, 1.0f); + setup->quad[0].inout.mask |= MASK_TOP_LEFT; } dx = (ix + 1.5f) - x; @@ -1397,8 +1167,8 @@ setup_point( struct setup_context *setup, dist2 = dx * dx + dy * dy; if (dist2 <= rmax2) { cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.input.coverage[QUAD_TOP_RIGHT] = MIN2(cover, 1.0f); - setup->quad.inout.mask |= MASK_TOP_RIGHT; + setup->quad[0].input.coverage[QUAD_TOP_RIGHT] = MIN2(cover, 1.0f); + setup->quad[0].inout.mask |= MASK_TOP_RIGHT; } dx = (ix + 0.5f) - x; @@ -1406,8 +1176,8 @@ setup_point( struct setup_context *setup, dist2 = dx * dx + dy * dy; if (dist2 <= rmax2) { cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.input.coverage[QUAD_BOTTOM_LEFT] = MIN2(cover, 1.0f); - setup->quad.inout.mask |= MASK_BOTTOM_LEFT; + setup->quad[0].input.coverage[QUAD_BOTTOM_LEFT] = MIN2(cover, 1.0f); + setup->quad[0].inout.mask |= MASK_BOTTOM_LEFT; } dx = (ix + 1.5f) - x; @@ -1415,14 +1185,14 @@ setup_point( struct setup_context *setup, dist2 = dx * dx + dy * dy; if (dist2 <= rmax2) { cover = 1.0F - (dist2 - rmin2) * cscale; - setup->quad.input.coverage[QUAD_BOTTOM_RIGHT] = MIN2(cover, 1.0f); - setup->quad.inout.mask |= MASK_BOTTOM_RIGHT; + setup->quad[0].input.coverage[QUAD_BOTTOM_RIGHT] = MIN2(cover, 1.0f); + setup->quad[0].inout.mask |= MASK_BOTTOM_RIGHT; } - if (setup->quad.inout.mask) { - setup->quad.input.x0 = ix; - setup->quad.input.y0 = iy; - CLIP_EMIT_QUAD(setup); + if (setup->quad[0].inout.mask) { + setup->quad[0].input.x0 = ix; + setup->quad[0].input.y0 = iy; + clip_emit_quad( setup, &setup->quad[0] ); } } } @@ -1466,33 +1236,25 @@ setup_point( struct setup_context *setup, mask &= (MASK_BOTTOM_LEFT | MASK_TOP_LEFT); } - setup->quad.inout.mask = mask; - setup->quad.input.x0 = ix; - setup->quad.input.y0 = iy; - CLIP_EMIT_QUAD(setup); + setup->quad[0].inout.mask = mask; + setup->quad[0].input.x0 = ix; + setup->quad[0].input.y0 = iy; + clip_emit_quad( setup, &setup->quad[0] ); } } } } - - WAIT_FOR_COMPLETION(setup); } void setup_prepare( struct setup_context *setup ) { struct softpipe_context *sp = setup->softpipe; - unsigned i; if (sp->dirty) { softpipe_update_derived(sp); } - /* Note: nr_attrs is only used for debugging (vertex printing) */ - setup->quad.nr_attrs = draw_num_vs_outputs(sp->draw); - - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - sp->quad[i].first->begin( sp->quad[i].first ); - } + sp->quad.first->begin( sp->quad.first ); if (sp->reduced_api_prim == PIPE_PRIM_TRIANGLES && sp->rasterizer->fill_cw == PIPE_POLYGON_MODE_FILL && @@ -1520,30 +1282,17 @@ void setup_destroy_context( struct setup_context *setup ) struct setup_context *setup_create_context( struct softpipe_context *softpipe ) { struct setup_context *setup = CALLOC_STRUCT(setup_context); -#if SP_NUM_QUAD_THREADS > 1 - uint i; -#endif + unsigned i; setup->softpipe = softpipe; - setup->quad.coef = setup->coef; - setup->quad.posCoef = &setup->posCoef; - -#if SP_NUM_QUAD_THREADS > 1 - setup->que.first = 0; - setup->que.last = 0; - pipe_mutex_init( setup->que.que_mutex ); - pipe_condvar_init( setup->que.que_notfull_condvar ); - pipe_condvar_init( setup->que.que_notempty_condvar ); - setup->que.jobs_added = 0; - setup->que.jobs_done = 0; - pipe_condvar_init( setup->que.que_done_condvar ); - for (i = 0; i < SP_NUM_QUAD_THREADS; i++) { - setup->threads[i].setup = setup; - setup->threads[i].id = i; - setup->threads[i].handle = pipe_thread_create( quad_thread, &setup->threads[i] ); + for (i = 0; i < MAX_QUADS; i++) { + setup->quad[i].coef = setup->coef; + setup->quad[i].posCoef = &setup->posCoef; } -#endif + + setup->span.left[0] = 1000000; /* greater than right[0] */ + setup->span.left[1] = 1000000; /* greater than right[1] */ return setup; } diff --git a/src/gallium/drivers/softpipe/sp_state.h b/src/gallium/drivers/softpipe/sp_state.h index 9776e978e3e..77ee3c1136b 100644 --- a/src/gallium/drivers/softpipe/sp_state.h +++ b/src/gallium/drivers/softpipe/sp_state.h @@ -87,6 +87,7 @@ struct sp_fragment_shader { struct sp_vertex_shader { struct pipe_shader_state shader; struct draw_vertex_shader *draw_data; + int max_sampler; /* -1 if no samplers */ }; diff --git a/src/gallium/drivers/softpipe/sp_state_blend.c b/src/gallium/drivers/softpipe/sp_state_blend.c index 384fe559afd..efed082f823 100644 --- a/src/gallium/drivers/softpipe/sp_state_blend.c +++ b/src/gallium/drivers/softpipe/sp_state_blend.c @@ -45,7 +45,7 @@ void softpipe_bind_blend_state( struct pipe_context *pipe, { struct softpipe_context *softpipe = softpipe_context(pipe); - softpipe->blend = (const struct pipe_blend_state *)blend; + softpipe->blend = (struct pipe_blend_state *)blend; softpipe->dirty |= SP_NEW_BLEND; } @@ -86,7 +86,7 @@ softpipe_bind_depth_stencil_state(struct pipe_context *pipe, { struct softpipe_context *softpipe = softpipe_context(pipe); - softpipe->depth_stencil = (const struct pipe_depth_stencil_alpha_state *)depth_stencil; + softpipe->depth_stencil = (struct pipe_depth_stencil_alpha_state *)depth_stencil; softpipe->dirty |= SP_NEW_DEPTH_STENCIL_ALPHA; } diff --git a/src/gallium/drivers/softpipe/sp_state_derived.c b/src/gallium/drivers/softpipe/sp_state_derived.c index 9258f641091..1faeca1c2a3 100644 --- a/src/gallium/drivers/softpipe/sp_state_derived.c +++ b/src/gallium/drivers/softpipe/sp_state_derived.c @@ -32,7 +32,10 @@ #include "draw/draw_vertex.h" #include "draw/draw_private.h" #include "sp_context.h" +#include "sp_screen.h" #include "sp_state.h" +#include "sp_texture.h" +#include "sp_tex_tile_cache.h" /** @@ -65,24 +68,19 @@ softpipe_get_vertex_info(struct softpipe_context *softpipe) const struct sp_fragment_shader *spfs = softpipe->fs; const enum interp_mode colorInterp = softpipe->rasterizer->flatshade ? INTERP_CONSTANT : INTERP_LINEAR; + struct vertex_info *vinfo_vbuf = &softpipe->vertex_info_vbuf; + const uint num = draw_num_vs_outputs(softpipe->draw); uint i; - if (softpipe->vbuf) { - /* if using the post-transform vertex buffer, tell draw_vbuf to - * simply emit the whole post-xform vertex as-is: - */ - struct vertex_info *vinfo_vbuf = &softpipe->vertex_info_vbuf; - const uint num = draw_num_vs_outputs(softpipe->draw); - uint i; - - /* No longer any need to try and emit draw vertex_header info. - */ - vinfo_vbuf->num_attribs = 0; - for (i = 0; i < num; i++) { - draw_emit_vertex_attr(vinfo_vbuf, EMIT_4F, INTERP_PERSPECTIVE, i); - } - draw_compute_vertex_size(vinfo_vbuf); + /* Tell draw_vbuf to simply emit the whole post-xform vertex + * as-is. No longer any need to try and emit draw vertex_header + * info. + */ + vinfo_vbuf->num_attribs = 0; + for (i = 0; i < num; i++) { + draw_emit_vertex_attr(vinfo_vbuf, EMIT_4F, INTERP_PERSPECTIVE, i); } + draw_compute_vertex_size(vinfo_vbuf); /* * Loop over fragment shader inputs, searching for the matching output @@ -181,11 +179,19 @@ softpipe_get_vbuf_vertex_info(struct softpipe_context *softpipe) static void compute_cliprect(struct softpipe_context *sp) { + /* SP_NEW_FRAMEBUFFER + */ uint surfWidth = sp->framebuffer.width; uint surfHeight = sp->framebuffer.height; + /* SP_NEW_RASTERIZER + */ if (sp->rasterizer->scissor) { - /* clip to scissor rect */ + + /* SP_NEW_SCISSOR + * + * clip to scissor rect: + */ sp->cliprect.minx = MAX2(sp->scissor.minx, 0); sp->cliprect.miny = MAX2(sp->scissor.miny, 0); sp->cliprect.maxx = MIN2(sp->scissor.maxx, surfWidth); @@ -201,27 +207,63 @@ compute_cliprect(struct softpipe_context *sp) } +static void +update_tgsi_samplers( struct softpipe_context *softpipe ) +{ + unsigned i; + + softpipe_reset_sampler_varients( softpipe ); + + for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { + struct softpipe_tex_tile_cache *tc = softpipe->tex_cache[i]; + if (tc->texture) { + struct softpipe_texture *spt = softpipe_texture(tc->texture); + if (spt->timestamp != tc->timestamp) { + sp_tex_tile_cache_validate_texture( tc ); + /* + _debug_printf("INV %d %d\n", tc->timestamp, spt->timestamp); + */ + tc->timestamp = spt->timestamp; + } + } + } +} + + /* Hopefully this will remain quite simple, otherwise need to pull in * something like the state tracker mechanism. */ void softpipe_update_derived( struct softpipe_context *softpipe ) { + struct softpipe_screen *sp_screen = softpipe_screen(softpipe->pipe.screen); + + /* Check for updated textures. + */ + if (softpipe->tex_timestamp != sp_screen->timestamp) { + softpipe->tex_timestamp = sp_screen->timestamp; + softpipe->dirty |= SP_NEW_TEXTURE; + } + + if (softpipe->dirty & (SP_NEW_SAMPLER | + SP_NEW_TEXTURE | + SP_NEW_FS | + SP_NEW_VS)) + update_tgsi_samplers( softpipe ); + if (softpipe->dirty & (SP_NEW_RASTERIZER | SP_NEW_FS | SP_NEW_VS)) invalidate_vertex_layout( softpipe ); if (softpipe->dirty & (SP_NEW_SCISSOR | - SP_NEW_DEPTH_STENCIL_ALPHA | + SP_NEW_RASTERIZER | SP_NEW_FRAMEBUFFER)) compute_cliprect(softpipe); if (softpipe->dirty & (SP_NEW_BLEND | SP_NEW_DEPTH_STENCIL_ALPHA | SP_NEW_FRAMEBUFFER | - SP_NEW_RASTERIZER | - SP_NEW_FS | - SP_NEW_QUERY)) + SP_NEW_FS)) sp_build_quad_pipeline(softpipe); softpipe->dirty = 0; diff --git a/src/gallium/drivers/softpipe/sp_state_fs.c b/src/gallium/drivers/softpipe/sp_state_fs.c index 4330c203935..256faa94b84 100644 --- a/src/gallium/drivers/softpipe/sp_state_fs.c +++ b/src/gallium/drivers/softpipe/sp_state_fs.c @@ -31,9 +31,8 @@ #include "pipe/p_defines.h" #include "util/u_memory.h" -#include "pipe/internal/p_winsys_screen.h" -#include "pipe/p_shader_tokens.h" #include "draw/draw_context.h" +#include "draw/draw_vs.h" #include "tgsi/tgsi_dump.h" #include "tgsi/tgsi_scan.h" #include "tgsi/tgsi_parse.h" @@ -51,12 +50,9 @@ softpipe_create_fs_state(struct pipe_context *pipe, tgsi_dump(templ->tokens, 0); /* codegen */ - state = softpipe_create_fs_llvm( softpipe, templ ); + state = softpipe_create_fs_sse( softpipe, templ ); if (!state) { - state = softpipe_create_fs_sse( softpipe, templ ); - if (!state) { - state = softpipe_create_fs_exec( softpipe, templ ); - } + state = softpipe_create_fs_exec( softpipe, templ ); } assert(state); @@ -111,6 +107,8 @@ softpipe_create_vs_state(struct pipe_context *pipe, if (state->draw_data == NULL) goto fail; + state->max_sampler = state->draw_data->info.file_max[TGSI_FILE_SAMPLER]; + return state; fail: @@ -128,7 +126,7 @@ softpipe_bind_vs_state(struct pipe_context *pipe, void *vs) { struct softpipe_context *softpipe = softpipe_context(pipe); - softpipe->vs = (const struct sp_vertex_shader *)vs; + softpipe->vs = (struct sp_vertex_shader *) vs; draw_bind_vertex_shader(softpipe->draw, (softpipe->vs ? softpipe->vs->draw_data : NULL)); @@ -142,8 +140,7 @@ softpipe_delete_vs_state(struct pipe_context *pipe, void *vs) { struct softpipe_context *softpipe = softpipe_context(pipe); - struct sp_vertex_shader *state = - (struct sp_vertex_shader *)vs; + struct sp_vertex_shader *state = (struct sp_vertex_shader *) vs; draw_delete_vertex_shader(softpipe->draw, state->draw_data); FREE( state ); diff --git a/src/gallium/drivers/softpipe/sp_state_sampler.c b/src/gallium/drivers/softpipe/sp_state_sampler.c index cb517b02e44..db0b8ab76b1 100644 --- a/src/gallium/drivers/softpipe/sp_state_sampler.c +++ b/src/gallium/drivers/softpipe/sp_state_sampler.c @@ -32,21 +32,37 @@ #include "util/u_memory.h" #include "draw/draw_context.h" +#include "draw/draw_context.h" #include "sp_context.h" -#include "sp_context.h" #include "sp_state.h" #include "sp_texture.h" -#include "sp_tile_cache.h" -#include "draw/draw_context.h" +#include "sp_tex_sample.h" +#include "sp_tex_tile_cache.h" +struct sp_sampler { + struct pipe_sampler_state base; + struct sp_sampler_varient *varients; + struct sp_sampler_varient *current; +}; + +static struct sp_sampler *sp_sampler( struct pipe_sampler_state *sampler ) +{ + return (struct sp_sampler *)sampler; +} + void * softpipe_create_sampler_state(struct pipe_context *pipe, const struct pipe_sampler_state *sampler) { - return mem_dup(sampler, sizeof(*sampler)); + struct sp_sampler *sp_sampler = CALLOC_STRUCT(sp_sampler); + + sp_sampler->base = *sampler; + sp_sampler->varients = NULL; + + return (void *)sp_sampler; } @@ -97,7 +113,7 @@ softpipe_set_sampler_textures(struct pipe_context *pipe, struct pipe_texture *tex = i < num ? texture[i] : NULL; pipe_texture_reference(&softpipe->texture[i], tex); - sp_tile_cache_set_texture(pipe, softpipe->tex_cache[i], tex); + sp_tex_tile_cache_set_texture(softpipe->tex_cache[i], tex); } softpipe->num_textures = num; @@ -106,10 +122,111 @@ softpipe_set_sampler_textures(struct pipe_context *pipe, } +/** + * Find/create an sp_sampler_varient object for sampling the given texture, + * sampler and tex unit. + * + * Note that the tex unit is significant. We can't re-use a sampler + * varient for multiple texture units because the sampler varient contains + * the texture object pointer. If the texture object pointer were stored + * somewhere outside the sampler varient, we could re-use samplers for + * multiple texture units. + */ +static struct sp_sampler_varient * +get_sampler_varient( unsigned unit, + struct sp_sampler *sampler, + struct pipe_texture *texture, + unsigned processor ) +{ + struct softpipe_texture *sp_texture = softpipe_texture(texture); + struct sp_sampler_varient *v = NULL; + union sp_sampler_key key; + + /* if this fails, widen the key.unit field and update this assertion */ + assert(PIPE_MAX_SAMPLERS <= 16); + + key.bits.target = sp_texture->base.target; + key.bits.is_pot = sp_texture->pot; + key.bits.processor = processor; + key.bits.unit = unit; + key.bits.pad = 0; + + if (sampler->current && + key.value == sampler->current->key.value) { + v = sampler->current; + } + + if (v == NULL) { + for (v = sampler->varients; v; v = v->next) + if (v->key.value == key.value) + break; + + if (v == NULL) { + v = sp_create_sampler_varient( &sampler->base, key ); + v->next = sampler->varients; + sampler->varients = v; + } + } + + sampler->current = v; + return v; +} + + + + +void +softpipe_reset_sampler_varients(struct softpipe_context *softpipe) +{ + int i; + + /* It's a bit hard to build these samplers ahead of time -- don't + * really know which samplers are going to be used for vertex and + * fragment programs. + */ + for (i = 0; i <= softpipe->vs->max_sampler; i++) { + if (softpipe->sampler[i]) { + softpipe->tgsi.vert_samplers_list[i] = + get_sampler_varient( i, + sp_sampler(softpipe->sampler[i]), + softpipe->texture[i], + TGSI_PROCESSOR_VERTEX ); + + sp_sampler_varient_bind_texture( softpipe->tgsi.vert_samplers_list[i], + softpipe->tex_cache[i], + softpipe->texture[i] ); + } + } + + for (i = 0; i <= softpipe->fs->info.file_max[TGSI_FILE_SAMPLER]; i++) { + if (softpipe->sampler[i]) { + softpipe->tgsi.frag_samplers_list[i] = + get_sampler_varient( i, + sp_sampler(softpipe->sampler[i]), + softpipe->texture[i], + TGSI_PROCESSOR_FRAGMENT ); + + sp_sampler_varient_bind_texture( softpipe->tgsi.frag_samplers_list[i], + softpipe->tex_cache[i], + softpipe->texture[i] ); + } + } +} + + + void softpipe_delete_sampler_state(struct pipe_context *pipe, void *sampler) { + struct sp_sampler *sp_sampler = (struct sp_sampler *)sampler; + struct sp_sampler_varient *v, *tmp; + + for (v = sp_sampler->varients; v; v = tmp) { + tmp = v->next; + sp_sampler_varient_destroy(v); + } + FREE( sampler ); } diff --git a/src/gallium/drivers/softpipe/sp_state_surface.c b/src/gallium/drivers/softpipe/sp_state_surface.c index 7c06d864a75..bc0e2011300 100644 --- a/src/gallium/drivers/softpipe/sp_state_surface.c +++ b/src/gallium/drivers/softpipe/sp_state_surface.c @@ -53,10 +53,10 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, /* check if changing cbuf */ if (sp->framebuffer.cbufs[i] != fb->cbufs[i]) { /* flush old */ - sp_flush_tile_cache(sp, sp->cbuf_cache[i]); + sp_flush_tile_cache(sp->cbuf_cache[i]); /* assign new */ - sp->framebuffer.cbufs[i] = fb->cbufs[i]; + pipe_surface_reference(&sp->framebuffer.cbufs[i], fb->cbufs[i]); /* update cache */ sp_tile_cache_set_surface(sp->cbuf_cache[i], fb->cbufs[i]); @@ -68,58 +68,28 @@ softpipe_set_framebuffer_state(struct pipe_context *pipe, /* zbuf changing? */ if (sp->framebuffer.zsbuf != fb->zsbuf) { /* flush old */ - sp_flush_tile_cache(sp, sp->zsbuf_cache); + sp_flush_tile_cache(sp->zsbuf_cache); /* assign new */ - sp->framebuffer.zsbuf = fb->zsbuf; + pipe_surface_reference(&sp->framebuffer.zsbuf, fb->zsbuf); /* update cache */ sp_tile_cache_set_surface(sp->zsbuf_cache, fb->zsbuf); - } - -#if 0 - /* XXX combined depth/stencil here */ - - /* sbuf changing? */ - if (sp->framebuffer.sbuf != fb->sbuf) { - /* flush old */ - sp_flush_tile_cache(sp, sp->sbuf_cache_sep); - - /* assign new */ - sp->framebuffer.sbuf = fb->sbuf; - - /* update cache */ - if (fb->sbuf != fb->zbuf) { - /* separate stencil buf */ - sp->sbuf_cache = sp->sbuf_cache_sep; - sp_tile_cache_set_surface(sp->sbuf_cache, fb->sbuf); - } - else { - /* combined depth/stencil */ - sp->sbuf_cache = sp->zbuf_cache; - sp_tile_cache_set_surface(sp->sbuf_cache, fb->sbuf); - } - } -#endif - /* Tell draw module how deep the Z/depth buffer is */ - { - int depth_bits; - double mrd; + /* Tell draw module how deep the Z/depth buffer is */ if (sp->framebuffer.zsbuf) { + int depth_bits; + double mrd; depth_bits = pf_get_component_bits(sp->framebuffer.zsbuf->format, PIPE_FORMAT_COMP_Z); + if (depth_bits > 16) { + mrd = 0.0000001; + } + else { + mrd = 0.00002; + } + draw_set_mrd(sp->draw, mrd); } - else { - depth_bits = 0; - } - if (depth_bits > 16) { - mrd = 0.0000001; - } - else { - mrd = 0.00002; - } - draw_set_mrd(sp->draw, mrd); } sp->framebuffer.width = fb->width; diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c index f99a30277dd..c22ee86b66c 100644 --- a/src/gallium/drivers/softpipe/sp_tex_sample.c +++ b/src/gallium/drivers/softpipe/sp_tex_sample.c @@ -31,29 +31,33 @@ * * Authors: * Brian Paul + * Keith Whitwell */ -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_texture.h" -#include "sp_tex_sample.h" -#include "sp_tile_cache.h" #include "pipe/p_context.h" #include "pipe/p_defines.h" +#include "pipe/p_shader_tokens.h" #include "util/u_math.h" #include "util/u_memory.h" +#include "sp_quad.h" /* only for #define QUAD_* tokens */ +#include "sp_tex_sample.h" +#include "sp_tex_tile_cache.h" /* - * Note, the FRAC macro has to work perfectly. Otherwise you'll sometimes - * see 1-pixel bands of improperly weighted linear-filtered textures. + * Return fractional part of 'f'. Used for computing interpolation weights. + * Need to be careful with negative values. + * Note, if this function isn't perfect you'll sometimes see 1-pixel bands + * of improperly weighted linear-filtered textures. * The tests/texwrap.c demo is a good test. - * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0. - * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x). */ -#define FRAC(f) ((f) - util_ifloor(f)) +static INLINE float +frac(float f) +{ + return f - util_ifloor(f); +} + /** @@ -100,10 +104,16 @@ lerp_3d(float a, float b, float c, /** - * If A is a signed integer, A % B doesn't give the right value for A < 0 - * (in terms of texture repeat). Just casting to unsigned fixes that. + * Compute coord % size for repeat wrap modes. + * Note that if coord is a signed integer, coord % size doesn't give + * the right value for coord < 0 (in terms of texture repeat). Just + * casting to unsigned fixes that. */ -#define REMAINDER(A, B) ((unsigned) (A) % (unsigned) (B)) +static INLINE int +repeat(int coord, unsigned size) +{ + return (int) ((unsigned) coord % size); +} /** @@ -115,133 +125,153 @@ lerp_3d(float a, float b, float c, * \param icoord returns the integer texcoords * \return integer texture index */ -static INLINE void -nearest_texcoord_4(unsigned wrapMode, const float s[4], unsigned size, - int icoord[4]) +static void +wrap_nearest_repeat(const float s[4], unsigned size, int icoord[4]) { uint ch; - switch (wrapMode) { - case PIPE_TEX_WRAP_REPEAT: - /* s limited to [0,1) */ - /* i limited to [0,size-1] */ - for (ch = 0; ch < 4; ch++) { - int i = util_ifloor(s[ch] * size); - icoord[ch] = REMAINDER(i, size); - } - return; - case PIPE_TEX_WRAP_CLAMP: + /* s limited to [0,1) */ + /* i limited to [0,size-1] */ + for (ch = 0; ch < 4; ch++) { + int i = util_ifloor(s[ch] * size); + icoord[ch] = repeat(i, size); + } +} + + +static void +wrap_nearest_clamp(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + /* s limited to [0,1] */ + /* i limited to [0,size-1] */ + for (ch = 0; ch < 4; ch++) { + if (s[ch] <= 0.0F) + icoord[ch] = 0; + else if (s[ch] >= 1.0F) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(s[ch] * size); + } +} + + +static void +wrap_nearest_clamp_to_edge(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + /* s limited to [min,max] */ + /* i limited to [0, size-1] */ + const float min = 1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + if (s[ch] < min) + icoord[ch] = 0; + else if (s[ch] > max) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(s[ch] * size); + } +} + + +static void +wrap_nearest_clamp_to_border(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + /* s limited to [min,max] */ + /* i limited to [-1, size] */ + const float min = -1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + if (s[ch] <= min) + icoord[ch] = -1; + else if (s[ch] >= max) + icoord[ch] = size; + else + icoord[ch] = util_ifloor(s[ch] * size); + } +} + + +static void +wrap_nearest_mirror_repeat(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + const float min = 1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + const int flr = util_ifloor(s[ch]); + float u; + if (flr & 1) + u = 1.0F - (s[ch] - (float) flr); + else + u = s[ch] - (float) flr; + if (u < min) + icoord[ch] = 0; + else if (u > max) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(u * size); + } +} + + +static void +wrap_nearest_mirror_clamp(const float s[4], unsigned size, int icoord[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { /* s limited to [0,1] */ /* i limited to [0,size-1] */ - for (ch = 0; ch < 4; ch++) { - if (s[ch] <= 0.0F) - icoord[ch] = 0; - else if (s[ch] >= 1.0F) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(s[ch] * size); - } - return; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - { - /* s limited to [min,max] */ - /* i limited to [0, size-1] */ - const float min = 1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - if (s[ch] < min) - icoord[ch] = 0; - else if (s[ch] > max) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(s[ch] * size); - } - } - return; - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - { - /* s limited to [min,max] */ - /* i limited to [-1, size] */ - const float min = -1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - if (s[ch] <= min) - icoord[ch] = -1; - else if (s[ch] >= max) - icoord[ch] = size; - else - icoord[ch] = util_ifloor(s[ch] * size); - } - } - return; - case PIPE_TEX_WRAP_MIRROR_REPEAT: - { - const float min = 1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - const int flr = util_ifloor(s[ch]); - float u; - if (flr & 1) - u = 1.0F - (s[ch] - (float) flr); - else - u = s[ch] - (float) flr; - if (u < min) - icoord[ch] = 0; - else if (u > max) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(u * size); - } - } - return; - case PIPE_TEX_WRAP_MIRROR_CLAMP: - for (ch = 0; ch < 4; ch++) { - /* s limited to [0,1] */ - /* i limited to [0,size-1] */ - const float u = fabsf(s[ch]); - if (u <= 0.0F) - icoord[ch] = 0; - else if (u >= 1.0F) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(u * size); - } - return; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: - { - /* s limited to [min,max] */ - /* i limited to [0, size-1] */ - const float min = 1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - const float u = fabsf(s[ch]); - if (u < min) - icoord[ch] = 0; - else if (u > max) - icoord[ch] = size - 1; - else - icoord[ch] = util_ifloor(u * size); - } - } - return; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: - { - /* s limited to [min,max] */ - /* i limited to [0, size-1] */ - const float min = -1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - const float u = fabsf(s[ch]); - if (u < min) - icoord[ch] = -1; - else if (u > max) - icoord[ch] = size; - else - icoord[ch] = util_ifloor(u * size); - } - } - return; - default: - assert(0); + const float u = fabsf(s[ch]); + if (u <= 0.0F) + icoord[ch] = 0; + else if (u >= 1.0F) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(u * size); + } +} + + +static void +wrap_nearest_mirror_clamp_to_edge(const float s[4], unsigned size, + int icoord[4]) +{ + uint ch; + /* s limited to [min,max] */ + /* i limited to [0, size-1] */ + const float min = 1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + const float u = fabsf(s[ch]); + if (u < min) + icoord[ch] = 0; + else if (u > max) + icoord[ch] = size - 1; + else + icoord[ch] = util_ifloor(u * size); + } +} + + +static void +wrap_nearest_mirror_clamp_to_border(const float s[4], unsigned size, + int icoord[4]) +{ + uint ch; + /* s limited to [min,max] */ + /* i limited to [0, size-1] */ + const float min = -1.0F / (2.0F * size); + const float max = 1.0F - min; + for (ch = 0; ch < 4; ch++) { + const float u = fabsf(s[ch]); + if (u < min) + icoord[ch] = -1; + else if (u > max) + icoord[ch] = size; + else + icoord[ch] = util_ifloor(u * size); } } @@ -256,125 +286,156 @@ nearest_texcoord_4(unsigned wrapMode, const float s[4], unsigned size, * \param w returns blend factor/weight between texture indexes * \param icoord returns the computed integer texture coords */ -static INLINE void -linear_texcoord_4(unsigned wrapMode, const float s[4], unsigned size, +static void +wrap_linear_repeat(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = s[ch] * size - 0.5F; + icoord0[ch] = repeat(util_ifloor(u), size); + icoord1[ch] = repeat(icoord0[ch] + 1, size); + w[ch] = frac(u); + } +} + + +static void +wrap_linear_clamp(const float s[4], unsigned size, int icoord0[4], int icoord1[4], float w[4]) { uint ch; + for (ch = 0; ch < 4; ch++) { + float u = CLAMP(s[ch], 0.0F, 1.0F); + u = u * size - 0.5f; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); + } +} - switch (wrapMode) { - case PIPE_TEX_WRAP_REPEAT: - for (ch = 0; ch < 4; ch++) { - float u = s[ch] * size - 0.5F; - icoord0[ch] = REMAINDER(util_ifloor(u), size); - icoord1[ch] = REMAINDER(icoord0[ch] + 1, size); - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_CLAMP: - for (ch = 0; ch < 4; ch++) { - float u = CLAMP(s[ch], 0.0F, 1.0F); - u = u * size - 0.5f; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - for (ch = 0; ch < 4; ch++) { - float u = CLAMP(s[ch], 0.0F, 1.0F); - u = u * size - 0.5f; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - if (icoord0[ch] < 0) - icoord0[ch] = 0; - if (icoord1[ch] >= (int) size) - icoord1[ch] = size - 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - { - const float min = -1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - float u = CLAMP(s[ch], min, max); - u = u * size - 0.5f; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - } - break;; - case PIPE_TEX_WRAP_MIRROR_REPEAT: - for (ch = 0; ch < 4; ch++) { - const int flr = util_ifloor(s[ch]); - float u; - if (flr & 1) - u = 1.0F - (s[ch] - (float) flr); - else - u = s[ch] - (float) flr; - u = u * size - 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - if (icoord0[ch] < 0) - icoord0[ch] = 0; - if (icoord1[ch] >= (int) size) - icoord1[ch] = size - 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_MIRROR_CLAMP: - for (ch = 0; ch < 4; ch++) { - float u = fabsf(s[ch]); - if (u >= 1.0F) - u = (float) size; - else - u *= size; - u -= 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: - for (ch = 0; ch < 4; ch++) { - float u = fabsf(s[ch]); - if (u >= 1.0F) - u = (float) size; - else - u *= size; - u -= 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - if (icoord0[ch] < 0) - icoord0[ch] = 0; - if (icoord1[ch] >= (int) size) - icoord1[ch] = size - 1; - w[ch] = FRAC(u); - } - break;; - case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: - { - const float min = -1.0F / (2.0F * size); - const float max = 1.0F - min; - for (ch = 0; ch < 4; ch++) { - float u = fabsf(s[ch]); - if (u <= min) - u = min * size; - else if (u >= max) - u = max * size; - else - u *= size; - u -= 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - } - break;; - default: - assert(0); + +static void +wrap_linear_clamp_to_edge(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = CLAMP(s[ch], 0.0F, 1.0F); + u = u * size - 0.5f; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + if (icoord0[ch] < 0) + icoord0[ch] = 0; + if (icoord1[ch] >= (int) size) + icoord1[ch] = size - 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_clamp_to_border(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + const float min = -1.0F / (2.0F * size); + const float max = 1.0F - min; + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = CLAMP(s[ch], min, max); + u = u * size - 0.5f; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_mirror_repeat(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + const int flr = util_ifloor(s[ch]); + float u; + if (flr & 1) + u = 1.0F - (s[ch] - (float) flr); + else + u = s[ch] - (float) flr; + u = u * size - 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + if (icoord0[ch] < 0) + icoord0[ch] = 0; + if (icoord1[ch] >= (int) size) + icoord1[ch] = size - 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_mirror_clamp(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = fabsf(s[ch]); + if (u >= 1.0F) + u = (float) size; + else + u *= size; + u -= 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_mirror_clamp_to_edge(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = fabsf(s[ch]); + if (u >= 1.0F) + u = (float) size; + else + u *= size; + u -= 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + if (icoord0[ch] < 0) + icoord0[ch] = 0; + if (icoord1[ch] >= (int) size) + icoord1[ch] = size - 1; + w[ch] = frac(u); + } +} + + +static void +wrap_linear_mirror_clamp_to_border(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) +{ + const float min = -1.0F / (2.0F * size); + const float max = 1.0F - min; + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = fabsf(s[ch]); + if (u <= min) + u = min * size; + else if (u >= max) + u = max * size; + else + u *= size; + u -= 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); } } @@ -383,27 +444,27 @@ linear_texcoord_4(unsigned wrapMode, const float s[4], unsigned size, * For RECT textures / unnormalized texcoords * Only a subset of wrap modes supported. */ -static INLINE void -nearest_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size, - int icoord[4]) +static void +wrap_nearest_unorm_clamp(const float s[4], unsigned size, int icoord[4]) { uint ch; - switch (wrapMode) { - case PIPE_TEX_WRAP_CLAMP: - for (ch = 0; ch < 4; ch++) { - int i = util_ifloor(s[ch]); - icoord[ch]= CLAMP(i, 0, (int) size-1); - } - return; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - /* fall-through */ - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - for (ch = 0; ch < 4; ch++) { - icoord[ch]= util_ifloor( CLAMP(s[ch], 0.5F, (float) size - 0.5F) ); - } - return; - default: - assert(0); + for (ch = 0; ch < 4; ch++) { + int i = util_ifloor(s[ch]); + icoord[ch]= CLAMP(i, 0, (int) size-1); + } +} + + +/** + * Handles clamp_to_edge and clamp_to_border: + */ +static void +wrap_nearest_unorm_clamp_to_border(const float s[4], unsigned size, + int icoord[4]) +{ + uint ch; + for (ch = 0; ch < 4; ch++) { + icoord[ch]= util_ifloor( CLAMP(s[ch], 0.5F, (float) size - 0.5F) ); } } @@ -412,358 +473,971 @@ nearest_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size, * For RECT textures / unnormalized texcoords. * Only a subset of wrap modes supported. */ -static INLINE void -linear_texcoord_unnorm_4(unsigned wrapMode, const float s[4], unsigned size, - int icoord0[4], int icoord1[4], float w[4]) +static void +wrap_linear_unorm_clamp(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) { uint ch; - switch (wrapMode) { - case PIPE_TEX_WRAP_CLAMP: - for (ch = 0; ch < 4; ch++) { - /* Not exactly what the spec says, but it matches NVIDIA output */ - float u = CLAMP(s[ch] - 0.5F, 0.0f, (float) size - 1.0f); - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - w[ch] = FRAC(u); - } - return; - case PIPE_TEX_WRAP_CLAMP_TO_EDGE: - /* fall-through */ - case PIPE_TEX_WRAP_CLAMP_TO_BORDER: - for (ch = 0; ch < 4; ch++) { - float u = CLAMP(s[ch], 0.5F, (float) size - 0.5F); - u -= 0.5F; - icoord0[ch] = util_ifloor(u); - icoord1[ch] = icoord0[ch] + 1; - if (icoord1[ch] > (int) size - 1) - icoord1[ch] = size - 1; - w[ch] = FRAC(u); - } - break; - default: - assert(0); + for (ch = 0; ch < 4; ch++) { + /* Not exactly what the spec says, but it matches NVIDIA output */ + float u = CLAMP(s[ch] - 0.5F, 0.0f, (float) size - 1.0f); + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + w[ch] = frac(u); } } -static unsigned -choose_cube_face(float rx, float ry, float rz, float *newS, float *newT) +static void +wrap_linear_unorm_clamp_to_border(const float s[4], unsigned size, + int icoord0[4], int icoord1[4], float w[4]) { - /* - major axis - direction target sc tc ma - ---------- ------------------------------- --- --- --- - +rx TEXTURE_CUBE_MAP_POSITIVE_X_EXT -rz -ry rx - -rx TEXTURE_CUBE_MAP_NEGATIVE_X_EXT +rz -ry rx - +ry TEXTURE_CUBE_MAP_POSITIVE_Y_EXT +rx +rz ry - -ry TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT +rx -rz ry - +rz TEXTURE_CUBE_MAP_POSITIVE_Z_EXT +rx -ry rz - -rz TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT -rx -ry rz - */ - const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz); - unsigned face; - float sc, tc, ma; - - if (arx >= ary && arx >= arz) { - if (rx >= 0.0F) { - face = PIPE_TEX_FACE_POS_X; - sc = -rz; - tc = -ry; - ma = arx; + uint ch; + for (ch = 0; ch < 4; ch++) { + float u = CLAMP(s[ch], 0.5F, (float) size - 0.5F); + u -= 0.5F; + icoord0[ch] = util_ifloor(u); + icoord1[ch] = icoord0[ch] + 1; + if (icoord1[ch] > (int) size - 1) + icoord1[ch] = size - 1; + w[ch] = frac(u); + } +} + + + +/** + * Examine the quad's texture coordinates to compute the partial + * derivatives w.r.t X and Y, then compute lambda (level of detail). + */ +static float +compute_lambda_1d(const struct sp_sampler_varient *samp, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias) +{ + const struct pipe_texture *texture = samp->texture; + const struct pipe_sampler_state *sampler = samp->sampler; + float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]); + float dsdy = fabsf(s[QUAD_TOP_LEFT] - s[QUAD_BOTTOM_LEFT]); + float rho = MAX2(dsdx, dsdy) * texture->width[0]; + float lambda; + + lambda = util_fast_log2(rho); + lambda += lodbias + sampler->lod_bias; + lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod); + + return lambda; +} + + +static float +compute_lambda_2d(const struct sp_sampler_varient *samp, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias) +{ + const struct pipe_texture *texture = samp->texture; + const struct pipe_sampler_state *sampler = samp->sampler; + float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]); + float dsdy = fabsf(s[QUAD_TOP_LEFT] - s[QUAD_BOTTOM_LEFT]); + float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]); + float dtdy = fabsf(t[QUAD_TOP_LEFT] - t[QUAD_BOTTOM_LEFT]); + float maxx = MAX2(dsdx, dsdy) * texture->width[0]; + float maxy = MAX2(dtdx, dtdy) * texture->height[0]; + float rho = MAX2(maxx, maxy); + float lambda; + + lambda = util_fast_log2(rho); + lambda += lodbias + sampler->lod_bias; + lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod); + + return lambda; +} + + +static float +compute_lambda_3d(const struct sp_sampler_varient *samp, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias) +{ + const struct pipe_texture *texture = samp->texture; + const struct pipe_sampler_state *sampler = samp->sampler; + float dsdx = fabsf(s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]); + float dsdy = fabsf(s[QUAD_TOP_LEFT] - s[QUAD_BOTTOM_LEFT]); + float dtdx = fabsf(t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]); + float dtdy = fabsf(t[QUAD_TOP_LEFT] - t[QUAD_BOTTOM_LEFT]); + float dpdx = fabsf(p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT]); + float dpdy = fabsf(p[QUAD_TOP_LEFT] - p[QUAD_BOTTOM_LEFT]); + float maxx = MAX2(dsdx, dsdy) * texture->width[0]; + float maxy = MAX2(dtdx, dtdy) * texture->height[0]; + float maxz = MAX2(dpdx, dpdy) * texture->depth[0]; + float rho, lambda; + + rho = MAX2(maxx, maxy); + rho = MAX2(rho, maxz); + + lambda = util_fast_log2(rho); + lambda += lodbias + sampler->lod_bias; + lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod); + + return lambda; +} + + +/** + * Compute lambda for a vertex texture sampler. + * Since there aren't derivatives to use, just return the LOD bias. + */ +static float +compute_lambda_vert(const struct sp_sampler_varient *samp, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias) +{ + return lodbias; +} + + + +/** + * Get a texel from a texture, using the texture tile cache. + * + * \param addr the template tex address containing cube, z, face info. + * \param x the x coord of texel within 2D image + * \param y the y coord of texel within 2D image + * \param rgba the quad to put the texel/color into + * + * XXX maybe move this into sp_tex_tile_cache.c and merge with the + * sp_get_cached_tile_tex() function. Also, get 4 texels instead of 1... + */ + + + + +static INLINE const float * +get_texel_2d_no_border(const struct sp_sampler_varient *samp, + union tex_tile_address addr, int x, int y) +{ + const struct softpipe_tex_cached_tile *tile; + + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + y %= TILE_SIZE; + x %= TILE_SIZE; + + tile = sp_get_cached_tile_tex(samp->cache, addr); + + return &tile->data.color[y][x][0]; +} + + +static INLINE const float * +get_texel_2d(const struct sp_sampler_varient *samp, + union tex_tile_address addr, int x, int y) +{ + const struct pipe_texture *texture = samp->texture; + unsigned level = addr.bits.level; + + if (x < 0 || x >= (int) texture->width[level] || + y < 0 || y >= (int) texture->height[level]) { + return samp->sampler->border_color; + } + else { + return get_texel_2d_no_border( samp, addr, x, y ); + } +} + + +/* Gather a quad of adjacent texels within a tile: + */ +static INLINE void +get_texel_quad_2d_no_border_single_tile(const struct sp_sampler_varient *samp, + union tex_tile_address addr, + unsigned x, unsigned y, + const float *out[4]) +{ + const struct softpipe_tex_cached_tile *tile; + + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + y %= TILE_SIZE; + x %= TILE_SIZE; + + tile = sp_get_cached_tile_tex(samp->cache, addr); + + out[0] = &tile->data.color[y ][x ][0]; + out[1] = &tile->data.color[y ][x+1][0]; + out[2] = &tile->data.color[y+1][x ][0]; + out[3] = &tile->data.color[y+1][x+1][0]; +} + + +/* Gather a quad of potentially non-adjacent texels: + */ +static INLINE void +get_texel_quad_2d_no_border(const struct sp_sampler_varient *samp, + union tex_tile_address addr, + int x0, int y0, + int x1, int y1, + const float *out[4]) +{ + out[0] = get_texel_2d_no_border( samp, addr, x0, y0 ); + out[1] = get_texel_2d_no_border( samp, addr, x1, y0 ); + out[2] = get_texel_2d_no_border( samp, addr, x0, y1 ); + out[3] = get_texel_2d_no_border( samp, addr, x1, y1 ); +} + +/* Can involve a lot of unnecessary checks for border color: + */ +static INLINE void +get_texel_quad_2d(const struct sp_sampler_varient *samp, + union tex_tile_address addr, + int x0, int y0, + int x1, int y1, + const float *out[4]) +{ + out[0] = get_texel_2d( samp, addr, x0, y0 ); + out[1] = get_texel_2d( samp, addr, x1, y0 ); + out[3] = get_texel_2d( samp, addr, x1, y1 ); + out[2] = get_texel_2d( samp, addr, x0, y1 ); +} + + + +/* 3d varients: + */ +static INLINE const float * +get_texel_3d_no_border(const struct sp_sampler_varient *samp, + union tex_tile_address addr, int x, int y, int z) +{ + const struct softpipe_tex_cached_tile *tile; + + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + addr.bits.z = z; + y %= TILE_SIZE; + x %= TILE_SIZE; + + tile = sp_get_cached_tile_tex(samp->cache, addr); + + return &tile->data.color[y][x][0]; +} + + +static INLINE const float * +get_texel_3d(const struct sp_sampler_varient *samp, + union tex_tile_address addr, int x, int y, int z) +{ + const struct pipe_texture *texture = samp->texture; + unsigned level = addr.bits.level; + + if (x < 0 || x >= (int) texture->width[level] || + y < 0 || y >= (int) texture->height[level] || + z < 0 || z >= (int) texture->depth[level]) { + return samp->sampler->border_color; + } + else { + return get_texel_3d_no_border( samp, addr, x, y, z ); + } +} + + +/** + * Given the logbase2 of a mipmap's base level size and a mipmap level, + * return the size (in texels) of that mipmap level. + * For example, if level[0].width = 256 then base_pot will be 8. + * If level = 2, then we'll return 64 (the width at level=2). + * Return 1 if level > base_pot. + */ +static INLINE unsigned +pot_level_size(unsigned base_pot, unsigned level) +{ + return (base_pot >= level) ? (1 << (base_pot - level)) : 1; +} + + +/* Some image-filter fastpaths: + */ +static INLINE void +img_filter_2d_linear_repeat_POT(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + unsigned j; + unsigned level = samp->level; + unsigned xpot = pot_level_size(samp->xpot, level); + unsigned ypot = pot_level_size(samp->ypot, level); + unsigned xmax = (xpot - 1) & (TILE_SIZE - 1); /* MIN2(TILE_SIZE, xpot) - 1; */ + unsigned ymax = (ypot - 1) & (TILE_SIZE - 1); /* MIN2(TILE_SIZE, ypot) - 1; */ + union tex_tile_address addr; + + addr.value = 0; + addr.bits.level = samp->level; + + for (j = 0; j < QUAD_SIZE; j++) { + int c; + + float u = s[j] * xpot - 0.5F; + float v = t[j] * ypot - 0.5F; + + int uflr = util_ifloor(u); + int vflr = util_ifloor(v); + + float xw = u - (float)uflr; + float yw = v - (float)vflr; + + int x0 = uflr & (xpot - 1); + int y0 = vflr & (ypot - 1); + + const float *tx[4]; + + /* Can we fetch all four at once: + */ + if (x0 < xmax && y0 < ymax) { + get_texel_quad_2d_no_border_single_tile(samp, addr, x0, y0, tx); } else { - face = PIPE_TEX_FACE_NEG_X; - sc = rz; - tc = -ry; - ma = arx; + unsigned x1 = (x0 + 1) & (xpot - 1); + unsigned y1 = (y0 + 1) & (ypot - 1); + get_texel_quad_2d_no_border(samp, addr, x0, y0, x1, y1, tx); + } + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp_2d(xw, yw, + tx[0][c], tx[1][c], + tx[2][c], tx[3][c]); } } - else if (ary >= arx && ary >= arz) { - if (ry >= 0.0F) { - face = PIPE_TEX_FACE_POS_Y; - sc = rx; - tc = rz; - ma = ary; +} + + +static INLINE void +img_filter_2d_nearest_repeat_POT(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + unsigned j; + unsigned level = samp->level; + unsigned xpot = pot_level_size(samp->xpot, level); + unsigned ypot = pot_level_size(samp->ypot, level); + union tex_tile_address addr; + + addr.value = 0; + addr.bits.level = samp->level; + + for (j = 0; j < QUAD_SIZE; j++) { + int c; + + float u = s[j] * xpot; + float v = t[j] * ypot; + + int uflr = util_ifloor(u); + int vflr = util_ifloor(v); + + int x0 = uflr & (xpot - 1); + int y0 = vflr & (ypot - 1); + + const float *out = get_texel_2d_no_border(samp, addr, x0, y0); + + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; } - else { - face = PIPE_TEX_FACE_NEG_Y; - sc = rx; - tc = -rz; - ma = ary; + } +} + + +static INLINE void +img_filter_2d_nearest_clamp_POT(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + unsigned j; + unsigned level = samp->level; + unsigned xpot = pot_level_size(samp->xpot, level); + unsigned ypot = pot_level_size(samp->ypot, level); + union tex_tile_address addr; + + addr.value = 0; + addr.bits.level = samp->level; + + for (j = 0; j < QUAD_SIZE; j++) { + int c; + + float u = s[j] * xpot; + float v = t[j] * ypot; + + int x0, y0; + const float *out; + + x0 = util_ifloor(u); + if (x0 < 0) + x0 = 0; + else if (x0 > xpot - 1) + x0 = xpot - 1; + + y0 = util_ifloor(v); + if (y0 < 0) + y0 = 0; + else if (y0 > ypot - 1) + y0 = ypot - 1; + + out = get_texel_2d_no_border(samp, addr, x0, y0); + + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; } } - else { - if (rz > 0.0F) { - face = PIPE_TEX_FACE_POS_Z; - sc = rx; - tc = -ry; - ma = arz; +} + + +static void +img_filter_1d_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width; + int x[4]; + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + + assert(width > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->nearest_texcoord_s(s, width, x); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *out = get_texel_2d(samp, addr, x[j], 0); + int c; + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; } - else { - face = PIPE_TEX_FACE_NEG_Z; - sc = -rx; - tc = -ry; - ma = arz; + } +} + + +static void +img_filter_2d_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width, height; + int x[4], y[4]; + union tex_tile_address addr; + + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + + assert(width > 0); + assert(height > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->nearest_texcoord_s(s, width, x); + samp->nearest_texcoord_t(t, height, y); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *out = get_texel_2d(samp, addr, x[j], y[j]); + int c; + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; } } +} - *newS = ( sc / ma + 1.0F ) * 0.5F; - *newT = ( tc / ma + 1.0F ) * 0.5F; - return face; +static INLINE union tex_tile_address +face(union tex_tile_address addr, unsigned face ) +{ + addr.bits.face = face; + return addr; } -/** - * Examine the quad's texture coordinates to compute the partial - * derivatives w.r.t X and Y, then compute lambda (level of detail). - * - * This is only done for fragment shaders, not vertex shaders. - */ -static float -compute_lambda(const struct pipe_texture *tex, - const struct pipe_sampler_state *sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias) +static void +img_filter_cube_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) { - float rho, lambda; + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + const unsigned *faces = samp->faces; /* zero when not cube-mapping */ + unsigned level0, j; + int width, height; + int x[4], y[4]; + union tex_tile_address addr; - assert(sampler->normalized_coords); + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; - assert(s); - { - float dsdx = s[QUAD_BOTTOM_RIGHT] - s[QUAD_BOTTOM_LEFT]; - float dsdy = s[QUAD_TOP_LEFT] - s[QUAD_BOTTOM_LEFT]; - dsdx = fabsf(dsdx); - dsdy = fabsf(dsdy); - rho = MAX2(dsdx, dsdy) * tex->width[0]; - } - if (t) { - float dtdx = t[QUAD_BOTTOM_RIGHT] - t[QUAD_BOTTOM_LEFT]; - float dtdy = t[QUAD_TOP_LEFT] - t[QUAD_BOTTOM_LEFT]; - float max; - dtdx = fabsf(dtdx); - dtdy = fabsf(dtdy); - max = MAX2(dtdx, dtdy) * tex->height[0]; - rho = MAX2(rho, max); + assert(width > 0); + assert(height > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->nearest_texcoord_s(s, width, x); + samp->nearest_texcoord_t(t, height, y); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *out = get_texel_2d(samp, face(addr, faces[j]), x[j], y[j]); + int c; + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; + } } - if (p) { - float dpdx = p[QUAD_BOTTOM_RIGHT] - p[QUAD_BOTTOM_LEFT]; - float dpdy = p[QUAD_TOP_LEFT] - p[QUAD_BOTTOM_LEFT]; - float max; - dpdx = fabsf(dpdx); - dpdy = fabsf(dpdy); - max = MAX2(dpdx, dpdy) * tex->depth[0]; - rho = MAX2(rho, max); +} + + +static void +img_filter_3d_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width, height, depth; + int x[4], y[4], z[4]; + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + depth = texture->depth[level0]; + + assert(width > 0); + assert(height > 0); + assert(depth > 0); + + samp->nearest_texcoord_s(s, width, x); + samp->nearest_texcoord_t(t, height, y); + samp->nearest_texcoord_p(p, depth, z); + + addr.value = 0; + addr.bits.level = samp->level; + + for (j = 0; j < QUAD_SIZE; j++) { + const float *out = get_texel_3d(samp, addr, x[j], y[j], z[j]); + int c; + for (c = 0; c < 4; c++) { + rgba[c][j] = out[c]; + } } +} - lambda = util_fast_log2(rho); - lambda += lodbias + sampler->lod_bias; - lambda = CLAMP(lambda, sampler->min_lod, sampler->max_lod); - return lambda; +static void +img_filter_1d_linear(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width; + int x0[4], x1[4]; + float xw[4]; /* weights */ + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + + assert(width > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->linear_texcoord_s(s, width, x0, x1, xw); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *tx0 = get_texel_2d(samp, addr, x0[j], 0); + const float *tx1 = get_texel_2d(samp, addr, x1[j], 0); + int c; + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp(xw[j], tx0[c], tx1[c]); + } + } } -/** - * Do several things here: - * 1. Compute lambda from the texcoords, if needed - * 2. Determine if we're minifying or magnifying - * 3. If minifying, choose mipmap levels - * 4. Return image filter to use within mipmap images - * \param level0 Returns first mipmap level to sample from - * \param level1 Returns second mipmap level to sample from - * \param levelBlend Returns blend factor between levels, in [0,1] - * \param imgFilter Returns either the min or mag filter, depending on lambda - */ static void -choose_mipmap_levels(const struct pipe_texture *texture, - const struct pipe_sampler_state *sampler, +img_filter_2d_linear(struct tgsi_sampler *tgsi_sampler, const float s[QUAD_SIZE], const float t[QUAD_SIZE], const float p[QUAD_SIZE], - boolean computeLambda, float lodbias, - unsigned *level0, unsigned *level1, float *levelBlend, - unsigned *imgFilter) -{ - if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NONE) { - /* no mipmap selection needed */ - *level0 = *level1 = CLAMP((int) sampler->min_lod, - 0, (int) texture->last_level); - - if (sampler->min_img_filter != sampler->mag_img_filter) { - /* non-mipmapped texture, but still need to determine if doing - * minification or magnification. - */ - float lambda = compute_lambda(texture, sampler, s, t, p, lodbias); - if (lambda <= 0.0) { - *imgFilter = sampler->mag_img_filter; - } - else { - *imgFilter = sampler->min_img_filter; - } + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width, height; + int x0[4], y0[4], x1[4], y1[4]; + float xw[4], yw[4]; /* weights */ + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + + assert(width > 0); + assert(height > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->linear_texcoord_s(s, width, x0, x1, xw); + samp->linear_texcoord_t(t, height, y0, y1, yw); + + for (j = 0; j < QUAD_SIZE; j++) { + const float *tx0 = get_texel_2d(samp, addr, x0[j], y0[j]); + const float *tx1 = get_texel_2d(samp, addr, x1[j], y0[j]); + const float *tx2 = get_texel_2d(samp, addr, x0[j], y1[j]); + const float *tx3 = get_texel_2d(samp, addr, x1[j], y1[j]); + int c; + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp_2d(xw[j], yw[j], + tx0[c], tx1[c], + tx2[c], tx3[c]); } - else { - *imgFilter = sampler->mag_img_filter; + } +} + + +static void +img_filter_cube_linear(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + const unsigned *faces = samp->faces; /* zero when not cube-mapping */ + unsigned level0, j; + int width, height; + int x0[4], y0[4], x1[4], y1[4]; + float xw[4], yw[4]; /* weights */ + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + + assert(width > 0); + assert(height > 0); + + addr.value = 0; + addr.bits.level = samp->level; + + samp->linear_texcoord_s(s, width, x0, x1, xw); + samp->linear_texcoord_t(t, height, y0, y1, yw); + + for (j = 0; j < QUAD_SIZE; j++) { + union tex_tile_address addrj = face(addr, faces[j]); + const float *tx0 = get_texel_2d(samp, addrj, x0[j], y0[j]); + const float *tx1 = get_texel_2d(samp, addrj, x1[j], y0[j]); + const float *tx2 = get_texel_2d(samp, addrj, x0[j], y1[j]); + const float *tx3 = get_texel_2d(samp, addrj, x1[j], y1[j]); + int c; + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp_2d(xw[j], yw[j], + tx0[c], tx1[c], + tx2[c], tx3[c]); } } - else { - float lambda; +} - if (computeLambda) - /* fragment shader */ - lambda = compute_lambda(texture, sampler, s, t, p, lodbias); - else - /* vertex shader */ - lambda = lodbias; /* not really a bias, but absolute LOD */ - if (lambda <= 0.0) { /* XXX threshold depends on the filter */ - /* magnifying */ - *imgFilter = sampler->mag_img_filter; - *level0 = *level1 = 0; +static void +img_filter_3d_linear(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + unsigned level0, j; + int width, height, depth; + int x0[4], x1[4], y0[4], y1[4], z0[4], z1[4]; + float xw[4], yw[4], zw[4]; /* interpolation weights */ + union tex_tile_address addr; + + level0 = samp->level; + width = texture->width[level0]; + height = texture->height[level0]; + depth = texture->depth[level0]; + + addr.value = 0; + addr.bits.level = level0; + + assert(width > 0); + assert(height > 0); + assert(depth > 0); + + samp->linear_texcoord_s(s, width, x0, x1, xw); + samp->linear_texcoord_t(t, height, y0, y1, yw); + samp->linear_texcoord_p(p, depth, z0, z1, zw); + + for (j = 0; j < QUAD_SIZE; j++) { + int c; + + const float *tx00 = get_texel_3d(samp, addr, x0[j], y0[j], z0[j]); + const float *tx01 = get_texel_3d(samp, addr, x1[j], y0[j], z0[j]); + const float *tx02 = get_texel_3d(samp, addr, x0[j], y1[j], z0[j]); + const float *tx03 = get_texel_3d(samp, addr, x1[j], y1[j], z0[j]); + + const float *tx10 = get_texel_3d(samp, addr, x0[j], y0[j], z1[j]); + const float *tx11 = get_texel_3d(samp, addr, x1[j], y0[j], z1[j]); + const float *tx12 = get_texel_3d(samp, addr, x0[j], y1[j], z1[j]); + const float *tx13 = get_texel_3d(samp, addr, x1[j], y1[j], z1[j]); + + /* interpolate R, G, B, A */ + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp_3d(xw[j], yw[j], zw[j], + tx00[c], tx01[c], + tx02[c], tx03[c], + tx10[c], tx11[c], + tx12[c], tx13[c]); } - else { - /* minifying */ - *imgFilter = sampler->min_img_filter; - - /* choose mipmap level(s) and compute the blend factor between them */ - if (sampler->min_mip_filter == PIPE_TEX_MIPFILTER_NEAREST) { - /* Nearest mipmap level */ - const int lvl = (int) (lambda + 0.5); - *level0 = - *level1 = CLAMP(lvl, 0, (int) texture->last_level); - } - else { - /* Linear interpolation between mipmap levels */ - const int lvl = (int) lambda; - *level0 = CLAMP(lvl, 0, (int) texture->last_level); - *level1 = CLAMP(lvl + 1, 0, (int) texture->last_level); - *levelBlend = FRAC(lambda); /* blending weight between levels */ + } +} + + +static void +mip_filter_linear(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + int level0; + float lambda; + + lambda = samp->compute_lambda(samp, s, t, p, lodbias); + level0 = (int)lambda; + + if (lambda < 0.0) { + samp->level = 0; + samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba ); + } + else if (level0 >= texture->last_level) { + samp->level = texture->last_level; + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba ); + } + else { + float levelBlend = lambda - level0; + float rgba0[4][4]; + float rgba1[4][4]; + int c,j; + + samp->level = level0; + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba0 ); + + samp->level = level0+1; + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba1 ); + + for (j = 0; j < QUAD_SIZE; j++) { + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp(levelBlend, rgba0[c][j], rgba1[c][j]); } } } } -/** - * Get a texel from a texture, using the texture tile cache. - * - * \param face the cube face in 0..5 - * \param level the mipmap level - * \param x the x coord of texel within 2D image - * \param y the y coord of texel within 2D image - * \param z which slice of a 3D texture - * \param rgba the quad to put the texel/color into - * \param j which element of the rgba quad to write to - * - * XXX maybe move this into sp_tile_cache.c and merge with the - * sp_get_cached_tile_tex() function. Also, get 4 texels instead of 1... - */ static void -get_texel(const struct tgsi_sampler *tgsi_sampler, - unsigned face, unsigned level, int x, int y, int z, - float rgba[NUM_CHANNELS][QUAD_SIZE], unsigned j) +mip_filter_nearest(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + float lambda; - if (x < 0 || x >= (int) texture->width[level] || - y < 0 || y >= (int) texture->height[level] || - z < 0 || z >= (int) texture->depth[level]) { - rgba[0][j] = sampler->border_color[0]; - rgba[1][j] = sampler->border_color[1]; - rgba[2][j] = sampler->border_color[2]; - rgba[3][j] = sampler->border_color[3]; + lambda = samp->compute_lambda(samp, s, t, p, lodbias); + + if (lambda < 0.0) { + samp->level = 0; + samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba ); } else { - const int tx = x % TILE_SIZE; - const int ty = y % TILE_SIZE; - const struct softpipe_cached_tile *tile - = sp_get_cached_tile_tex(sp, samp->cache, - x, y, z, face, level); - rgba[0][j] = tile->data.color[ty][tx][0]; - rgba[1][j] = tile->data.color[ty][tx][1]; - rgba[2][j] = tile->data.color[ty][tx][2]; - rgba[3][j] = tile->data.color[ty][tx][3]; - if (0) - { - debug_printf("Get texel %f %f %f %f from %s\n", - rgba[0][j], rgba[1][j], rgba[2][j], rgba[3][j], - pf_name(texture->format)); - } + samp->level = (int)(lambda + 0.5) ; + samp->level = MIN2(samp->level, (int)texture->last_level); + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba ); + } + +#if 0 + printf("RGBA %g %g %g %g, %g %g %g %g, %g %g %g %g, %g %g %g %g\n", + rgba[0][0], rgba[1][0], rgba[2][0], rgba[3][0], + rgba[0][1], rgba[1][1], rgba[2][1], rgba[3][1], + rgba[0][2], rgba[1][2], rgba[2][2], rgba[3][2], + rgba[0][3], rgba[1][3], rgba[2][3], rgba[3][3]); +#endif +} + + +static void +mip_filter_none(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + float lambda = samp->compute_lambda(samp, s, t, p, lodbias); + + if (lambda < 0.0) { + samp->mag_img_filter( tgsi_sampler, s, t, p, 0, rgba ); + } + else { + samp->min_img_filter( tgsi_sampler, s, t, p, 0, rgba ); } } + /** - * Compare texcoord 'p' (aka R) against texture value 'rgba[0]' - * When we sampled the depth texture, the depth value was put into all - * RGBA channels. We look at the red channel here. - * \param rgba quad of (depth) texel values - * \param p texture 'P' components for four pixels in quad - * \param j which pixel in the quad to test [0..3] + * Specialized version of mip_filter_linear with hard-wired calls to + * 2d lambda calculation and 2d_linear_repeat_POT img filters. */ -static INLINE void -shadow_compare(const struct pipe_sampler_state *sampler, - float rgba[NUM_CHANNELS][QUAD_SIZE], - const float p[QUAD_SIZE], - uint j) +static void +mip_filter_linear_2d_linear_repeat_POT( + struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) { - int k; - switch (sampler->compare_func) { - case PIPE_FUNC_LESS: - k = p[j] < rgba[0][j]; - break; - case PIPE_FUNC_LEQUAL: - k = p[j] <= rgba[0][j]; - break; - case PIPE_FUNC_GREATER: - k = p[j] > rgba[0][j]; - break; - case PIPE_FUNC_GEQUAL: - k = p[j] >= rgba[0][j]; - break; - case PIPE_FUNC_EQUAL: - k = p[j] == rgba[0][j]; - break; - case PIPE_FUNC_NOTEQUAL: - k = p[j] != rgba[0][j]; - break; - case PIPE_FUNC_ALWAYS: - k = 1; - break; - case PIPE_FUNC_NEVER: - k = 0; - break; - default: - k = 0; - assert(0); - break; + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_texture *texture = samp->texture; + int level0; + float lambda; + + lambda = compute_lambda_2d(samp, s, t, p, lodbias); + level0 = (int)lambda; + + /* Catches both negative and large values of level0: + */ + if ((unsigned)level0 >= texture->last_level) { + if (level0 < 0) + samp->level = 0; + else + samp->level = texture->last_level; + + img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba ); } + else { + float levelBlend = lambda - level0; + float rgba0[4][4]; + float rgba1[4][4]; + int c,j; - /* XXX returning result for default GL_DEPTH_TEXTURE_MODE = GL_LUMINANCE */ - rgba[0][j] = rgba[1][j] = rgba[2][j] = (float) k; - rgba[3][j] = 1.0F; + samp->level = level0; + img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba0 ); + + samp->level = level0+1; + img_filter_2d_linear_repeat_POT( tgsi_sampler, s, t, p, 0, rgba1 ); + + for (j = 0; j < QUAD_SIZE; j++) { + for (c = 0; c < 4; c++) { + rgba[c][j] = lerp(levelBlend, rgba0[c][j], rgba1[c][j]); + } + } + } } + /** - * As above, but do four z/texture comparisons. + * Do shadow/depth comparisons. */ -static INLINE void -shadow_compare4(const struct pipe_sampler_state *sampler, - float rgba[NUM_CHANNELS][QUAD_SIZE], - const float p[QUAD_SIZE]) +static void +sample_compare(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) { + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct pipe_sampler_state *sampler = samp->sampler; int j, k0, k1, k2, k3; float val; + samp->mip_filter( tgsi_sampler, s, t, p, lodbias, rgba ); + + /** + * Compare texcoord 'p' (aka R) against texture value 'rgba[0]' + * When we sampled the depth texture, the depth value was put into all + * RGBA channels. We look at the red channel here. + */ + /* compare four texcoords vs. four texture samples */ switch (sampler->compare_func) { case PIPE_FUNC_LESS: @@ -826,470 +1500,392 @@ shadow_compare4(const struct pipe_sampler_state *sampler, /** - * Common code for sampling 1D/2D/cube textures. - * Could probably extend for 3D... + * Compute which cube face is referenced by each texcoord and put that + * info into the sampler faces[] array. Then sample the cube faces */ static void -sp_get_samples_2d_common(const struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE], - const unsigned faces[4]) -{ - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - const struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; - unsigned level0, level1, j, imgFilter; - int width, height; - float levelBlend; - - choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias, - &level0, &level1, &levelBlend, &imgFilter); - - assert(sampler->normalized_coords); - - width = texture->width[level0]; - height = texture->height[level0]; - - assert(width > 0); - - switch (imgFilter) { - case PIPE_TEX_FILTER_NEAREST: - { - int x[4], y[4]; - nearest_texcoord_4(sampler->wrap_s, s, width, x); - nearest_texcoord_4(sampler->wrap_t, t, height, y); - - for (j = 0; j < QUAD_SIZE; j++) { - get_texel(tgsi_sampler, faces[j], level0, x[j], y[j], 0, rgba, j); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - shadow_compare(sampler, rgba, p, j); - } +sample_cube(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + unsigned j; + float ssss[4], tttt[4]; - if (level0 != level1) { - /* get texels from second mipmap level and blend */ - float rgba2[4][4]; - unsigned c; - x[j] /= 2; - y[j] /= 2; - get_texel(tgsi_sampler, faces[j], level1, x[j], y[j], 0, - rgba2, j); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){ - shadow_compare(sampler, rgba2, p, j); - } - - for (c = 0; c < NUM_CHANNELS; c++) { - rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]); - } - } + /* + major axis + direction target sc tc ma + ---------- ------------------------------- --- --- --- + +rx TEXTURE_CUBE_MAP_POSITIVE_X_EXT -rz -ry rx + -rx TEXTURE_CUBE_MAP_NEGATIVE_X_EXT +rz -ry rx + +ry TEXTURE_CUBE_MAP_POSITIVE_Y_EXT +rx +rz ry + -ry TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT +rx -rz ry + +rz TEXTURE_CUBE_MAP_POSITIVE_Z_EXT +rx -ry rz + -rz TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT -rx -ry rz + */ + for (j = 0; j < QUAD_SIZE; j++) { + float rx = s[j]; + float ry = t[j]; + float rz = p[j]; + const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz); + unsigned face; + float sc, tc, ma; + + if (arx >= ary && arx >= arz) { + if (rx >= 0.0F) { + face = PIPE_TEX_FACE_POS_X; + sc = -rz; + tc = -ry; + ma = arx; + } + else { + face = PIPE_TEX_FACE_NEG_X; + sc = rz; + tc = -ry; + ma = arx; } } - break; - case PIPE_TEX_FILTER_LINEAR: - case PIPE_TEX_FILTER_ANISO: + else if (ary >= arx && ary >= arz) { + if (ry >= 0.0F) { + face = PIPE_TEX_FACE_POS_Y; + sc = rx; + tc = rz; + ma = ary; + } + else { + face = PIPE_TEX_FACE_NEG_Y; + sc = rx; + tc = -rz; + ma = ary; + } + } + else { + if (rz > 0.0F) { + face = PIPE_TEX_FACE_POS_Z; + sc = rx; + tc = -ry; + ma = arz; + } + else { + face = PIPE_TEX_FACE_NEG_Z; + sc = -rx; + tc = -ry; + ma = arz; + } + } + { - int x0[4], y0[4], x1[4], y1[4]; - float xw[4], yw[4]; /* weights */ - - linear_texcoord_4(sampler->wrap_s, s, width, x0, x1, xw); - linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw); - - for (j = 0; j < QUAD_SIZE; j++) { - float tx[4][4]; /* texels */ - int c; - get_texel(tgsi_sampler, faces[j], level0, x0[j], y0[j], 0, tx, 0); - get_texel(tgsi_sampler, faces[j], level0, x1[j], y0[j], 0, tx, 1); - get_texel(tgsi_sampler, faces[j], level0, x0[j], y1[j], 0, tx, 2); - get_texel(tgsi_sampler, faces[j], level0, x1[j], y1[j], 0, tx, 3); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - shadow_compare4(sampler, tx, p); - } + const float ima = 1.0 / ma; + ssss[j] = ( sc * ima + 1.0F ) * 0.5F; + tttt[j] = ( tc * ima + 1.0F ) * 0.5F; + samp->faces[j] = face; + } + } - /* interpolate R, G, B, A */ - for (c = 0; c < 4; c++) { - rgba[c][j] = lerp_2d(xw[j], yw[j], - tx[c][0], tx[c][1], - tx[c][2], tx[c][3]); - } + /* In our little pipeline, the compare stage is next. If compare + * is not active, this will point somewhere deeper into the + * pipeline, eg. to mip_filter or even img_filter. + */ + samp->compare(tgsi_sampler, ssss, tttt, NULL, lodbias, rgba); +} - if (level0 != level1) { - /* get texels from second mipmap level and blend */ - float rgba2[4][4]; - x0[j] /= 2; - y0[j] /= 2; - x1[j] /= 2; - y1[j] /= 2; - get_texel(tgsi_sampler, faces[j], level1, x0[j], y0[j], 0, tx, 0); - get_texel(tgsi_sampler, faces[j], level1, x1[j], y0[j], 0, tx, 1); - get_texel(tgsi_sampler, faces[j], level1, x0[j], y1[j], 0, tx, 2); - get_texel(tgsi_sampler, faces[j], level1, x1[j], y1[j], 0, tx, 3); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE){ - shadow_compare4(sampler, tx, p); - } - - /* interpolate R, G, B, A */ - for (c = 0; c < 4; c++) { - rgba2[c][j] = lerp_2d(xw[j], yw[j], - tx[c][0], tx[c][1], tx[c][2], tx[c][3]); - } - - for (c = 0; c < NUM_CHANNELS; c++) { - rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]); - } - } - } - } - break; + + +static wrap_nearest_func +get_nearest_unorm_wrap(unsigned mode) +{ + switch (mode) { + case PIPE_TEX_WRAP_CLAMP: + return wrap_nearest_unorm_clamp; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return wrap_nearest_unorm_clamp_to_border; default: assert(0); + return wrap_nearest_unorm_clamp; } } -static INLINE void -sp_get_samples_1d(const struct tgsi_sampler *sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +static wrap_nearest_func +get_nearest_wrap(unsigned mode) { - static const unsigned faces[4] = {0, 0, 0, 0}; - static const float tzero[4] = {0, 0, 0, 0}; - sp_get_samples_2d_common(sampler, s, tzero, NULL, - computeLambda, lodbias, rgba, faces); + switch (mode) { + case PIPE_TEX_WRAP_REPEAT: + return wrap_nearest_repeat; + case PIPE_TEX_WRAP_CLAMP: + return wrap_nearest_clamp; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + return wrap_nearest_clamp_to_edge; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return wrap_nearest_clamp_to_border; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + return wrap_nearest_mirror_repeat; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + return wrap_nearest_mirror_clamp; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + return wrap_nearest_mirror_clamp_to_edge; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + return wrap_nearest_mirror_clamp_to_border; + default: + assert(0); + return wrap_nearest_repeat; + } } -static INLINE void -sp_get_samples_2d(const struct tgsi_sampler *sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +static wrap_linear_func +get_linear_unorm_wrap(unsigned mode) { - static const unsigned faces[4] = {0, 0, 0, 0}; - sp_get_samples_2d_common(sampler, s, t, p, - computeLambda, lodbias, rgba, faces); + switch (mode) { + case PIPE_TEX_WRAP_CLAMP: + return wrap_linear_unorm_clamp; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return wrap_linear_unorm_clamp_to_border; + default: + assert(0); + return wrap_linear_unorm_clamp; + } } -static INLINE void -sp_get_samples_3d(const struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +static wrap_linear_func +get_linear_wrap(unsigned mode) { - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - const struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; - /* get/map pipe_surfaces corresponding to 3D tex slices */ - unsigned level0, level1, j, imgFilter; - int width, height, depth; - float levelBlend; - const uint face = 0; - - choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias, - &level0, &level1, &levelBlend, &imgFilter); - - assert(sampler->normalized_coords); - - width = texture->width[level0]; - height = texture->height[level0]; - depth = texture->depth[level0]; - - assert(width > 0); - assert(height > 0); - assert(depth > 0); - - switch (imgFilter) { - case PIPE_TEX_FILTER_NEAREST: - { - int x[4], y[4], z[4]; - nearest_texcoord_4(sampler->wrap_s, s, width, x); - nearest_texcoord_4(sampler->wrap_t, t, height, y); - nearest_texcoord_4(sampler->wrap_r, p, depth, z); - for (j = 0; j < QUAD_SIZE; j++) { - get_texel(tgsi_sampler, face, level0, x[j], y[j], z[j], rgba, j); - if (level0 != level1) { - /* get texels from second mipmap level and blend */ - float rgba2[4][4]; - unsigned c; - x[j] /= 2; - y[j] /= 2; - z[j] /= 2; - get_texel(tgsi_sampler, face, level1, x[j], y[j], z[j], rgba2, j); - for (c = 0; c < NUM_CHANNELS; c++) { - rgba[c][j] = lerp(levelBlend, rgba2[c][j], rgba[c][j]); - } - } - } - } - break; - case PIPE_TEX_FILTER_LINEAR: - case PIPE_TEX_FILTER_ANISO: - { - int x0[4], x1[4], y0[4], y1[4], z0[4], z1[4]; - float xw[4], yw[4], zw[4]; /* interpolation weights */ - linear_texcoord_4(sampler->wrap_s, s, width, x0, x1, xw); - linear_texcoord_4(sampler->wrap_t, t, height, y0, y1, yw); - linear_texcoord_4(sampler->wrap_r, p, depth, z0, z1, zw); - - for (j = 0; j < QUAD_SIZE; j++) { - int c; - float tx0[4][4], tx1[4][4]; - get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z0[j], tx0, 0); - get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z0[j], tx0, 1); - get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z0[j], tx0, 2); - get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z0[j], tx0, 3); - get_texel(tgsi_sampler, face, level0, x0[j], y0[j], z1[j], tx1, 0); - get_texel(tgsi_sampler, face, level0, x1[j], y0[j], z1[j], tx1, 1); - get_texel(tgsi_sampler, face, level0, x0[j], y1[j], z1[j], tx1, 2); - get_texel(tgsi_sampler, face, level0, x1[j], y1[j], z1[j], tx1, 3); - - /* interpolate R, G, B, A */ - for (c = 0; c < 4; c++) { - rgba[c][j] = lerp_3d(xw[j], yw[j], zw[j], - tx0[c][0], tx0[c][1], - tx0[c][2], tx0[c][3], - tx1[c][0], tx1[c][1], - tx1[c][2], tx1[c][3]); - } - - if (level0 != level1) { - /* get texels from second mipmap level and blend */ - float rgba2[4][4]; - x0[j] /= 2; - y0[j] /= 2; - z0[j] /= 2; - x1[j] /= 2; - y1[j] /= 2; - z1[j] /= 2; - get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z0[j], tx0, 0); - get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z0[j], tx0, 1); - get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z0[j], tx0, 2); - get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z0[j], tx0, 3); - get_texel(tgsi_sampler, face, level1, x0[j], y0[j], z1[j], tx1, 0); - get_texel(tgsi_sampler, face, level1, x1[j], y0[j], z1[j], tx1, 1); - get_texel(tgsi_sampler, face, level1, x0[j], y1[j], z1[j], tx1, 2); - get_texel(tgsi_sampler, face, level1, x1[j], y1[j], z1[j], tx1, 3); - - /* interpolate R, G, B, A */ - for (c = 0; c < 4; c++) { - rgba2[c][j] = lerp_3d(xw[j], yw[j], zw[j], - tx0[c][0], tx0[c][1], - tx0[c][2], tx0[c][3], - tx1[c][0], tx1[c][1], - tx1[c][2], tx1[c][3]); - } - - /* blend mipmap levels */ - for (c = 0; c < NUM_CHANNELS; c++) { - rgba[c][j] = lerp(levelBlend, rgba[c][j], rgba2[c][j]); - } - } - } - } - break; + switch (mode) { + case PIPE_TEX_WRAP_REPEAT: + return wrap_linear_repeat; + case PIPE_TEX_WRAP_CLAMP: + return wrap_linear_clamp; + case PIPE_TEX_WRAP_CLAMP_TO_EDGE: + return wrap_linear_clamp_to_edge; + case PIPE_TEX_WRAP_CLAMP_TO_BORDER: + return wrap_linear_clamp_to_border; + case PIPE_TEX_WRAP_MIRROR_REPEAT: + return wrap_linear_mirror_repeat; + case PIPE_TEX_WRAP_MIRROR_CLAMP: + return wrap_linear_mirror_clamp; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_EDGE: + return wrap_linear_mirror_clamp_to_edge; + case PIPE_TEX_WRAP_MIRROR_CLAMP_TO_BORDER: + return wrap_linear_mirror_clamp_to_border; default: assert(0); + return wrap_linear_repeat; } } -static void -sp_get_samples_cube(const struct tgsi_sampler *sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +static compute_lambda_func +get_lambda_func(const union sp_sampler_key key) { - unsigned faces[QUAD_SIZE], j; - float ssss[4], tttt[4]; - for (j = 0; j < QUAD_SIZE; j++) { - faces[j] = choose_cube_face(s[j], t[j], p[j], ssss + j, tttt + j); + if (key.bits.processor == TGSI_PROCESSOR_VERTEX) + return compute_lambda_vert; + + switch (key.bits.target) { + case PIPE_TEXTURE_1D: + return compute_lambda_1d; + case PIPE_TEXTURE_2D: + case PIPE_TEXTURE_CUBE: + return compute_lambda_2d; + case PIPE_TEXTURE_3D: + return compute_lambda_3d; + default: + assert(0); + return compute_lambda_1d; } - sp_get_samples_2d_common(sampler, ssss, tttt, NULL, - computeLambda, lodbias, rgba, faces); } -static void -sp_get_samples_rect(const struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) -{ - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - const struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; - const uint face = 0; - unsigned level0, level1, j, imgFilter; - int width, height; - float levelBlend; - - choose_mipmap_levels(texture, sampler, s, t, p, computeLambda, lodbias, - &level0, &level1, &levelBlend, &imgFilter); - - /* texture RECTS cannot be mipmapped */ - assert(level0 == level1); - - width = texture->width[level0]; - height = texture->height[level0]; - - assert(width > 0); - - switch (imgFilter) { - case PIPE_TEX_FILTER_NEAREST: - { - int x[4], y[4]; - nearest_texcoord_unnorm_4(sampler->wrap_s, s, width, x); - nearest_texcoord_unnorm_4(sampler->wrap_t, t, height, y); - for (j = 0; j < QUAD_SIZE; j++) { - get_texel(tgsi_sampler, face, level0, x[j], y[j], 0, rgba, j); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - shadow_compare(sampler, rgba, p, j); - } - } - } +static filter_func +get_img_filter(const union sp_sampler_key key, + unsigned filter, + const struct pipe_sampler_state *sampler) +{ + switch (key.bits.target) { + case PIPE_TEXTURE_1D: + if (filter == PIPE_TEX_FILTER_NEAREST) + return img_filter_1d_nearest; + else + return img_filter_1d_linear; break; - case PIPE_TEX_FILTER_LINEAR: - case PIPE_TEX_FILTER_ANISO: + case PIPE_TEXTURE_2D: + /* Try for fast path: + */ + if (key.bits.is_pot && + sampler->wrap_s == sampler->wrap_t && + sampler->normalized_coords) { - int x0[4], y0[4], x1[4], y1[4]; - float xw[4], yw[4]; /* weights */ - linear_texcoord_unnorm_4(sampler->wrap_s, s, width, x0, x1, xw); - linear_texcoord_unnorm_4(sampler->wrap_t, t, height, y0, y1, yw); - for (j = 0; j < QUAD_SIZE; j++) { - float tx[4][4]; /* texels */ - int c; - get_texel(tgsi_sampler, face, level0, x0[j], y0[j], 0, tx, 0); - get_texel(tgsi_sampler, face, level0, x1[j], y0[j], 0, tx, 1); - get_texel(tgsi_sampler, face, level0, x0[j], y1[j], 0, tx, 2); - get_texel(tgsi_sampler, face, level0, x1[j], y1[j], 0, tx, 3); - if (sampler->compare_mode == PIPE_TEX_COMPARE_R_TO_TEXTURE) { - shadow_compare4(sampler, tx, p); + switch (sampler->wrap_s) { + case PIPE_TEX_WRAP_REPEAT: + switch (filter) { + case PIPE_TEX_FILTER_NEAREST: + return img_filter_2d_nearest_repeat_POT; + case PIPE_TEX_FILTER_LINEAR: + return img_filter_2d_linear_repeat_POT; + default: + break; } - for (c = 0; c < 4; c++) { - rgba[c][j] = lerp_2d(xw[j], yw[j], - tx[c][0], tx[c][1], tx[c][2], tx[c][3]); + break; + case PIPE_TEX_WRAP_CLAMP: + switch (filter) { + case PIPE_TEX_FILTER_NEAREST: + return img_filter_2d_nearest_clamp_POT; + default: + break; } } } + /* Otherwise use default versions: + */ + if (filter == PIPE_TEX_FILTER_NEAREST) + return img_filter_2d_nearest; + else + return img_filter_2d_linear; + break; + case PIPE_TEXTURE_CUBE: + if (filter == PIPE_TEX_FILTER_NEAREST) + return img_filter_cube_nearest; + else + return img_filter_cube_linear; + break; + case PIPE_TEXTURE_3D: + if (filter == PIPE_TEX_FILTER_NEAREST) + return img_filter_3d_nearest; + else + return img_filter_3d_linear; break; default: assert(0); + return img_filter_1d_nearest; } } /** - * Common code for vertex/fragment program texture sampling. + * Bind the given texture object and texture cache to the sampler varient. */ -static INLINE void -sp_get_samples(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - boolean computeLambda, - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +void +sp_sampler_varient_bind_texture( struct sp_sampler_varient *samp, + struct softpipe_tex_tile_cache *tex_cache, + const struct pipe_texture *texture ) { - const struct sp_shader_sampler *samp = sp_shader_sampler(tgsi_sampler); - const struct softpipe_context *sp = samp->sp; - const uint unit = samp->unit; - const struct pipe_texture *texture = sp->texture[unit]; - const struct pipe_sampler_state *sampler = sp->sampler[unit]; - - if (!texture) - return; + const struct pipe_sampler_state *sampler = samp->sampler; - switch (texture->target) { - case PIPE_TEXTURE_1D: - assert(sampler->normalized_coords); - sp_get_samples_1d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - break; - case PIPE_TEXTURE_2D: - if (sampler->normalized_coords) - sp_get_samples_2d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - else - sp_get_samples_rect(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - break; - case PIPE_TEXTURE_3D: - assert(sampler->normalized_coords); - sp_get_samples_3d(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - break; - case PIPE_TEXTURE_CUBE: - assert(sampler->normalized_coords); - sp_get_samples_cube(tgsi_sampler, s, t, p, computeLambda, lodbias, rgba); - break; - default: - assert(0); - } - -#if 0 /* DEBUG */ - { - int i; - printf("Sampled at %f, %f, %f:\n", s[0], t[0], p[0]); - for (i = 0; i < 4; i++) { - printf("Frag %d: %f %f %f %f\n", i, - rgba[0][i], - rgba[1][i], - rgba[2][i], - rgba[3][i]); - } - } -#endif + samp->texture = texture; + samp->cache = tex_cache; + samp->xpot = util_unsigned_logbase2( texture->width[0] ); + samp->ypot = util_unsigned_logbase2( texture->height[0] ); + samp->level = CLAMP((int) sampler->min_lod, 0, (int) texture->last_level); } -/** - * Called via tgsi_sampler::get_samples() when running a fragment shader. - * Get four filtered RGBA values from the sampler's texture. - */ void -sp_get_samples_fragment(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +sp_sampler_varient_destroy( struct sp_sampler_varient *samp ) { - sp_get_samples(tgsi_sampler, s, t, p, TRUE, lodbias, rgba); + FREE(samp); } /** - * Called via tgsi_sampler::get_samples() when running a vertex shader. - * Get four filtered RGBA values from the sampler's texture. + * Create a sampler varient for a given set of non-orthogonal state. */ -void -sp_get_samples_vertex(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]) +struct sp_sampler_varient * +sp_create_sampler_varient( const struct pipe_sampler_state *sampler, + const union sp_sampler_key key ) { - sp_get_samples(tgsi_sampler, s, t, p, FALSE, lodbias, rgba); + struct sp_sampler_varient *samp = CALLOC_STRUCT(sp_sampler_varient); + if (!samp) + return NULL; + + samp->sampler = sampler; + samp->key = key; + + /* Note that (for instance) linear_texcoord_s and + * nearest_texcoord_s may be active at the same time, if the + * sampler min_img_filter differs from its mag_img_filter. + */ + if (sampler->normalized_coords) { + samp->linear_texcoord_s = get_linear_wrap( sampler->wrap_s ); + samp->linear_texcoord_t = get_linear_wrap( sampler->wrap_t ); + samp->linear_texcoord_p = get_linear_wrap( sampler->wrap_r ); + + samp->nearest_texcoord_s = get_nearest_wrap( sampler->wrap_s ); + samp->nearest_texcoord_t = get_nearest_wrap( sampler->wrap_t ); + samp->nearest_texcoord_p = get_nearest_wrap( sampler->wrap_r ); + } + else { + samp->linear_texcoord_s = get_linear_unorm_wrap( sampler->wrap_s ); + samp->linear_texcoord_t = get_linear_unorm_wrap( sampler->wrap_t ); + samp->linear_texcoord_p = get_linear_unorm_wrap( sampler->wrap_r ); + + samp->nearest_texcoord_s = get_nearest_unorm_wrap( sampler->wrap_s ); + samp->nearest_texcoord_t = get_nearest_unorm_wrap( sampler->wrap_t ); + samp->nearest_texcoord_p = get_nearest_unorm_wrap( sampler->wrap_r ); + } + + samp->compute_lambda = get_lambda_func( key ); + + samp->min_img_filter = get_img_filter(key, sampler->min_img_filter, sampler); + samp->mag_img_filter = get_img_filter(key, sampler->mag_img_filter, sampler); + + switch (sampler->min_mip_filter) { + case PIPE_TEX_MIPFILTER_NONE: + if (sampler->min_img_filter == sampler->mag_img_filter) + samp->mip_filter = samp->min_img_filter; + else + samp->mip_filter = mip_filter_none; + break; + + case PIPE_TEX_MIPFILTER_NEAREST: + samp->mip_filter = mip_filter_nearest; + break; + + case PIPE_TEX_MIPFILTER_LINEAR: + if (key.bits.is_pot && + sampler->min_img_filter == sampler->mag_img_filter && + sampler->normalized_coords && + sampler->wrap_s == PIPE_TEX_WRAP_REPEAT && + sampler->wrap_t == PIPE_TEX_WRAP_REPEAT && + sampler->min_img_filter == PIPE_TEX_FILTER_LINEAR) + { + samp->mip_filter = mip_filter_linear_2d_linear_repeat_POT; + } + else + { + samp->mip_filter = mip_filter_linear; + } + break; + } + + if (sampler->compare_mode != FALSE) { + samp->compare = sample_compare; + } + else { + /* Skip compare operation by promoting the mip_filter function + * pointer: + */ + samp->compare = samp->mip_filter; + } + + if (key.bits.target == PIPE_TEXTURE_CUBE) { + samp->base.get_samples = sample_cube; + } + else { + samp->faces[0] = 0; + samp->faces[1] = 0; + samp->faces[2] = 0; + samp->faces[3] = 0; + + /* Skip cube face determination by promoting the compare + * function pointer: + */ + samp->base.get_samples = samp->compare; + } + + return samp; } diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.h b/src/gallium/drivers/softpipe/sp_tex_sample.h index 40d8eb2c2a8..b0797711d37 100644 --- a/src/gallium/drivers/softpipe/sp_tex_sample.h +++ b/src/gallium/drivers/softpipe/sp_tex_sample.h @@ -31,43 +31,122 @@ #include "tgsi/tgsi_exec.h" +struct sp_sampler_varient; + +typedef void (*wrap_nearest_func)(const float s[4], + unsigned size, + int icoord[4]); + +typedef void (*wrap_linear_func)(const float s[4], + unsigned size, + int icoord0[4], + int icoord1[4], + float w[4]); + +typedef float (*compute_lambda_func)(const struct sp_sampler_varient *sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias); + +typedef void (*filter_func)(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]); + + +union sp_sampler_key { + struct { + unsigned target:3; + unsigned is_pot:1; + unsigned processor:2; + unsigned unit:4; + unsigned pad:22; + } bits; + unsigned value; +}; /** * Subclass of tgsi_sampler */ -struct sp_shader_sampler +struct sp_sampler_varient { struct tgsi_sampler base; /**< base class */ - uint unit; - struct softpipe_context *sp; - struct softpipe_tile_cache *cache; + union sp_sampler_key key; + + /* The owner of this struct: + */ + const struct pipe_sampler_state *sampler; + + + /* Currently bound texture: + */ + const struct pipe_texture *texture; + struct softpipe_tex_tile_cache *cache; + + unsigned processor; + + /* For sp_get_samples_2d_linear_POT: + */ + unsigned xpot; + unsigned ypot; + unsigned level; + + unsigned faces[4]; + + wrap_nearest_func nearest_texcoord_s; + wrap_nearest_func nearest_texcoord_t; + wrap_nearest_func nearest_texcoord_p; + + wrap_linear_func linear_texcoord_s; + wrap_linear_func linear_texcoord_t; + wrap_linear_func linear_texcoord_p; + + filter_func min_img_filter; + filter_func mag_img_filter; + + compute_lambda_func compute_lambda; + + filter_func mip_filter; + filter_func compare; + + /* Linked list: + */ + struct sp_sampler_varient *next; }; +struct sp_sampler; +/* Create a sampler varient for a given set of non-orthogonal state. Currently the + */ +struct sp_sampler_varient * +sp_create_sampler_varient( const struct pipe_sampler_state *sampler, + const union sp_sampler_key key ); -static INLINE const struct sp_shader_sampler * -sp_shader_sampler(const struct tgsi_sampler *sampler) -{ - return (const struct sp_shader_sampler *) sampler; -} +void sp_sampler_varient_bind_texture( struct sp_sampler_varient *varient, + struct softpipe_tex_tile_cache *tex_cache, + const struct pipe_texture *tex ); +void sp_sampler_varient_destroy( struct sp_sampler_varient * ); -extern void -sp_get_samples_fragment(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]); + + +static INLINE struct sp_sampler_varient * +sp_sampler_varient(const struct tgsi_sampler *sampler) +{ + return (struct sp_sampler_varient *) sampler; +} extern void -sp_get_samples_vertex(struct tgsi_sampler *tgsi_sampler, - const float s[QUAD_SIZE], - const float t[QUAD_SIZE], - const float p[QUAD_SIZE], - float lodbias, - float rgba[NUM_CHANNELS][QUAD_SIZE]); +sp_get_samples(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + float lodbias, + float rgba[NUM_CHANNELS][QUAD_SIZE]); #endif /* SP_TEX_SAMPLE_H */ diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c new file mode 100644 index 00000000000..407a22a9f4b --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c @@ -0,0 +1,273 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * Texture tile caching. + * + * Author: + * Brian Paul + */ + +#include "pipe/p_inlines.h" +#include "util/u_memory.h" +#include "util/u_tile.h" +#include "sp_context.h" +#include "sp_surface.h" +#include "sp_texture.h" +#include "sp_tex_tile_cache.h" + + + +struct softpipe_tex_tile_cache * +sp_create_tex_tile_cache( struct pipe_screen *screen ) +{ + struct softpipe_tex_tile_cache *tc; + uint pos; + + tc = CALLOC_STRUCT( softpipe_tex_tile_cache ); + if (tc) { + tc->screen = screen; + for (pos = 0; pos < NUM_ENTRIES; pos++) { + tc->entries[pos].addr.bits.invalid = 1; + } + tc->last_tile = &tc->entries[0]; /* any tile */ + } + return tc; +} + + +void +sp_destroy_tex_tile_cache(struct softpipe_tex_tile_cache *tc) +{ + struct pipe_screen *screen; + uint pos; + + for (pos = 0; pos < NUM_ENTRIES; pos++) { + /*assert(tc->entries[pos].x < 0);*/ + } + if (tc->transfer) { + screen = tc->transfer->texture->screen; + screen->tex_transfer_destroy(tc->transfer); + } + if (tc->tex_trans) { + screen = tc->tex_trans->texture->screen; + screen->tex_transfer_destroy(tc->tex_trans); + } + + FREE( tc ); +} + + + + +void +sp_tex_tile_cache_map_transfers(struct softpipe_tex_tile_cache *tc) +{ + if (tc->tex_trans && !tc->tex_trans_map) + tc->tex_trans_map = tc->screen->transfer_map(tc->screen, tc->tex_trans); +} + + +void +sp_tex_tile_cache_unmap_transfers(struct softpipe_tex_tile_cache *tc) +{ + if (tc->tex_trans_map) { + tc->screen->transfer_unmap(tc->screen, tc->tex_trans); + tc->tex_trans_map = NULL; + } +} + +/** + * Invalidate all cached tiles for the cached texture. + * Should be called when the texture is modified. + */ +void +sp_tex_tile_cache_validate_texture(struct softpipe_tex_tile_cache *tc) +{ + unsigned i; + + assert(tc); + assert(tc->texture); + + for (i = 0; i < NUM_ENTRIES; i++) { + tc->entries[i].addr.bits.invalid = 1; + } +} + +/** + * Specify the texture to cache. + */ +void +sp_tex_tile_cache_set_texture(struct softpipe_tex_tile_cache *tc, + struct pipe_texture *texture) +{ + uint i; + + assert(!tc->transfer); + + if (tc->texture != texture) { + pipe_texture_reference(&tc->texture, texture); + + if (tc->tex_trans) { + struct pipe_screen *screen = tc->tex_trans->texture->screen; + + if (tc->tex_trans_map) { + screen->transfer_unmap(screen, tc->tex_trans); + tc->tex_trans_map = NULL; + } + + screen->tex_transfer_destroy(tc->tex_trans); + tc->tex_trans = NULL; + } + + /* mark as entries as invalid/empty */ + /* XXX we should try to avoid this when the teximage hasn't changed */ + for (i = 0; i < NUM_ENTRIES; i++) { + tc->entries[i].addr.bits.invalid = 1; + } + + tc->tex_face = -1; /* any invalid value here */ + } +} + + + + +/** + * Flush the tile cache: write all dirty tiles back to the transfer. + * any tiles "flagged" as cleared will be "really" cleared. + */ +void +sp_flush_tex_tile_cache(struct softpipe_tex_tile_cache *tc) +{ + int pos; + + if (tc->texture) { + /* caching a texture, mark all entries as empty */ + for (pos = 0; pos < NUM_ENTRIES; pos++) { + tc->entries[pos].addr.bits.invalid = 1; + } + tc->tex_face = -1; + } + +} + + +/** + * Given the texture face, level, zslice, x and y values, compute + * the cache entry position/index where we'd hope to find the + * cached texture tile. + * This is basically a direct-map cache. + * XXX There's probably lots of ways in which we can improve this. + */ +static INLINE uint +tex_cache_pos( union tex_tile_address addr ) +{ + uint entry = (addr.bits.x + + addr.bits.y * 9 + + addr.bits.z * 3 + + addr.bits.face + + addr.bits.level * 7); + + return entry % NUM_ENTRIES; +} + +/** + * Similar to sp_get_cached_tile() but for textures. + * Tiles are read-only and indexed with more params. + */ +const struct softpipe_tex_cached_tile * +sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, + union tex_tile_address addr ) +{ + struct pipe_screen *screen = tc->screen; + struct softpipe_tex_cached_tile *tile; + + tile = tc->entries + tex_cache_pos( addr ); + + if (addr.value != tile->addr.value) { + + /* cache miss. Most misses are because we've invaldiated the + * texture cache previously -- most commonly on binding a new + * texture. Currently we effectively flush the cache on texture + * bind. + */ +#if 0 + _debug_printf("miss at %u: x=%d y=%d z=%d face=%d level=%d\n" + " tile %u: x=%d y=%d z=%d face=%d level=%d\n", + pos, x/TILE_SIZE, y/TILE_SIZE, z, face, level, + pos, tile->addr.bits.x, tile->addr.bits.y, tile->z, tile->face, tile->level); +#endif + + /* check if we need to get a new transfer */ + if (!tc->tex_trans || + tc->tex_face != addr.bits.face || + tc->tex_level != addr.bits.level || + tc->tex_z != addr.bits.z) { + /* get new transfer (view into texture) */ + + if (tc->tex_trans) { + if (tc->tex_trans_map) { + tc->screen->transfer_unmap(tc->screen, tc->tex_trans); + tc->tex_trans_map = NULL; + } + + screen->tex_transfer_destroy(tc->tex_trans); + tc->tex_trans = NULL; + } + + tc->tex_trans = + screen->get_tex_transfer(screen, tc->texture, + addr.bits.face, + addr.bits.level, + addr.bits.z, + PIPE_TRANSFER_READ, 0, 0, + tc->texture->width[addr.bits.level], + tc->texture->height[addr.bits.level]); + + tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans); + + tc->tex_face = addr.bits.face; + tc->tex_level = addr.bits.level; + tc->tex_z = addr.bits.z; + } + + /* get tile from the transfer (view into texture) */ + pipe_get_tile_rgba(tc->tex_trans, + addr.bits.x * TILE_SIZE, + addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, + (float *) tile->data.color); + tile->addr = addr; + } + + tc->last_tile = tile; + return tile; +} + + + diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.h b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h new file mode 100644 index 00000000000..ac6886a3df1 --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h @@ -0,0 +1,155 @@ +/************************************************************************** + * + * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef SP_TEX_TILE_CACHE_H +#define SP_TEX_TILE_CACHE_H + + +#include "pipe/p_compiler.h" + + +struct softpipe_context; +struct softpipe_tex_tile_cache; + + +/** + * Cache tile size (width and height). This needs to be a power of two. + */ +#define TILE_SIZE 64 + + +/* If we need to support > 4096, just expand this to be a 64 bit + * union, or consider tiling in Z as well. + */ +union tex_tile_address { + struct { + unsigned x:6; /* 4096 / TILE_SIZE */ + unsigned y:6; /* 4096 / TILE_SIZE */ + unsigned z:12; /* 4096 -- z not tiled */ + unsigned face:3; + unsigned level:4; + unsigned invalid:1; + } bits; + unsigned value; +}; + + +struct softpipe_tex_cached_tile +{ + union tex_tile_address addr; + union { + float color[TILE_SIZE][TILE_SIZE][4]; + } data; +}; + +#define NUM_ENTRIES 50 + +struct softpipe_tex_tile_cache +{ + struct pipe_screen *screen; + struct pipe_transfer *transfer; + void *transfer_map; + + struct pipe_texture *texture; /**< if caching a texture */ + unsigned timestamp; + + struct softpipe_tex_cached_tile entries[NUM_ENTRIES]; + + struct pipe_transfer *tex_trans; + void *tex_trans_map; + int tex_face, tex_level, tex_z; + + struct softpipe_tex_cached_tile *last_tile; /**< most recently retrieved tile */ +}; + + +extern struct softpipe_tex_tile_cache * +sp_create_tex_tile_cache( struct pipe_screen *screen ); + +extern void +sp_destroy_tex_tile_cache(struct softpipe_tex_tile_cache *tc); + + +extern void +sp_tex_tile_cache_map_transfers(struct softpipe_tex_tile_cache *tc); + +extern void +sp_tex_tile_cache_unmap_transfers(struct softpipe_tex_tile_cache *tc); + +extern void +sp_tex_tile_cache_set_texture(struct softpipe_tex_tile_cache *tc, + struct pipe_texture *texture); + +void +sp_tex_tile_cache_validate_texture(struct softpipe_tex_tile_cache *tc); + +extern void +sp_flush_tex_tile_cache(struct softpipe_tex_tile_cache *tc); + + + +extern const struct softpipe_tex_cached_tile * +sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, + union tex_tile_address addr ); + +static INLINE union tex_tile_address +tex_tile_address( unsigned x, + unsigned y, + unsigned z, + unsigned face, + unsigned level ) +{ + union tex_tile_address addr; + + addr.value = 0; + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + addr.bits.z = z; + addr.bits.face = face; + addr.bits.level = level; + + return addr; +} + +/* Quickly retrieve tile if it matches last lookup. + */ +static INLINE const struct softpipe_tex_cached_tile * +sp_get_cached_tile_tex(struct softpipe_tex_tile_cache *tc, + union tex_tile_address addr ) +{ + if (tc->last_tile->addr.value == addr.value) + return tc->last_tile; + + return sp_find_cached_tile_tex( tc, addr ); +} + + + + + +#endif /* SP_TEX_TILE_CACHE_H */ + diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c index 548e40c4a8c..2e6c43c7ef2 100644 --- a/src/gallium/drivers/softpipe/sp_texture.c +++ b/src/gallium/drivers/softpipe/sp_texture.c @@ -113,16 +113,20 @@ softpipe_displaytarget_layout(struct pipe_screen *screen, static struct pipe_texture * softpipe_texture_create(struct pipe_screen *screen, - const struct pipe_texture *templat) + const struct pipe_texture *template) { struct softpipe_texture *spt = CALLOC_STRUCT(softpipe_texture); if (!spt) return NULL; - spt->base = *templat; + spt->base = *template; pipe_reference_init(&spt->base.reference, 1); spt->base.screen = screen; + spt->pot = (util_is_power_of_two(template->width[0]) && + util_is_power_of_two(template->height[0]) && + util_is_power_of_two(template->depth[0])); + if (spt->base.tex_usage & (PIPE_TEXTURE_USAGE_DISPLAY_TARGET | PIPE_TEXTURE_USAGE_PRIMARY)) { if (!softpipe_displaytarget_layout(screen, spt)) @@ -218,6 +222,13 @@ softpipe_get_tex_surface(struct pipe_screen *screen, if (ps->usage & PIPE_BUFFER_USAGE_GPU_READ) ps->usage |= PIPE_BUFFER_USAGE_CPU_READ; + if (ps->usage & (PIPE_BUFFER_USAGE_CPU_WRITE | + PIPE_BUFFER_USAGE_GPU_WRITE)) { + /* Mark the surface as dirty. The tile cache will look for this. */ + spt->timestamp++; + softpipe_screen(screen)->timestamp++; + } + ps->face = face; ps->level = level; ps->zslice = zslice; @@ -318,27 +329,18 @@ softpipe_transfer_map( struct pipe_screen *screen, { ubyte *map, *xfer_map; struct softpipe_texture *spt; - unsigned flags = 0; assert(transfer->texture); spt = softpipe_texture(transfer->texture); - if (transfer->usage != PIPE_TRANSFER_READ) { - flags |= PIPE_BUFFER_USAGE_CPU_WRITE; - } - - if (transfer->usage != PIPE_TRANSFER_WRITE) { - flags |= PIPE_BUFFER_USAGE_CPU_READ; - } - - map = pipe_buffer_map(screen, spt->buffer, flags); + map = pipe_buffer_map(screen, spt->buffer, pipe_transfer_buffer_flags(transfer)); if (map == NULL) return NULL; /* May want to different things here depending on read/write nature * of the map: */ - if (transfer->texture && transfer->usage != PIPE_TRANSFER_READ) { + if (transfer->texture && (transfer->usage & PIPE_TRANSFER_WRITE)) { /* Do something to notify sharing contexts of a texture change. * In softpipe, that would mean flushing the texture cache. */ @@ -364,12 +366,64 @@ softpipe_transfer_unmap(struct pipe_screen *screen, pipe_buffer_unmap( screen, spt->buffer ); - if (transfer->usage != PIPE_TRANSFER_READ) { + if (transfer->usage & PIPE_TRANSFER_WRITE) { /* Mark the texture as dirty to expire the tile caches. */ - spt->modified = TRUE; + spt->timestamp++; } } +static struct pipe_video_surface* +softpipe_video_surface_create(struct pipe_screen *screen, + enum pipe_video_chroma_format chroma_format, + unsigned width, unsigned height) +{ + struct softpipe_video_surface *sp_vsfc; + struct pipe_texture template; + + assert(screen); + assert(width && height); + + sp_vsfc = CALLOC_STRUCT(softpipe_video_surface); + if (!sp_vsfc) + return NULL; + + pipe_reference_init(&sp_vsfc->base.reference, 1); + sp_vsfc->base.screen = screen; + sp_vsfc->base.chroma_format = chroma_format; + /*sp_vsfc->base.surface_format = PIPE_VIDEO_SURFACE_FORMAT_VUYA;*/ + sp_vsfc->base.width = width; + sp_vsfc->base.height = height; + + memset(&template, 0, sizeof(struct pipe_texture)); + template.target = PIPE_TEXTURE_2D; + template.format = PIPE_FORMAT_X8R8G8B8_UNORM; + template.last_level = 0; + /* vl_mpeg12_mc_renderer expects this when it's initialized with pot_buffers=true */ + template.width[0] = util_next_power_of_two(width); + template.height[0] = util_next_power_of_two(height); + template.depth[0] = 1; + pf_get_block(template.format, &template.block); + template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET; + + sp_vsfc->tex = screen->texture_create(screen, &template); + if (!sp_vsfc->tex) { + FREE(sp_vsfc); + return NULL; + } + + return &sp_vsfc->base; +} + + +static void +softpipe_video_surface_destroy(struct pipe_video_surface *vsfc) +{ + struct softpipe_video_surface *sp_vsfc = softpipe_video_surface(vsfc); + + pipe_texture_reference(&sp_vsfc->tex, NULL); + FREE(sp_vsfc); +} + void softpipe_init_screen_texture_funcs(struct pipe_screen *screen) @@ -385,6 +439,9 @@ softpipe_init_screen_texture_funcs(struct pipe_screen *screen) screen->tex_transfer_destroy = softpipe_tex_transfer_destroy; screen->transfer_map = softpipe_transfer_map; screen->transfer_unmap = softpipe_transfer_unmap; + + screen->video_surface_create = softpipe_video_surface_create; + screen->video_surface_destroy = softpipe_video_surface_destroy; } diff --git a/src/gallium/drivers/softpipe/sp_texture.h b/src/gallium/drivers/softpipe/sp_texture.h index b1b6130b22c..2ef64e1e7c3 100644 --- a/src/gallium/drivers/softpipe/sp_texture.h +++ b/src/gallium/drivers/softpipe/sp_texture.h @@ -30,6 +30,7 @@ #include "pipe/p_state.h" +#include "pipe/p_video_state.h" struct pipe_context; @@ -48,7 +49,11 @@ struct softpipe_texture */ struct pipe_buffer *buffer; - boolean modified; + /* True if texture images are power-of-two in all dimensions: + */ + boolean pot; + + unsigned timestamp; }; struct softpipe_transfer @@ -58,6 +63,15 @@ struct softpipe_transfer unsigned long offset; }; +struct softpipe_video_surface +{ + struct pipe_video_surface base; + + /* The data is held here: + */ + struct pipe_texture *tex; +}; + /** cast wrappers */ static INLINE struct softpipe_texture * @@ -72,6 +86,12 @@ softpipe_transfer(struct pipe_transfer *pt) return (struct softpipe_transfer *) pt; } +static INLINE struct softpipe_video_surface * +softpipe_video_surface(struct pipe_video_surface *pvs) +{ + return (struct softpipe_video_surface *) pvs; +} + extern void softpipe_init_screen_texture_funcs(struct pipe_screen *screen); diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.c b/src/gallium/drivers/softpipe/sp_tile_cache.c index 1f9b8f1f4fb..83fb4e0d151 100644 --- a/src/gallium/drivers/softpipe/sp_tile_cache.c +++ b/src/gallium/drivers/softpipe/sp_tile_cache.c @@ -26,7 +26,7 @@ **************************************************************************/ /** - * Texture tile caching. + * Render target tile caching. * * Author: * Brian Paul @@ -35,38 +35,8 @@ #include "pipe/p_inlines.h" #include "util/u_memory.h" #include "util/u_tile.h" -#include "sp_context.h" -#include "sp_surface.h" -#include "sp_texture.h" #include "sp_tile_cache.h" -#define NUM_ENTRIES 50 - - -/** XXX move these */ -#define MAX_WIDTH 2048 -#define MAX_HEIGHT 2048 - - -struct softpipe_tile_cache -{ - struct pipe_screen *screen; - struct pipe_surface *surface; /**< the surface we're caching */ - struct pipe_transfer *transfer; - void *transfer_map; - struct pipe_texture *texture; /**< if caching a texture */ - struct softpipe_cached_tile entries[NUM_ENTRIES]; - uint clear_flags[(MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32]; - float clear_color[4]; /**< for color bufs */ - uint clear_val; /**< for z+stencil, or packed color clear value */ - boolean depth_stencil; /**< Is the surface a depth/stencil format? */ - - struct pipe_transfer *tex_trans; - void *tex_trans_map; - int tex_face, tex_level, tex_z; - - struct softpipe_cached_tile tile; /**< scratch tile for clears */ -}; /** @@ -76,7 +46,7 @@ struct softpipe_tile_cache * a LRU replacement policy. */ #define CACHE_POS(x, y) \ - (((x) / TILE_SIZE + ((y) / TILE_SIZE) * 5) % NUM_ENTRIES) + (((x) + (y) * 5) % NUM_ENTRIES) @@ -84,12 +54,10 @@ struct softpipe_tile_cache * Is the tile at (x,y) in cleared state? */ static INLINE uint -is_clear_flag_set(const uint *bitvec, int x, int y) +is_clear_flag_set(const uint *bitvec, union tile_address addr) { int pos, bit; - x /= TILE_SIZE; - y /= TILE_SIZE; - pos = y * (MAX_WIDTH / TILE_SIZE) + x; + pos = addr.bits.y * (MAX_WIDTH / TILE_SIZE) + addr.bits.x; assert(pos / 32 < (MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32); bit = bitvec[pos / 32] & (1 << (pos & 31)); return bit; @@ -100,12 +68,10 @@ is_clear_flag_set(const uint *bitvec, int x, int y) * Mark the tile at (x,y) as not cleared. */ static INLINE void -clear_clear_flag(uint *bitvec, int x, int y) +clear_clear_flag(uint *bitvec, union tile_address addr) { int pos; - x /= TILE_SIZE; - y /= TILE_SIZE; - pos = y * (MAX_WIDTH / TILE_SIZE) + x; + pos = addr.bits.y * (MAX_WIDTH / TILE_SIZE) + addr.bits.x; assert(pos / 32 < (MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32); bitvec[pos / 32] &= ~(1 << (pos & 31)); } @@ -116,14 +82,25 @@ sp_create_tile_cache( struct pipe_screen *screen ) { struct softpipe_tile_cache *tc; uint pos; + int maxLevels, maxTexSize; + + /* sanity checking: max sure MAX_WIDTH/HEIGHT >= largest texture image */ + maxLevels = screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS); + maxTexSize = 1 << (maxLevels - 1); + assert(MAX_WIDTH >= maxTexSize); tc = CALLOC_STRUCT( softpipe_tile_cache ); if (tc) { tc->screen = screen; for (pos = 0; pos < NUM_ENTRIES; pos++) { - tc->entries[pos].x = - tc->entries[pos].y = -1; + tc->entries[pos].addr.bits.invalid = 1; } + tc->last_tile = &tc->entries[0]; /* any tile */ + +#if TILE_CLEAR_OPTIMIZATION + /* set flags to indicate all the tiles are cleared */ + memset(tc->clear_flags, 255, sizeof(tc->clear_flags)); +#endif } return tc; } @@ -142,10 +119,6 @@ sp_destroy_tile_cache(struct softpipe_tile_cache *tc) screen = tc->transfer->texture->screen; screen->tex_transfer_destroy(tc->transfer); } - if (tc->tex_trans) { - screen = tc->tex_trans->texture->screen; - screen->tex_transfer_destroy(tc->tex_trans); - } FREE( tc ); } @@ -158,8 +131,6 @@ void sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, struct pipe_surface *ps) { - assert(!tc->texture); - if (tc->transfer) { struct pipe_screen *screen = tc->transfer->texture->screen; @@ -211,9 +182,6 @@ sp_tile_cache_map_transfers(struct softpipe_tile_cache *tc) { if (tc->transfer && !tc->transfer_map) tc->transfer_map = tc->screen->transfer_map(tc->screen, tc->transfer); - - if (tc->tex_trans && !tc->tex_trans_map) - tc->tex_trans_map = tc->screen->transfer_map(tc->screen, tc->tex_trans); } @@ -224,47 +192,6 @@ sp_tile_cache_unmap_transfers(struct softpipe_tile_cache *tc) tc->screen->transfer_unmap(tc->screen, tc->transfer); tc->transfer_map = NULL; } - - if (tc->tex_trans_map) { - tc->screen->transfer_unmap(tc->screen, tc->tex_trans); - tc->tex_trans_map = NULL; - } -} - - -/** - * Specify the texture to cache. - */ -void -sp_tile_cache_set_texture(struct pipe_context *pipe, - struct softpipe_tile_cache *tc, - struct pipe_texture *texture) -{ - uint i; - - assert(!tc->transfer); - - pipe_texture_reference(&tc->texture, texture); - - if (tc->tex_trans) { - struct pipe_screen *screen = tc->tex_trans->texture->screen; - - if (tc->tex_trans_map) { - screen->transfer_unmap(screen, tc->tex_trans); - tc->tex_trans_map = NULL; - } - - screen->tex_transfer_destroy(tc->tex_trans); - tc->tex_trans = NULL; - } - - /* mark as entries as invalid/empty */ - /* XXX we should try to avoid this when the teximage hasn't changed */ - for (i = 0; i < NUM_ENTRIES; i++) { - tc->entries[i].x = -1; - } - - tc->tex_face = -1; /* any invalid value here */ } @@ -308,7 +235,7 @@ clear_tile(struct softpipe_cached_tile *tile, switch (pf_get_size(format)) { case 1: - memset(tile->data.any, 0, TILE_SIZE * TILE_SIZE); + memset(tile->data.any, clear_value, TILE_SIZE * TILE_SIZE); break; case 2: if (clear_value == 0) { @@ -344,8 +271,7 @@ clear_tile(struct softpipe_cached_tile *tile, * Actually clear the tiles which were flagged as being in a clear state. */ static void -sp_tile_cache_flush_clear(struct pipe_context *pipe, - struct softpipe_tile_cache *tc) +sp_tile_cache_flush_clear(struct softpipe_tile_cache *tc) { struct pipe_transfer *pt = tc->transfer; const uint w = tc->transfer->width; @@ -359,13 +285,15 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe, /* push the tile to all positions marked as clear */ for (y = 0; y < h; y += TILE_SIZE) { for (x = 0; x < w; x += TILE_SIZE) { - if (is_clear_flag_set(tc->clear_flags, x, y)) { + union tile_address addr = tile_address(x, y); + + if (is_clear_flag_set(tc->clear_flags, addr)) { pipe_put_tile_raw(pt, x, y, TILE_SIZE, TILE_SIZE, tc->tile.data.color32, 0/*STRIDE*/); /* do this? */ - clear_clear_flag(tc->clear_flags, x, y); + clear_clear_flag(tc->clear_flags, addr); numCleared++; } @@ -382,8 +310,7 @@ sp_tile_cache_flush_clear(struct pipe_context *pipe, * any tiles "flagged" as cleared will be "really" cleared. */ void -sp_flush_tile_cache(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc) +sp_flush_tile_cache(struct softpipe_tile_cache *tc) { struct pipe_transfer *pt = tc->transfer; int inuse = 0, pos; @@ -392,33 +319,30 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, /* caching a drawing transfer */ for (pos = 0; pos < NUM_ENTRIES; pos++) { struct softpipe_cached_tile *tile = tc->entries + pos; - if (tile->x >= 0) { + if (!tile->addr.bits.invalid) { if (tc->depth_stencil) { pipe_put_tile_raw(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { pipe_put_tile_rgba(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, (float *) tile->data.color); } - tile->x = tile->y = -1; /* mark as empty */ + tile->addr.bits.invalid = 1; /* mark as empty */ inuse++; } } #if TILE_CLEAR_OPTIMIZATION - sp_tile_cache_flush_clear(&softpipe->pipe, tc); + sp_tile_cache_flush_clear(tc); #endif } - else if (tc->texture) { - /* caching a texture, mark all entries as empty */ - for (pos = 0; pos < NUM_ENTRIES; pos++) { - tc->entries[pos].x = -1; - } - tc->tex_face = -1; - } #if 0 debug_printf("flushed tiles in use: %d\n", inuse); @@ -431,40 +355,39 @@ sp_flush_tile_cache(struct softpipe_context *softpipe, * \param x, y position of tile, in pixels */ struct softpipe_cached_tile * -sp_get_cached_tile(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc, int x, int y) +sp_find_cached_tile(struct softpipe_tile_cache *tc, + union tile_address addr ) { struct pipe_transfer *pt = tc->transfer; - - /* tile pos in framebuffer: */ - const int tile_x = x & ~(TILE_SIZE - 1); - const int tile_y = y & ~(TILE_SIZE - 1); - + /* cache pos/entry: */ - const int pos = CACHE_POS(x, y); + const int pos = CACHE_POS(addr.bits.x, + addr.bits.y); struct softpipe_cached_tile *tile = tc->entries + pos; - if (tile_x != tile->x || - tile_y != tile->y) { + if (addr.value != tile->addr.value) { - if (tile->x != -1) { + if (tile->addr.bits.invalid == 0) { /* put dirty tile back in framebuffer */ if (tc->depth_stencil) { pipe_put_tile_raw(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { pipe_put_tile_rgba(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, (float *) tile->data.color); } } - tile->x = tile_x; - tile->y = tile_y; + tile->addr = addr; - if (is_clear_flag_set(tc->clear_flags, x, y)) { + if (is_clear_flag_set(tc->clear_flags, addr)) { /* don't get tile from framebuffer, just clear it */ if (tc->depth_stencil) { clear_tile(tile, pt->format, tc->clear_val); @@ -472,125 +395,33 @@ sp_get_cached_tile(struct softpipe_context *softpipe, else { clear_tile_rgba(tile, pt->format, tc->clear_color); } - clear_clear_flag(tc->clear_flags, x, y); + clear_clear_flag(tc->clear_flags, addr); } else { /* get new tile data from transfer */ if (tc->depth_stencil) { pipe_get_tile_raw(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { pipe_get_tile_rgba(pt, - tile->x, tile->y, TILE_SIZE, TILE_SIZE, + tile->addr.bits.x * TILE_SIZE, + tile->addr.bits.y * TILE_SIZE, + TILE_SIZE, TILE_SIZE, (float *) tile->data.color); } } } + tc->last_tile = tile; return tile; } -/** - * Given the texture face, level, zslice, x and y values, compute - * the cache entry position/index where we'd hope to find the - * cached texture tile. - * This is basically a direct-map cache. - * XXX There's probably lots of ways in which we can improve this. - */ -static INLINE uint -tex_cache_pos(int x, int y, int z, int face, int level) -{ - uint entry = x + y * 9 + z * 3 + face + level * 7; - return entry % NUM_ENTRIES; -} - - -/** - * Similar to sp_get_cached_tile() but for textures. - * Tiles are read-only and indexed with more params. - */ -const struct softpipe_cached_tile * -sp_get_cached_tile_tex(struct softpipe_context *sp, - struct softpipe_tile_cache *tc, int x, int y, int z, - int face, int level) -{ - struct pipe_screen *screen = sp->pipe.screen; - /* tile pos in framebuffer: */ - const int tile_x = x & ~(TILE_SIZE - 1); - const int tile_y = y & ~(TILE_SIZE - 1); - /* cache pos/entry: */ - const uint pos = tex_cache_pos(x / TILE_SIZE, y / TILE_SIZE, z, - face, level); - struct softpipe_cached_tile *tile = tc->entries + pos; - if (tc->texture) { - struct softpipe_texture *spt = softpipe_texture(tc->texture); - if (spt->modified) { - /* texture was modified, invalidate all cached tiles */ - uint p; - for (p = 0; p < NUM_ENTRIES; p++) { - tile = tc->entries + p; - tile->x = -1; - } - spt->modified = FALSE; - } - } - - if (tile_x != tile->x || - tile_y != tile->y || - z != tile->z || - face != tile->face || - level != tile->level) { - /* cache miss */ - -#if 0 - printf("miss at %u x=%d y=%d z=%d face=%d level=%d\n", pos, - x/TILE_SIZE, y/TILE_SIZE, z, face, level); -#endif - /* check if we need to get a new transfer */ - if (!tc->tex_trans || - tc->tex_face != face || - tc->tex_level != level || - tc->tex_z != z) { - /* get new transfer (view into texture) */ - - if (tc->tex_trans) { - if (tc->tex_trans_map) { - tc->screen->transfer_unmap(tc->screen, tc->tex_trans); - tc->tex_trans_map = NULL; - } - - screen->tex_transfer_destroy(tc->tex_trans); - tc->tex_trans = NULL; - } - - tc->tex_trans = screen->get_tex_transfer(screen, tc->texture, face, level, z, - PIPE_TRANSFER_READ, 0, 0, - tc->texture->width[level], - tc->texture->height[level]); - tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans); - - tc->tex_face = face; - tc->tex_level = level; - tc->tex_z = z; - } - - /* get tile from the transfer (view into texture) */ - pipe_get_tile_rgba(tc->tex_trans, - tile_x, tile_y, TILE_SIZE, TILE_SIZE, - (float *) tile->data.color); - tile->x = tile_x; - tile->y = tile_y; - tile->z = z; - tile->face = face; - tile->level = level; - } - - return tile; -} /** @@ -621,6 +452,6 @@ sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float *rgba, for (pos = 0; pos < NUM_ENTRIES; pos++) { struct softpipe_cached_tile *tile = tc->entries + pos; - tile->x = tile->y = -1; + tile->addr.bits.invalid = 1; } } diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.h b/src/gallium/drivers/softpipe/sp_tile_cache.h index 8f247d0e580..a12092702a6 100644 --- a/src/gallium/drivers/softpipe/sp_tile_cache.h +++ b/src/gallium/drivers/softpipe/sp_tile_cache.h @@ -34,7 +34,6 @@ #include "pipe/p_compiler.h" -struct softpipe_context; struct softpipe_tile_cache; @@ -44,11 +43,23 @@ struct softpipe_tile_cache; #define TILE_SIZE 64 +/* If we need to support > 4096, just expand this to be a 64 bit + * union, or consider tiling in Z as well. + */ +union tile_address { + struct { + unsigned x:6; /* 4096 / TILE_SIZE */ + unsigned y:6; /* 4096 / TILE_SIZE */ + unsigned invalid:1; + unsigned pad:19; + } bits; + unsigned value; +}; + struct softpipe_cached_tile { - int x, y; /**< pos of tile in window coords */ - int z, face, level; /**< Extra texture indexes */ + union tile_address addr; union { float color[TILE_SIZE][TILE_SIZE][4]; uint color32[TILE_SIZE][TILE_SIZE]; @@ -59,6 +70,32 @@ struct softpipe_cached_tile } data; }; +#define NUM_ENTRIES 50 + + +/** XXX move these */ +#define MAX_WIDTH 4096 +#define MAX_HEIGHT 4096 + + +struct softpipe_tile_cache +{ + struct pipe_screen *screen; + struct pipe_surface *surface; /**< the surface we're caching */ + struct pipe_transfer *transfer; + void *transfer_map; + + struct softpipe_cached_tile entries[NUM_ENTRIES]; + uint clear_flags[(MAX_WIDTH / TILE_SIZE) * (MAX_HEIGHT / TILE_SIZE) / 32]; + float clear_color[4]; /**< for color bufs */ + uint clear_val; /**< for z+stencil, or packed color clear value */ + boolean depth_stencil; /**< Is the surface a depth/stencil format? */ + + struct softpipe_cached_tile tile; /**< scratch tile for clears */ + + struct softpipe_cached_tile *last_tile; /**< most recently retrieved tile */ +}; + extern struct softpipe_tile_cache * sp_create_tile_cache( struct pipe_screen *screen ); @@ -80,26 +117,45 @@ extern void sp_tile_cache_unmap_transfers(struct softpipe_tile_cache *tc); extern void -sp_tile_cache_set_texture(struct pipe_context *pipe, - struct softpipe_tile_cache *tc, - struct pipe_texture *texture); - -extern void -sp_flush_tile_cache(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc); +sp_flush_tile_cache(struct softpipe_tile_cache *tc); extern void sp_tile_cache_clear(struct softpipe_tile_cache *tc, const float *rgba, uint clearValue); extern struct softpipe_cached_tile * -sp_get_cached_tile(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc, int x, int y); +sp_find_cached_tile(struct softpipe_tile_cache *tc, + union tile_address addr ); + + +static INLINE union tile_address +tile_address( unsigned x, + unsigned y ) +{ + union tile_address addr; + + addr.value = 0; + addr.bits.x = x / TILE_SIZE; + addr.bits.y = y / TILE_SIZE; + + return addr; +} + +/* Quickly retrieve tile if it matches last lookup. + */ +static INLINE struct softpipe_cached_tile * +sp_get_cached_tile(struct softpipe_tile_cache *tc, + int x, int y ) +{ + union tile_address addr = tile_address( x, y ); + + if (tc->last_tile->addr.value == addr.value) + return tc->last_tile; + + return sp_find_cached_tile( tc, addr ); +} + -extern const struct softpipe_cached_tile * -sp_get_cached_tile_tex(struct softpipe_context *softpipe, - struct softpipe_tile_cache *tc, int x, int y, int z, - int face, int level); #endif /* SP_TILE_CACHE_H */ diff --git a/src/gallium/drivers/softpipe/sp_video_context.c b/src/gallium/drivers/softpipe/sp_video_context.c new file mode 100644 index 00000000000..cae2d3efc58 --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_video_context.c @@ -0,0 +1,304 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include "sp_video_context.h" +#include <pipe/p_inlines.h> +#include <util/u_memory.h> +#include "softpipe/sp_winsys.h" +#include "softpipe/sp_texture.h" + +static void +sp_mpeg12_destroy(struct pipe_video_context *vpipe) +{ + struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe; + + assert(vpipe); + + /* Asserted in softpipe_delete_fs_state() for some reason */ + ctx->pipe->bind_vs_state(ctx->pipe, NULL); + ctx->pipe->bind_fs_state(ctx->pipe, NULL); + + ctx->pipe->delete_blend_state(ctx->pipe, ctx->blend); + ctx->pipe->delete_rasterizer_state(ctx->pipe, ctx->rast); + ctx->pipe->delete_depth_stencil_alpha_state(ctx->pipe, ctx->dsa); + + pipe_video_surface_reference(&ctx->decode_target, NULL); + vl_compositor_cleanup(&ctx->compositor); + vl_mpeg12_mc_renderer_cleanup(&ctx->mc_renderer); + ctx->pipe->destroy(ctx->pipe); + + FREE(ctx); +} + +static void +sp_mpeg12_decode_macroblocks(struct pipe_video_context *vpipe, + struct pipe_video_surface *past, + struct pipe_video_surface *future, + unsigned num_macroblocks, + struct pipe_macroblock *macroblocks, + struct pipe_fence_handle **fence) +{ + struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe; + struct pipe_mpeg12_macroblock *mpeg12_macroblocks = (struct pipe_mpeg12_macroblock*)macroblocks; + + assert(vpipe); + assert(num_macroblocks); + assert(macroblocks); + assert(macroblocks->codec == PIPE_VIDEO_CODEC_MPEG12); + assert(ctx->decode_target); + + vl_mpeg12_mc_renderer_render_macroblocks(&ctx->mc_renderer, + softpipe_video_surface(ctx->decode_target)->tex, + past ? softpipe_video_surface(past)->tex : NULL, + future ? softpipe_video_surface(future)->tex : NULL, + num_macroblocks, mpeg12_macroblocks, fence); +} + +static void +sp_mpeg12_clear_surface(struct pipe_video_context *vpipe, + unsigned x, unsigned y, + unsigned width, unsigned height, + unsigned value, + struct pipe_surface *surface) +{ + struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe; + + assert(vpipe); + assert(surface); + + ctx->pipe->surface_fill(ctx->pipe, surface, x, y, width, height, value); +} + +static void +sp_mpeg12_render_picture(struct pipe_video_context *vpipe, + /*struct pipe_surface *backround, + struct pipe_video_rect *backround_area,*/ + struct pipe_video_surface *src_surface, + enum pipe_mpeg12_picture_type picture_type, + /*unsigned num_past_surfaces, + struct pipe_video_surface *past_surfaces, + unsigned num_future_surfaces, + struct pipe_video_surface *future_surfaces,*/ + struct pipe_video_rect *src_area, + struct pipe_surface *dst_surface, + struct pipe_video_rect *dst_area, + /*unsigned num_layers, + struct pipe_surface *layers, + struct pipe_video_rect *layer_src_areas, + struct pipe_video_rect *layer_dst_areas*/ + struct pipe_fence_handle **fence) +{ + struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe; + + assert(vpipe); + assert(src_surface); + assert(src_area); + assert(dst_surface); + assert(dst_area); + + vl_compositor_render(&ctx->compositor, softpipe_video_surface(src_surface)->tex, + picture_type, src_area, dst_surface->texture, dst_area, fence); +} + +static void +sp_mpeg12_set_decode_target(struct pipe_video_context *vpipe, + struct pipe_video_surface *dt) +{ + struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe; + + assert(vpipe); + assert(dt); + + pipe_video_surface_reference(&ctx->decode_target, dt); +} + +static void sp_mpeg12_set_csc_matrix(struct pipe_video_context *vpipe, const float *mat) +{ + struct sp_mpeg12_context *ctx = (struct sp_mpeg12_context*)vpipe; + + assert(vpipe); + + vl_compositor_set_csc_matrix(&ctx->compositor, mat); +} + +static bool +init_pipe_state(struct sp_mpeg12_context *ctx) +{ + struct pipe_rasterizer_state rast; + struct pipe_blend_state blend; + struct pipe_depth_stencil_alpha_state dsa; + unsigned i; + + assert(ctx); + + rast.flatshade = 1; + rast.flatshade_first = 0; + rast.light_twoside = 0; + rast.front_winding = PIPE_WINDING_CCW; + rast.cull_mode = PIPE_WINDING_CW; + rast.fill_cw = PIPE_POLYGON_MODE_FILL; + rast.fill_ccw = PIPE_POLYGON_MODE_FILL; + rast.offset_cw = 0; + rast.offset_ccw = 0; + rast.scissor = 0; + rast.poly_smooth = 0; + rast.poly_stipple_enable = 0; + rast.point_sprite = 0; + rast.point_size_per_vertex = 0; + rast.multisample = 0; + rast.line_smooth = 0; + rast.line_stipple_enable = 0; + rast.line_stipple_factor = 0; + rast.line_stipple_pattern = 0; + rast.line_last_pixel = 0; + rast.bypass_vs_clip_and_viewport = 0; + rast.line_width = 1; + rast.point_smooth = 0; + rast.point_size = 1; + rast.offset_units = 1; + rast.offset_scale = 1; + /*rast.sprite_coord_mode[i] = ;*/ + ctx->rast = ctx->pipe->create_rasterizer_state(ctx->pipe, &rast); + ctx->pipe->bind_rasterizer_state(ctx->pipe, ctx->rast); + + blend.blend_enable = 0; + blend.rgb_func = PIPE_BLEND_ADD; + blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE; + blend.rgb_dst_factor = PIPE_BLENDFACTOR_ONE; + blend.alpha_func = PIPE_BLEND_ADD; + blend.alpha_src_factor = PIPE_BLENDFACTOR_ONE; + blend.alpha_dst_factor = PIPE_BLENDFACTOR_ONE; + blend.logicop_enable = 0; + blend.logicop_func = PIPE_LOGICOP_CLEAR; + /* Needed to allow color writes to FB, even if blending disabled */ + blend.colormask = PIPE_MASK_RGBA; + blend.dither = 0; + ctx->blend = ctx->pipe->create_blend_state(ctx->pipe, &blend); + ctx->pipe->bind_blend_state(ctx->pipe, ctx->blend); + + dsa.depth.enabled = 0; + dsa.depth.writemask = 0; + dsa.depth.func = PIPE_FUNC_ALWAYS; + for (i = 0; i < 2; ++i) { + dsa.stencil[i].enabled = 0; + dsa.stencil[i].func = PIPE_FUNC_ALWAYS; + dsa.stencil[i].fail_op = PIPE_STENCIL_OP_KEEP; + dsa.stencil[i].zpass_op = PIPE_STENCIL_OP_KEEP; + dsa.stencil[i].zfail_op = PIPE_STENCIL_OP_KEEP; + dsa.stencil[i].ref_value = 0; + dsa.stencil[i].valuemask = 0; + dsa.stencil[i].writemask = 0; + } + dsa.alpha.enabled = 0; + dsa.alpha.func = PIPE_FUNC_ALWAYS; + dsa.alpha.ref_value = 0; + ctx->dsa = ctx->pipe->create_depth_stencil_alpha_state(ctx->pipe, &dsa); + ctx->pipe->bind_depth_stencil_alpha_state(ctx->pipe, ctx->dsa); + + return true; +} + +static struct pipe_video_context * +sp_mpeg12_create(struct pipe_screen *screen, enum pipe_video_profile profile, + enum pipe_video_chroma_format chroma_format, + unsigned width, unsigned height) +{ + struct sp_mpeg12_context *ctx; + + assert(u_reduce_video_profile(profile) == PIPE_VIDEO_CODEC_MPEG12); + + ctx = CALLOC_STRUCT(sp_mpeg12_context); + + if (!ctx) + return NULL; + + ctx->base.profile = profile; + ctx->base.chroma_format = chroma_format; + ctx->base.width = width; + ctx->base.height = height; + + ctx->base.screen = screen; + ctx->base.destroy = sp_mpeg12_destroy; + ctx->base.decode_macroblocks = sp_mpeg12_decode_macroblocks; + ctx->base.clear_surface = sp_mpeg12_clear_surface; + ctx->base.render_picture = sp_mpeg12_render_picture; + ctx->base.set_decode_target = sp_mpeg12_set_decode_target; + ctx->base.set_csc_matrix = sp_mpeg12_set_csc_matrix; + + ctx->pipe = softpipe_create(screen); + if (!ctx->pipe) { + FREE(ctx); + return NULL; + } + + /* TODO: Use slice buffering for softpipe when implemented, no advantage to buffering an entire picture */ + if (!vl_mpeg12_mc_renderer_init(&ctx->mc_renderer, ctx->pipe, + width, height, chroma_format, + VL_MPEG12_MC_RENDERER_BUFFER_PICTURE, + /* TODO: Use XFER_NONE when implemented */ + VL_MPEG12_MC_RENDERER_EMPTY_BLOCK_XFER_ONE, + true)) { + ctx->pipe->destroy(ctx->pipe); + FREE(ctx); + return NULL; + } + + if (!vl_compositor_init(&ctx->compositor, ctx->pipe)) { + vl_mpeg12_mc_renderer_cleanup(&ctx->mc_renderer); + ctx->pipe->destroy(ctx->pipe); + FREE(ctx); + return NULL; + } + + if (!init_pipe_state(ctx)) { + vl_compositor_cleanup(&ctx->compositor); + vl_mpeg12_mc_renderer_cleanup(&ctx->mc_renderer); + ctx->pipe->destroy(ctx->pipe); + FREE(ctx); + return NULL; + } + + return &ctx->base; +} + +struct pipe_video_context * +sp_video_create(struct pipe_screen *screen, enum pipe_video_profile profile, + enum pipe_video_chroma_format chroma_format, + unsigned width, unsigned height) +{ + assert(screen); + assert(width && height); + + switch (u_reduce_video_profile(profile)) { + case PIPE_VIDEO_CODEC_MPEG12: + return sp_mpeg12_create(screen, profile, + chroma_format, + width, height); + default: + return NULL; + } +} diff --git a/src/gallium/drivers/softpipe/sp_video_context.h b/src/gallium/drivers/softpipe/sp_video_context.h new file mode 100644 index 00000000000..ccbd1ffe4c8 --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_video_context.h @@ -0,0 +1,57 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef SP_VIDEO_CONTEXT_H +#define SP_VIDEO_CONTEXT_H + +#include <pipe/p_video_context.h> +#include <vl/vl_mpeg12_mc_renderer.h> +#include <vl/vl_compositor.h> + +struct pipe_screen; +struct pipe_context; +struct pipe_video_surface; + +struct sp_mpeg12_context +{ + struct pipe_video_context base; + struct pipe_context *pipe; + struct pipe_video_surface *decode_target; + struct vl_mpeg12_mc_renderer mc_renderer; + struct vl_compositor compositor; + + void *rast; + void *dsa; + void *blend; +}; + +struct pipe_video_context * +sp_video_create(struct pipe_screen *screen, enum pipe_video_profile profile, + enum pipe_video_chroma_format chroma_format, + unsigned width, unsigned height); + +#endif /* SP_VIDEO_CONTEXT_H */ diff --git a/src/gallium/drivers/trace/tr_context.c b/src/gallium/drivers/trace/tr_context.c index ae0af4d0557..bf470b46ae1 100644 --- a/src/gallium/drivers/trace/tr_context.c +++ b/src/gallium/drivers/trace/tr_context.c @@ -125,11 +125,11 @@ trace_context_draw_block(struct trace_context *tr_ctx, int flag) } else if ((tr_ctx->draw_rule.blocker & flag) && (tr_ctx->draw_blocker & 4)) { boolean block = FALSE; - debug_printf("%s (%lu %lu) (%lu %lu) (%lu %u) (%lu %u)\n", __FUNCTION__, - tr_ctx->draw_rule.fs, tr_ctx->curr.fs, - tr_ctx->draw_rule.vs, tr_ctx->curr.vs, - tr_ctx->draw_rule.surf, 0, - tr_ctx->draw_rule.tex, 0); + debug_printf("%s (%p %p) (%p %p) (%p %u) (%p %u)\n", __FUNCTION__, + (void *) tr_ctx->draw_rule.fs, (void *) tr_ctx->curr.fs, + (void *) tr_ctx->draw_rule.vs, (void *) tr_ctx->curr.vs, + (void *) tr_ctx->draw_rule.surf, 0, + (void *) tr_ctx->draw_rule.tex, 0); if (tr_ctx->draw_rule.fs && tr_ctx->draw_rule.fs == tr_ctx->curr.fs) block = TRUE; diff --git a/src/gallium/drivers/trace/tr_rbug.c b/src/gallium/drivers/trace/tr_rbug.c index e85ac15edca..81e0a6f3b00 100644 --- a/src/gallium/drivers/trace/tr_rbug.c +++ b/src/gallium/drivers/trace/tr_rbug.c @@ -44,7 +44,7 @@ #if defined(PIPE_SUBSYSTEM_WINDOWS_USER) # define sleep Sleep -#elif defined(PIPE_OS_LINUX) +#elif defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD) void usleep(int); # define sleep usleep #else diff --git a/src/gallium/drivers/trace/tr_screen.c b/src/gallium/drivers/trace/tr_screen.c index 26f1c04594f..ab605c7fc87 100644 --- a/src/gallium/drivers/trace/tr_screen.c +++ b/src/gallium/drivers/trace/tr_screen.c @@ -403,7 +403,7 @@ trace_screen_transfer_map(struct pipe_screen *_screen, map = screen->transfer_map(screen, transfer); if(map) { - if(transfer->usage != PIPE_TRANSFER_READ) { + if(transfer->usage & PIPE_TRANSFER_WRITE) { assert(!tr_trans->map); tr_trans->map = map; } diff --git a/src/gallium/include/pipe/p_config.h b/src/gallium/include/pipe/p_config.h index de99957d9d0..f6feea5f74d 100644 --- a/src/gallium/include/pipe/p_config.h +++ b/src/gallium/include/pipe/p_config.h @@ -122,18 +122,35 @@ #if defined(__linux__) #define PIPE_OS_LINUX +#define PIPE_OS_UNIX #endif #if defined(__FreeBSD__) +#define PIPE_OS_FREEBSD #define PIPE_OS_BSD +#define PIPE_OS_UNIX +#endif + +#if defined(__OpenBSD__) +#define PIPE_OS_OPENBSD +#define PIPE_OS_BSD +#define PIPE_OS_UNIX +#endif + +#if defined(__NetBSD__) +#define PIPE_OS_NETBSD +#define PIPE_OS_BSD +#define PIPE_OS_UNIX #endif #if defined(__sun) #define PIPE_OS_SOLARIS +#define PIPE_OS_UNIX #endif #if defined(__APPLE__) #define PIPE_OS_APPLE +#define PIPE_OS_UNIX #endif #if defined(_WIN32) || defined(WIN32) @@ -142,6 +159,7 @@ #if defined(__HAIKU__) #define PIPE_OS_HAIKU +#define PIPE_OS_UNIX #endif /* diff --git a/src/gallium/include/pipe/p_defines.h b/src/gallium/include/pipe/p_defines.h index f252d6df00d..f8fa1e3f49d 100644 --- a/src/gallium/include/pipe/p_defines.h +++ b/src/gallium/include/pipe/p_defines.h @@ -193,7 +193,18 @@ enum pipe_texture_target { enum pipe_transfer_usage { PIPE_TRANSFER_READ = (1 << 0), PIPE_TRANSFER_WRITE = (1 << 1), - PIPE_TRANSFER_READ_WRITE = PIPE_TRANSFER_READ | PIPE_TRANSFER_WRITE /**< Read/modify/write */ + /** Read/modify/write */ + PIPE_TRANSFER_READ_WRITE = PIPE_TRANSFER_READ | PIPE_TRANSFER_WRITE, + /** + * The transfer should map the texture storage directly. The driver may + * return NULL if that isn't possible, and the state tracker needs to cope + * with that and use an alternative path without this flag. + * + * E.g. the state tracker could have a simpler path which maps textures and + * does read/modify/write cycles on them directly, and a more complicated + * path which uses minimal read and write transfers. + */ + PIPE_TRANSFER_MAP_DIRECTLY = (1 << 2) }; @@ -315,6 +326,31 @@ enum pipe_transfer_usage { #define PIPE_REFERENCED_FOR_READ (1 << 0) #define PIPE_REFERENCED_FOR_WRITE (1 << 1) + +enum pipe_video_codec +{ + PIPE_VIDEO_CODEC_UNKNOWN = 0, + PIPE_VIDEO_CODEC_MPEG12, /**< MPEG1, MPEG2 */ + PIPE_VIDEO_CODEC_MPEG4, /**< DIVX, XVID */ + PIPE_VIDEO_CODEC_VC1, /**< WMV */ + PIPE_VIDEO_CODEC_MPEG4_AVC /**< H.264 */ +}; + +enum pipe_video_profile +{ + PIPE_VIDEO_PROFILE_MPEG1, + PIPE_VIDEO_PROFILE_MPEG2_SIMPLE, + PIPE_VIDEO_PROFILE_MPEG2_MAIN, + PIPE_VIDEO_PROFILE_MPEG4_SIMPLE, + PIPE_VIDEO_PROFILE_MPEG4_ADVANCED_SIMPLE, + PIPE_VIDEO_PROFILE_VC1_SIMPLE, + PIPE_VIDEO_PROFILE_VC1_MAIN, + PIPE_VIDEO_PROFILE_VC1_ADVANCED, + PIPE_VIDEO_PROFILE_MPEG4_AVC_BASELINE, + PIPE_VIDEO_PROFILE_MPEG4_AVC_MAIN, + PIPE_VIDEO_PROFILE_MPEG4_AVC_HIGH +}; + #ifdef __cplusplus } #endif diff --git a/src/gallium/include/pipe/p_format.h b/src/gallium/include/pipe/p_format.h index c4469d4a9e9..af230809201 100644 --- a/src/gallium/include/pipe/p_format.h +++ b/src/gallium/include/pipe/p_format.h @@ -613,6 +613,24 @@ pf_has_alpha( enum pipe_format format ) } } +enum pipe_video_chroma_format +{ + PIPE_VIDEO_CHROMA_FORMAT_420, + PIPE_VIDEO_CHROMA_FORMAT_422, + PIPE_VIDEO_CHROMA_FORMAT_444 +}; + +#if 0 +enum pipe_video_surface_format +{ + PIPE_VIDEO_SURFACE_FORMAT_NV12, /**< Planar; Y plane, UV plane */ + PIPE_VIDEO_SURFACE_FORMAT_YV12, /**< Planar; Y plane, U plane, V plane */ + PIPE_VIDEO_SURFACE_FORMAT_YUYV, /**< Interleaved; Y,U,Y,V,Y,U,Y,V */ + PIPE_VIDEO_SURFACE_FORMAT_UYVY, /**< Interleaved; U,Y,V,Y,U,Y,V,Y */ + PIPE_VIDEO_SURFACE_FORMAT_VUYA /**< Packed; A31-24|Y23-16|U15-8|V7-0 */ +}; +#endif + #ifdef __cplusplus } #endif diff --git a/src/gallium/include/pipe/p_inlines.h b/src/gallium/include/pipe/p_inlines.h index 30a4aaf4093..5fbd62a03d2 100644 --- a/src/gallium/include/pipe/p_inlines.h +++ b/src/gallium/include/pipe/p_inlines.h @@ -176,6 +176,22 @@ pipe_transfer_destroy( struct pipe_transfer *transf ) screen->tex_transfer_destroy(transf); } +static INLINE unsigned +pipe_transfer_buffer_flags( struct pipe_transfer *transf ) +{ + switch (transf->usage & PIPE_TRANSFER_READ_WRITE) { + case PIPE_TRANSFER_READ_WRITE: + return PIPE_BUFFER_USAGE_CPU_READ | PIPE_BUFFER_USAGE_CPU_WRITE; + case PIPE_TRANSFER_READ: + return PIPE_BUFFER_USAGE_CPU_READ; + case PIPE_TRANSFER_WRITE: + return PIPE_BUFFER_USAGE_CPU_WRITE; + default: + debug_assert(0); + return 0; + } +} + #ifdef __cplusplus } #endif diff --git a/src/gallium/include/pipe/p_screen.h b/src/gallium/include/pipe/p_screen.h index 3f30c52a169..f0a4de5df33 100644 --- a/src/gallium/include/pipe/p_screen.h +++ b/src/gallium/include/pipe/p_screen.h @@ -53,7 +53,10 @@ extern "C" { struct pipe_fence_handle; struct pipe_winsys; struct pipe_buffer; - +struct pipe_texture; +struct pipe_surface; +struct pipe_video_surface; +struct pipe_transfer; /** @@ -252,6 +255,17 @@ struct pipe_screen { void (*buffer_destroy)( struct pipe_buffer *buf ); + /** + * Create a video surface suitable for use as a decoding target by the + * driver's pipe_video_context. + */ + struct pipe_video_surface* + (*video_surface_create)( struct pipe_screen *screen, + enum pipe_video_chroma_format chroma_format, + unsigned width, unsigned height ); + + void (*video_surface_destroy)( struct pipe_video_surface *vsfc ); + /** * Do any special operations to ensure frontbuffer contents are diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h index 2187f5b3677..287b424e4ac 100644 --- a/src/gallium/include/pipe/p_state.h +++ b/src/gallium/include/pipe/p_state.h @@ -114,11 +114,29 @@ struct pipe_rasterizer_state * the vertex shader, clipping and viewport processing. Note that * a vertex shader is still needed though, to indicate the mapping * from vertex elements to fragment shader input semantics. + * + * XXX: considered for removal. */ unsigned bypass_vs_clip_and_viewport:1; - unsigned flatshade_first:1; /**< take color attribute from the first vertex of a primitive */ - unsigned gl_rasterization_rules:1; /**< enable tweaks for GL rasterization? */ + /** + * Use the first vertex of a primitive as the provoking vertex for + * flat shading. + */ + unsigned flatshade_first:1; + + /** + * When true, triangle rasterization uses (0.5, 0.5) pixel centers + * for determining pixel ownership. + * + * When false, triangle rasterization uses (0,0) pixel centers for + * determining pixel ownership. + * + * Triangle rasterization always uses a 'top,left' rule for pixel + * ownership, this just alters which point we consider the pixel + * center for that test. + */ + unsigned gl_rasterization_rules:1; float line_width; float point_size; /**< used when no per-vertex size */ @@ -179,7 +197,6 @@ struct pipe_depth_state unsigned enabled:1; /**< depth test enabled? */ unsigned writemask:1; /**< allow depth buffer writes? */ unsigned func:3; /**< depth test func (PIPE_FUNC_x) */ - unsigned occlusion_count:1; /**< do occlusion counting? */ }; diff --git a/src/gallium/include/pipe/p_video_context.h b/src/gallium/include/pipe/p_video_context.h new file mode 100644 index 00000000000..6ae31418fa8 --- /dev/null +++ b/src/gallium/include/pipe/p_video_context.h @@ -0,0 +1,121 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef PIPE_VIDEO_CONTEXT_H +#define PIPE_VIDEO_CONTEXT_H + +#ifdef __cplusplus +extern "C" { +#endif + +#include <pipe/p_video_state.h> + +struct pipe_screen; +struct pipe_buffer; +struct pipe_surface; +struct pipe_video_surface; +struct pipe_macroblock; +struct pipe_picture_desc; +struct pipe_fence_handle; + +/** + * Gallium video rendering context + */ +struct pipe_video_context +{ + struct pipe_screen *screen; + enum pipe_video_profile profile; + enum pipe_video_chroma_format chroma_format; + unsigned width; + unsigned height; + + void *priv; /**< context private data (for DRI for example) */ + + void (*destroy)(struct pipe_video_context *vpipe); + + /** + * Picture decoding and displaying + */ + /*@{*/ + void (*decode_bitstream)(struct pipe_video_context *vpipe, + unsigned num_bufs, + struct pipe_buffer **bitstream_buf); + + void (*decode_macroblocks)(struct pipe_video_context *vpipe, + struct pipe_video_surface *past, + struct pipe_video_surface *future, + unsigned num_macroblocks, + struct pipe_macroblock *macroblocks, + struct pipe_fence_handle **fence); + + void (*clear_surface)(struct pipe_video_context *vpipe, + unsigned x, unsigned y, + unsigned width, unsigned height, + unsigned value, + struct pipe_surface *surface); + + void (*render_picture)(struct pipe_video_context *vpipe, + /*struct pipe_surface *backround, + struct pipe_video_rect *backround_area,*/ + struct pipe_video_surface *src_surface, + enum pipe_mpeg12_picture_type picture_type, + /*unsigned num_past_surfaces, + struct pipe_video_surface *past_surfaces, + unsigned num_future_surfaces, + struct pipe_video_surface *future_surfaces,*/ + struct pipe_video_rect *src_area, + struct pipe_surface *dst_surface, + struct pipe_video_rect *dst_area, + /*unsigned num_layers, + struct pipe_texture *layers, + struct pipe_video_rect *layer_src_areas, + struct pipe_video_rect *layer_dst_areas,*/ + struct pipe_fence_handle **fence); + /*@}*/ + + /** + * Parameter-like states (or properties) + */ + /*@{*/ + void (*set_picture_desc)(struct pipe_video_context *vpipe, + const struct pipe_picture_desc *desc); + + void (*set_decode_target)(struct pipe_video_context *vpipe, + struct pipe_video_surface *dt); + + void (*set_csc_matrix)(struct pipe_video_context *vpipe, const float *mat); + + /* TODO: Interface for scaling modes, post-processing, etc. */ + /*@}*/ +}; + + +#ifdef __cplusplus +} +#endif + +#endif /* PIPE_VIDEO_CONTEXT_H */ diff --git a/src/gallium/include/pipe/p_video_state.h b/src/gallium/include/pipe/p_video_state.h new file mode 100644 index 00000000000..4da26d608cf --- /dev/null +++ b/src/gallium/include/pipe/p_video_state.h @@ -0,0 +1,184 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef PIPE_VIDEO_STATE_H +#define PIPE_VIDEO_STATE_H + +/* u_reduce_video_profile() needs these */ +#include <pipe/p_compiler.h> +#include <util/u_debug.h> + +#include <pipe/p_defines.h> +#include <pipe/p_format.h> +#include <pipe/p_refcnt.h> +#include <pipe/p_screen.h> + +#ifdef __cplusplus +extern "C" { +#endif + +struct pipe_video_surface +{ + struct pipe_reference reference; + struct pipe_screen *screen; + enum pipe_video_chroma_format chroma_format; + /*enum pipe_video_surface_format surface_format;*/ + unsigned width; + unsigned height; +}; + +static INLINE void +pipe_video_surface_reference(struct pipe_video_surface **ptr, struct pipe_video_surface *surf) +{ + struct pipe_video_surface *old_surf = *ptr; + + if (pipe_reference((struct pipe_reference **)ptr, &surf->reference)) + old_surf->screen->video_surface_destroy(old_surf); +} + +struct pipe_video_rect +{ + unsigned x, y, w, h; +}; + +static INLINE enum pipe_video_codec +u_reduce_video_profile(enum pipe_video_profile profile) +{ + switch (profile) + { + case PIPE_VIDEO_PROFILE_MPEG1: + case PIPE_VIDEO_PROFILE_MPEG2_SIMPLE: + case PIPE_VIDEO_PROFILE_MPEG2_MAIN: + return PIPE_VIDEO_CODEC_MPEG12; + + case PIPE_VIDEO_PROFILE_MPEG4_SIMPLE: + case PIPE_VIDEO_PROFILE_MPEG4_ADVANCED_SIMPLE: + return PIPE_VIDEO_CODEC_MPEG4; + + case PIPE_VIDEO_PROFILE_VC1_SIMPLE: + case PIPE_VIDEO_PROFILE_VC1_MAIN: + case PIPE_VIDEO_PROFILE_VC1_ADVANCED: + return PIPE_VIDEO_CODEC_VC1; + + case PIPE_VIDEO_PROFILE_MPEG4_AVC_BASELINE: + case PIPE_VIDEO_PROFILE_MPEG4_AVC_MAIN: + case PIPE_VIDEO_PROFILE_MPEG4_AVC_HIGH: + return PIPE_VIDEO_CODEC_MPEG4_AVC; + + default: + assert(0); + return PIPE_VIDEO_CODEC_UNKNOWN; + } +} + +enum pipe_mpeg12_picture_type +{ + PIPE_MPEG12_PICTURE_TYPE_FIELD_TOP, + PIPE_MPEG12_PICTURE_TYPE_FIELD_BOTTOM, + PIPE_MPEG12_PICTURE_TYPE_FRAME +}; + +enum pipe_mpeg12_macroblock_type +{ + PIPE_MPEG12_MACROBLOCK_TYPE_INTRA, + PIPE_MPEG12_MACROBLOCK_TYPE_FWD, + PIPE_MPEG12_MACROBLOCK_TYPE_BKWD, + PIPE_MPEG12_MACROBLOCK_TYPE_BI, + + PIPE_MPEG12_MACROBLOCK_NUM_TYPES +}; + +enum pipe_mpeg12_motion_type +{ + PIPE_MPEG12_MOTION_TYPE_FIELD, + PIPE_MPEG12_MOTION_TYPE_FRAME, + PIPE_MPEG12_MOTION_TYPE_DUALPRIME, + PIPE_MPEG12_MOTION_TYPE_16x8 +}; + +enum pipe_mpeg12_dct_type +{ + PIPE_MPEG12_DCT_TYPE_FIELD, + PIPE_MPEG12_DCT_TYPE_FRAME +}; + +struct pipe_macroblock +{ + enum pipe_video_codec codec; +}; + +struct pipe_mpeg12_macroblock +{ + struct pipe_macroblock base; + + unsigned mbx; + unsigned mby; + enum pipe_mpeg12_macroblock_type mb_type; + enum pipe_mpeg12_motion_type mo_type; + enum pipe_mpeg12_dct_type dct_type; + signed pmv[2][2][2]; + unsigned cbp; + void *blocks; +}; + +#if 0 +struct pipe_picture_desc +{ + enum pipe_video_format format; +}; + +struct pipe_mpeg12_picture_desc +{ + struct pipe_picture_desc base; + + /* TODO: Use bitfields where possible? */ + struct pipe_surface *forward_reference; + struct pipe_surface *backward_reference; + unsigned picture_coding_type; + unsigned fcode; + unsigned intra_dc_precision; + unsigned picture_structure; + unsigned top_field_first; + unsigned frame_pred_frame_dct; + unsigned concealment_motion_vectors; + unsigned q_scale_type; + unsigned intra_vlc_format; + unsigned alternate_scan; + unsigned full_pel_forward_vector; + unsigned full_pel_backward_vector; + struct pipe_buffer *intra_quantizer_matrix; + struct pipe_buffer *non_intra_quantizer_matrix; + struct pipe_buffer *chroma_intra_quantizer_matrix; + struct pipe_buffer *chroma_non_intra_quantizer_matrix; +}; +#endif + +#ifdef __cplusplus +} +#endif + +#endif /* PIPE_VIDEO_STATE_H */ diff --git a/src/gallium/state_trackers/dri/dri_extensions.c b/src/gallium/state_trackers/dri/dri_extensions.c index 4349a4d1d2f..f39a305531e 100644 --- a/src/gallium/state_trackers/dri/dri_extensions.c +++ b/src/gallium/state_trackers/dri/dri_extensions.c @@ -39,6 +39,7 @@ #define need_GL_ARB_point_parameters #define need_GL_ARB_shader_objects #define need_GL_ARB_texture_compression +#define need_GL_ARB_vertex_array_object #define need_GL_ARB_vertex_buffer_object #define need_GL_ARB_vertex_program #define need_GL_ARB_vertex_shader @@ -52,6 +53,7 @@ #define need_GL_EXT_framebuffer_object #define need_GL_EXT_multi_draw_arrays #define need_GL_EXT_secondary_color +#define need_GL_APPLE_vertex_array_object #define need_GL_NV_vertex_program #define need_GL_VERSION_2_0 #define need_GL_VERSION_2_1 @@ -79,6 +81,7 @@ const struct dri_extension card_extensions[] = { {"GL_ARB_texture_env_dot3", NULL}, {"GL_ARB_texture_mirrored_repeat", NULL}, {"GL_ARB_texture_rectangle", NULL}, + {"GL_ARB_vertex_array_object", GL_ARB_vertex_array_object_functions}, {"GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions}, {"GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions}, {"GL_ARB_vertex_program", GL_ARB_vertex_program_functions}, @@ -103,6 +106,7 @@ const struct dri_extension card_extensions[] = { {"GL_EXT_texture_lod_bias", NULL}, {"GL_3DFX_texture_compression_FXT1", NULL}, {"GL_APPLE_client_storage", NULL}, + {"GL_APPLE_vertex_array_object", GL_APPLE_vertex_array_object_functions}, {"GL_MESA_pack_invert", NULL}, {"GL_MESA_ycbcr_texture", NULL}, {"GL_NV_blend_square", NULL}, diff --git a/src/gallium/state_trackers/egl/egl_context.c b/src/gallium/state_trackers/egl/egl_context.c index c4f7361ca0b..e21a4a1095b 100644 --- a/src/gallium/state_trackers/egl/egl_context.c +++ b/src/gallium/state_trackers/egl/egl_context.c @@ -138,7 +138,6 @@ drm_destroy_context(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *context) struct drm_context *c = lookup_drm_context(context); if (!_eglIsContextBound(&c->base)) { st_destroy_context(c->st); - c->pipe->destroy(c->pipe); free(c); } return EGL_TRUE; @@ -160,18 +159,12 @@ drm_make_current(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *draw, _EGLSurfa if (!drawSurf || !readSurf) return EGL_FALSE; - drawSurf->user = ctx; - readSurf->user = ctx; - st_make_current(ctx->st, drawSurf->stfb, readSurf->stfb); /* st_resize_framebuffer needs a bound context to work */ st_resize_framebuffer(drawSurf->stfb, drawSurf->w, drawSurf->h); st_resize_framebuffer(readSurf->stfb, readSurf->w, readSurf->h); } else { - drawSurf->user = NULL; - readSurf->user = NULL; - st_make_current(NULL, NULL, NULL); } diff --git a/src/gallium/state_trackers/egl/egl_surface.c b/src/gallium/state_trackers/egl/egl_surface.c index 69e2d6b7081..7911a8834e0 100644 --- a/src/gallium/state_trackers/egl/egl_surface.c +++ b/src/gallium/state_trackers/egl/egl_surface.c @@ -152,7 +152,6 @@ drm_takedown_shown_screen(_EGLDisplay *dpy, struct drm_screen *screen) pipe_surface_reference(&screen->surface, NULL); pipe_texture_reference(&screen->tex, NULL); - pipe_buffer_reference(&screen->buffer, NULL); screen->shown = 0; } @@ -250,8 +249,8 @@ drm_show_screen_surface_mesa(_EGLDriver *drv, _EGLDisplay *dpy, drm_create_texture(dpy, scrn, mode->Width, mode->Height); - if (!scrn->buffer) - return EGL_FALSE; + if (!scrn->tex) + goto err_tex; ret = drmModeAddFB(dev->drmFD, scrn->front.width, scrn->front.height, @@ -325,8 +324,8 @@ err_fb: err_bo: pipe_surface_reference(&scrn->surface, NULL); pipe_texture_reference(&scrn->tex, NULL); - pipe_buffer_reference(&scrn->buffer, NULL); +err_tex: return EGL_FALSE; } @@ -353,24 +352,21 @@ drm_swap_buffers(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *draw) if (!surf) return EGL_FALSE; - /* error checking */ - if (!_eglSwapBuffers(drv, dpy, draw)) - return EGL_FALSE; - st_get_framebuffer_surface(surf->stfb, ST_SURFACE_BACK_LEFT, &back_surf); if (back_surf) { + struct drm_context *ctx = lookup_drm_context(draw->Binding); st_notify_swapbuffers(surf->stfb); - if (surf->screen) { - surf->user->pipe->surface_copy(surf->user->pipe, + if (ctx && surf->screen) { + ctx->pipe->surface_copy(ctx->pipe, surf->screen->surface, 0, 0, back_surf, 0, 0, surf->w, surf->h); - surf->user->pipe->flush(surf->user->pipe, PIPE_FLUSH_RENDER_CACHE | PIPE_FLUSH_TEXTURE_CACHE, NULL); + ctx->pipe->flush(ctx->pipe, PIPE_FLUSH_RENDER_CACHE | PIPE_FLUSH_TEXTURE_CACHE, NULL); #ifdef DRM_MODE_FEATURE_DIRTYFB /* TODO query connector property to see if this is needed */ diff --git a/src/gallium/state_trackers/egl/egl_tracker.h b/src/gallium/state_trackers/egl/egl_tracker.h index dd4730f9579..73eb1a1226e 100644 --- a/src/gallium/state_trackers/egl/egl_tracker.h +++ b/src/gallium/state_trackers/egl/egl_tracker.h @@ -69,7 +69,6 @@ struct drm_surface * drm */ - struct drm_context *user; struct drm_screen *screen; int w; @@ -94,7 +93,6 @@ struct drm_screen * pipe */ - struct pipe_buffer *buffer; struct pipe_texture *tex; struct pipe_surface *surface; diff --git a/src/gallium/state_trackers/g3dvl/Makefile b/src/gallium/state_trackers/g3dvl/Makefile deleted file mode 100644 index f9f4d6be3c3..00000000000 --- a/src/gallium/state_trackers/g3dvl/Makefile +++ /dev/null @@ -1,21 +0,0 @@ -TARGET = libg3dvl.a -OBJECTS = vl_display.o vl_screen.o vl_context.o vl_surface.o vl_shader_build.o vl_util.o vl_basic_csc.o \ - vl_r16snorm_mc_buf.o -GALLIUMDIR = ../.. - -CFLAGS += -g -Wall -Werror-implicit-function-declaration -fPIC \ - -I${GALLIUMDIR}/include \ - -I${GALLIUMDIR}/auxiliary \ - -I${GALLIUMDIR}/winsys/g3dvl \ - -############################################# - -.PHONY = all clean - -all: ${TARGET} - -${TARGET}: ${OBJECTS} - ar rcs $@ $^ - -clean: - rm -rf ${OBJECTS} ${TARGET} diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c b/src/gallium/state_trackers/g3dvl/vl_basic_csc.c deleted file mode 100644 index b1683b891be..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_basic_csc.c +++ /dev/null @@ -1,720 +0,0 @@ -#define VL_INTERNAL -#include "vl_basic_csc.h" -#include <assert.h> -#include <pipe/p_context.h> -#include <pipe/p_state.h> -#include <pipe/p_inlines.h> -#include <tgsi/tgsi_parse.h> -#include <tgsi/tgsi_build.h> -#include <util/u_memory.h> -#include "vl_csc.h" -#include "vl_surface.h" -#include "vl_shader_build.h" -#include "vl_types.h" - -struct vlVertexShaderConsts -{ - struct vlVertex4f dst_scale; - struct vlVertex4f dst_trans; - struct vlVertex4f src_scale; - struct vlVertex4f src_trans; -}; - -struct vlFragmentShaderConsts -{ - struct vlVertex4f bias; - float matrix[16]; -}; - -struct vlBasicCSC -{ - struct vlCSC base; - - struct pipe_context *pipe; - struct pipe_viewport_state viewport; - struct pipe_framebuffer_state framebuffer; - struct pipe_texture *framebuffer_tex; - void *sampler; - void *vertex_shader, *fragment_shader; - struct pipe_vertex_buffer vertex_bufs[2]; - struct pipe_vertex_element vertex_elems[2]; - struct pipe_constant_buffer vs_const_buf, fs_const_buf; -}; - -static int vlResizeFrameBuffer -( - struct vlCSC *csc, - unsigned int width, - unsigned int height -) -{ - struct vlBasicCSC *basic_csc; - struct pipe_context *pipe; - struct pipe_texture template; - float clear_color[4]; - - assert(csc); - - basic_csc = (struct vlBasicCSC*)csc; - pipe = basic_csc->pipe; - - if (basic_csc->framebuffer.width == width && basic_csc->framebuffer.height == height) - return 0; - - basic_csc->viewport.scale[0] = width; - basic_csc->viewport.scale[1] = height; - basic_csc->viewport.scale[2] = 1; - basic_csc->viewport.scale[3] = 1; - basic_csc->viewport.translate[0] = 0; - basic_csc->viewport.translate[1] = 0; - basic_csc->viewport.translate[2] = 0; - basic_csc->viewport.translate[3] = 0; - - clear_color[0] = 0.0f; - clear_color[1] = 0.0f; - clear_color[2] = 0.0f; - clear_color[3] = 0.0f; - - if (basic_csc->framebuffer_tex) - { - pipe_surface_reference(&basic_csc->framebuffer.cbufs[0], NULL); - pipe_texture_reference(&basic_csc->framebuffer_tex, NULL); - } - - memset(&template, 0, sizeof(struct pipe_texture)); - template.target = PIPE_TEXTURE_2D; - template.format = PIPE_FORMAT_A8R8G8B8_UNORM; - template.last_level = 0; - template.width[0] = width; - template.height[0] = height; - template.depth[0] = 1; - pf_get_block(template.format, &template.block); - template.tex_usage = PIPE_TEXTURE_USAGE_DISPLAY_TARGET; - - basic_csc->framebuffer_tex = pipe->screen->texture_create(pipe->screen, &template); - - basic_csc->framebuffer.width = width; - basic_csc->framebuffer.height = height; - basic_csc->framebuffer.cbufs[0] = pipe->screen->get_tex_surface - ( - pipe->screen, - basic_csc->framebuffer_tex, - 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE - ); - - /* Clear to black, in case video doesn't fill the entire window */ - pipe->set_framebuffer_state(pipe, &basic_csc->framebuffer); - pipe->clear(pipe, PIPE_CLEAR_COLOR, clear_color, 0.0f, 0); - - return 0; -} - -static int vlBegin -( - struct vlCSC *csc -) -{ - struct vlBasicCSC *basic_csc; - struct pipe_context *pipe; - - assert(csc); - - basic_csc = (struct vlBasicCSC*)csc; - pipe = basic_csc->pipe; - - pipe->set_framebuffer_state(pipe, &basic_csc->framebuffer); - pipe->set_viewport_state(pipe, &basic_csc->viewport); - pipe->bind_sampler_states(pipe, 1, (void**)&basic_csc->sampler); - /* Source texture set in vlPutPictureCSC() */ - pipe->bind_vs_state(pipe, basic_csc->vertex_shader); - pipe->bind_fs_state(pipe, basic_csc->fragment_shader); - pipe->set_vertex_buffers(pipe, 2, basic_csc->vertex_bufs); - pipe->set_vertex_elements(pipe, 2, basic_csc->vertex_elems); - pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &basic_csc->vs_const_buf); - pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &basic_csc->fs_const_buf); - - return 0; -} - -static int vlPutPictureCSC -( - struct vlCSC *csc, - struct vlSurface *surface, - int srcx, - int srcy, - int srcw, - int srch, - int destx, - int desty, - int destw, - int desth, - enum vlPictureType picture_type -) -{ - struct vlBasicCSC *basic_csc; - struct pipe_context *pipe; - struct vlVertexShaderConsts *vs_consts; - - assert(csc); - assert(surface); - - basic_csc = (struct vlBasicCSC*)csc; - pipe = basic_csc->pipe; - - vs_consts = pipe_buffer_map - ( - pipe->screen, - basic_csc->vs_const_buf.buffer, - PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD - ); - - vs_consts->dst_scale.x = destw / (float)basic_csc->framebuffer.cbufs[0]->width; - vs_consts->dst_scale.y = desth / (float)basic_csc->framebuffer.cbufs[0]->height; - vs_consts->dst_scale.z = 1; - vs_consts->dst_scale.w = 1; - vs_consts->dst_trans.x = destx / (float)basic_csc->framebuffer.cbufs[0]->width; - vs_consts->dst_trans.y = desty / (float)basic_csc->framebuffer.cbufs[0]->height; - vs_consts->dst_trans.z = 0; - vs_consts->dst_trans.w = 0; - - vs_consts->src_scale.x = srcw / (float)surface->texture->width[0]; - vs_consts->src_scale.y = srch / (float)surface->texture->height[0]; - vs_consts->src_scale.z = 1; - vs_consts->src_scale.w = 1; - vs_consts->src_trans.x = srcx / (float)surface->texture->width[0]; - vs_consts->src_trans.y = srcy / (float)surface->texture->height[0]; - vs_consts->src_trans.z = 0; - vs_consts->src_trans.w = 0; - - pipe_buffer_unmap(pipe->screen, basic_csc->vs_const_buf.buffer); - - pipe->set_sampler_textures(pipe, 1, &surface->texture); - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLE_STRIP, 0, 4); - - return 0; -} - -static int vlEnd -( - struct vlCSC *csc -) -{ - assert(csc); - - return 0; -} - -static struct pipe_surface* vlGetFrameBuffer -( - struct vlCSC *csc -) -{ - struct vlBasicCSC *basic_csc; - - assert(csc); - - basic_csc = (struct vlBasicCSC*)csc; - - return basic_csc->framebuffer.cbufs[0]; -} - -static int vlDestroy -( - struct vlCSC *csc -) -{ - struct vlBasicCSC *basic_csc; - struct pipe_context *pipe; - unsigned int i; - - assert(csc); - - basic_csc = (struct vlBasicCSC*)csc; - pipe = basic_csc->pipe; - - if (basic_csc->framebuffer_tex) - { - pipe_surface_reference(&basic_csc->framebuffer.cbufs[0], NULL); - pipe_texture_reference(&basic_csc->framebuffer_tex, NULL); - } - - pipe->delete_sampler_state(pipe, basic_csc->sampler); - pipe->delete_vs_state(pipe, basic_csc->vertex_shader); - pipe->delete_fs_state(pipe, basic_csc->fragment_shader); - - for (i = 0; i < 2; ++i) - pipe_buffer_reference(&basic_csc->vertex_bufs[i].buffer, NULL); - - pipe_buffer_reference(&basic_csc->vs_const_buf.buffer, NULL); - pipe_buffer_reference(&basic_csc->fs_const_buf.buffer, NULL); - - FREE(basic_csc); - - return 0; -} - -/* - * Represents 2 triangles in a strip in normalized coords. - * Used to render the surface onto the frame buffer. - */ -static const struct vlVertex2f surface_verts[4] = -{ - {0.0f, 0.0f}, - {0.0f, 1.0f}, - {1.0f, 0.0f}, - {1.0f, 1.0f} -}; - -/* - * Represents texcoords for the above. We can use the position values directly. - * TODO: Duplicate these in the shader, no need to create a buffer. - */ -static const struct vlVertex2f *surface_texcoords = surface_verts; - -/* - * Identity color conversion constants, for debugging - */ -static const struct vlFragmentShaderConsts identity = -{ - { - 0.0f, 0.0f, 0.0f, 0.0f - }, - { - 1.0f, 0.0f, 0.0f, 0.0f, - 0.0f, 1.0f, 0.0f, 0.0f, - 0.0f, 0.0f, 1.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -/* - * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: - * Y is in [16,235], Cb and Cr are in [16,240] - * R, G, and B are in [16,235] - */ -static const struct vlFragmentShaderConsts bt_601 = -{ - { - 0.0f, 0.501960784f, 0.501960784f, 0.0f - }, - { - 1.0f, 0.0f, 1.371f, 0.0f, - 1.0f, -0.336f, -0.698f, 0.0f, - 1.0f, 1.732f, 0.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -/* - * Converts ITU-R BT.601 YCbCr pixels to RGB pixels where: - * Y is in [16,235], Cb and Cr are in [16,240] - * R, G, and B are in [0,255] - */ -static const struct vlFragmentShaderConsts bt_601_full = -{ - { - 0.062745098f, 0.501960784f, 0.501960784f, 0.0f - }, - { - 1.164f, 0.0f, 1.596f, 0.0f, - 1.164f, -0.391f, -0.813f, 0.0f, - 1.164f, 2.018f, 0.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -/* - * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: - * Y is in [16,235], Cb and Cr are in [16,240] - * R, G, and B are in [16,235] - */ -static const struct vlFragmentShaderConsts bt_709 = -{ - { - 0.0f, 0.501960784f, 0.501960784f, 0.0f - }, - { - 1.0f, 0.0f, 1.540f, 0.0f, - 1.0f, -0.183f, -0.459f, 0.0f, - 1.0f, 1.816f, 0.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -/* - * Converts ITU-R BT.709 YCbCr pixels to RGB pixels where: - * Y is in [16,235], Cb and Cr are in [16,240] - * R, G, and B are in [0,255] - */ -const struct vlFragmentShaderConsts bt_709_full = -{ - { - 0.062745098f, 0.501960784f, 0.501960784f, 0.0f - }, - { - 1.164f, 0.0f, 1.793f, 0.0f, - 1.164f, -0.213f, -0.534f, 0.0f, - 1.164f, 2.115f, 0.0f, 0.0f, - 0.0f, 0.0f, 0.0f, 1.0f - } -}; - -static int vlCreateVertexShader -( - struct vlBasicCSC *csc -) -{ - const unsigned int max_tokens = 50; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(csc); - - pipe = csc->pipe; - tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Vertex texcoords - */ - for (i = 0; i < 2; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling vector to scale vertex pos rect to destination size - * decl c1 ; Translation vector to move vertex pos rect into position - * decl c2 ; Scaling vector to scale texcoord rect to source size - * decl c3 ; Translation vector to move texcoord rect into position - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 3); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Vertex texcoords - */ - for (i = 0; i < 2; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl t0, t1 */ - decl = vl_decl_temps(0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * madd o0, i0, c0, c1 ; Scale and translate unit output rect to destination size and pos - * madd o1, i1, c2, c3 ; Scale and translate unit texcoord rect to source size and pos - */ - for (i = 0; i < 2; ++i) - { - inst = vl_inst4(TGSI_OPCODE_MAD, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i, TGSI_FILE_CONSTANT, i * 2, TGSI_FILE_CONSTANT, i * 2 + 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - csc->vertex_shader = pipe->create_vs_state(pipe, &vs); - FREE(tokens); - - return 0; -} - -static int vlCreateFragmentShader -( - struct vlBasicCSC *csc -) -{ - const unsigned int max_tokens = 50; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(csc); - - pipe = csc->pipe; - tokens = (struct tgsi_token*)MALLOC(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* decl i0 ; Texcoords for s0 */ - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, 1, 0, 0, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl c0 ; Bias vector for CSC - * decl c1-c4 ; CSC matrix c1-c4 - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 4); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0 */ - decl = vl_decl_temps(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl s0 ; Sampler for tex containing picture to display */ - decl = vl_decl_samplers(0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* tex2d t0, i0, s0 ; Read src pixel */ - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_INPUT, 0, TGSI_FILE_SAMPLER, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t0, t0, c0 ; Subtract bias vector from pixel */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * dp4 o0.x, t0, c1 ; Multiply pixel by the color conversion matrix - * dp4 o0.y, t0, c2 - * dp4 o0.z, t0, c3 - */ - for (i = 0; i < 3; ++i) - { - inst = vl_inst3(TGSI_OPCODE_DP4, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, i + 1); - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - csc->fragment_shader = pipe->create_fs_state(pipe, &fs); - FREE(tokens); - - return 0; -} - -static int vlCreateDataBufs -( - struct vlBasicCSC *csc -) -{ - struct pipe_context *pipe; - - assert(csc); - - pipe = csc->pipe; - - /* - * Create our vertex buffer and vertex buffer element - * VB contains 4 vertices that render a quad covering the entire window - * to display a rendered surface - * Quad is rendered as a tri strip - */ - csc->vertex_bufs[0].stride = sizeof(struct vlVertex2f); - csc->vertex_bufs[0].max_index = 3; - csc->vertex_bufs[0].buffer_offset = 0; - csc->vertex_bufs[0].buffer = pipe_buffer_create - ( - pipe->screen, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct vlVertex2f) * 4 - ); - - memcpy - ( - pipe_buffer_map(pipe->screen, csc->vertex_bufs[0].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - surface_verts, - sizeof(struct vlVertex2f) * 4 - ); - - pipe_buffer_unmap(pipe->screen, csc->vertex_bufs[0].buffer); - - csc->vertex_elems[0].src_offset = 0; - csc->vertex_elems[0].vertex_buffer_index = 0; - csc->vertex_elems[0].nr_components = 2; - csc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* - * Create our texcoord buffer and texcoord buffer element - * Texcoord buffer contains the TCs for mapping the rendered surface to the 4 vertices - */ - csc->vertex_bufs[1].stride = sizeof(struct vlVertex2f); - csc->vertex_bufs[1].max_index = 3; - csc->vertex_bufs[1].buffer_offset = 0; - csc->vertex_bufs[1].buffer = pipe_buffer_create - ( - pipe->screen, - 1, - PIPE_BUFFER_USAGE_VERTEX, - sizeof(struct vlVertex2f) * 4 - ); - - memcpy - ( - pipe_buffer_map(pipe->screen, csc->vertex_bufs[1].buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - surface_texcoords, - sizeof(struct vlVertex2f) * 4 - ); - - pipe_buffer_unmap(pipe->screen, csc->vertex_bufs[1].buffer); - - csc->vertex_elems[1].src_offset = 0; - csc->vertex_elems[1].vertex_buffer_index = 1; - csc->vertex_elems[1].nr_components = 2; - csc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* - * Create our vertex shader's constant buffer - * Const buffer contains scaling and translation vectors - */ - csc->vs_const_buf.buffer = pipe_buffer_create - ( - pipe->screen, - 1, - PIPE_BUFFER_USAGE_CONSTANT | PIPE_BUFFER_USAGE_DISCARD, - sizeof(struct vlVertexShaderConsts) - ); - - /* - * Create our fragment shader's constant buffer - * Const buffer contains the color conversion matrix and bias vectors - */ - csc->fs_const_buf.buffer = pipe_buffer_create - ( - pipe->screen, - 1, - PIPE_BUFFER_USAGE_CONSTANT, - sizeof(struct vlFragmentShaderConsts) - ); - - /* - * TODO: Refactor this into a seperate function, - * allow changing the CSC matrix at runtime to switch between regular & full versions - */ - memcpy - ( - pipe_buffer_map(pipe->screen, csc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - &bt_601_full, - sizeof(struct vlFragmentShaderConsts) - ); - - pipe_buffer_unmap(pipe->screen, csc->fs_const_buf.buffer); - - return 0; -} - -static int vlInit -( - struct vlBasicCSC *csc -) -{ - struct pipe_context *pipe; - struct pipe_sampler_state sampler; - - assert(csc); - - pipe = csc->pipe; - - /* Delay creating the FB until vlPutPictureCSC() so we know window size */ - csc->framebuffer_tex = NULL; - csc->framebuffer.width = 0; - csc->framebuffer.height = 0; - csc->framebuffer.nr_cbufs = 1; - csc->framebuffer.cbufs[0] = NULL; - csc->framebuffer.zsbuf = NULL; - - sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.min_img_filter = PIPE_TEX_FILTER_LINEAR; - sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; - sampler.mag_img_filter = PIPE_TEX_FILTER_LINEAR; - sampler.compare_mode = PIPE_TEX_COMPARE_NONE; - sampler.compare_func = PIPE_FUNC_ALWAYS; - sampler.normalized_coords = 1; - /*sampler.prefilter = ;*/ - /*sampler.lod_bias = ;*/ - /*sampler.min_lod = ;*/ - /*sampler.max_lod = ;*/ - /*sampler.border_color[i] = ;*/ - /*sampler.max_anisotropy = ;*/ - csc->sampler = pipe->create_sampler_state(pipe, &sampler); - - vlCreateVertexShader(csc); - vlCreateFragmentShader(csc); - vlCreateDataBufs(csc); - - return 0; -} - -int vlCreateBasicCSC -( - struct pipe_context *pipe, - struct vlCSC **csc -) -{ - struct vlBasicCSC *basic_csc; - - assert(pipe); - assert(csc); - - basic_csc = CALLOC_STRUCT(vlBasicCSC); - - if (!basic_csc) - return 1; - - basic_csc->base.vlResizeFrameBuffer = &vlResizeFrameBuffer; - basic_csc->base.vlBegin = &vlBegin; - basic_csc->base.vlPutPicture = &vlPutPictureCSC; - basic_csc->base.vlEnd = &vlEnd; - basic_csc->base.vlGetFrameBuffer = &vlGetFrameBuffer; - basic_csc->base.vlDestroy = &vlDestroy; - basic_csc->pipe = pipe; - - vlInit(basic_csc); - - *csc = &basic_csc->base; - - return 0; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_basic_csc.h b/src/gallium/state_trackers/g3dvl/vl_basic_csc.h deleted file mode 100644 index 2e17f1d814a..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_basic_csc.h +++ /dev/null @@ -1,13 +0,0 @@ -#ifndef vl_basic_csc_h -#define vl_basic_csc_h - -struct pipe_context; -struct vlCSC; - -int vlCreateBasicCSC -( - struct pipe_context *pipe, - struct vlCSC **csc -); - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_context.c b/src/gallium/state_trackers/g3dvl/vl_context.c deleted file mode 100644 index 5cfd233c4c1..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_context.c +++ /dev/null @@ -1,205 +0,0 @@ -#define VL_INTERNAL -#include "vl_context.h" -#include <assert.h> -#include <pipe/p_context.h> -#include <pipe/p_state.h> -#include <util/u_memory.h> -#include "vl_render.h" -#include "vl_r16snorm_mc_buf.h" -#include "vl_csc.h" -#include "vl_basic_csc.h" - -static int vlInitCommon(struct vlContext *context) -{ - struct pipe_context *pipe; - struct pipe_rasterizer_state rast; - struct pipe_blend_state blend; - struct pipe_depth_stencil_alpha_state dsa; - unsigned int i; - - assert(context); - - pipe = context->pipe; - - rast.flatshade = 1; - rast.flatshade_first = 0; - rast.light_twoside = 0; - rast.front_winding = PIPE_WINDING_CCW; - rast.cull_mode = PIPE_WINDING_CW; - rast.fill_cw = PIPE_POLYGON_MODE_FILL; - rast.fill_ccw = PIPE_POLYGON_MODE_FILL; - rast.offset_cw = 0; - rast.offset_ccw = 0; - rast.scissor = 0; - rast.poly_smooth = 0; - rast.poly_stipple_enable = 0; - rast.point_sprite = 0; - rast.point_size_per_vertex = 0; - rast.multisample = 0; - rast.line_smooth = 0; - rast.line_stipple_enable = 0; - rast.line_stipple_factor = 0; - rast.line_stipple_pattern = 0; - rast.line_last_pixel = 0; - rast.bypass_vs_clip_and_viewport = 0; - rast.line_width = 1; - rast.point_smooth = 0; - rast.point_size = 1; - rast.offset_units = 1; - rast.offset_scale = 1; - /*rast.sprite_coord_mode[i] = ;*/ - context->raster = pipe->create_rasterizer_state(pipe, &rast); - pipe->bind_rasterizer_state(pipe, context->raster); - - blend.blend_enable = 0; - blend.rgb_func = PIPE_BLEND_ADD; - blend.rgb_src_factor = PIPE_BLENDFACTOR_ONE; - blend.rgb_dst_factor = PIPE_BLENDFACTOR_ONE; - blend.alpha_func = PIPE_BLEND_ADD; - blend.alpha_src_factor = PIPE_BLENDFACTOR_ONE; - blend.alpha_dst_factor = PIPE_BLENDFACTOR_ONE; - blend.logicop_enable = 0; - blend.logicop_func = PIPE_LOGICOP_CLEAR; - /* Needed to allow color writes to FB, even if blending disabled */ - blend.colormask = PIPE_MASK_RGBA; - blend.dither = 0; - context->blend = pipe->create_blend_state(pipe, &blend); - pipe->bind_blend_state(pipe, context->blend); - - dsa.depth.enabled = 0; - dsa.depth.writemask = 0; - dsa.depth.func = PIPE_FUNC_ALWAYS; - dsa.depth.occlusion_count = 0; - for (i = 0; i < 2; ++i) - { - dsa.stencil[i].enabled = 0; - dsa.stencil[i].func = PIPE_FUNC_ALWAYS; - dsa.stencil[i].fail_op = PIPE_STENCIL_OP_KEEP; - dsa.stencil[i].zpass_op = PIPE_STENCIL_OP_KEEP; - dsa.stencil[i].zfail_op = PIPE_STENCIL_OP_KEEP; - dsa.stencil[i].ref_value = 0; - dsa.stencil[i].valuemask = 0; - dsa.stencil[i].writemask = 0; - } - dsa.alpha.enabled = 0; - dsa.alpha.func = PIPE_FUNC_ALWAYS; - dsa.alpha.ref_value = 0; - context->dsa = pipe->create_depth_stencil_alpha_state(pipe, &dsa); - pipe->bind_depth_stencil_alpha_state(pipe, context->dsa); - - return 0; -} - -int vlCreateContext -( - struct vlScreen *screen, - struct pipe_context *pipe, - unsigned int picture_width, - unsigned int picture_height, - enum vlFormat picture_format, - enum vlProfile profile, - enum vlEntryPoint entry_point, - struct vlContext **context -) -{ - struct vlContext *ctx; - - assert(screen); - assert(context); - assert(pipe); - - ctx = CALLOC_STRUCT(vlContext); - - if (!ctx) - return 1; - - ctx->screen = screen; - ctx->pipe = pipe; - ctx->picture_width = picture_width; - ctx->picture_height = picture_height; - ctx->picture_format = picture_format; - ctx->profile = profile; - ctx->entry_point = entry_point; - - vlInitCommon(ctx); - - vlCreateR16SNormBufferedMC(pipe, picture_width, picture_height, picture_format, &ctx->render); - vlCreateBasicCSC(pipe, &ctx->csc); - - *context = ctx; - - return 0; -} - -int vlDestroyContext -( - struct vlContext *context -) -{ - assert(context); - - /* XXX: Must unbind shaders before we can delete them for some reason */ - context->pipe->bind_vs_state(context->pipe, NULL); - context->pipe->bind_fs_state(context->pipe, NULL); - - context->render->vlDestroy(context->render); - context->csc->vlDestroy(context->csc); - - context->pipe->delete_blend_state(context->pipe, context->blend); - context->pipe->delete_rasterizer_state(context->pipe, context->raster); - context->pipe->delete_depth_stencil_alpha_state(context->pipe, context->dsa); - - FREE(context); - - return 0; -} - -struct vlScreen* vlContextGetScreen -( - struct vlContext *context -) -{ - assert(context); - - return context->screen; -} - -struct pipe_context* vlGetPipeContext -( - struct vlContext *context -) -{ - assert(context); - - return context->pipe; -} - -unsigned int vlGetPictureWidth -( - struct vlContext *context -) -{ - assert(context); - - return context->picture_width; -} - -unsigned int vlGetPictureHeight -( - struct vlContext *context -) -{ - assert(context); - - return context->picture_height; -} - -enum vlFormat vlGetPictureFormat -( - struct vlContext *context -) -{ - assert(context); - - return context->picture_format; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_context.h b/src/gallium/state_trackers/g3dvl/vl_context.h deleted file mode 100644 index 3d14634c44e..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_context.h +++ /dev/null @@ -1,73 +0,0 @@ -#ifndef vl_context_h -#define vl_context_h - -#include "vl_types.h" - -struct pipe_context; - -#ifdef VL_INTERNAL -struct vlRender; -struct vlCSC; - -struct vlContext -{ - struct vlScreen *screen; - struct pipe_context *pipe; - unsigned int picture_width; - unsigned int picture_height; - enum vlFormat picture_format; - enum vlProfile profile; - enum vlEntryPoint entry_point; - - void *raster; - void *dsa; - void *blend; - - struct vlRender *render; - struct vlCSC *csc; -}; -#endif - -int vlCreateContext -( - struct vlScreen *screen, - struct pipe_context *pipe, - unsigned int picture_width, - unsigned int picture_height, - enum vlFormat picture_format, - enum vlProfile profile, - enum vlEntryPoint entry_point, - struct vlContext **context -); - -int vlDestroyContext -( - struct vlContext *context -); - -struct vlScreen* vlContextGetScreen -( - struct vlContext *context -); - -struct pipe_context* vlGetPipeContext -( - struct vlContext *context -); - -unsigned int vlGetPictureWidth -( - struct vlContext *context -); - -unsigned int vlGetPictureHeight -( - struct vlContext *context -); - -enum vlFormat vlGetPictureFormat -( - struct vlContext *context -); - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_csc.h b/src/gallium/state_trackers/g3dvl/vl_csc.h deleted file mode 100644 index 36417a27929..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_csc.h +++ /dev/null @@ -1,53 +0,0 @@ -#ifndef vl_csc_h -#define vl_csc_h - -#include "vl_types.h" - -struct pipe_surface; - -struct vlCSC -{ - int (*vlResizeFrameBuffer) - ( - struct vlCSC *csc, - unsigned int width, - unsigned int height - ); - - int (*vlBegin) - ( - struct vlCSC *csc - ); - - int (*vlPutPicture) - ( - struct vlCSC *csc, - struct vlSurface *surface, - int srcx, - int srcy, - int srcw, - int srch, - int destx, - int desty, - int destw, - int desth, - enum vlPictureType picture_type - ); - - int (*vlEnd) - ( - struct vlCSC *csc - ); - - struct pipe_surface* (*vlGetFrameBuffer) - ( - struct vlCSC *csc - ); - - int (*vlDestroy) - ( - struct vlCSC *csc - ); -}; - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_defs.h b/src/gallium/state_trackers/g3dvl/vl_defs.h deleted file mode 100644 index d612d02502f..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_defs.h +++ /dev/null @@ -1,11 +0,0 @@ -#ifndef vl_defs_h -#define vl_defs_h - -#define VL_BLOCK_WIDTH 8 -#define VL_BLOCK_HEIGHT 8 -#define VL_BLOCK_SIZE (VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT) -#define VL_MACROBLOCK_WIDTH 16 -#define VL_MACROBLOCK_HEIGHT 16 -#define VL_MACROBLOCK_SIZE (VL_MACROBLOCK_WIDTH * VL_MACROBLOCK_HEIGHT) - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_display.c b/src/gallium/state_trackers/g3dvl/vl_display.c deleted file mode 100644 index dce06de7583..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_display.c +++ /dev/null @@ -1,48 +0,0 @@ -#define VL_INTERNAL -#include "vl_display.h" -#include <assert.h> -#include <util/u_memory.h> - -int vlCreateDisplay -( - vlNativeDisplay native_display, - struct vlDisplay **display -) -{ - struct vlDisplay *dpy; - - assert(native_display); - assert(display); - - dpy = CALLOC_STRUCT(vlDisplay); - - if (!dpy) - return 1; - - dpy->native = native_display; - *display = dpy; - - return 0; -} - -int vlDestroyDisplay -( - struct vlDisplay *display -) -{ - assert(display); - - FREE(display); - - return 0; -} - -vlNativeDisplay vlGetNativeDisplay -( - struct vlDisplay *display -) -{ - assert(display); - - return display->native; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_display.h b/src/gallium/state_trackers/g3dvl/vl_display.h deleted file mode 100644 index e11fd407993..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_display.h +++ /dev/null @@ -1,29 +0,0 @@ -#ifndef vl_display_h -#define vl_display_h - -#include "vl_types.h" - -#ifdef VL_INTERNAL -struct vlDisplay -{ - vlNativeDisplay native; -}; -#endif - -int vlCreateDisplay -( - vlNativeDisplay native_display, - struct vlDisplay **display -); - -int vlDestroyDisplay -( - struct vlDisplay *display -); - -vlNativeDisplay vlGetNativeDisplay -( - struct vlDisplay *display -); - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c deleted file mode 100644 index 23631adb693..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.c +++ /dev/null @@ -1,1155 +0,0 @@ -#define VL_INTERNAL -#include "vl_r16snorm_mc_buf.h" -#include <assert.h> -#include <pipe/p_context.h> -#include <pipe/p_screen.h> -#include <pipe/p_state.h> -#include <pipe/p_inlines.h> -#include <tgsi/tgsi_parse.h> -#include <tgsi/tgsi_build.h> -#include <util/u_math.h> -#include <util/u_memory.h> -#include "vl_render.h" -#include "vl_shader_build.h" -#include "vl_surface.h" -#include "vl_util.h" -#include "vl_types.h" -#include "vl_defs.h" - -const unsigned int DEFAULT_BUF_ALIGNMENT = 1; - -enum vlMacroBlockTypeEx -{ - vlMacroBlockExTypeIntra, - vlMacroBlockExTypeFwdPredictedFrame, - vlMacroBlockExTypeFwdPredictedField, - vlMacroBlockExTypeBkwdPredictedFrame, - vlMacroBlockExTypeBkwdPredictedField, - vlMacroBlockExTypeBiPredictedFrame, - vlMacroBlockExTypeBiPredictedField, - - vlNumMacroBlockExTypes -}; - -struct vlVertexShaderConsts -{ - struct vlVertex4f denorm; -}; - -struct vlFragmentShaderConsts -{ - struct vlVertex4f multiplier; - struct vlVertex4f div; -}; - -struct vlMacroBlockVertexStream0 -{ - struct vlVertex2f pos; - struct vlVertex2f luma_tc; - struct vlVertex2f cb_tc; - struct vlVertex2f cr_tc; -}; - -struct vlR16SnormBufferedMC -{ - struct vlRender base; - - unsigned int picture_width; - unsigned int picture_height; - enum vlFormat picture_format; - unsigned int macroblocks_per_picture; - - struct vlSurface *buffered_surface; - struct vlSurface *past_surface; - struct vlSurface *future_surface; - struct vlVertex2f surface_tex_inv_size; - struct vlVertex2f zero_block[3]; - unsigned int num_macroblocks; - struct vlMpeg2MacroBlock *macroblocks; - struct pipe_transfer *tex_transfer[3]; - short *texels[3]; - - struct pipe_context *pipe; - struct pipe_viewport_state viewport; - struct pipe_framebuffer_state render_target; - - union - { - void *all[5]; - struct - { - void *y; - void *cb; - void *cr; - void *ref[2]; - }; - } samplers; - - union - { - struct pipe_texture *all[5]; - struct - { - struct pipe_texture *y; - struct pipe_texture *cb; - struct pipe_texture *cr; - struct pipe_texture *ref[2]; - }; - } textures; - - union - { - struct pipe_vertex_buffer all[3]; - struct - { - struct pipe_vertex_buffer ycbcr; - struct pipe_vertex_buffer ref[2]; - }; - } vertex_bufs; - - void *i_vs, *p_vs[2], *b_vs[2]; - void *i_fs, *p_fs[2], *b_fs[2]; - struct pipe_vertex_element vertex_elems[8]; - struct pipe_constant_buffer vs_const_buf; - struct pipe_constant_buffer fs_const_buf; -}; - -static inline int vlBegin -( - struct vlRender *render -) -{ - assert(render); - - return 0; -} - -static inline int vlGrabFrameCodedBlock(short *src, short *dst, unsigned int dst_pitch) -{ - unsigned int y; - - for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - memcpy - ( - dst + y * dst_pitch, - src + y * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - - return 0; -} - -static inline int vlGrabFieldCodedBlock(short *src, short *dst, unsigned int dst_pitch) -{ - unsigned int y; - - for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - memcpy - ( - dst + y * dst_pitch * 2, - src + y * VL_BLOCK_WIDTH, - VL_BLOCK_WIDTH * 2 - ); - - return 0; -} - -static inline int vlGrabNoBlock(short *dst, unsigned int dst_pitch) -{ - unsigned int y; - - for (y = 0; y < VL_BLOCK_HEIGHT; ++y) - memset - ( - dst + y * dst_pitch, - 0, - VL_BLOCK_WIDTH * 2 - ); - - return 0; -} - -static inline int vlGrabBlocks -( - struct vlR16SnormBufferedMC *mc, - unsigned int mbx, - unsigned int mby, - enum vlDCTType dct_type, - unsigned int coded_block_pattern, - short *blocks -) -{ - short *texels; - unsigned int tex_pitch; - unsigned int x, y, tb = 0, sb = 0; - unsigned int mbpx = mbx * VL_MACROBLOCK_WIDTH, mbpy = mby * VL_MACROBLOCK_HEIGHT; - - assert(mc); - assert(blocks); - - tex_pitch = mc->tex_transfer[0]->stride / mc->tex_transfer[0]->block.size; - texels = mc->texels[0] + mbpy * tex_pitch + mbpx; - - for (y = 0; y < 2; ++y) - { - for (x = 0; x < 2; ++x, ++tb) - { - if ((coded_block_pattern >> (5 - tb)) & 1) - { - short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; - - if (dct_type == vlDCTTypeFrameCoded) - { - vlGrabFrameCodedBlock - ( - cur_block, - texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH, - tex_pitch - ); - } - else - { - vlGrabFieldCodedBlock - ( - cur_block, - texels + y * tex_pitch + x * VL_BLOCK_WIDTH, - tex_pitch - ); - } - - ++sb; - } - else if (mc->zero_block[0].x < 0.0f) - { - vlGrabNoBlock(texels + y * tex_pitch * VL_BLOCK_HEIGHT + x * VL_BLOCK_WIDTH, tex_pitch); - - mc->zero_block[0].x = (mbpx + x * 8) * mc->surface_tex_inv_size.x; - mc->zero_block[0].y = (mbpy + y * 8) * mc->surface_tex_inv_size.y; - } - } - } - - /* TODO: Implement 422, 444 */ - mbpx >>= 1; - mbpy >>= 1; - - for (tb = 0; tb < 2; ++tb) - { - tex_pitch = mc->tex_transfer[tb + 1]->stride / mc->tex_transfer[tb + 1]->block.size; - texels = mc->texels[tb + 1] + mbpy * tex_pitch + mbpx; - - if ((coded_block_pattern >> (1 - tb)) & 1) - { - short *cur_block = blocks + sb * VL_BLOCK_WIDTH * VL_BLOCK_HEIGHT; - - vlGrabFrameCodedBlock - ( - cur_block, - texels, - tex_pitch - ); - - ++sb; - } - else if (mc->zero_block[tb + 1].x < 0.0f) - { - vlGrabNoBlock(texels, tex_pitch); - - mc->zero_block[tb + 1].x = (mbpx << 1) * mc->surface_tex_inv_size.x; - mc->zero_block[tb + 1].y = (mbpy << 1) * mc->surface_tex_inv_size.y; - } - } - - return 0; -} - -static inline enum vlMacroBlockTypeEx vlGetMacroBlockTypeEx(struct vlMpeg2MacroBlock *mb) -{ - assert(mb); - - switch (mb->mb_type) - { - case vlMacroBlockTypeIntra: - return vlMacroBlockExTypeIntra; - case vlMacroBlockTypeFwdPredicted: - return mb->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeFwdPredictedFrame : vlMacroBlockExTypeFwdPredictedField; - case vlMacroBlockTypeBkwdPredicted: - return mb->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeBkwdPredictedFrame : vlMacroBlockExTypeBkwdPredictedField; - case vlMacroBlockTypeBiPredicted: - return mb->mo_type == vlMotionTypeFrame ? - vlMacroBlockExTypeBiPredictedFrame : vlMacroBlockExTypeBiPredictedField; - default: - assert(0); - } - - /* Unreachable */ - return -1; -} - -static inline int vlGrabMacroBlock -( - struct vlR16SnormBufferedMC *mc, - struct vlMpeg2MacroBlock *macroblock -) -{ - assert(mc); - assert(macroblock); - assert(mc->num_macroblocks < mc->macroblocks_per_picture); - - mc->macroblocks[mc->num_macroblocks].mbx = macroblock->mbx; - mc->macroblocks[mc->num_macroblocks].mby = macroblock->mby; - mc->macroblocks[mc->num_macroblocks].mb_type = macroblock->mb_type; - mc->macroblocks[mc->num_macroblocks].mo_type = macroblock->mo_type; - mc->macroblocks[mc->num_macroblocks].dct_type = macroblock->dct_type; - mc->macroblocks[mc->num_macroblocks].PMV[0][0][0] = macroblock->PMV[0][0][0]; - mc->macroblocks[mc->num_macroblocks].PMV[0][0][1] = macroblock->PMV[0][0][1]; - mc->macroblocks[mc->num_macroblocks].PMV[0][1][0] = macroblock->PMV[0][1][0]; - mc->macroblocks[mc->num_macroblocks].PMV[0][1][1] = macroblock->PMV[0][1][1]; - mc->macroblocks[mc->num_macroblocks].PMV[1][0][0] = macroblock->PMV[1][0][0]; - mc->macroblocks[mc->num_macroblocks].PMV[1][0][1] = macroblock->PMV[1][0][1]; - mc->macroblocks[mc->num_macroblocks].PMV[1][1][0] = macroblock->PMV[1][1][0]; - mc->macroblocks[mc->num_macroblocks].PMV[1][1][1] = macroblock->PMV[1][1][1]; - mc->macroblocks[mc->num_macroblocks].cbp = macroblock->cbp; - mc->macroblocks[mc->num_macroblocks].blocks = macroblock->blocks; - - vlGrabBlocks - ( - mc, - macroblock->mbx, - macroblock->mby, - macroblock->dct_type, - macroblock->cbp, - macroblock->blocks - ); - - mc->num_macroblocks++; - - return 0; -} - -#define SET_BLOCK(vb, cbp, mbx, mby, unitx, unity, ofsx, ofsy, hx, hy, lm, cbm, crm, zb) \ - do { \ - (vb)[0].pos.x = (mbx) * (unitx) + (ofsx); (vb)[0].pos.y = (mby) * (unity) + (ofsy); \ - (vb)[1].pos.x = (mbx) * (unitx) + (ofsx); (vb)[1].pos.y = (mby) * (unity) + (ofsy) + (hy); \ - (vb)[2].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].pos.y = (mby) * (unity) + (ofsy); \ - (vb)[3].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].pos.y = (mby) * (unity) + (ofsy); \ - (vb)[4].pos.x = (mbx) * (unitx) + (ofsx); (vb)[4].pos.y = (mby) * (unity) + (ofsy) + (hy); \ - (vb)[5].pos.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].pos.y = (mby) * (unity) + (ofsy) + (hy); \ - \ - if ((cbp) & (lm)) \ - { \ - (vb)[0].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].luma_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[1].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - (vb)[2].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].luma_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[3].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].luma_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[4].luma_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - (vb)[5].luma_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].luma_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - } \ - else \ - { \ - (vb)[0].luma_tc.x = (zb)[0].x; (vb)[0].luma_tc.y = (zb)[0].y; \ - (vb)[1].luma_tc.x = (zb)[0].x; (vb)[1].luma_tc.y = (zb)[0].y + (hy); \ - (vb)[2].luma_tc.x = (zb)[0].x + (hx); (vb)[2].luma_tc.y = (zb)[0].y; \ - (vb)[3].luma_tc.x = (zb)[0].x + (hx); (vb)[3].luma_tc.y = (zb)[0].y; \ - (vb)[4].luma_tc.x = (zb)[0].x; (vb)[4].luma_tc.y = (zb)[0].y + (hy); \ - (vb)[5].luma_tc.x = (zb)[0].x + (hx); (vb)[5].luma_tc.y = (zb)[0].y + (hy); \ - } \ - \ - if ((cbp) & (cbm)) \ - { \ - (vb)[0].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].cb_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[1].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - (vb)[2].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].cb_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[3].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].cb_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[4].cb_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - (vb)[5].cb_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].cb_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - } \ - else \ - { \ - (vb)[0].cb_tc.x = (zb)[1].x; (vb)[0].cb_tc.y = (zb)[1].y; \ - (vb)[1].cb_tc.x = (zb)[1].x; (vb)[1].cb_tc.y = (zb)[1].y + (hy); \ - (vb)[2].cb_tc.x = (zb)[1].x + (hx); (vb)[2].cb_tc.y = (zb)[1].y; \ - (vb)[3].cb_tc.x = (zb)[1].x + (hx); (vb)[3].cb_tc.y = (zb)[1].y; \ - (vb)[4].cb_tc.x = (zb)[1].x; (vb)[4].cb_tc.y = (zb)[1].y + (hy); \ - (vb)[5].cb_tc.x = (zb)[1].x + (hx); (vb)[5].cb_tc.y = (zb)[1].y + (hy); \ - } \ - \ - if ((cbp) & (crm)) \ - { \ - (vb)[0].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[0].cr_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[1].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[1].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - (vb)[2].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[2].cr_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[3].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[3].cr_tc.y = (mby) * (unity) + (ofsy); \ - (vb)[4].cr_tc.x = (mbx) * (unitx) + (ofsx); (vb)[4].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - (vb)[5].cr_tc.x = (mbx) * (unitx) + (ofsx) + (hx); (vb)[5].cr_tc.y = (mby) * (unity) + (ofsy) + (hy); \ - } \ - else \ - { \ - (vb)[0].cr_tc.x = (zb)[2].x; (vb)[0].cr_tc.y = (zb)[2].y; \ - (vb)[1].cr_tc.x = (zb)[2].x; (vb)[1].cr_tc.y = (zb)[2].y + (hy); \ - (vb)[2].cr_tc.x = (zb)[2].x + (hx); (vb)[2].cr_tc.y = (zb)[2].y; \ - (vb)[3].cr_tc.x = (zb)[2].x + (hx); (vb)[3].cr_tc.y = (zb)[2].y; \ - (vb)[4].cr_tc.x = (zb)[2].x; (vb)[4].cr_tc.y = (zb)[2].y + (hy); \ - (vb)[5].cr_tc.x = (zb)[2].x + (hx); (vb)[5].cr_tc.y = (zb)[2].y + (hy); \ - } \ - } while (0) - -static inline int vlGenMacroblockVerts -( - struct vlR16SnormBufferedMC *mc, - struct vlMpeg2MacroBlock *macroblock, - unsigned int pos, - struct vlMacroBlockVertexStream0 *ycbcr_vb, - struct vlVertex2f **ref_vb -) -{ - struct vlVertex2f mo_vec[2]; - unsigned int i; - - assert(mc); - assert(macroblock); - assert(ycbcr_vb); - assert(pos < mc->macroblocks_per_picture); - - switch (macroblock->mb_type) - { - case vlMacroBlockTypeBiPredicted: - { - struct vlVertex2f *vb; - - assert(ref_vb && ref_vb[1]); - - vb = ref_vb[1] + pos * 2 * 24; - - mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; - - if (macroblock->mo_type == vlMotionTypeFrame) - { - for (i = 0; i < 24 * 2; i += 2) - { - vb[i].x = mo_vec[0].x; - vb[i].y = mo_vec[0].y; - } - } - else - { - mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y; - - for (i = 0; i < 24 * 2; i += 2) - { - vb[i].x = mo_vec[0].x; - vb[i].y = mo_vec[0].y; - vb[i + 1].x = mo_vec[1].x; - vb[i + 1].y = mo_vec[1].y; - } - } - - /* fall-through */ - } - case vlMacroBlockTypeFwdPredicted: - case vlMacroBlockTypeBkwdPredicted: - { - struct vlVertex2f *vb; - - assert(ref_vb && ref_vb[0]); - - vb = ref_vb[0] + pos * 2 * 24; - - if (macroblock->mb_type == vlMacroBlockTypeBkwdPredicted) - { - mo_vec[0].x = macroblock->PMV[0][1][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[0].y = macroblock->PMV[0][1][1] * 0.5f * mc->surface_tex_inv_size.y; - - if (macroblock->mo_type == vlMotionTypeField) - { - mo_vec[1].x = macroblock->PMV[1][1][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[1].y = macroblock->PMV[1][1][1] * 0.5f * mc->surface_tex_inv_size.y; - } - } - else - { - mo_vec[0].x = macroblock->PMV[0][0][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[0].y = macroblock->PMV[0][0][1] * 0.5f * mc->surface_tex_inv_size.y; - - if (macroblock->mo_type == vlMotionTypeField) - { - mo_vec[1].x = macroblock->PMV[1][0][0] * 0.5f * mc->surface_tex_inv_size.x; - mo_vec[1].y = macroblock->PMV[1][0][1] * 0.5f * mc->surface_tex_inv_size.y; - } - } - - if (macroblock->mo_type == vlMotionTypeFrame) - { - for (i = 0; i < 24 * 2; i += 2) - { - vb[i].x = mo_vec[0].x; - vb[i].y = mo_vec[0].y; - } - } - else - { - for (i = 0; i < 24 * 2; i += 2) - { - vb[i].x = mo_vec[0].x; - vb[i].y = mo_vec[0].y; - vb[i + 1].x = mo_vec[1].x; - vb[i + 1].y = mo_vec[1].y; - } - } - - /* fall-through */ - } - case vlMacroBlockTypeIntra: - { - const struct vlVertex2f unit = - { - mc->surface_tex_inv_size.x * VL_MACROBLOCK_WIDTH, - mc->surface_tex_inv_size.y * VL_MACROBLOCK_HEIGHT - }; - const struct vlVertex2f half = - { - mc->surface_tex_inv_size.x * (VL_MACROBLOCK_WIDTH / 2), - mc->surface_tex_inv_size.y * (VL_MACROBLOCK_HEIGHT / 2) - }; - - struct vlMacroBlockVertexStream0 *vb; - - vb = ycbcr_vb + pos * 24; - - SET_BLOCK - ( - vb, - macroblock->cbp, macroblock->mbx, macroblock->mby, - unit.x, unit.y, 0, 0, half.x, half.y, - 32, 2, 1, mc->zero_block - ); - - SET_BLOCK - ( - vb + 6, - macroblock->cbp, macroblock->mbx, macroblock->mby, - unit.x, unit.y, half.x, 0, half.x, half.y, - 16, 2, 1, mc->zero_block - ); - - SET_BLOCK - ( - vb + 12, - macroblock->cbp, macroblock->mbx, macroblock->mby, - unit.x, unit.y, 0, half.y, half.x, half.y, - 8, 2, 1, mc->zero_block - ); - - SET_BLOCK - ( - vb + 18, - macroblock->cbp, macroblock->mbx, macroblock->mby, - unit.x, unit.y, half.x, half.y, half.x, half.y, - 4, 2, 1, mc->zero_block - ); - - break; - } - default: - assert(0); - } - - return 0; -} - -static int vlFlush -( - struct vlRender *render -) -{ - struct vlR16SnormBufferedMC *mc; - struct pipe_context *pipe; - struct vlVertexShaderConsts *vs_consts; - unsigned int num_macroblocks[vlNumMacroBlockExTypes] = {0}; - unsigned int offset[vlNumMacroBlockExTypes]; - unsigned int vb_start = 0; - unsigned int i; - - assert(render); - - mc = (struct vlR16SnormBufferedMC*)render; - - if (!mc->buffered_surface) - return 0; - - if (mc->num_macroblocks < mc->macroblocks_per_picture) - return 0; - - assert(mc->num_macroblocks <= mc->macroblocks_per_picture); - - pipe = mc->pipe; - - for (i = 0; i < mc->num_macroblocks; ++i) - { - enum vlMacroBlockTypeEx mb_type_ex = vlGetMacroBlockTypeEx(&mc->macroblocks[i]); - - num_macroblocks[mb_type_ex]++; - } - - offset[0] = 0; - - for (i = 1; i < vlNumMacroBlockExTypes; ++i) - offset[i] = offset[i - 1] + num_macroblocks[i - 1]; - - { - struct vlMacroBlockVertexStream0 *ycbcr_vb; - struct vlVertex2f *ref_vb[2]; - - ycbcr_vb = (struct vlMacroBlockVertexStream0*)pipe_buffer_map - ( - pipe->screen, - mc->vertex_bufs.ycbcr.buffer, - PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD - ); - - for (i = 0; i < 2; ++i) - ref_vb[i] = (struct vlVertex2f*)pipe_buffer_map - ( - pipe->screen, - mc->vertex_bufs.ref[i].buffer, - PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD - ); - - for (i = 0; i < mc->num_macroblocks; ++i) - { - enum vlMacroBlockTypeEx mb_type_ex = vlGetMacroBlockTypeEx(&mc->macroblocks[i]); - - vlGenMacroblockVerts(mc, &mc->macroblocks[i], offset[mb_type_ex], ycbcr_vb, ref_vb); - - offset[mb_type_ex]++; - } - - pipe_buffer_unmap(pipe->screen, mc->vertex_bufs.ycbcr.buffer); - for (i = 0; i < 2; ++i) - pipe_buffer_unmap(pipe->screen, mc->vertex_bufs.ref[i].buffer); - } - - for (i = 0; i < 3; ++i) - { - pipe->screen->transfer_unmap(pipe->screen, mc->tex_transfer[i]); - pipe->screen->tex_transfer_destroy(mc->tex_transfer[i]); - } - - mc->render_target.cbufs[0] = pipe->screen->get_tex_surface - ( - pipe->screen, - mc->buffered_surface->texture, - 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE - ); - - pipe->set_framebuffer_state(pipe, &mc->render_target); - pipe->set_viewport_state(pipe, &mc->viewport); - vs_consts = pipe_buffer_map - ( - pipe->screen, - mc->vs_const_buf.buffer, - PIPE_BUFFER_USAGE_CPU_WRITE | PIPE_BUFFER_USAGE_DISCARD - ); - - vs_consts->denorm.x = mc->buffered_surface->texture->width[0]; - vs_consts->denorm.y = mc->buffered_surface->texture->height[0]; - - pipe_buffer_unmap(pipe->screen, mc->vs_const_buf.buffer); - pipe->set_constant_buffer(pipe, PIPE_SHADER_VERTEX, 0, &mc->vs_const_buf); - pipe->set_constant_buffer(pipe, PIPE_SHADER_FRAGMENT, 0, &mc->fs_const_buf); - - if (num_macroblocks[vlMacroBlockExTypeIntra] > 0) - { - pipe->set_vertex_buffers(pipe, 1, mc->vertex_bufs.all); - pipe->set_vertex_elements(pipe, 4, mc->vertex_elems); - pipe->set_sampler_textures(pipe, 3, mc->textures.all); - pipe->bind_sampler_states(pipe, 3, mc->samplers.all); - pipe->bind_vs_state(pipe, mc->i_vs); - pipe->bind_fs_state(pipe, mc->i_fs); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeIntra] * 24); - vb_start += num_macroblocks[vlMacroBlockExTypeIntra] * 24; - } - - if (num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] > 0) - { - pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs.all); - pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); - mc->textures.ref[0] = mc->past_surface->texture; - pipe->set_sampler_textures(pipe, 4, mc->textures.all); - pipe->bind_sampler_states(pipe, 4, mc->samplers.all); - pipe->bind_vs_state(pipe, mc->p_vs[0]); - pipe->bind_fs_state(pipe, mc->p_fs[0]); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24); - vb_start += num_macroblocks[vlMacroBlockExTypeFwdPredictedFrame] * 24; - } - - if (num_macroblocks[vlMacroBlockExTypeFwdPredictedField] > 0) - { - pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs.all); - pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); - mc->textures.ref[0] = mc->past_surface->texture; - pipe->set_sampler_textures(pipe, 4, mc->textures.all); - pipe->bind_sampler_states(pipe, 4, mc->samplers.all); - pipe->bind_vs_state(pipe, mc->p_vs[1]); - pipe->bind_fs_state(pipe, mc->p_fs[1]); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24); - vb_start += num_macroblocks[vlMacroBlockExTypeFwdPredictedField] * 24; - } - - if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] > 0) - { - pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs.all); - pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); - mc->textures.ref[0] = mc->future_surface->texture; - pipe->set_sampler_textures(pipe, 4, mc->textures.all); - pipe->bind_sampler_states(pipe, 4, mc->samplers.all); - pipe->bind_vs_state(pipe, mc->p_vs[0]); - pipe->bind_fs_state(pipe, mc->p_fs[0]); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24); - vb_start += num_macroblocks[vlMacroBlockExTypeBkwdPredictedFrame] * 24; - } - - if (num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] > 0) - { - pipe->set_vertex_buffers(pipe, 2, mc->vertex_bufs.all); - pipe->set_vertex_elements(pipe, 6, mc->vertex_elems); - mc->textures.ref[0] = mc->future_surface->texture; - pipe->set_sampler_textures(pipe, 4, mc->textures.all); - pipe->bind_sampler_states(pipe, 4, mc->samplers.all); - pipe->bind_vs_state(pipe, mc->p_vs[1]); - pipe->bind_fs_state(pipe, mc->p_fs[1]); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24); - vb_start += num_macroblocks[vlMacroBlockExTypeBkwdPredictedField] * 24; - } - - if (num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] > 0) - { - pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs.all); - pipe->set_vertex_elements(pipe, 8, mc->vertex_elems); - mc->textures.ref[0] = mc->past_surface->texture; - mc->textures.ref[1] = mc->future_surface->texture; - pipe->set_sampler_textures(pipe, 5, mc->textures.all); - pipe->bind_sampler_states(pipe, 5, mc->samplers.all); - pipe->bind_vs_state(pipe, mc->b_vs[0]); - pipe->bind_fs_state(pipe, mc->b_fs[0]); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24); - vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedFrame] * 24; - } - - if (num_macroblocks[vlMacroBlockExTypeBiPredictedField] > 0) - { - pipe->set_vertex_buffers(pipe, 3, mc->vertex_bufs.all); - pipe->set_vertex_elements(pipe, 8, mc->vertex_elems); - mc->textures.ref[0] = mc->past_surface->texture; - mc->textures.ref[1] = mc->future_surface->texture; - pipe->set_sampler_textures(pipe, 5, mc->textures.all); - pipe->bind_sampler_states(pipe, 5, mc->samplers.all); - pipe->bind_vs_state(pipe, mc->b_vs[1]); - pipe->bind_fs_state(pipe, mc->b_fs[1]); - - pipe->draw_arrays(pipe, PIPE_PRIM_TRIANGLES, vb_start, num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24); - vb_start += num_macroblocks[vlMacroBlockExTypeBiPredictedField] * 24; - } - - pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, &mc->buffered_surface->render_fence); - pipe_surface_reference(&mc->render_target.cbufs[0], NULL); - - for (i = 0; i < 3; ++i) - mc->zero_block[i].x = -1.0f; - - mc->buffered_surface = NULL; - mc->num_macroblocks = 0; - - return 0; -} - -static int vlRenderMacroBlocksMpeg2R16SnormBuffered -( - struct vlRender *render, - struct vlMpeg2MacroBlockBatch *batch, - struct vlSurface *surface -) -{ - struct vlR16SnormBufferedMC *mc; - bool new_surface = false; - unsigned int i; - - assert(render); - - mc = (struct vlR16SnormBufferedMC*)render; - - if (mc->buffered_surface) - { - if (mc->buffered_surface != surface) - { - vlFlush(&mc->base); - new_surface = true; - } - } - else - new_surface = true; - - if (new_surface) - { - mc->buffered_surface = surface; - mc->past_surface = batch->past_surface; - mc->future_surface = batch->future_surface; - mc->surface_tex_inv_size.x = 1.0f / surface->texture->width[0]; - mc->surface_tex_inv_size.y = 1.0f / surface->texture->height[0]; - - for (i = 0; i < 3; ++i) - { - mc->tex_transfer[i] = mc->pipe->screen->get_tex_transfer - ( - mc->pipe->screen, - mc->textures.all[i], - 0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0, - surface->texture->width[0], - surface->texture->height[0] - ); - - mc->texels[i] = mc->pipe->screen->transfer_map(mc->pipe->screen, mc->tex_transfer[i]); - } - } - - for (i = 0; i < batch->num_macroblocks; ++i) - vlGrabMacroBlock(mc, &batch->macroblocks[i]); - - return 0; -} - -static inline int vlEnd -( - struct vlRender *render -) -{ - assert(render); - - return 0; -} - -static int vlDestroy -( - struct vlRender *render -) -{ - struct vlR16SnormBufferedMC *mc; - struct pipe_context *pipe; - unsigned int i; - - assert(render); - - mc = (struct vlR16SnormBufferedMC*)render; - pipe = mc->pipe; - - for (i = 0; i < 5; ++i) - pipe->delete_sampler_state(pipe, mc->samplers.all[i]); - - for (i = 0; i < 3; ++i) - pipe_buffer_reference(&mc->vertex_bufs.all[i].buffer, NULL); - - /* Textures 3 & 4 are not created directly, no need to release them here */ - for (i = 0; i < 3; ++i) - pipe_texture_reference(&mc->textures.all[i], NULL); - - pipe->delete_vs_state(pipe, mc->i_vs); - pipe->delete_fs_state(pipe, mc->i_fs); - - for (i = 0; i < 2; ++i) - { - pipe->delete_vs_state(pipe, mc->p_vs[i]); - pipe->delete_fs_state(pipe, mc->p_fs[i]); - pipe->delete_vs_state(pipe, mc->b_vs[i]); - pipe->delete_fs_state(pipe, mc->b_fs[i]); - } - - pipe_buffer_reference(&mc->vs_const_buf.buffer, NULL); - pipe_buffer_reference(&mc->fs_const_buf.buffer, NULL); - - FREE(mc->macroblocks); - FREE(mc); - - return 0; -} - -/* - * Muliplier renormalizes block samples from 16 bits to 12 bits. - * Divider is used when calculating Y % 2 for choosing top or bottom - * field for P or B macroblocks. - * TODO: Use immediates. - */ -static const struct vlFragmentShaderConsts fs_consts = -{ - {32767.0f / 255.0f, 32767.0f / 255.0f, 32767.0f / 255.0f, 0.0f}, - {0.5f, 2.0f, 0.0f, 0.0f} -}; - -#include "vl_r16snorm_mc_buf_shaders.inc" - -static int vlCreateDataBufs -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int mbw = align(mc->picture_width, VL_MACROBLOCK_WIDTH) / VL_MACROBLOCK_WIDTH; - const unsigned int mbh = align(mc->picture_height, VL_MACROBLOCK_HEIGHT) / VL_MACROBLOCK_HEIGHT; - - struct pipe_context *pipe; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - mc->macroblocks_per_picture = mbw * mbh; - - /* Create our vertex buffers */ - mc->vertex_bufs.ycbcr.stride = sizeof(struct vlVertex2f) * 4; - mc->vertex_bufs.ycbcr.max_index = 24 * mc->macroblocks_per_picture - 1; - mc->vertex_bufs.ycbcr.buffer_offset = 0; - mc->vertex_bufs.ycbcr.buffer = pipe_buffer_create - ( - pipe->screen, - DEFAULT_BUF_ALIGNMENT, - PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_DISCARD, - sizeof(struct vlVertex2f) * 4 * 24 * mc->macroblocks_per_picture - ); - - for (i = 1; i < 3; ++i) - { - mc->vertex_bufs.all[i].stride = sizeof(struct vlVertex2f) * 2; - mc->vertex_bufs.all[i].max_index = 24 * mc->macroblocks_per_picture - 1; - mc->vertex_bufs.all[i].buffer_offset = 0; - mc->vertex_bufs.all[i].buffer = pipe_buffer_create - ( - pipe->screen, - DEFAULT_BUF_ALIGNMENT, - PIPE_BUFFER_USAGE_VERTEX | PIPE_BUFFER_USAGE_DISCARD, - sizeof(struct vlVertex2f) * 2 * 24 * mc->macroblocks_per_picture - ); - } - - /* Position element */ - mc->vertex_elems[0].src_offset = 0; - mc->vertex_elems[0].vertex_buffer_index = 0; - mc->vertex_elems[0].nr_components = 2; - mc->vertex_elems[0].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* Luma, texcoord element */ - mc->vertex_elems[1].src_offset = sizeof(struct vlVertex2f); - mc->vertex_elems[1].vertex_buffer_index = 0; - mc->vertex_elems[1].nr_components = 2; - mc->vertex_elems[1].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* Chroma Cr texcoord element */ - mc->vertex_elems[2].src_offset = sizeof(struct vlVertex2f) * 2; - mc->vertex_elems[2].vertex_buffer_index = 0; - mc->vertex_elems[2].nr_components = 2; - mc->vertex_elems[2].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* Chroma Cb texcoord element */ - mc->vertex_elems[3].src_offset = sizeof(struct vlVertex2f) * 3; - mc->vertex_elems[3].vertex_buffer_index = 0; - mc->vertex_elems[3].nr_components = 2; - mc->vertex_elems[3].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* First ref surface top field texcoord element */ - mc->vertex_elems[4].src_offset = 0; - mc->vertex_elems[4].vertex_buffer_index = 1; - mc->vertex_elems[4].nr_components = 2; - mc->vertex_elems[4].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* First ref surface bottom field texcoord element */ - mc->vertex_elems[5].src_offset = sizeof(struct vlVertex2f); - mc->vertex_elems[5].vertex_buffer_index = 1; - mc->vertex_elems[5].nr_components = 2; - mc->vertex_elems[5].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* Second ref surface top field texcoord element */ - mc->vertex_elems[6].src_offset = 0; - mc->vertex_elems[6].vertex_buffer_index = 2; - mc->vertex_elems[6].nr_components = 2; - mc->vertex_elems[6].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* Second ref surface bottom field texcoord element */ - mc->vertex_elems[7].src_offset = sizeof(struct vlVertex2f); - mc->vertex_elems[7].vertex_buffer_index = 2; - mc->vertex_elems[7].nr_components = 2; - mc->vertex_elems[7].src_format = PIPE_FORMAT_R32G32_FLOAT; - - /* Create our constant buffer */ - mc->vs_const_buf.buffer = pipe_buffer_create - ( - pipe->screen, - DEFAULT_BUF_ALIGNMENT, - PIPE_BUFFER_USAGE_CONSTANT | PIPE_BUFFER_USAGE_DISCARD, - sizeof(struct vlVertexShaderConsts) - ); - - mc->fs_const_buf.buffer = pipe_buffer_create - ( - pipe->screen, - DEFAULT_BUF_ALIGNMENT, - PIPE_BUFFER_USAGE_CONSTANT, - sizeof(struct vlFragmentShaderConsts) - ); - - memcpy - ( - pipe_buffer_map(pipe->screen, mc->fs_const_buf.buffer, PIPE_BUFFER_USAGE_CPU_WRITE), - &fs_consts, - sizeof(struct vlFragmentShaderConsts) - ); - - pipe_buffer_unmap(pipe->screen, mc->fs_const_buf.buffer); - - mc->macroblocks = MALLOC(sizeof(struct vlMpeg2MacroBlock) * mc->macroblocks_per_picture); - - return 0; -} - -static int vlInit -( - struct vlR16SnormBufferedMC *mc -) -{ - struct pipe_context *pipe; - struct pipe_sampler_state sampler; - struct pipe_texture template; - unsigned int filters[5]; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - - mc->buffered_surface = NULL; - mc->past_surface = NULL; - mc->future_surface = NULL; - for (i = 0; i < 3; ++i) - mc->zero_block[i].x = -1.0f; - mc->num_macroblocks = 0; - - /* For MC we render to textures, which are rounded up to nearest POT */ - mc->viewport.scale[0] = vlRoundUpPOT(mc->picture_width); - mc->viewport.scale[1] = vlRoundUpPOT(mc->picture_height); - mc->viewport.scale[2] = 1; - mc->viewport.scale[3] = 1; - mc->viewport.translate[0] = 0; - mc->viewport.translate[1] = 0; - mc->viewport.translate[2] = 0; - mc->viewport.translate[3] = 0; - - mc->render_target.width = vlRoundUpPOT(mc->picture_width); - mc->render_target.height = vlRoundUpPOT(mc->picture_height); - mc->render_target.nr_cbufs = 1; - /* FB for MC stage is a vlSurface created by the user, set at render time */ - mc->render_target.zsbuf = NULL; - - filters[0] = PIPE_TEX_FILTER_NEAREST; - /* FIXME: Linear causes discoloration around block edges */ - filters[1] = /*mc->picture_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/; - filters[2] = /*mc->picture_format == vlFormatYCbCr444 ?*/ PIPE_TEX_FILTER_NEAREST /*: PIPE_TEX_FILTER_LINEAR*/; - filters[3] = PIPE_TEX_FILTER_LINEAR; - filters[4] = PIPE_TEX_FILTER_LINEAR; - - for (i = 0; i < 5; ++i) - { - sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.wrap_r = PIPE_TEX_WRAP_CLAMP_TO_EDGE; - sampler.min_img_filter = filters[i]; - sampler.min_mip_filter = PIPE_TEX_MIPFILTER_NONE; - sampler.mag_img_filter = filters[i]; - sampler.compare_mode = PIPE_TEX_COMPARE_NONE; - sampler.compare_func = PIPE_FUNC_ALWAYS; - sampler.normalized_coords = 1; - /*sampler.prefilter = ;*/ - /*sampler.lod_bias = ;*/ - sampler.min_lod = 0; - /*sampler.max_lod = ;*/ - /*sampler.border_color[i] = ;*/ - /*sampler.max_anisotropy = ;*/ - mc->samplers.all[i] = pipe->create_sampler_state(pipe, &sampler); - } - - memset(&template, 0, sizeof(struct pipe_texture)); - template.target = PIPE_TEXTURE_2D; - template.format = PIPE_FORMAT_R16_SNORM; - template.last_level = 0; - template.width[0] = vlRoundUpPOT(mc->picture_width); - template.height[0] = vlRoundUpPOT(mc->picture_height); - template.depth[0] = 1; - pf_get_block(template.format, &template.block); - template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_DYNAMIC; - - mc->textures.y = pipe->screen->texture_create(pipe->screen, &template); - - if (mc->picture_format == vlFormatYCbCr420) - { - template.width[0] = vlRoundUpPOT(mc->picture_width / 2); - template.height[0] = vlRoundUpPOT(mc->picture_height / 2); - } - else if (mc->picture_format == vlFormatYCbCr422) - template.height[0] = vlRoundUpPOT(mc->picture_height / 2); - - mc->textures.cb = pipe->screen->texture_create(pipe->screen, &template); - mc->textures.cr = pipe->screen->texture_create(pipe->screen, &template); - - /* textures.all[3] & textures.all[4] are assigned from vlSurfaces for P and B macroblocks at render time */ - - vlCreateVertexShaderIMB(mc); - vlCreateFragmentShaderIMB(mc); - vlCreateVertexShaderFramePMB(mc); - vlCreateVertexShaderFieldPMB(mc); - vlCreateFragmentShaderFramePMB(mc); - vlCreateFragmentShaderFieldPMB(mc); - vlCreateVertexShaderFrameBMB(mc); - vlCreateVertexShaderFieldBMB(mc); - vlCreateFragmentShaderFrameBMB(mc); - vlCreateFragmentShaderFieldBMB(mc); - vlCreateDataBufs(mc); - - return 0; -} - -int vlCreateR16SNormBufferedMC -( - struct pipe_context *pipe, - unsigned int picture_width, - unsigned int picture_height, - enum vlFormat picture_format, - struct vlRender **render -) -{ - struct vlR16SnormBufferedMC *mc; - - assert(pipe); - assert(render); - - mc = CALLOC_STRUCT(vlR16SnormBufferedMC); - - mc->base.vlBegin = &vlBegin; - mc->base.vlRenderMacroBlocksMpeg2 = &vlRenderMacroBlocksMpeg2R16SnormBuffered; - mc->base.vlEnd = &vlEnd; - mc->base.vlFlush = &vlFlush; - mc->base.vlDestroy = &vlDestroy; - mc->pipe = pipe; - mc->picture_width = picture_width; - mc->picture_height = picture_height; - - vlInit(mc); - - *render = &mc->base; - - return 0; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h deleted file mode 100644 index 27177d64cad..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf.h +++ /dev/null @@ -1,18 +0,0 @@ -#ifndef vl_r16snorm_mc_buf_h -#define vl_r16snorm_mc_buf_h - -#include "vl_types.h" - -struct pipe_context; -struct vlRender; - -int vlCreateR16SNormBufferedMC -( - struct pipe_context *pipe, - unsigned int picture_width, - unsigned int picture_height, - enum vlFormat picture_format, - struct vlRender **render -); - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc b/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc deleted file mode 100644 index 34d93e1df0a..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_r16snorm_mc_buf_shaders.inc +++ /dev/null @@ -1,1185 +0,0 @@ -static int vlCreateVertexShaderIMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 50; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma Cb texcoords - * decl i3 ; Chroma Cr texcoords - */ - for (i = 0; i < 4; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma Cb texcoords - * decl o3 ; Chroma Cr texcoords - */ - for (i = 0; i < 4; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma Cb texcoords to output - * mov o3, i3 ; Move input chroma Cr texcoords to output - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - mc->i_vs = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderIMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Luma texcoords - * decl i1 ; Chroma Cb texcoords - * decl i2 ; Chroma Cr texcoords - */ - for (i = 0; i < 3; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0, t1 */ - decl = vl_decl_temps(0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - */ - for (i = 0; i < 3; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header,max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i2, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul o0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - mc->i_fs = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateVertexShaderFramePMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma Cb texcoords - * decl i3 ; Chroma Cr texcoords - * decl i4 ; Ref surface top field texcoords - * decl i5 ; Ref surface bottom field texcoords (unused, packed in the same stream) - */ - for (i = 0; i < 6; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma Cb texcoords - * decl o3 ; Chroma Cr texcoords - * decl o4 ; Ref macroblock texcoords - */ - for (i = 0; i < 5; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma Cb texcoords to output - * mov o3, i3 ; Move input chroma Cr texcoords to output - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* add o4, i0, i4 ; Translate vertex pos by motion vec to form ref macroblock texcoords */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, 4); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - mc->p_vs[0] = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateVertexShaderFieldPMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma Cb texcoords - * decl i3 ; Chroma Cr texcoords - * decl i4 ; Ref macroblock top field texcoords - * decl i5 ; Ref macroblock bottom field texcoords - */ - for (i = 0; i < 6; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl c0 ; Render target dimensions */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma Cb texcoords - * decl o3 ; Chroma Cr texcoords - * decl o4 ; Ref macroblock top field texcoords - * decl o5 ; Ref macroblock bottom field texcoords - * decl o6 ; Denormalized vertex pos - */ - for (i = 0; i < 7; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma Cb texcoords to output - * mov o3, i3 ; Move input chroma Cr texcoords to output - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* - * add o4, i0, i4 ; Translate vertex pos by motion vec to form top field macroblock texcoords - * add o5, i0, i5 ; Translate vertex pos by motion vec to form bottom field macroblock texcoords - */ - for (i = 0; i < 2; ++i) - { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 4); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul o6, i0, c0 ; Denorm vertex pos */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 6, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - mc->p_vs[1] = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFramePMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Luma texcoords - * decl i1 ; Chroma Cb texcoords - * decl i2 ; Chroma Cr texcoords - * decl i3 ; Ref macroblock texcoords - */ - for (i = 0; i < 4; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0, t1 */ - decl = vl_decl_temps(0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for ref surface texture - */ - for (i = 0; i < 4; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i2, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul t0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* tex2d t1, i3, s3 ; Read texel from ref macroblock */ - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, 3, TGSI_FILE_SAMPLER, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, t1 ; Add ref and differential to form final output */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - mc->p_fs[0] = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFieldPMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 200; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Luma texcoords - * decl i1 ; Chroma Cb texcoords - * decl i2 ; Chroma Cr texcoords - * decl i3 ; Ref macroblock top field texcoords - * decl i4 ; Ref macroblock bottom field texcoords - * decl i5 ; Denormalized vertex pos - */ - for (i = 0; i < 6; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm - * decl c1 ; Constants 1/2 & 2 in .x, .y channels for Y-mod-2 top/bottom field selection - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0-t4 */ - decl = vl_decl_temps(0, 4); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for ref surface texture - */ - for (i = 0; i < 4; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i2, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul t0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * tex2d t1, i3, s3 ; Read texel from ref macroblock top field - * tex2d t2, i4, s3 ; Read texel from ref macroblock bottom field - */ - for (i = 0; i < 2; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* XXX: Pos values off by 0.5? */ - /* sub t4, i5.y, c1.x ; Sub 0.5 from denormalized pos */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 5, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* floor t3, t3 ; Get rid of fractional part */ - inst = vl_inst2(TGSI_OPCODE_FLR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t3, t3, c1.y ; Multiply by 2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* TODO: Move to conditional tex fetch on t3 instead of lerp */ - /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, t1 ; Add ref and differential to form final output */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - mc->p_fs[1] = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateVertexShaderFrameBMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma Cb texcoords - * decl i3 ; Chroma Cr texcoords - * decl i4 ; First ref macroblock top field texcoords - * decl i5 ; First ref macroblock bottom field texcoords (unused, packed in the same stream) - * decl i6 ; Second ref macroblock top field texcoords - * decl i7 ; Second ref macroblock bottom field texcoords (unused, packed in the same stream) - */ - for (i = 0; i < 8; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma Cb texcoords - * decl o3 ; Chroma Cr texcoords - * decl o4 ; First ref macroblock texcoords - * decl o5 ; Second ref macroblock texcoords - */ - for (i = 0; i < 6; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma Cb texcoords to output - * mov o3, i3 ; Move input chroma Cr texcoords to output - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* - * add o4, i0, i4 ; Translate vertex pos by motion vec to form first ref macroblock texcoords - * add o5, i0, i6 ; Translate vertex pos by motion vec to form second ref macroblock texcoords - */ - for (i = 0; i < 2; ++i) - { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, (i + 2) * 2); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - mc->b_vs[0] = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateVertexShaderFieldBMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state vs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_VERTEX, header); - - ti = 3; - - /* - * decl i0 ; Vertex pos - * decl i1 ; Luma texcoords - * decl i2 ; Chroma Cb texcoords - * decl i3 ; Chroma Cr texcoords - * decl i4 ; First ref macroblock top field texcoords - * decl i5 ; First ref macroblock bottom field texcoords - * decl i6 ; Second ref macroblock top field texcoords - * decl i7 ; Second ref macroblock bottom field texcoords - */ - for (i = 0; i < 8; i++) - { - decl = vl_decl_input(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl c0 ; Render target dimensions */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl o0 ; Vertex pos - * decl o1 ; Luma texcoords - * decl o2 ; Chroma Cb texcoords - * decl o3 ; Chroma Cr texcoords - * decl o4 ; First ref macroblock top field texcoords - * decl o5 ; First ref macroblock Bottom field texcoords - * decl o6 ; Second ref macroblock top field texcoords - * decl o7 ; Second ref macroblock Bottom field texcoords - * decl o8 ; Denormalized vertex pos - */ - for (i = 0; i < 9; i++) - { - decl = vl_decl_output(i == 0 ? TGSI_SEMANTIC_POSITION : TGSI_SEMANTIC_GENERIC, i, i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* decl t0, t1 */ - decl = vl_decl_temps(0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * mov o0, i0 ; Move input vertex pos to output - * mov o1, i1 ; Move input luma texcoords to output - * mov o2, i2 ; Move input chroma Cb texcoords to output - * mov o3, i3 ; Move input chroma Cr texcoords to output - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_OUTPUT, i, TGSI_FILE_INPUT, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* - * add o4, i0, i4 ; Translate vertex pos by motion vec to form first top field macroblock texcoords - * add o5, i0, i5 ; Translate vertex pos by motion vec to form first bottom field macroblock texcoords - * add o6, i0, i6 ; Translate vertex pos by motion vec to form second top field macroblock texcoords - * add o7, i0, i7 ; Translate vertex pos by motion vec to form second bottom field macroblock texcoords - */ - for (i = 0; i < 4; ++i) - { - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, i + 4, TGSI_FILE_INPUT, 0, TGSI_FILE_INPUT, i + 4); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul o8, i0, c0 ; Denorm vertex pos */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_OUTPUT, 8, TGSI_FILE_INPUT, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - vs.tokens = tokens; - mc->b_vs[1] = pipe->create_vs_state(pipe, &vs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFrameBMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 100; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Luma texcoords - * decl i1 ; Chroma Cb texcoords - * decl i2 ; Chroma Cr texcoords - * decl i3 ; First ref macroblock texcoords - * decl i4 ; Second ref macroblock texcoords - */ - for (i = 0; i < 5; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm - * decl c1 ; Constant 1/2 in .x channel to use as weight to blend past and future texels - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0-t2 */ - decl = vl_decl_temps(0, 2); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for first ref surface texture - * decl s4 ; Sampler for second ref surface texture - */ - for (i = 0; i < 5; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i2, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul t0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * tex2d t1, i3, s3 ; Read texel from first ref macroblock - * tex2d t2, i4, s4 ; Read texel from second ref macroblock - */ - for (i = 0; i < 2; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, i + 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ - inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - mc->b_fs[0] = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} - -static int vlCreateFragmentShaderFieldBMB -( - struct vlR16SnormBufferedMC *mc -) -{ - const unsigned int max_tokens = 200; - - struct pipe_context *pipe; - struct pipe_shader_state fs; - struct tgsi_token *tokens; - struct tgsi_header *header; - - struct tgsi_full_declaration decl; - struct tgsi_full_instruction inst; - - unsigned int ti; - unsigned int i; - - assert(mc); - - pipe = mc->pipe; - tokens = (struct tgsi_token*)malloc(max_tokens * sizeof(struct tgsi_token)); - - /* Version */ - *(struct tgsi_version*)&tokens[0] = tgsi_build_version(); - /* Header */ - header = (struct tgsi_header*)&tokens[1]; - *header = tgsi_build_header(); - /* Processor */ - *(struct tgsi_processor*)&tokens[2] = tgsi_build_processor(TGSI_PROCESSOR_FRAGMENT, header); - - ti = 3; - - /* - * decl i0 ; Luma texcoords - * decl i1 ; Chroma Cb texcoords - * decl i2 ; Chroma Cr texcoords - * decl i3 ; First ref macroblock top field texcoords - * decl i4 ; First ref macroblock bottom field texcoords - * decl i5 ; Second ref macroblock top field texcoords - * decl i6 ; Second ref macroblock bottom field texcoords - * decl i7 ; Denormalized vertex pos - */ - for (i = 0; i < 8; ++i) - { - decl = vl_decl_interpolated_input(TGSI_SEMANTIC_GENERIC, i + 1, i, i, TGSI_INTERPOLATE_LINEAR); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * decl c0 ; Scaling factor, rescales 16-bit snorm to 9-bit snorm - * decl c1 ; Constants 1/2 & 2 in .x, .y channels to use as weight to blend past and future texels - * ; and for Y-mod-2 top/bottom field selection - */ - decl = vl_decl_constants(TGSI_SEMANTIC_GENERIC, 0, 0, 1); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl o0 ; Fragment color */ - decl = vl_decl_output(TGSI_SEMANTIC_COLOR, 0, 0, 0); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* decl t0-t5 */ - decl = vl_decl_temps(0, 5); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - - /* - * decl s0 ; Sampler for luma texture - * decl s1 ; Sampler for chroma Cb texture - * decl s2 ; Sampler for chroma Cr texture - * decl s3 ; Sampler for first ref surface texture - * decl s4 ; Sampler for second ref surface texture - */ - for (i = 0; i < 5; ++i) - { - decl = vl_decl_samplers(i, i); - ti += tgsi_build_full_declaration(&decl, &tokens[ti], header, max_tokens - ti); - } - - /* - * tex2d t1, i0, s0 ; Read texel from luma texture - * mov t0.x, t1.x ; Move luma sample into .x component - * tex2d t1, i1, s1 ; Read texel from chroma Cb texture - * mov t0.y, t1.x ; Move Cb sample into .y component - * tex2d t1, i2, s2 ; Read texel from chroma Cr texture - * mov t0.z, t1.x ; Move Cr sample into .z component - */ - for (i = 0; i < 3; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_INPUT, i, TGSI_FILE_SAMPLER, i); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - inst = vl_inst2(TGSI_OPCODE_MOV, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullDstRegisters[0].DstRegister.WriteMask = TGSI_WRITEMASK_X << i; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* mul t0, t0, c0 ; Rescale texel to correct range */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_CONSTANT, 0); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* XXX: Pos values off by 0.5? */ - /* sub t4, i7.y, c1.x ; Sub 0.5 from denormalized pos */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_INPUT, 7, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t3, t4, c1.x ; Multiply pos Y-coord by 1/2 */ - inst = vl_inst3(TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* floor t3, t3 ; Get rid of fractional part */ - inst = vl_inst2(TGSI_OPCODE_FLR, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* mul t3, t3, c1.y ; Multiply by 2 */ - inst = vl_inst3( TGSI_OPCODE_MUL, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_CONSTANT, 1); - inst.FullSrcRegisters[1].SrcRegister.SwizzleX = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleY = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleZ = TGSI_SWIZZLE_Y; - inst.FullSrcRegisters[1].SrcRegister.SwizzleW = TGSI_SWIZZLE_Y; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* sub t3, t4, t3 ; Subtract from original Y to get Y % 2 */ - inst = vl_inst3(TGSI_OPCODE_SUB, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * tex2d t1, i3, s3 ; Read texel from past ref macroblock top field - * tex2d t2, i4, s3 ; Read texel from past ref macroblock bottom field - */ - for (i = 0; i < 2; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 1, TGSI_FILE_INPUT, i + 3, TGSI_FILE_SAMPLER, 3); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* TODO: Move to conditional tex fetch on t3 instead of lerp */ - /* lerp t1, t3, t1, t2 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* - * tex2d t4, i5, s4 ; Read texel from future ref macroblock top field - * tex2d t5, i6, s4 ; Read texel from future ref macroblock bottom field - */ - for (i = 0; i < 2; ++i) - { - inst = vl_tex(TGSI_TEXTURE_2D, TGSI_FILE_TEMPORARY, i + 4, TGSI_FILE_INPUT, i + 5, TGSI_FILE_SAMPLER, 4); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - } - - /* TODO: Move to conditional tex fetch on t3 instead of lerp */ - /* lerp t2, t3, t4, t5 ; Choose between top and bottom fields based on Y % 2 */ - inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 2, TGSI_FILE_TEMPORARY, 3, TGSI_FILE_TEMPORARY, 4, TGSI_FILE_TEMPORARY, 5); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* lerp t1, c1.x, t1, t2 ; Blend past and future texels */ - inst = vl_inst4(TGSI_OPCODE_LRP, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_CONSTANT, 1, TGSI_FILE_TEMPORARY, 1, TGSI_FILE_TEMPORARY, 2); - inst.FullSrcRegisters[0].SrcRegister.SwizzleX = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleY = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleZ = TGSI_SWIZZLE_X; - inst.FullSrcRegisters[0].SrcRegister.SwizzleW = TGSI_SWIZZLE_X; - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* add o0, t0, t1 ; Add past/future ref and differential to form final output */ - inst = vl_inst3(TGSI_OPCODE_ADD, TGSI_FILE_OUTPUT, 0, TGSI_FILE_TEMPORARY, 0, TGSI_FILE_TEMPORARY, 1); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - /* end */ - inst = vl_end(); - ti += tgsi_build_full_instruction(&inst, &tokens[ti], header, max_tokens - ti); - - fs.tokens = tokens; - mc->b_fs[1] = pipe->create_fs_state(pipe, &fs); - free(tokens); - - return 0; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_render.h b/src/gallium/state_trackers/g3dvl/vl_render.h deleted file mode 100644 index 166030b4988..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_render.h +++ /dev/null @@ -1,38 +0,0 @@ -#ifndef vl_render_h -#define vl_render_h - -#include "vl_types.h" - -struct pipe_surface; - -struct vlRender -{ - int (*vlBegin) - ( - struct vlRender *render - ); - - int (*vlRenderMacroBlocksMpeg2) - ( - struct vlRender *render, - struct vlMpeg2MacroBlockBatch *batch, - struct vlSurface *surface - ); - - int (*vlEnd) - ( - struct vlRender *render - ); - - int (*vlFlush) - ( - struct vlRender *render - ); - - int (*vlDestroy) - ( - struct vlRender *render - ); -}; - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_screen.c b/src/gallium/state_trackers/g3dvl/vl_screen.c deleted file mode 100644 index ade8643a66a..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_screen.c +++ /dev/null @@ -1,115 +0,0 @@ -#define VL_INTERNAL -#include "vl_screen.h" -#include <assert.h> -#include <util/u_memory.h> - -int vlCreateScreen -( - struct vlDisplay *display, - int screen, - struct pipe_screen *pscreen, - struct vlScreen **vl_screen -) -{ - struct vlScreen *scrn; - - assert(display); - assert(pscreen); - assert(vl_screen); - - scrn = CALLOC_STRUCT(vlScreen); - - if (!scrn) - return 1; - - scrn->display = display; - scrn->ordinal = screen; - scrn->pscreen = pscreen; - *vl_screen = scrn; - - return 0; -} - -int vlDestroyScreen -( - struct vlScreen *screen -) -{ - assert(screen); - - FREE(screen); - - return 0; -} - -struct vlDisplay* vlGetDisplay -( - struct vlScreen *screen -) -{ - assert(screen); - - return screen->display; -} - -struct pipe_screen* vlGetPipeScreen -( - struct vlScreen *screen -) -{ - assert(screen); - - return screen->pscreen; -} - -unsigned int vlGetMaxProfiles -( - struct vlScreen *screen -) -{ - assert(screen); - - return vlProfileCount; -} - -int vlQueryProfiles -( - struct vlScreen *screen, - enum vlProfile *profiles -) -{ - assert(screen); - assert(profiles); - - profiles[0] = vlProfileMpeg2Simple; - profiles[1] = vlProfileMpeg2Main; - - return 0; -} - -unsigned int vlGetMaxEntryPoints -( - struct vlScreen *screen -) -{ - assert(screen); - - return vlEntryPointCount; -} - -int vlQueryEntryPoints -( - struct vlScreen *screen, - enum vlProfile profile, - enum vlEntryPoint *entry_points -) -{ - assert(screen); - assert(entry_points); - - entry_points[0] = vlEntryPointIDCT; - entry_points[1] = vlEntryPointMC; - entry_points[2] = vlEntryPointCSC; - - return 0; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_screen.h b/src/gallium/state_trackers/g3dvl/vl_screen.h deleted file mode 100644 index 98f3d429b61..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_screen.h +++ /dev/null @@ -1,63 +0,0 @@ -#ifndef vl_screen_h -#define vl_screen_h - -#include "vl_types.h" - -struct pipe_screen; - -#ifdef VL_INTERNAL -struct vlScreen -{ - struct vlDisplay *display; - unsigned int ordinal; - struct pipe_screen *pscreen; -}; -#endif - -int vlCreateScreen -( - struct vlDisplay *display, - int screen, - struct pipe_screen *pscreen, - struct vlScreen **vl_screen -); - -int vlDestroyScreen -( - struct vlScreen *screen -); - -struct vlDisplay* vlGetDisplay -( - struct vlScreen *screen -); - -struct pipe_screen* vlGetPipeScreen -( - struct vlScreen *screen -); - -unsigned int vlGetMaxProfiles -( - struct vlScreen *screen -); - -int vlQueryProfiles -( - struct vlScreen *screen, - enum vlProfile *profiles -); - -unsigned int vlGetMaxEntryPoints -( - struct vlScreen *screen -); - -int vlQueryEntryPoints -( - struct vlScreen *screen, - enum vlProfile profile, - enum vlEntryPoint *entry_points -); - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.c b/src/gallium/state_trackers/g3dvl/vl_shader_build.c deleted file mode 100644 index 51f1721a332..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_shader_build.c +++ /dev/null @@ -1,215 +0,0 @@ -#include "vl_shader_build.h" -#include <assert.h> -#include <tgsi/tgsi_parse.h> -#include <tgsi/tgsi_build.h> - -struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last) -{ - struct tgsi_full_declaration decl = tgsi_default_full_declaration(); - - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = name; - decl.Semantic.SemanticIndex = index; - decl.DeclarationRange.First = first; - decl.DeclarationRange.Last = last; - - return decl; -} - -struct tgsi_full_declaration vl_decl_interpolated_input -( - unsigned int name, - unsigned int index, - unsigned int first, - unsigned int last, - int interpolation -) -{ - struct tgsi_full_declaration decl = tgsi_default_full_declaration(); - - assert - ( - interpolation == TGSI_INTERPOLATE_CONSTANT || - interpolation == TGSI_INTERPOLATE_LINEAR || - interpolation == TGSI_INTERPOLATE_PERSPECTIVE - ); - - decl.Declaration.File = TGSI_FILE_INPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = name; - decl.Semantic.SemanticIndex = index; - decl.Declaration.Interpolate = interpolation;; - decl.DeclarationRange.First = first; - decl.DeclarationRange.Last = last; - - return decl; -} - -struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last) -{ - struct tgsi_full_declaration decl = tgsi_default_full_declaration(); - - decl.Declaration.File = TGSI_FILE_CONSTANT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = name; - decl.Semantic.SemanticIndex = index; - decl.DeclarationRange.First = first; - decl.DeclarationRange.Last = last; - - return decl; -} - -struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last) -{ - struct tgsi_full_declaration decl = tgsi_default_full_declaration(); - - decl.Declaration.File = TGSI_FILE_OUTPUT; - decl.Declaration.Semantic = 1; - decl.Semantic.SemanticName = name; - decl.Semantic.SemanticIndex = index; - decl.DeclarationRange.First = first; - decl.DeclarationRange.Last = last; - - return decl; -} - -struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last) -{ - struct tgsi_full_declaration decl = tgsi_default_full_declaration(); - - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_TEMPORARY; - decl.DeclarationRange.First = first; - decl.DeclarationRange.Last = last; - - return decl; -} - -struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last) -{ - struct tgsi_full_declaration decl = tgsi_default_full_declaration(); - - decl = tgsi_default_full_declaration(); - decl.Declaration.File = TGSI_FILE_SAMPLER; - decl.DeclarationRange.First = first; - decl.DeclarationRange.Last = last; - - return decl; -} - -struct tgsi_full_instruction vl_inst2 -( - int opcode, - enum tgsi_file_type dst_file, - unsigned int dst_index, - enum tgsi_file_type src_file, - unsigned int src_index -) -{ - struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - - inst.Instruction.Opcode = opcode; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = dst_file; - inst.FullDstRegisters[0].DstRegister.Index = dst_index; - inst.Instruction.NumSrcRegs = 1; - inst.FullSrcRegisters[0].SrcRegister.File = src_file; - inst.FullSrcRegisters[0].SrcRegister.Index = src_index; - - return inst; -} - -struct tgsi_full_instruction vl_inst3 -( - int opcode, - enum tgsi_file_type dst_file, - unsigned int dst_index, - enum tgsi_file_type src1_file, - unsigned int src1_index, - enum tgsi_file_type src2_file, - unsigned int src2_index -) -{ - struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - - inst.Instruction.Opcode = opcode; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = dst_file; - inst.FullDstRegisters[0].DstRegister.Index = dst_index; - inst.Instruction.NumSrcRegs = 2; - inst.FullSrcRegisters[0].SrcRegister.File = src1_file; - inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; - inst.FullSrcRegisters[1].SrcRegister.File = src2_file; - inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; - - return inst; -} - -struct tgsi_full_instruction vl_tex -( - int tex, - enum tgsi_file_type dst_file, - unsigned int dst_index, - enum tgsi_file_type src1_file, - unsigned int src1_index, - enum tgsi_file_type src2_file, - unsigned int src2_index -) -{ - struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - - inst.Instruction.Opcode = TGSI_OPCODE_TEX; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = dst_file; - inst.FullDstRegisters[0].DstRegister.Index = dst_index; - inst.Instruction.NumSrcRegs = 2; - inst.InstructionExtTexture.Texture = tex; - inst.FullSrcRegisters[0].SrcRegister.File = src1_file; - inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; - inst.FullSrcRegisters[1].SrcRegister.File = src2_file; - inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; - - return inst; -} - -struct tgsi_full_instruction vl_inst4 -( - int opcode, - enum tgsi_file_type dst_file, - unsigned int dst_index, - enum tgsi_file_type src1_file, - unsigned int src1_index, - enum tgsi_file_type src2_file, - unsigned int src2_index, - enum tgsi_file_type src3_file, - unsigned int src3_index -) -{ - struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - - inst.Instruction.Opcode = opcode; - inst.Instruction.NumDstRegs = 1; - inst.FullDstRegisters[0].DstRegister.File = dst_file; - inst.FullDstRegisters[0].DstRegister.Index = dst_index; - inst.Instruction.NumSrcRegs = 3; - inst.FullSrcRegisters[0].SrcRegister.File = src1_file; - inst.FullSrcRegisters[0].SrcRegister.Index = src1_index; - inst.FullSrcRegisters[1].SrcRegister.File = src2_file; - inst.FullSrcRegisters[1].SrcRegister.Index = src2_index; - inst.FullSrcRegisters[2].SrcRegister.File = src3_file; - inst.FullSrcRegisters[2].SrcRegister.Index = src3_index; - - return inst; -} - -struct tgsi_full_instruction vl_end(void) -{ - struct tgsi_full_instruction inst = tgsi_default_full_instruction(); - - inst.Instruction.Opcode = TGSI_OPCODE_END; - inst.Instruction.NumDstRegs = 0; - inst.Instruction.NumSrcRegs = 0; - - return inst; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_shader_build.h b/src/gallium/state_trackers/g3dvl/vl_shader_build.h deleted file mode 100644 index dc615cb1566..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_shader_build.h +++ /dev/null @@ -1,61 +0,0 @@ -#ifndef vl_shader_build_h -#define vl_shader_build_h - -#include <pipe/p_shader_tokens.h> - -struct tgsi_full_declaration vl_decl_input(unsigned int name, unsigned int index, unsigned int first, unsigned int last); -struct tgsi_full_declaration vl_decl_interpolated_input -( - unsigned int name, - unsigned int index, - unsigned int first, - unsigned int last, - int interpolation -); -struct tgsi_full_declaration vl_decl_constants(unsigned int name, unsigned int index, unsigned int first, unsigned int last); -struct tgsi_full_declaration vl_decl_output(unsigned int name, unsigned int index, unsigned int first, unsigned int last); -struct tgsi_full_declaration vl_decl_temps(unsigned int first, unsigned int last); -struct tgsi_full_declaration vl_decl_samplers(unsigned int first, unsigned int last); -struct tgsi_full_instruction vl_inst2 -( - int opcode, - enum tgsi_file_type dst_file, - unsigned int dst_index, - enum tgsi_file_type src_file, - unsigned int src_index -); -struct tgsi_full_instruction vl_inst3 -( - int opcode, - enum tgsi_file_type dst_file, - unsigned int dst_index, - enum tgsi_file_type src1_file, - unsigned int src1_index, - enum tgsi_file_type src2_file, - unsigned int src2_index -); -struct tgsi_full_instruction vl_tex -( - int tex, - enum tgsi_file_type dst_file, - unsigned int dst_index, - enum tgsi_file_type src1_file, - unsigned int src1_index, - enum tgsi_file_type src2_file, - unsigned int src2_index -); -struct tgsi_full_instruction vl_inst4 -( - int opcode, - enum tgsi_file_type dst_file, - unsigned int dst_index, - enum tgsi_file_type src1_file, - unsigned int src1_index, - enum tgsi_file_type src2_file, - unsigned int src2_index, - enum tgsi_file_type src3_file, - unsigned int src3_index -); -struct tgsi_full_instruction vl_end(void); - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.c b/src/gallium/state_trackers/g3dvl/vl_surface.c deleted file mode 100644 index 7f60852cae8..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_surface.c +++ /dev/null @@ -1,242 +0,0 @@ -#define VL_INTERNAL -#include "vl_surface.h" -#include <assert.h> -#include <string.h> -#include <pipe/p_screen.h> -#include <pipe/p_state.h> -#include <pipe/p_inlines.h> -#include <util/u_memory.h> -#include <vl_winsys.h> -#include "vl_screen.h" -#include "vl_context.h" -#include "vl_render.h" -#include "vl_csc.h" -#include "vl_util.h" - -int vlCreateSurface -( - struct vlScreen *screen, - unsigned int width, - unsigned int height, - enum vlFormat format, - struct vlSurface **surface -) -{ - struct vlSurface *sfc; - struct pipe_texture template; - - assert(screen); - assert(surface); - - sfc = CALLOC_STRUCT(vlSurface); - - if (!sfc) - return 1; - - sfc->screen = screen; - sfc->width = width; - sfc->height = height; - sfc->format = format; - - memset(&template, 0, sizeof(struct pipe_texture)); - template.target = PIPE_TEXTURE_2D; - template.format = PIPE_FORMAT_A8R8G8B8_UNORM; - template.last_level = 0; - template.width[0] = vlRoundUpPOT(sfc->width); - template.height[0] = vlRoundUpPOT(sfc->height); - template.depth[0] = 1; - pf_get_block(template.format, &template.block); - template.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER | PIPE_TEXTURE_USAGE_RENDER_TARGET; - - sfc->texture = vlGetPipeScreen(screen)->texture_create(vlGetPipeScreen(screen), &template); - - if (!sfc->texture) - { - FREE(sfc); - return 1; - } - - *surface = sfc; - - return 0; -} - -int vlDestroySurface -( - struct vlSurface *surface -) -{ - assert(surface); - - pipe_texture_reference(&surface->texture, NULL); - FREE(surface); - - return 0; -} - -int vlRenderMacroBlocksMpeg2 -( - struct vlMpeg2MacroBlockBatch *batch, - struct vlSurface *surface -) -{ - assert(batch); - assert(surface); - assert(surface->context); - - surface->context->render->vlBegin(surface->context->render); - - surface->context->render->vlRenderMacroBlocksMpeg2 - ( - surface->context->render, - batch, - surface - ); - - surface->context->render->vlEnd(surface->context->render); - - return 0; -} - -int vlPutPicture -( - struct vlSurface *surface, - vlNativeDrawable drawable, - int srcx, - int srcy, - int srcw, - int srch, - int destx, - int desty, - int destw, - int desth, - int drawable_w, - int drawable_h, - enum vlPictureType picture_type -) -{ - struct vlCSC *csc; - struct pipe_context *pipe; - - assert(surface); - assert(surface->context); - - surface->context->render->vlFlush(surface->context->render); - - csc = surface->context->csc; - pipe = surface->context->pipe; - - csc->vlResizeFrameBuffer(csc, drawable_w, drawable_h); - - csc->vlBegin(csc); - - csc->vlPutPicture - ( - csc, - surface, - srcx, - srcy, - srcw, - srch, - destx, - desty, - destw, - desth, - picture_type - ); - - csc->vlEnd(csc); - - pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, &surface->disp_fence); - - bind_pipe_drawable(pipe, drawable); - - pipe->screen->flush_frontbuffer - ( - pipe->screen, - csc->vlGetFrameBuffer(csc), - pipe->priv - ); - - return 0; -} - -int vlSurfaceGetStatus -( - struct vlSurface *surface, - enum vlResourceStatus *status -) -{ - assert(surface); - assert(surface->context); - assert(status); - - if (surface->render_fence && !surface->context->pipe->screen->fence_signalled(surface->context->pipe->screen, surface->render_fence, 0)) - { - *status = vlResourceStatusRendering; - return 0; - } - - if (surface->disp_fence && !surface->context->pipe->screen->fence_signalled(surface->context->pipe->screen, surface->disp_fence, 0)) - { - *status = vlResourceStatusDisplaying; - return 0; - } - - *status = vlResourceStatusFree; - - return 0; -} - -int vlSurfaceFlush -( - struct vlSurface *surface -) -{ - assert(surface); - assert(surface->context); - - surface->context->render->vlFlush(surface->context->render); - - return 0; -} - -int vlSurfaceSync -( - struct vlSurface *surface -) -{ - assert(surface); - assert(surface->context); - assert(surface->render_fence); - - surface->context->pipe->screen->fence_finish(surface->context->pipe->screen, surface->render_fence, 0); - - return 0; -} - -struct vlScreen* vlSurfaceGetScreen -( - struct vlSurface *surface -) -{ - assert(surface); - - return surface->screen; -} - -struct vlContext* vlBindToContext -( - struct vlSurface *surface, - struct vlContext *context -) -{ - struct vlContext *old; - - assert(surface); - - old = surface->context; - surface->context = context; - - return old; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_surface.h b/src/gallium/state_trackers/g3dvl/vl_surface.h deleted file mode 100644 index 133e1515ef3..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_surface.h +++ /dev/null @@ -1,86 +0,0 @@ -#ifndef vl_surface_h -#define vl_surface_h - -#include "vl_types.h" - -#ifdef VL_INTERNAL -struct pipe_texture; - -struct vlSurface -{ - struct vlScreen *screen; - struct vlContext *context; - unsigned int width; - unsigned int height; - enum vlFormat format; - struct pipe_texture *texture; - struct pipe_fence_handle *render_fence; - struct pipe_fence_handle *disp_fence; -}; -#endif - -int vlCreateSurface -( - struct vlScreen *screen, - unsigned int width, - unsigned int height, - enum vlFormat format, - struct vlSurface **surface -); - -int vlDestroySurface -( - struct vlSurface *surface -); - -int vlRenderMacroBlocksMpeg2 -( - struct vlMpeg2MacroBlockBatch *batch, - struct vlSurface *surface -); - -int vlPutPicture -( - struct vlSurface *surface, - vlNativeDrawable drawable, - int srcx, - int srcy, - int srcw, - int srch, - int destx, - int desty, - int destw, - int desth, - int drawable_w, - int drawable_h, - enum vlPictureType picture_type -); - -int vlSurfaceGetStatus -( - struct vlSurface *surface, - enum vlResourceStatus *status -); - -int vlSurfaceFlush -( - struct vlSurface *surface -); - -int vlSurfaceSync -( - struct vlSurface *surface -); - -struct vlScreen* vlSurfaceGetScreen -( - struct vlSurface *surface -); - -struct vlContext* vlBindToContext -( - struct vlSurface *surface, - struct vlContext *context -); - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_types.h b/src/gallium/state_trackers/g3dvl/vl_types.h deleted file mode 100644 index 274e1f74377..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_types.h +++ /dev/null @@ -1,115 +0,0 @@ -#ifndef vl_types_h -#define vl_types_h - -#if 1 /*#ifdef X11*/ -#include <X11/Xlib.h> - -typedef Display* vlNativeDisplay; -typedef Drawable vlNativeDrawable; -#endif - -struct vlDisplay; -struct vlScreen; -struct vlContext; -struct vlSurface; - -enum vlResourceStatus -{ - vlResourceStatusFree, - vlResourceStatusRendering, - vlResourceStatusDisplaying -}; - -enum vlProfile -{ - vlProfileMpeg2Simple, - vlProfileMpeg2Main, - - vlProfileCount -}; - -enum vlEntryPoint -{ - vlEntryPointIDCT, - vlEntryPointMC, - vlEntryPointCSC, - - vlEntryPointCount -}; - -enum vlFormat -{ - vlFormatYCbCr420, - vlFormatYCbCr422, - vlFormatYCbCr444 -}; - -enum vlPictureType -{ - vlPictureTypeTopField, - vlPictureTypeBottomField, - vlPictureTypeFrame -}; - -enum vlMotionType -{ - vlMotionTypeField, - vlMotionTypeFrame, - vlMotionTypeDualPrime, - vlMotionType16x8 -}; - -enum vlFieldOrder -{ - vlFieldOrderFirst, - vlFieldOrderSecond -}; - -enum vlDCTType -{ - vlDCTTypeFrameCoded, - vlDCTTypeFieldCoded -}; - -struct vlVertex2f -{ - float x, y; -}; - -struct vlVertex4f -{ - float x, y, z, w; -}; - -enum vlMacroBlockType -{ - vlMacroBlockTypeIntra, - vlMacroBlockTypeFwdPredicted, - vlMacroBlockTypeBkwdPredicted, - vlMacroBlockTypeBiPredicted, - - vlNumMacroBlockTypes -}; - -struct vlMpeg2MacroBlock -{ - unsigned int mbx, mby; - enum vlMacroBlockType mb_type; - enum vlMotionType mo_type; - enum vlDCTType dct_type; - int PMV[2][2][2]; - unsigned int cbp; - short *blocks; -}; - -struct vlMpeg2MacroBlockBatch -{ - struct vlSurface *past_surface; - struct vlSurface *future_surface; - enum vlPictureType picture_type; - enum vlFieldOrder field_order; - unsigned int num_macroblocks; - struct vlMpeg2MacroBlock *macroblocks; -}; - -#endif diff --git a/src/gallium/state_trackers/g3dvl/vl_util.c b/src/gallium/state_trackers/g3dvl/vl_util.c deleted file mode 100644 index 50aa9af66f2..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_util.c +++ /dev/null @@ -1,16 +0,0 @@ -#include "vl_util.h" -#include <assert.h> - -unsigned int vlRoundUpPOT(unsigned int x) -{ - unsigned int i; - - assert(x > 0); - - --x; - - for (i = 1; i < sizeof(unsigned int) * 8; i <<= 1) - x |= x >> i; - - return x + 1; -} diff --git a/src/gallium/state_trackers/g3dvl/vl_util.h b/src/gallium/state_trackers/g3dvl/vl_util.h deleted file mode 100644 index bc98e79df47..00000000000 --- a/src/gallium/state_trackers/g3dvl/vl_util.h +++ /dev/null @@ -1,6 +0,0 @@ -#ifndef vl_util_h -#define vl_util_h - -unsigned int vlRoundUpPOT(unsigned int x); - -#endif diff --git a/src/gallium/state_trackers/glx/xlib/glx_api.c b/src/gallium/state_trackers/glx/xlib/glx_api.c index d1a98f89917..65012e92538 100644 --- a/src/gallium/state_trackers/glx/xlib/glx_api.c +++ b/src/gallium/state_trackers/glx/xlib/glx_api.c @@ -1538,9 +1538,9 @@ get_config( XMesaVisual xmvis, int attrib, int *value, GLboolean fbconfig ) if (!fbconfig) return GLX_BAD_ATTRIBUTE; if (xmvis->mesa_visual.rgbMode) - *value = GLX_RGBA_BIT; + *value = GLX_RGBA_TYPE; else - *value = GLX_COLOR_INDEX_BIT; + *value = GLX_COLOR_INDEX_TYPE; break; case GLX_X_RENDERABLE_SGIX: if (!fbconfig) @@ -2120,9 +2120,9 @@ glXQueryContext( Display *dpy, GLXContext ctx, int attribute, int *value ) break; case GLX_RENDER_TYPE: if (xmctx->xm_visual->mesa_visual.rgbMode) - *value = GLX_RGBA_BIT; + *value = GLX_RGBA_TYPE; else - *value = GLX_COLOR_INDEX_BIT; + *value = GLX_COLOR_INDEX_TYPE; break; case GLX_SCREEN: *value = 0; diff --git a/src/gallium/state_trackers/python/retrace/interpreter.py b/src/gallium/state_trackers/python/retrace/interpreter.py index 6f0bd6ae52a..f4ed2fde4dc 100755 --- a/src/gallium/state_trackers/python/retrace/interpreter.py +++ b/src/gallium/state_trackers/python/retrace/interpreter.py @@ -314,7 +314,7 @@ class Screen(Object): if texture is None: return None transfer = Transfer(texture.get_surface(face, level, zslice), x, y, w, h) - if transfer and usage != gallium.PIPE_TRANSFER_WRITE: + if transfer and usage & gallium.PIPE_TRANSFER_READ if self.interpreter.options.all: self.interpreter.present(transfer.surface, 'transf_read', x, y, w, h) return transfer diff --git a/src/gallium/state_trackers/python/st_softpipe_winsys.c b/src/gallium/state_trackers/python/st_softpipe_winsys.c index f0a4826a001..f0abd12e3dc 100644 --- a/src/gallium/state_trackers/python/st_softpipe_winsys.c +++ b/src/gallium/state_trackers/python/st_softpipe_winsys.c @@ -172,6 +172,7 @@ st_softpipe_surface_buffer_create(struct pipe_winsys *winsys, unsigned width, unsigned height, enum pipe_format format, unsigned usage, + unsigned tex_usage, unsigned *stride) { const unsigned alignment = 64; diff --git a/src/gallium/state_trackers/vega/st_inlines.h b/src/gallium/state_trackers/vega/st_inlines.h index 1f331dfcdb7..610755e0636 100644 --- a/src/gallium/state_trackers/vega/st_inlines.h +++ b/src/gallium/state_trackers/vega/st_inlines.h @@ -57,8 +57,7 @@ st_cond_flush_get_tex_transfer(struct vg_context *st, pipe->is_texture_referenced(pipe, pt, face, level); if (referenced && ((referenced & PIPE_REFERENCED_FOR_WRITE) || - usage == PIPE_TRANSFER_WRITE || - usage == PIPE_TRANSFER_READ_WRITE)) + (usage & PIPE_TRANSFER_WRITE))) vgFlush(); return screen->get_tex_transfer(screen, pt, face, level, zslice, usage, diff --git a/src/gallium/state_trackers/wgl/SConscript b/src/gallium/state_trackers/wgl/SConscript index 69b88618ecb..b05944a33b3 100644 --- a/src/gallium/state_trackers/wgl/SConscript +++ b/src/gallium/state_trackers/wgl/SConscript @@ -18,20 +18,17 @@ if env['platform'] in ['windows']: ]) sources = [ - 'icd/stw_icd.c', - - 'wgl/stw_wgl.c', - - 'shared/stw_context.c', - 'shared/stw_device.c', - 'shared/stw_framebuffer.c', - 'shared/stw_pixelformat.c', - 'shared/stw_extensionsstring.c', - 'shared/stw_extswapinterval.c', - 'shared/stw_getprocaddress.c', - 'shared/stw_extgallium.c', - 'shared/stw_arbpixelformat.c', - 'shared/stw_tls.c', + 'stw_context.c', + 'stw_device.c', + 'stw_ext_extensionsstring.c', + 'stw_ext_gallium.c', + 'stw_ext_pixelformat.c', + 'stw_ext_swapinterval.c', + 'stw_framebuffer.c', + 'stw_getprocaddress.c', + 'stw_pixelformat.c', + 'stw_tls.c', + 'stw_wgl.c', ] wgl = env.ConvenienceLibrary( diff --git a/src/gallium/state_trackers/wgl/opengl32.def b/src/gallium/state_trackers/wgl/opengl32.def index 596417ed844..5daa6ddd413 100644 --- a/src/gallium/state_trackers/wgl/opengl32.def +++ b/src/gallium/state_trackers/wgl/opengl32.def @@ -376,6 +376,7 @@ EXPORTS DrvDescribePixelFormat DrvGetLayerPaletteEntries DrvGetProcAddress + DrvPresentBuffers DrvRealizeLayerPalette DrvReleaseContext DrvSetCallbackProcs diff --git a/src/gallium/state_trackers/wgl/opengl32.mingw.def b/src/gallium/state_trackers/wgl/opengl32.mingw.def index 1f03ea3b375..6ebb31a6f1b 100644 --- a/src/gallium/state_trackers/wgl/opengl32.mingw.def +++ b/src/gallium/state_trackers/wgl/opengl32.mingw.def @@ -375,6 +375,7 @@ EXPORTS DrvDescribePixelFormat = DrvDescribePixelFormat@16 DrvGetLayerPaletteEntries = DrvGetLayerPaletteEntries@20 DrvGetProcAddress = DrvGetProcAddress@4 + DrvPresentBuffers = DrvPresentBuffers@8 DrvRealizeLayerPalette = DrvRealizeLayerPalette@12 DrvReleaseContext = DrvReleaseContext@4 DrvSetCallbackProcs = DrvSetCallbackProcs@8 diff --git a/src/gallium/state_trackers/wgl/shared/stw_context.c b/src/gallium/state_trackers/wgl/shared/stw_context.c deleted file mode 100644 index 4968ecc692d..00000000000 --- a/src/gallium/state_trackers/wgl/shared/stw_context.c +++ /dev/null @@ -1,382 +0,0 @@ -/************************************************************************** - * - * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sub license, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial portions - * of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS - * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR - * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, - * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE - * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - **************************************************************************/ - -#include <windows.h> - -#include "main/mtypes.h" -#include "main/context.h" -#include "pipe/p_compiler.h" -#include "pipe/p_context.h" -#include "state_tracker/st_context.h" -#include "state_tracker/st_public.h" - -#ifdef DEBUG -#include "trace/tr_screen.h" -#include "trace/tr_context.h" -#endif - -#include "shared/stw_device.h" -#include "shared/stw_winsys.h" -#include "shared/stw_framebuffer.h" -#include "shared/stw_pixelformat.h" -#include "stw_public.h" -#include "stw_context.h" -#include "stw_tls.h" - - -static INLINE struct stw_context * -stw_context(GLcontext *glctx) -{ - if(!glctx) - return NULL; - assert(glctx->DriverCtx); - return (struct stw_context *)glctx->DriverCtx; -} - -static INLINE struct stw_context * -stw_current_context(void) -{ - /* We must check if multiple threads are being used or GET_CURRENT_CONTEXT - * might return the current context of the thread first seen. */ - _glapi_check_multithread(); - - { - GET_CURRENT_CONTEXT( glctx ); - return stw_context(glctx); - } -} - -BOOL -stw_copy_context( - UINT_PTR hglrcSrc, - UINT_PTR hglrcDst, - UINT mask ) -{ - struct stw_context *src; - struct stw_context *dst; - BOOL ret = FALSE; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - - src = stw_lookup_context_locked( hglrcSrc ); - dst = stw_lookup_context_locked( hglrcDst ); - - if (src && dst) { - /* FIXME */ - assert(0); - (void) src; - (void) dst; - (void) mask; - } - - pipe_mutex_unlock( stw_dev->ctx_mutex ); - - return ret; -} - -BOOL -stw_share_lists( - UINT_PTR hglrc1, - UINT_PTR hglrc2 ) -{ - struct stw_context *ctx1; - struct stw_context *ctx2; - BOOL ret = FALSE; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - - ctx1 = stw_lookup_context_locked( hglrc1 ); - ctx2 = stw_lookup_context_locked( hglrc2 ); - - if (ctx1 && ctx2 && - ctx1->iPixelFormat == ctx2->iPixelFormat) { - ret = _mesa_share_state(ctx2->st->ctx, ctx1->st->ctx); - } - - pipe_mutex_unlock( stw_dev->ctx_mutex ); - - return ret; -} - -static void -stw_viewport(GLcontext * glctx, GLint x, GLint y, - GLsizei width, GLsizei height) -{ - struct stw_context *ctx = (struct stw_context *)glctx->DriverCtx; - struct stw_framebuffer *fb; - - fb = stw_framebuffer_from_hdc( ctx->hdc ); - if(fb) { - stw_framebuffer_update(fb); - stw_framebuffer_release(fb); - } -} - -UINT_PTR -stw_create_layer_context( - HDC hdc, - int iLayerPlane ) -{ - int iPixelFormat; - const struct stw_pixelformat_info *pfi; - GLvisual visual; - struct stw_context *ctx = NULL; - struct pipe_screen *screen = NULL; - struct pipe_context *pipe = NULL; - - if(!stw_dev) - return 0; - - if (iLayerPlane != 0) - return 0; - - iPixelFormat = GetPixelFormat(hdc); - if(!iPixelFormat) - return 0; - - pfi = stw_pixelformat_get_info( iPixelFormat - 1 ); - stw_pixelformat_visual(&visual, pfi); - - ctx = CALLOC_STRUCT( stw_context ); - if (ctx == NULL) - goto no_ctx; - - ctx->hdc = hdc; - ctx->iPixelFormat = iPixelFormat; - - screen = stw_dev->screen; - -#ifdef DEBUG - /* Unwrap screen */ - if(stw_dev->trace_running) - screen = trace_screen(screen)->screen; -#endif - - pipe = stw_dev->stw_winsys->create_context( screen ); - if (pipe == NULL) - goto no_pipe; - -#ifdef DEBUG - /* Wrap context */ - if(stw_dev->trace_running) - pipe = trace_context_create(stw_dev->screen, pipe); -#endif - - /* pass to stw_flush_frontbuffer as context_private */ - assert(!pipe->priv); - pipe->priv = hdc; - - ctx->st = st_create_context( pipe, &visual, NULL ); - if (ctx->st == NULL) - goto no_st_ctx; - - ctx->st->ctx->DriverCtx = ctx; - ctx->st->ctx->Driver.Viewport = stw_viewport; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - ctx->hglrc = handle_table_add(stw_dev->ctx_table, ctx); - pipe_mutex_unlock( stw_dev->ctx_mutex ); - if (!ctx->hglrc) - goto no_hglrc; - - return ctx->hglrc; - -no_hglrc: - st_destroy_context(ctx->st); - goto no_pipe; /* st_context_destroy already destroys pipe */ -no_st_ctx: - pipe->destroy( pipe ); -no_pipe: - FREE(ctx); -no_ctx: - return 0; -} - -BOOL -stw_delete_context( - UINT_PTR hglrc ) -{ - struct stw_context *ctx ; - BOOL ret = FALSE; - - if (!stw_dev) - return FALSE; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - ctx = stw_lookup_context_locked(hglrc); - handle_table_remove(stw_dev->ctx_table, hglrc); - pipe_mutex_unlock( stw_dev->ctx_mutex ); - - if (ctx) { - struct stw_context *curctx = stw_current_context(); - - /* Unbind current if deleting current context. */ - if (curctx == ctx) - st_make_current( NULL, NULL, NULL ); - - st_destroy_context(ctx->st); - FREE(ctx); - - ret = TRUE; - } - - return ret; -} - -BOOL -stw_release_context( - UINT_PTR hglrc ) -{ - struct stw_context *ctx; - - if (!stw_dev) - return FALSE; - - pipe_mutex_lock( stw_dev->ctx_mutex ); - ctx = stw_lookup_context_locked( hglrc ); - pipe_mutex_unlock( stw_dev->ctx_mutex ); - - if (!ctx) - return FALSE; - - /* The expectation is that ctx is the same context which is - * current for this thread. We should check that and return False - * if not the case. - */ - if (ctx != stw_current_context()) - return FALSE; - - if (stw_make_current( NULL, 0 ) == FALSE) - return FALSE; - - return TRUE; -} - - -UINT_PTR -stw_get_current_context( void ) -{ - struct stw_context *ctx; - - ctx = stw_current_context(); - if(!ctx) - return 0; - - return ctx->hglrc; -} - -HDC -stw_get_current_dc( void ) -{ - struct stw_context *ctx; - - ctx = stw_current_context(); - if(!ctx) - return NULL; - - return ctx->hdc; -} - -BOOL -stw_make_current( - HDC hdc, - UINT_PTR hglrc ) -{ - struct stw_context *curctx = NULL; - struct stw_context *ctx = NULL; - struct stw_framebuffer *fb = NULL; - - if (!stw_dev) - goto fail; - - curctx = stw_current_context(); - if (curctx != NULL) { - if (curctx->hglrc != hglrc) - st_flush(curctx->st, PIPE_FLUSH_RENDER_CACHE, NULL); - - /* Return if already current. */ - if (curctx->hglrc == hglrc && curctx->hdc == hdc) { - ctx = curctx; - fb = stw_framebuffer_from_hdc( hdc ); - goto success; - } - } - - if (hdc == NULL || hglrc == 0) { - return st_make_current( NULL, NULL, NULL ); - } - - pipe_mutex_lock( stw_dev->ctx_mutex ); - ctx = stw_lookup_context_locked( hglrc ); - pipe_mutex_unlock( stw_dev->ctx_mutex ); - if(!ctx) - goto fail; - - fb = stw_framebuffer_from_hdc( hdc ); - if(!fb) { - /* Applications should call SetPixelFormat before creating a context, - * but not all do, and the opengl32 runtime seems to use a default pixel - * format in some cases, so we must create a framebuffer for those here - */ - int iPixelFormat = GetPixelFormat(hdc); - if(iPixelFormat) - fb = stw_framebuffer_create( hdc, iPixelFormat ); - if(!fb) - goto fail; - } - - if(fb->iPixelFormat != ctx->iPixelFormat) - goto fail; - - /* Lazy allocation of the frame buffer */ - if(!stw_framebuffer_allocate(fb)) - goto fail; - - /* Bind the new framebuffer */ - ctx->hdc = hdc; - - /* pass to stw_flush_frontbuffer as context_private */ - ctx->st->pipe->priv = hdc; - - if(!st_make_current( ctx->st, fb->stfb, fb->stfb )) - goto fail; - -success: - assert(fb); - if(fb) { - stw_framebuffer_update(fb); - stw_framebuffer_release(fb); - } - - return TRUE; - -fail: - if(fb) - stw_framebuffer_release(fb); - st_make_current( NULL, NULL, NULL ); - return FALSE; -} diff --git a/src/gallium/state_trackers/wgl/icd/stw_icd.c b/src/gallium/state_trackers/wgl/stw_context.c index 347f40aa06b..f2f0264844a 100644 --- a/src/gallium/state_trackers/wgl/icd/stw_icd.c +++ b/src/gallium/state_trackers/wgl/stw_context.c @@ -26,18 +26,49 @@ **************************************************************************/ #include <windows.h> -#include <stdio.h> -#include "GL/gl.h" - -#include "util/u_debug.h" -#include "pipe/p_thread.h" - -#include "shared/stw_public.h" -#include "icd/stw_icd.h" +#include "main/mtypes.h" +#include "main/context.h" +#include "pipe/p_compiler.h" +#include "pipe/p_context.h" +#include "state_tracker/st_context.h" +#include "state_tracker/st_public.h" + +#ifdef DEBUG +#include "trace/tr_screen.h" +#include "trace/tr_context.h" +#endif + +#include "stw_icd.h" +#include "stw_device.h" +#include "stw_winsys.h" +#include "stw_framebuffer.h" +#include "stw_pixelformat.h" +#include "stw_context.h" +#include "stw_tls.h" + + +static INLINE struct stw_context * +stw_context(GLcontext *glctx) +{ + if(!glctx) + return NULL; + assert(glctx->DriverCtx); + return (struct stw_context *)glctx->DriverCtx; +} -#define DBG 0 +static INLINE struct stw_context * +stw_current_context(void) +{ + /* We must check if multiple threads are being used or GET_CURRENT_CONTEXT + * might return the current context of the thread first seen. */ + _glapi_check_multithread(); + { + GET_CURRENT_CONTEXT( glctx ); + return stw_context(glctx); + } +} BOOL APIENTRY DrvCopyContext( @@ -45,24 +76,64 @@ DrvCopyContext( DHGLRC dhrcDest, UINT fuMask ) { - return stw_copy_context(dhrcSource, dhrcDest, fuMask); -} + struct stw_context *src; + struct stw_context *dst; + BOOL ret = FALSE; + pipe_mutex_lock( stw_dev->ctx_mutex ); + + src = stw_lookup_context_locked( dhrcSource ); + dst = stw_lookup_context_locked( dhrcDest ); + + if (src && dst) { + /* FIXME */ + assert(0); + (void) src; + (void) dst; + (void) fuMask; + } -DHGLRC APIENTRY -DrvCreateLayerContext( - HDC hdc, - INT iLayerPlane ) + pipe_mutex_unlock( stw_dev->ctx_mutex ); + + return ret; +} + +BOOL APIENTRY +DrvShareLists( + DHGLRC dhglrc1, + DHGLRC dhglrc2 ) { - DHGLRC r; + struct stw_context *ctx1; + struct stw_context *ctx2; + BOOL ret = FALSE; + + pipe_mutex_lock( stw_dev->ctx_mutex ); - r = stw_create_layer_context( hdc, iLayerPlane ); + ctx1 = stw_lookup_context_locked( dhglrc1 ); + ctx2 = stw_lookup_context_locked( dhglrc2 ); + + if (ctx1 && ctx2 && + ctx1->iPixelFormat == ctx2->iPixelFormat) { + ret = _mesa_share_state(ctx2->st->ctx, ctx1->st->ctx); + } + + pipe_mutex_unlock( stw_dev->ctx_mutex ); - if (DBG) - debug_printf( "%s( %p, %i ) = %u\n", - __FUNCTION__, hdc, iLayerPlane, r ); + return ret; +} + +static void +stw_viewport(GLcontext * glctx, GLint x, GLint y, + GLsizei width, GLsizei height) +{ + struct stw_context *ctx = (struct stw_context *)glctx->DriverCtx; + struct stw_framebuffer *fb; - return r; + fb = stw_framebuffer_from_hdc( ctx->hdc ); + if(fb) { + stw_framebuffer_update(fb); + stw_framebuffer_release(fb); + } } DHGLRC APIENTRY @@ -72,114 +143,253 @@ DrvCreateContext( return DrvCreateLayerContext( hdc, 0 ); } -BOOL APIENTRY -DrvDeleteContext( - DHGLRC dhglrc ) +DHGLRC APIENTRY +DrvCreateLayerContext( + HDC hdc, + INT iLayerPlane ) { - BOOL r; + int iPixelFormat; + const struct stw_pixelformat_info *pfi; + GLvisual visual; + struct stw_context *ctx = NULL; + struct pipe_screen *screen = NULL; + struct pipe_context *pipe = NULL; - r = stw_delete_context( dhglrc ); + if(!stw_dev) + return 0; - if (DBG) - debug_printf( "%s( %u ) = %u\n", - __FUNCTION__, dhglrc, r ); + if (iLayerPlane != 0) + return 0; + + iPixelFormat = GetPixelFormat(hdc); + if(!iPixelFormat) + return 0; - return r; + pfi = stw_pixelformat_get_info( iPixelFormat - 1 ); + stw_pixelformat_visual(&visual, pfi); + + ctx = CALLOC_STRUCT( stw_context ); + if (ctx == NULL) + goto no_ctx; + + ctx->hdc = hdc; + ctx->iPixelFormat = iPixelFormat; + + screen = stw_dev->screen; + +#ifdef DEBUG + /* Unwrap screen */ + if(stw_dev->trace_running) + screen = trace_screen(screen)->screen; +#endif + + pipe = stw_dev->stw_winsys->create_context( screen ); + if (pipe == NULL) + goto no_pipe; + +#ifdef DEBUG + /* Wrap context */ + if(stw_dev->trace_running) + pipe = trace_context_create(stw_dev->screen, pipe); +#endif + + /* pass to stw_flush_frontbuffer as context_private */ + assert(!pipe->priv); + pipe->priv = hdc; + + ctx->st = st_create_context( pipe, &visual, NULL ); + if (ctx->st == NULL) + goto no_st_ctx; + + ctx->st->ctx->DriverCtx = ctx; + ctx->st->ctx->Driver.Viewport = stw_viewport; + + pipe_mutex_lock( stw_dev->ctx_mutex ); + ctx->dhglrc = handle_table_add(stw_dev->ctx_table, ctx); + pipe_mutex_unlock( stw_dev->ctx_mutex ); + if (!ctx->dhglrc) + goto no_hglrc; + + return ctx->dhglrc; + +no_hglrc: + st_destroy_context(ctx->st); + goto no_pipe; /* st_context_destroy already destroys pipe */ +no_st_ctx: + pipe->destroy( pipe ); +no_pipe: + FREE(ctx); +no_ctx: + return 0; } BOOL APIENTRY -DrvDescribeLayerPlane( - HDC hdc, - INT iPixelFormat, - INT iLayerPlane, - UINT nBytes, - LPLAYERPLANEDESCRIPTOR plpd ) +DrvDeleteContext( + DHGLRC dhglrc ) { - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); + struct stw_context *ctx ; + BOOL ret = FALSE; + + if (!stw_dev) + return FALSE; - return FALSE; -} + pipe_mutex_lock( stw_dev->ctx_mutex ); + ctx = stw_lookup_context_locked(dhglrc); + handle_table_remove(stw_dev->ctx_table, dhglrc); + pipe_mutex_unlock( stw_dev->ctx_mutex ); -LONG APIENTRY -DrvDescribePixelFormat( - HDC hdc, - INT iPixelFormat, - ULONG cjpfd, - PIXELFORMATDESCRIPTOR *ppfd ) -{ - LONG r; + if (ctx) { + struct stw_context *curctx = stw_current_context(); + + /* Unbind current if deleting current context. */ + if (curctx == ctx) + st_make_current( NULL, NULL, NULL ); - r = stw_pixelformat_describe( hdc, iPixelFormat, cjpfd, ppfd ); + st_destroy_context(ctx->st); + FREE(ctx); - if (DBG) - debug_printf( "%s( %p, %d, %u, %p ) = %d\n", - __FUNCTION__, hdc, iPixelFormat, cjpfd, ppfd, r ); + ret = TRUE; + } - return r; + return ret; } -int APIENTRY -DrvGetLayerPaletteEntries( - HDC hdc, - INT iLayerPlane, - INT iStart, - INT cEntries, - COLORREF *pcr ) +BOOL APIENTRY +DrvReleaseContext( + DHGLRC dhglrc ) { - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); + struct stw_context *ctx; - return 0; -} + if (!stw_dev) + return FALSE; -PROC APIENTRY -DrvGetProcAddress( - LPCSTR lpszProc ) -{ - PROC r; + pipe_mutex_lock( stw_dev->ctx_mutex ); + ctx = stw_lookup_context_locked( dhglrc ); + pipe_mutex_unlock( stw_dev->ctx_mutex ); - r = stw_get_proc_address( lpszProc ); + if (!ctx) + return FALSE; + + /* The expectation is that ctx is the same context which is + * current for this thread. We should check that and return False + * if not the case. + */ + if (ctx != stw_current_context()) + return FALSE; - if (DBG) - debug_printf( "%s( \"%s\" ) = %p\n", __FUNCTION__, lpszProc, r ); + if (stw_make_current( NULL, 0 ) == FALSE) + return FALSE; - return r; + return TRUE; } -BOOL APIENTRY -DrvRealizeLayerPalette( - HDC hdc, - INT iLayerPlane, - BOOL bRealize ) + +DHGLRC +stw_get_current_context( void ) { - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); + struct stw_context *ctx; - return FALSE; + ctx = stw_current_context(); + if(!ctx) + return 0; + + return ctx->dhglrc; } -BOOL APIENTRY -DrvReleaseContext( - DHGLRC dhglrc ) +HDC +stw_get_current_dc( void ) { - return stw_release_context(dhglrc); + struct stw_context *ctx; + + ctx = stw_current_context(); + if(!ctx) + return NULL; + + return ctx->hdc; } -void APIENTRY -DrvSetCallbackProcs( - INT nProcs, - PROC *pProcs ) +BOOL +stw_make_current( + HDC hdc, + DHGLRC dhglrc ) { - if (DBG) - debug_printf( "%s( %d, %p )\n", __FUNCTION__, nProcs, pProcs ); + struct stw_context *curctx = NULL; + struct stw_context *ctx = NULL; + struct stw_framebuffer *fb = NULL; - return; -} + if (!stw_dev) + goto fail; + + curctx = stw_current_context(); + if (curctx != NULL) { + if (curctx->dhglrc != dhglrc) + st_flush(curctx->st, PIPE_FLUSH_RENDER_CACHE, NULL); + + /* Return if already current. */ + if (curctx->dhglrc == dhglrc && curctx->hdc == hdc) { + ctx = curctx; + fb = stw_framebuffer_from_hdc( hdc ); + goto success; + } + } + + if (hdc == NULL || dhglrc == 0) { + return st_make_current( NULL, NULL, NULL ); + } + + pipe_mutex_lock( stw_dev->ctx_mutex ); + ctx = stw_lookup_context_locked( dhglrc ); + pipe_mutex_unlock( stw_dev->ctx_mutex ); + if(!ctx) + goto fail; + + fb = stw_framebuffer_from_hdc( hdc ); + if(!fb) { + /* Applications should call SetPixelFormat before creating a context, + * but not all do, and the opengl32 runtime seems to use a default pixel + * format in some cases, so we must create a framebuffer for those here + */ + int iPixelFormat = GetPixelFormat(hdc); + if(iPixelFormat) + fb = stw_framebuffer_create( hdc, iPixelFormat ); + if(!fb) + goto fail; + } + + if(fb->iPixelFormat != ctx->iPixelFormat) + goto fail; + + /* Lazy allocation of the frame buffer */ + if(!stw_framebuffer_allocate(fb)) + goto fail; + + /* Bind the new framebuffer */ + ctx->hdc = hdc; + + /* pass to stw_flush_frontbuffer as context_private */ + ctx->st->pipe->priv = hdc; + + if(!st_make_current( ctx->st, fb->stfb, fb->stfb )) + goto fail; + +success: + assert(fb); + if(fb) { + stw_framebuffer_update(fb); + stw_framebuffer_release(fb); + } + + return TRUE; +fail: + if(fb) + stw_framebuffer_release(fb); + st_make_current( NULL, NULL, NULL ); + return FALSE; +} /** - * Although WGL allows different dispatch entrypoints per context + * Although WGL allows different dispatch entrypoints per context */ static const GLCLTPROCTABLE cpt = { @@ -524,7 +734,6 @@ static const GLCLTPROCTABLE cpt = } }; - PGLCLTPROCTABLE APIENTRY DrvSetContext( HDC hdc, @@ -532,86 +741,9 @@ DrvSetContext( PFN_SETPROCTABLE pfnSetProcTable ) { PGLCLTPROCTABLE r = (PGLCLTPROCTABLE)&cpt; - + if (!stw_make_current( hdc, dhglrc )) r = NULL; - - if (DBG) - debug_printf( "%s( 0x%p, %u, 0x%p ) = %p\n", - __FUNCTION__, hdc, dhglrc, pfnSetProcTable, r ); - - return r; -} - -int APIENTRY -DrvSetLayerPaletteEntries( - HDC hdc, - INT iLayerPlane, - INT iStart, - INT cEntries, - CONST COLORREF *pcr ) -{ - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); - - return 0; -} - -BOOL APIENTRY -DrvSetPixelFormat( - HDC hdc, - LONG iPixelFormat ) -{ - BOOL r; - - r = stw_pixelformat_set( hdc, iPixelFormat ); - - if (DBG) - debug_printf( "%s( %p, %d ) = %s\n", __FUNCTION__, hdc, iPixelFormat, r ? "TRUE" : "FALSE" ); return r; } - -BOOL APIENTRY -DrvShareLists( - DHGLRC dhglrc1, - DHGLRC dhglrc2 ) -{ - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); - - return stw_share_lists(dhglrc1, dhglrc2); -} - -BOOL APIENTRY -DrvSwapBuffers( - HDC hdc ) -{ - if (DBG) - debug_printf( "%s( %p )\n", __FUNCTION__, hdc ); - - return stw_swap_buffers( hdc ); -} - -BOOL APIENTRY -DrvSwapLayerBuffers( - HDC hdc, - UINT fuPlanes ) -{ - if (DBG) - debug_printf( "%s\n", __FUNCTION__ ); - - return stw_swap_layer_buffers( hdc, fuPlanes ); -} - -BOOL APIENTRY -DrvValidateVersion( - ULONG ulVersion ) -{ - if (DBG) - debug_printf( "%s( %u )\n", __FUNCTION__, ulVersion ); - - /* TODO: get the expected version from the winsys */ - - return ulVersion == 1; -} diff --git a/src/gallium/state_trackers/wgl/shared/stw_context.h b/src/gallium/state_trackers/wgl/stw_context.h index 166471de5eb..256c27e21ef 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_context.h +++ b/src/gallium/state_trackers/wgl/stw_context.h @@ -35,9 +35,15 @@ struct st_context; struct stw_context { struct st_context *st; - UINT_PTR hglrc; + DHGLRC dhglrc; int iPixelFormat; HDC hdc; }; +DHGLRC stw_get_current_context( void ); + +HDC stw_get_current_dc( void ); + +BOOL stw_make_current( HDC hdc, DHGLRC dhglrc ); + #endif /* STW_CONTEXT_H */ diff --git a/src/gallium/state_trackers/wgl/shared/stw_device.c b/src/gallium/state_trackers/wgl/stw_device.c index 0b6954915a6..985b8f0456a 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_device.c +++ b/src/gallium/state_trackers/wgl/stw_device.c @@ -29,6 +29,7 @@ #include "glapi/glthread.h" #include "util/u_debug.h" +#include "util/u_math.h" #include "pipe/p_screen.h" #include "state_tracker/st_public.h" @@ -37,12 +38,12 @@ #include "trace/tr_texture.h" #endif -#include "shared/stw_device.h" -#include "shared/stw_winsys.h" -#include "shared/stw_pixelformat.h" -#include "shared/stw_public.h" -#include "shared/stw_tls.h" -#include "shared/stw_framebuffer.h" +#include "stw_device.h" +#include "stw_winsys.h" +#include "stw_pixelformat.h" +#include "stw_icd.h" +#include "stw_tls.h" +#include "stw_framebuffer.h" #ifdef WIN32_THREADS extern _glthread_Mutex OneTimeLock; @@ -62,38 +63,28 @@ stw_flush_frontbuffer(struct pipe_screen *screen, struct pipe_surface *surface, void *context_private ) { - const struct stw_winsys *stw_winsys = stw_dev->stw_winsys; HDC hdc = (HDC)context_private; struct stw_framebuffer *fb; fb = stw_framebuffer_from_hdc( hdc ); - /* fb can be NULL if window was destroyed already */ - if (fb) { + if (!fb) { + /* fb can be NULL if window was destroyed already */ + return; + } + #if DEBUG - { - struct pipe_surface *surface2; - - if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_FRONT_LEFT, &surface2 )) - assert(0); - else - assert(surface2 == surface); - } -#endif + { + /* ensure that a random surface was not passed to us */ + struct pipe_surface *surface2; -#ifdef DEBUG - if(stw_dev->trace_running) { - screen = trace_screen(screen)->screen; - surface = trace_surface(surface)->surface; - } -#endif - } - - stw_winsys->flush_frontbuffer(screen, surface, hdc); - - if(fb) { - stw_framebuffer_update(fb); - stw_framebuffer_release(fb); + if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_FRONT_LEFT, &surface2 )) + assert(0); + else + assert(surface2 == surface); } +#endif + + stw_framebuffer_present_locked(hdc, fb, ST_SURFACE_FRONT_LEFT); } @@ -126,6 +117,9 @@ stw_init(const struct stw_winsys *stw_winsys) if(!screen) goto error1; + if(stw_winsys->get_adapter_luid) + stw_winsys->get_adapter_luid(screen, &stw_dev->AdapterLuid); + #ifdef DEBUG stw_dev->screen = trace_screen_create(screen); stw_dev->trace_running = stw_dev->screen != screen ? TRUE : FALSE; @@ -182,7 +176,7 @@ stw_cleanup(void) /* Ensure all contexts are destroyed */ i = handle_table_get_first_handle(stw_dev->ctx_table); while (i) { - stw_delete_context(i); + DrvDeleteContext(i); i = handle_table_get_next_handle(stw_dev->ctx_table, i); } handle_table_destroy(stw_dev->ctx_table); @@ -212,7 +206,7 @@ stw_cleanup(void) struct stw_context * -stw_lookup_context_locked( UINT_PTR dhglrc ) +stw_lookup_context_locked( DHGLRC dhglrc ) { if (dhglrc == 0) return NULL; @@ -223,3 +217,28 @@ stw_lookup_context_locked( UINT_PTR dhglrc ) return (struct stw_context *) handle_table_get(stw_dev->ctx_table, dhglrc); } + +void APIENTRY +DrvSetCallbackProcs( + INT nProcs, + PROC *pProcs ) +{ + size_t size; + + if (stw_dev == NULL) + return; + + size = MIN2(nProcs * sizeof *pProcs, sizeof stw_dev->callbacks); + memcpy(&stw_dev->callbacks, pProcs, size); + + return; +} + + +BOOL APIENTRY +DrvValidateVersion( + ULONG ulVersion ) +{ + /* TODO: get the expected version from the winsys */ + return ulVersion == 1; +} diff --git a/src/gallium/state_trackers/wgl/shared/stw_device.h b/src/gallium/state_trackers/wgl/stw_device.h index e1bb9518dd1..0bf3b0da825 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_device.h +++ b/src/gallium/state_trackers/wgl/stw_device.h @@ -29,11 +29,10 @@ #define STW_DEVICE_H_ -#include <windows.h> - #include "pipe/p_compiler.h" #include "pipe/p_thread.h" #include "util/u_handle_table.h" +#include "stw_icd.h" #include "stw_pixelformat.h" @@ -53,10 +52,14 @@ struct stw_device boolean trace_running; #endif + LUID AdapterLuid; + struct stw_pixelformat_info pixelformats[STW_MAX_PIXELFORMATS]; unsigned pixelformat_count; unsigned pixelformat_extended_count; + GLCALLBACKTABLE callbacks; + pipe_mutex ctx_mutex; struct handle_table *ctx_table; @@ -69,7 +72,7 @@ struct stw_device }; struct stw_context * -stw_lookup_context_locked( UINT_PTR hglrc ); +stw_lookup_context_locked( DHGLRC hglrc ); extern struct stw_device *stw_dev; diff --git a/src/gallium/state_trackers/wgl/shared/stw_extensionsstring.c b/src/gallium/state_trackers/wgl/stw_ext_extensionsstring.c index 62c859e1f92..62c859e1f92 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_extensionsstring.c +++ b/src/gallium/state_trackers/wgl/stw_ext_extensionsstring.c diff --git a/src/gallium/state_trackers/wgl/shared/stw_extgallium.c b/src/gallium/state_trackers/wgl/stw_ext_gallium.c index fc22737d7e3..fb30ec5dba9 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_extgallium.c +++ b/src/gallium/state_trackers/wgl/stw_ext_gallium.c @@ -27,9 +27,9 @@ #include "pipe/p_screen.h" -#include "stw_public.h" #include "stw_device.h" #include "stw_winsys.h" +#include "stw_ext_gallium.h" #ifdef DEBUG #include "trace/tr_screen.h" diff --git a/src/gallium/state_trackers/wgl/shared/stw_extgallium.h b/src/gallium/state_trackers/wgl/stw_ext_gallium.h index cc35f2bb7fe..cc35f2bb7fe 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_extgallium.h +++ b/src/gallium/state_trackers/wgl/stw_ext_gallium.h diff --git a/src/gallium/state_trackers/wgl/shared/stw_arbpixelformat.c b/src/gallium/state_trackers/wgl/stw_ext_pixelformat.c index 0e2d4076993..8a9995aba8e 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_arbpixelformat.c +++ b/src/gallium/state_trackers/wgl/stw_ext_pixelformat.c @@ -43,7 +43,6 @@ #include "pipe/p_compiler.h" #include "util/u_memory.h" -#include "stw_public.h" #include "stw_pixelformat.h" diff --git a/src/gallium/state_trackers/wgl/shared/stw_extswapinterval.c b/src/gallium/state_trackers/wgl/stw_ext_swapinterval.c index 9eac6a1d09d..9eac6a1d09d 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_extswapinterval.c +++ b/src/gallium/state_trackers/wgl/stw_ext_swapinterval.c diff --git a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.c b/src/gallium/state_trackers/wgl/stw_framebuffer.c index b8956bb5509..8a3e11b6b40 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.c +++ b/src/gallium/state_trackers/wgl/stw_framebuffer.c @@ -1,8 +1,8 @@ /************************************************************************** - * - * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * + * Copyright 2008-2009 Vmware, Inc. * All Rights Reserved. - * + * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the * "Software"), to deal in the Software without restriction, including @@ -10,19 +10,19 @@ * distribute, sub license, and/or sell copies of the Software, and to * permit persons to whom the Software is furnished to do so, subject to * the following conditions: - * + * * The above copyright notice and this permission notice (including the * next paragraph) shall be included in all copies or substantial portions * of the Software. - * + * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * + * **************************************************************************/ #include <windows.h> @@ -38,9 +38,9 @@ #include "trace/tr_texture.h" #endif +#include "stw_icd.h" #include "stw_framebuffer.h" #include "stw_device.h" -#include "stw_public.h" #include "stw_winsys.h" #include "stw_tls.h" @@ -83,6 +83,9 @@ stw_framebuffer_destroy_locked( *link = fb->next; fb->next = NULL; + if(fb->shared_surface) + stw_dev->stw_winsys->shared_surface_close(stw_dev->screen, fb->shared_surface); + st_unreference_framebuffer(fb->stfb); pipe_mutex_unlock( fb->mutex ); @@ -106,13 +109,18 @@ static INLINE void stw_framebuffer_get_size( struct stw_framebuffer *fb ) { unsigned width, height; - RECT rect; + RECT client_rect; + RECT window_rect; + POINT client_pos; assert(fb->hWnd); - GetClientRect( fb->hWnd, &rect ); - width = rect.right - rect.left; - height = rect.bottom - rect.top; + /* Get the client area size. */ + GetClientRect( fb->hWnd, &client_rect ); + assert(client_rect.left == 0); + assert(client_rect.top == 0); + width = client_rect.right - client_rect.left; + height = client_rect.bottom - client_rect.top; if(width < 1) width = 1; @@ -124,6 +132,31 @@ stw_framebuffer_get_size( struct stw_framebuffer *fb ) fb->width = width; fb->height = height; } + + client_pos.x = 0; + client_pos.y = 0; + ClientToScreen(fb->hWnd, &client_pos); + + GetWindowRect(fb->hWnd, &window_rect); + + fb->client_rect.left = client_pos.x - window_rect.left; + fb->client_rect.top = client_pos.y - window_rect.top; + fb->client_rect.right = fb->client_rect.left + fb->width; + fb->client_rect.bottom = fb->client_rect.top + fb->height; + +#if 0 + debug_printf("\n"); + debug_printf("%s: client_position = (%i, %i)\n", + __FUNCTION__, client_pos.x, client_pos.y); + debug_printf("%s: window_rect = (%i, %i) - (%i, %i)\n", + __FUNCTION__, + window_rect.left, window_rect.top, + window_rect.right, window_rect.bottom); + debug_printf("%s: client_rect = (%i, %i) - (%i, %i)\n", + __FUNCTION__, + fb->client_rect.left, fb->client_rect.top, + fb->client_rect.right, fb->client_rect.bottom); +#endif } @@ -155,6 +188,7 @@ stw_call_window_proc( * can be masked out by the application. */ LPWINDOWPOS lpWindowPos = (LPWINDOWPOS)pParams->lParam; if((lpWindowPos->flags & SWP_SHOWWINDOW) || + !(lpWindowPos->flags & SWP_NOMOVE) || !(lpWindowPos->flags & SWP_NOSIZE)) { fb = stw_framebuffer_from_hwnd( pParams->hwnd ); if(fb) { @@ -379,10 +413,10 @@ stw_framebuffer_from_hwnd( } -BOOL -stw_pixelformat_set( +BOOL APIENTRY +DrvSetPixelFormat( HDC hdc, - int iPixelFormat ) + LONG iPixelFormat ) { uint count; uint index; @@ -435,35 +469,24 @@ stw_pixelformat_get( } -BOOL -stw_swap_buffers( - HDC hdc ) +BOOL APIENTRY +DrvPresentBuffers(HDC hdc, PGLPRESENTBUFFERSDATA data) { struct stw_framebuffer *fb; struct pipe_screen *screen; struct pipe_surface *surface; + unsigned surface_index; + BOOL ret = FALSE; fb = stw_framebuffer_from_hdc( hdc ); if (fb == NULL) return FALSE; - if (!(fb->pfi->pfd.dwFlags & PFD_DOUBLEBUFFER)) { - stw_framebuffer_release(fb); - return TRUE; - } - - /* If we're swapping the buffer associated with the current context - * we have to flush any pending rendering commands first. - */ - st_notify_swapbuffers( fb->stfb ); - screen = stw_dev->screen; - - if(!st_get_framebuffer_surface( fb->stfb, ST_SURFACE_BACK_LEFT, &surface )) { - /* FIXME: this shouldn't happen, but does on glean */ - stw_framebuffer_release(fb); - return FALSE; - } + + surface_index = (unsigned)(uintptr_t)data->pPrivateData; + if(!st_get_framebuffer_surface( fb->stfb, surface_index, &surface )) + goto fail; #ifdef DEBUG if(stw_dev->trace_running) { @@ -472,22 +495,127 @@ stw_swap_buffers( } #endif - stw_dev->stw_winsys->flush_frontbuffer( screen, surface, hdc ); - + if(data->hSharedSurface != fb->hSharedSurface) { + if(fb->shared_surface) { + stw_dev->stw_winsys->shared_surface_close(screen, fb->shared_surface); + fb->shared_surface = NULL; + } + + fb->hSharedSurface = data->hSharedSurface; + + if(data->hSharedSurface && + stw_dev->stw_winsys->shared_surface_open) { + fb->shared_surface = stw_dev->stw_winsys->shared_surface_open(screen, fb->hSharedSurface); + } + } + + if(fb->shared_surface) { + stw_dev->stw_winsys->compose(screen, + surface, + fb->shared_surface, + &fb->client_rect, + data->PresentHistoryToken); + } + else { + stw_dev->stw_winsys->present( screen, surface, hdc ); + } + + ret = TRUE; + +fail: + stw_framebuffer_update(fb); + stw_framebuffer_release(fb); - - return TRUE; + + return ret; } +/** + * Queue a composition. + * + * It will drop the lock on success. + */ BOOL -stw_swap_layer_buffers( +stw_framebuffer_present_locked(HDC hdc, + struct stw_framebuffer *fb, + unsigned surface_index) +{ + if(stw_dev->callbacks.wglCbPresentBuffers && + stw_dev->stw_winsys->compose) { + GLCBPRESENTBUFFERSDATA data; + + memset(&data, 0, sizeof data); + data.magic1 = 2; + data.magic2 = 0; + data.AdapterLuid = stw_dev->AdapterLuid; + data.rect = fb->client_rect; + data.pPrivateData = (void *)(uintptr_t)surface_index; + + stw_framebuffer_release(fb); + + return stw_dev->callbacks.wglCbPresentBuffers(hdc, &data); + } + else { + struct pipe_screen *screen = stw_dev->screen; + struct pipe_surface *surface; + + if(!st_get_framebuffer_surface( fb->stfb, surface_index, &surface )) { + /* FIXME: this shouldn't happen, but does on glean */ + stw_framebuffer_release(fb); + return FALSE; + } + +#ifdef DEBUG + if(stw_dev->trace_running) { + screen = trace_screen(screen)->screen; + surface = trace_surface(surface)->surface; + } +#endif + + stw_dev->stw_winsys->present( screen, surface, hdc ); + + stw_framebuffer_update(fb); + + stw_framebuffer_release(fb); + + return TRUE; + } +} + + +BOOL APIENTRY +DrvSwapBuffers( + HDC hdc ) +{ + struct stw_framebuffer *fb; + + fb = stw_framebuffer_from_hdc( hdc ); + if (fb == NULL) + return FALSE; + + if (!(fb->pfi->pfd.dwFlags & PFD_DOUBLEBUFFER)) { + stw_framebuffer_release(fb); + return TRUE; + } + + /* If we're swapping the buffer associated with the current context + * we have to flush any pending rendering commands first. + */ + st_notify_swapbuffers( fb->stfb ); + + return stw_framebuffer_present_locked(hdc, fb, ST_SURFACE_BACK_LEFT); +} + + +BOOL APIENTRY +DrvSwapLayerBuffers( HDC hdc, UINT fuPlanes ) { if(fuPlanes & WGL_SWAP_MAIN_PLANE) - return stw_swap_buffers(hdc); + return DrvSwapBuffers(hdc); return FALSE; } diff --git a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.h b/src/gallium/state_trackers/wgl/stw_framebuffer.h index 13d29f37e48..5afbe749086 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_framebuffer.h +++ b/src/gallium/state_trackers/wgl/stw_framebuffer.h @@ -73,9 +73,20 @@ struct stw_framebuffer /* FIXME: Make this work for multiple contexts bound to the same framebuffer */ boolean must_resize; + unsigned width; unsigned height; + /** + * Client area rectangle, relative to the window upper-left corner. + * + * @sa GLCBPRESENTBUFFERSDATA::rect. + */ + RECT client_rect; + + HANDLE hSharedSurface; + struct stw_shared_surface *shared_surface; + /** * This is protected by stw_device::fb_mutex, not the mutex above. * @@ -126,6 +137,11 @@ BOOL stw_framebuffer_allocate( struct stw_framebuffer *fb ); +BOOL +stw_framebuffer_present_locked(HDC hdc, + struct stw_framebuffer *fb, + unsigned surface_index); + void stw_framebuffer_update( struct stw_framebuffer *fb); diff --git a/src/gallium/state_trackers/wgl/shared/stw_getprocaddress.c b/src/gallium/state_trackers/wgl/stw_getprocaddress.c index 879ced925a5..8875dc22f3d 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_getprocaddress.c +++ b/src/gallium/state_trackers/wgl/stw_getprocaddress.c @@ -33,8 +33,7 @@ #include <GL/wglext.h> #include "glapi/glapi.h" -#include "stw_public.h" -#include "stw_extgallium.h" +#include "stw_ext_gallium.h" struct stw_extension_entry { @@ -68,8 +67,8 @@ static const struct stw_extension_entry stw_extension_entries[] = { { NULL, NULL } }; -PROC -stw_get_proc_address( +PROC APIENTRY +DrvGetProcAddress( LPCSTR lpszProc ) { const struct stw_extension_entry *entry; diff --git a/src/gallium/state_trackers/wgl/icd/stw_icd.h b/src/gallium/state_trackers/wgl/stw_icd.h index cbc1a665481..02eb543fef0 100644 --- a/src/gallium/state_trackers/wgl/icd/stw_icd.h +++ b/src/gallium/state_trackers/wgl/stw_icd.h @@ -1,6 +1,6 @@ /************************************************************************** * - * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas. + * Copyright 2008-2009 Vmware, Inc. * All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a @@ -18,7 +18,7 @@ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. @@ -388,6 +388,113 @@ typedef struct _GLCLTPROCTABLE typedef VOID (APIENTRY * PFN_SETPROCTABLE)(PGLCLTPROCTABLE); +/** + * Presentation data passed to opengl32!wglCbPresentBuffers. + * + * Pure software drivers don't need to worry about this -- if they stick to the + * GDI API then will integrate with the Desktop Window Manager (DWM) without + * problems. Hardware drivers, however, cannot present directly to the primary + * surface while the DWM is active, as DWM gets exclusive access to the primary + * surface. + * + * Proper DWM integration requires: + * - advertise the PFD_SUPPORT_COMPOSITION flag + * - redirect glFlush/glfinish/wglSwapBuffers into a surface shared with the + * DWM process. + * + * @sa http://www.opengl.org/pipeline/article/vol003_7/ + * @sa http://blogs.msdn.com/greg_schechter/archive/2006/05/02/588934.aspx + */ +typedef struct _GLCBPRESENTBUFFERSDATA +{ + /** + * wglCbPresentBuffers enforces this to be 2. + */ + DWORD magic1; + + /** + * wglCbPresentBuffers enforces to be 0 or 1, but it is most commonly + * set to 0. + */ + DWORD magic2; + + /** + * Locally unique identifier (LUID) of the graphics adapter. + * + * This should contain the value returned by D3DKMTOpenAdapterFromHdc. It + * is passed to dwmapi!DwmpDxGetWindowSharedSurface in order to obtain + * the shared surface handle for the bound drawable (window). + * + * @sa http://msdn.microsoft.com/en-us/library/ms799177.aspx + */ + LUID AdapterLuid; + + /** + * This is passed unmodified to DrvPresentBuffers + */ + LPVOID pPrivateData; + + /** + * Client area rectangle to update, relative to the window upper-left corner. + */ + RECT rect; +} GLCBPRESENTBUFFERSDATA, *PGLCBPRESENTBUFFERSDATA; + +/** + * Callbacks supplied to DrvSetCallbackProcs by the OpenGL runtime. + * + * Pointers to several callback functions in opengl32.dll. + */ +typedef struct _GLCALLBACKTABLE +{ + /** Unused */ + PROC wglCbSetCurrentValue; + + /** Unused */ + PROC wglCbGetCurrentValue; + + /** Unused */ + PROC wglCbGetDhglrc; + + /** Unused */ + PROC wglCbGetDdHandle; + + /** + * Queue a present composition. + * + * Makes the runtime call DrvPresentBuffers with the composition information. + */ + BOOL (APIENTRY *wglCbPresentBuffers)(HDC hdc, PGLCBPRESENTBUFFERSDATA data); + +} GLCALLBACKTABLE; + +typedef struct _GLPRESENTBUFFERSDATA +{ + /** + * The shared surface handle. + * + * Return by dwmapi!DwmpDxGetWindowSharedSurface. + * + * @sa http://channel9.msdn.com/forums/TechOff/251261-Help-Getting-the-shared-window-texture-out-of-DWM-/ + */ + HANDLE hSharedSurface; + + LUID AdapterLuid; + + /** + * Present history token. + * + * This is returned by dwmapi!DwmpDxGetWindowSharedSurface and + * should be passed to D3DKMTRender in D3DKMT_RENDER::PresentHistoryToken. + * + * @sa http://msdn.microsoft.com/en-us/library/ms799176.aspx + */ + ULONGLONG PresentHistoryToken; + + /** Same as GLCBPRESENTBUFFERSDATA::pPrivateData */ + LPVOID pPrivateData; +} GLPRESENTBUFFERSDATA, *PGLPRESENTBUFFERSDATA; + BOOL APIENTRY DrvCopyContext( DHGLRC dhrcSource, @@ -435,6 +542,9 @@ DrvGetProcAddress( LPCSTR lpszProc ); BOOL APIENTRY +DrvPresentBuffers(HDC hdc, PGLPRESENTBUFFERSDATA data); + +BOOL APIENTRY DrvRealizeLayerPalette( HDC hdc, INT iLayerPlane, diff --git a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.c b/src/gallium/state_trackers/wgl/stw_pixelformat.c index c296744838b..7abe5d9f7fa 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.c +++ b/src/gallium/state_trackers/wgl/stw_pixelformat.c @@ -34,9 +34,9 @@ #include "util/u_debug.h" +#include "stw_icd.h" #include "stw_device.h" #include "stw_pixelformat.h" -#include "stw_public.h" #include "stw_tls.h" @@ -154,8 +154,11 @@ stw_pixelformat_add( pfi->pfd.dwFlags = PFD_SUPPORT_OPENGL; /* TODO: also support non-native pixel formats */ - pfi->pfd.dwFlags |= PFD_DRAW_TO_WINDOW ; - + pfi->pfd.dwFlags |= PFD_DRAW_TO_WINDOW; + + /* See http://www.opengl.org/pipeline/article/vol003_7/ */ + pfi->pfd.dwFlags |= PFD_SUPPORT_COMPOSITION; + if (doublebuffer) pfi->pfd.dwFlags |= PFD_DOUBLEBUFFER | PFD_SWAP_COPY; @@ -288,12 +291,12 @@ stw_pixelformat_visual(GLvisual *visual, } -int -stw_pixelformat_describe( +LONG APIENTRY +DrvDescribePixelFormat( HDC hdc, - int iPixelFormat, - UINT nBytes, - LPPIXELFORMATDESCRIPTOR ppfd ) + INT iPixelFormat, + ULONG cjpfd, + PIXELFORMATDESCRIPTOR *ppfd ) { uint count; uint index; @@ -306,7 +309,7 @@ stw_pixelformat_describe( if (ppfd == NULL) return count; - if (index >= count || nBytes != sizeof( PIXELFORMATDESCRIPTOR )) + if (index >= count || cjpfd != sizeof( PIXELFORMATDESCRIPTOR )) return 0; pfi = stw_pixelformat_get_info( index ); @@ -316,6 +319,52 @@ stw_pixelformat_describe( return count; } +BOOL APIENTRY +DrvDescribeLayerPlane( + HDC hdc, + INT iPixelFormat, + INT iLayerPlane, + UINT nBytes, + LPLAYERPLANEDESCRIPTOR plpd ) +{ + assert(0); + return FALSE; +} + +int APIENTRY +DrvGetLayerPaletteEntries( + HDC hdc, + INT iLayerPlane, + INT iStart, + INT cEntries, + COLORREF *pcr ) +{ + assert(0); + return 0; +} + +int APIENTRY +DrvSetLayerPaletteEntries( + HDC hdc, + INT iLayerPlane, + INT iStart, + INT cEntries, + CONST COLORREF *pcr ) +{ + assert(0); + return 0; +} + +BOOL APIENTRY +DrvRealizeLayerPalette( + HDC hdc, + INT iLayerPlane, + BOOL bRealize ) +{ + assert(0); + return FALSE; +} + /* Only used by the wgl code, but have it here to avoid exporting the * pixelformat.h functionality. */ diff --git a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.h b/src/gallium/state_trackers/wgl/stw_pixelformat.h index bec429231b2..3a690b35bad 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_pixelformat.h +++ b/src/gallium/state_trackers/wgl/stw_pixelformat.h @@ -30,6 +30,10 @@ #include <windows.h> +#ifndef PFD_SUPPORT_COMPOSITION +#define PFD_SUPPORT_COMPOSITION 0x00008000 +#endif + #include "main/mtypes.h" #include "pipe/p_compiler.h" @@ -62,4 +66,11 @@ void stw_pixelformat_visual(GLvisual *visual, const struct stw_pixelformat_info *pfi ); +int +stw_pixelformat_choose( HDC hdc, + CONST PIXELFORMATDESCRIPTOR *ppfd ); + +int +stw_pixelformat_get(HDC hdc); + #endif /* STW_PIXELFORMAT_H */ diff --git a/src/gallium/state_trackers/wgl/shared/stw_tls.c b/src/gallium/state_trackers/wgl/stw_tls.c index 4bd6a9289c9..4bd6a9289c9 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_tls.c +++ b/src/gallium/state_trackers/wgl/stw_tls.c diff --git a/src/gallium/state_trackers/wgl/shared/stw_tls.h b/src/gallium/state_trackers/wgl/stw_tls.h index fbf8b1cbee4..fbf8b1cbee4 100644 --- a/src/gallium/state_trackers/wgl/shared/stw_tls.h +++ b/src/gallium/state_trackers/wgl/stw_tls.h diff --git a/src/gallium/state_trackers/wgl/wgl/stw_wgl.c b/src/gallium/state_trackers/wgl/stw_wgl.c index a131292f7ae..bb199fdd252 100644 --- a/src/gallium/state_trackers/wgl/wgl/stw_wgl.c +++ b/src/gallium/state_trackers/wgl/stw_wgl.c @@ -28,7 +28,9 @@ #include <windows.h> #include "util/u_debug.h" -#include "shared/stw_public.h" +#include "stw_icd.h" +#include "stw_context.h" +#include "stw_pixelformat.h" #include "stw_wgl.h" @@ -38,16 +40,16 @@ wglCopyContext( HGLRC hglrcDst, UINT mask ) { - return stw_copy_context( (UINT_PTR)hglrcSrc, - (UINT_PTR)hglrcDst, - mask ); + return DrvCopyContext( (DHGLRC)(UINT_PTR)hglrcSrc, + (DHGLRC)(UINT_PTR)hglrcDst, + mask ); } WINGDIAPI HGLRC APIENTRY wglCreateContext( HDC hdc ) { - return wglCreateLayerContext(hdc, 0); + return (HGLRC) DrvCreateContext(hdc); } WINGDIAPI HGLRC APIENTRY @@ -55,21 +57,21 @@ wglCreateLayerContext( HDC hdc, int iLayerPlane ) { - return (HGLRC) stw_create_layer_context( hdc, iLayerPlane ); + return (HGLRC) DrvCreateLayerContext( hdc, iLayerPlane ); } WINGDIAPI BOOL APIENTRY wglDeleteContext( HGLRC hglrc ) { - return stw_delete_context( (UINT_PTR)hglrc ); + return DrvDeleteContext((DHGLRC)(UINT_PTR)hglrc ); } WINGDIAPI HGLRC APIENTRY wglGetCurrentContext( VOID ) { - return (HGLRC)stw_get_current_context(); + return (HGLRC)(UINT_PTR)stw_get_current_context(); } WINGDIAPI HDC APIENTRY @@ -83,7 +85,7 @@ wglMakeCurrent( HDC hdc, HGLRC hglrc ) { - return stw_make_current( hdc, (UINT_PTR)hglrc ); + return DrvSetContext( hdc, (DHGLRC)(UINT_PTR)hglrc, NULL ) ? TRUE : FALSE; } @@ -91,7 +93,7 @@ WINGDIAPI BOOL APIENTRY wglSwapBuffers( HDC hdc ) { - return stw_swap_buffers( hdc ); + return DrvSwapBuffers( hdc ); } @@ -100,14 +102,14 @@ wglSwapLayerBuffers( HDC hdc, UINT fuPlanes ) { - return stw_swap_layer_buffers( hdc, fuPlanes ); + return DrvSwapLayerBuffers( hdc, fuPlanes ); } WINGDIAPI PROC APIENTRY wglGetProcAddress( LPCSTR lpszProc ) { - return stw_get_proc_address( lpszProc ); + return DrvGetProcAddress( lpszProc ); } @@ -141,7 +143,7 @@ wglDescribePixelFormat( UINT nBytes, LPPIXELFORMATDESCRIPTOR ppfd ) { - return stw_pixelformat_describe( hdc, iPixelFormat, nBytes, ppfd ); + return DrvDescribePixelFormat( hdc, iPixelFormat, nBytes, ppfd ); } WINGDIAPI int APIENTRY @@ -160,7 +162,7 @@ wglSetPixelFormat( if (ppfd->nSize != sizeof( PIXELFORMATDESCRIPTOR )) return FALSE; - return stw_pixelformat_set( hdc, iPixelFormat ); + return DrvSetPixelFormat( hdc, iPixelFormat ); } @@ -186,7 +188,8 @@ wglShareLists( HGLRC hglrc1, HGLRC hglrc2 ) { - return stw_share_lists( (UINT_PTR)hglrc1, (UINT_PTR)hglrc2);; + return DrvShareLists((DHGLRC)(UINT_PTR)hglrc1, + (DHGLRC)(UINT_PTR)hglrc2); } WINGDIAPI BOOL APIENTRY @@ -264,15 +267,7 @@ wglDescribeLayerPlane( UINT nBytes, LPLAYERPLANEDESCRIPTOR plpd ) { - (void) hdc; - (void) iPixelFormat; - (void) iLayerPlane; - (void) nBytes; - (void) plpd; - - assert( 0 ); - - return FALSE; + return DrvDescribeLayerPlane(hdc, iPixelFormat, iLayerPlane, nBytes, plpd); } WINGDIAPI int APIENTRY @@ -283,15 +278,7 @@ wglSetLayerPaletteEntries( int cEntries, CONST COLORREF *pcr ) { - (void) hdc; - (void) iLayerPlane; - (void) iStart; - (void) cEntries; - (void) pcr; - - assert( 0 ); - - return 0; + return DrvSetLayerPaletteEntries(hdc, iLayerPlane, iStart, cEntries, pcr); } WINGDIAPI int APIENTRY @@ -302,15 +289,7 @@ wglGetLayerPaletteEntries( int cEntries, COLORREF *pcr ) { - (void) hdc; - (void) iLayerPlane; - (void) iStart; - (void) cEntries; - (void) pcr; - - assert( 0 ); - - return 0; + return DrvGetLayerPaletteEntries(hdc, iLayerPlane, iStart, cEntries, pcr); } WINGDIAPI BOOL APIENTRY diff --git a/src/gallium/state_trackers/wgl/wgl/stw_wgl.h b/src/gallium/state_trackers/wgl/stw_wgl.h index a98179944aa..a98179944aa 100644 --- a/src/gallium/state_trackers/wgl/wgl/stw_wgl.h +++ b/src/gallium/state_trackers/wgl/stw_wgl.h diff --git a/src/gallium/state_trackers/wgl/stw_winsys.h b/src/gallium/state_trackers/wgl/stw_winsys.h new file mode 100644 index 00000000000..1ead47d6e63 --- /dev/null +++ b/src/gallium/state_trackers/wgl/stw_winsys.h @@ -0,0 +1,109 @@ +/************************************************************************** + * + * Copyright 2008-2009 Vmware, Inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef STW_WINSYS_H +#define STW_WINSYS_H + +#include <windows.h> /* for HDC */ + +#include "pipe/p_compiler.h" + +struct pipe_screen; +struct pipe_context; +struct pipe_surface; + +struct stw_shared_surface; + +struct stw_winsys +{ + struct pipe_screen * + (*create_screen)( void ); + + struct pipe_context * + (*create_context)( struct pipe_screen *screen ); + + /** + * Present the color buffer to the window associated with the device context. + */ + void + (*present)( struct pipe_screen *screen, + struct pipe_surface *surf, + HDC hDC ); + + /** + * Locally unique identifier (LUID) of the graphics adapter. + * + * @sa GLCBPRESENTBUFFERSDATA::AdapterLuid; + */ + boolean + (*get_adapter_luid)( struct pipe_screen *screen, + LUID *pAdapterLuid ); + + /** + * Open a shared surface (optional). + * + * @sa GLCBPRESENTBUFFERSDATA::hSharedSurface; + */ + struct stw_shared_surface * + (*shared_surface_open)(struct pipe_screen *screen, + HANDLE hSharedSurface); + + /** + * Open a shared surface (optional). + */ + void + (*shared_surface_close)(struct pipe_screen *screen, + struct stw_shared_surface *surface); + + /** + * Compose into a shared (optional). + * + * Blit the color buffer into a shared surface. + * + * @sa GLPRESENTBUFFERSDATA::PresentHistoryToken. + */ + void + (*compose)( struct pipe_screen *screen, + struct pipe_surface *src, + struct stw_shared_surface *dest, + LPCRECT pRect, + ULONGLONG PresentHistoryToken ); +}; + +boolean +stw_init(const struct stw_winsys *stw_winsys); + +boolean +stw_init_thread(void); + +void +stw_cleanup_thread(void); + +void +stw_cleanup(void); + +#endif /* STW_WINSYS_H */ diff --git a/src/gallium/state_trackers/xorg/xorg_composite.c b/src/gallium/state_trackers/xorg/xorg_composite.c index 66ca4cb5905..7037d17e43f 100644 --- a/src/gallium/state_trackers/xorg/xorg_composite.c +++ b/src/gallium/state_trackers/xorg/xorg_composite.c @@ -57,24 +57,6 @@ pixel_to_float4(Pixel pixel, float *color) color[3] = ((float)a) / 255.; } -static INLINE void -render_pixel_to_float4(PictFormatPtr format, - CARD32 pixel, float *color) -{ - CARD32 r, g, b, a; - - debug_assert(format->type == PictTypeDirect); - - r = (pixel >> format->direct.red) & format->direct.redMask; - g = (pixel >> format->direct.green) & format->direct.greenMask; - b = (pixel >> format->direct.blue) & format->direct.blueMask; - a = (pixel >> format->direct.alpha) & format->direct.alphaMask; - color[0] = ((float)r) / ((float)format->direct.redMask); - color[1] = ((float)g) / ((float)format->direct.greenMask); - color[2] = ((float)b) / ((float)format->direct.blueMask); - color[3] = ((float)a) / ((float)format->direct.alphaMask); -} - struct acceleration_info { int op : 16; int with_mask : 1; @@ -150,24 +132,22 @@ setup_vertex_data0(struct exa_context *ctx, int srcX, int srcY, int maskX, int maskY, int dstX, int dstY, int width, int height) { - float vertices[4][2][4]; - /* 1st vertex */ - setup_vertex0(vertices[0], dstX, dstY, + setup_vertex0(ctx->vertices2[0], dstX, dstY, ctx->solid_color); /* 2nd vertex */ - setup_vertex0(vertices[1], dstX + width, dstY, + setup_vertex0(ctx->vertices2[1], dstX + width, dstY, ctx->solid_color); /* 3rd vertex */ - setup_vertex0(vertices[2], dstX + width, dstY + height, + setup_vertex0(ctx->vertices2[2], dstX + width, dstY + height, ctx->solid_color); /* 4th vertex */ - setup_vertex0(vertices[3], dstX, dstY + height, + setup_vertex0(ctx->vertices2[3], dstX, dstY + height, ctx->solid_color); return pipe_user_buffer_create(ctx->pipe->screen, - vertices, - sizeof(vertices)); + ctx->vertices2, + sizeof(ctx->vertices2)); } static INLINE void @@ -189,7 +169,6 @@ setup_vertex_data1(struct exa_context *ctx, int srcX, int srcY, int maskX, int maskY, int dstX, int dstY, int width, int height) { - float vertices[4][2][4]; float s0, t0, s1, t1; struct pipe_texture *src = ctx->bound_textures[0]; @@ -199,21 +178,21 @@ setup_vertex_data1(struct exa_context *ctx, t1 = srcY + height / src->height[0]; /* 1st vertex */ - setup_vertex1(vertices[0], dstX, dstY, + setup_vertex1(ctx->vertices2[0], dstX, dstY, s0, t0); /* 2nd vertex */ - setup_vertex1(vertices[1], dstX + width, dstY, + setup_vertex1(ctx->vertices2[1], dstX + width, dstY, s1, t0); /* 3rd vertex */ - setup_vertex1(vertices[2], dstX + width, dstY + height, + setup_vertex1(ctx->vertices2[2], dstX + width, dstY + height, s1, t1); /* 4th vertex */ - setup_vertex1(vertices[3], dstX, dstY + height, + setup_vertex1(ctx->vertices2[3], dstX, dstY + height, s0, t1); return pipe_user_buffer_create(ctx->pipe->screen, - vertices, - sizeof(vertices)); + ctx->vertices2, + sizeof(ctx->vertices2)); } static struct pipe_buffer * @@ -222,24 +201,22 @@ setup_vertex_data_tex(struct exa_context *ctx, float s0, float t0, float s1, float t1, float z) { - float vertices[4][2][4]; - /* 1st vertex */ - setup_vertex1(vertices[0], x0, y0, + setup_vertex1(ctx->vertices2[0], x0, y0, s0, t0); /* 2nd vertex */ - setup_vertex1(vertices[1], x1, y0, + setup_vertex1(ctx->vertices2[1], x1, y0, s1, t0); /* 3rd vertex */ - setup_vertex1(vertices[2], x1, y1, + setup_vertex1(ctx->vertices2[2], x1, y1, s1, t1); /* 4th vertex */ - setup_vertex1(vertices[3], x0, y1, + setup_vertex1(ctx->vertices2[3], x0, y1, s0, t1); return pipe_user_buffer_create(ctx->pipe->screen, - vertices, - sizeof(vertices)); + ctx->vertices2, + sizeof(ctx->vertices2)); } @@ -269,7 +246,6 @@ setup_vertex_data2(struct exa_context *ctx, int srcX, int srcY, int maskX, int maskY, int dstX, int dstY, int width, int height) { - float vertices[4][3][4]; float st0[4], st1[4]; struct pipe_texture *src = ctx->bound_textures[0]; struct pipe_texture *mask = ctx->bound_textures[0]; @@ -285,21 +261,21 @@ setup_vertex_data2(struct exa_context *ctx, st1[3] = maskY + height / mask->height[0]; /* 1st vertex */ - setup_vertex2(vertices[0], dstX, dstY, + setup_vertex2(ctx->vertices3[0], dstX, dstY, st0[0], st0[1], st1[0], st1[1]); /* 2nd vertex */ - setup_vertex2(vertices[1], dstX + width, dstY, + setup_vertex2(ctx->vertices3[1], dstX + width, dstY, st0[2], st0[1], st1[2], st1[1]); /* 3rd vertex */ - setup_vertex2(vertices[2], dstX + width, dstY + height, + setup_vertex2(ctx->vertices3[2], dstX + width, dstY + height, st0[2], st0[3], st1[2], st1[3]); /* 4th vertex */ - setup_vertex2(vertices[3], dstX, dstY + height, + setup_vertex2(ctx->vertices3[3], dstX, dstY + height, st0[0], st0[3], st1[0], st1[3]); return pipe_user_buffer_create(ctx->pipe->screen, - vertices, - sizeof(vertices)); + ctx->vertices3, + sizeof(ctx->vertices3)); } boolean xorg_composite_accelerated(int op, @@ -311,22 +287,25 @@ boolean xorg_composite_accelerated(int op, unsigned accel_ops_count = sizeof(accelerated_ops)/sizeof(struct acceleration_info); - - /*FIXME: currently accel is disabled */ - return FALSE; - - if (pSrcPicture) { - /* component alpha not supported */ - if (pSrcPicture->componentAlpha) + if (pSrcPicture->pSourcePict) { + /* Gradients not yet supported */ + if (pSrcPicture->pSourcePict->type != SourcePictTypeSolidFill) return FALSE; - /* fills not supported */ - if (pSrcPicture->pSourcePict) + + /* Solid source with mask not yet handled properly */ + if (pMaskPicture) return FALSE; } for (i = 0; i < accel_ops_count; ++i) { if (op == accelerated_ops[i].op) { - if (pMaskPicture && !accelerated_ops[i].with_mask) + /* Check for unsupported component alpha */ + if ((pSrcPicture->componentAlpha && + !accelerated_ops[i].component_alpha) || + (pMaskPicture && + (!accelerated_ops[i].with_mask || + (pMaskPicture->componentAlpha && + !accelerated_ops[i].component_alpha)))) return FALSE; return TRUE; } @@ -359,6 +338,9 @@ bind_framebuffer_state(struct exa_context *exa, struct exa_pixmap_priv *pDst) state.zsbuf = 0; cso_set_framebuffer(exa->cso, &state); + + /* we do fire and forget for the framebuffer, this is the forget part */ + pipe_surface_reference(&surface, NULL); } enum AxisOrientation { @@ -391,7 +373,7 @@ bind_viewport_state(struct exa_context *exa, struct exa_pixmap_priv *pDst) int width = pDst->tex->width[0]; int height = pDst->tex->height[0]; - debug_printf("Bind viewport (%d, %d)\n", width, height); + /*debug_printf("Bind viewport (%d, %d)\n", width, height);*/ set_viewport(exa, width, height, Y0_TOP); } @@ -400,14 +382,9 @@ static void bind_blend_state(struct exa_context *exa, int op, PicturePtr pSrcPicture, PicturePtr pMaskPicture) { - boolean component_alpha = (pSrcPicture) ? - pSrcPicture->componentAlpha : FALSE; struct xorg_composite_blend blend_opt; struct pipe_blend_state blend; - if (component_alpha) { - op = PictOpOver; - } blend_opt = blend_for_op(op); memset(&blend, 0, sizeof(struct pipe_blend_state)); @@ -448,9 +425,9 @@ bind_shaders(struct exa_context *exa, int op, if (pSrcPicture->pSourcePict->type == SourcePictTypeSolidFill) { fs_traits |= FS_SOLID_FILL; vs_traits |= VS_SOLID_FILL; - render_pixel_to_float4(pSrcPicture->pFormat, - pSrcPicture->pSourcePict->solidFill.color, - exa->solid_color); + debug_assert(pSrcPicture->format == PICT_a8r8g8b8); + pixel_to_float4(pSrcPicture->pSourcePict->solidFill.color, + exa->solid_color); exa->has_solid_color = TRUE; } else { debug_assert("!gradients not supported"); @@ -488,6 +465,12 @@ bind_samplers(struct exa_context *exa, int op, memset(&src_sampler, 0, sizeof(struct pipe_sampler_state)); memset(&mask_sampler, 0, sizeof(struct pipe_sampler_state)); + if ((pSrc && exa->pipe->is_texture_referenced(exa->pipe, pSrc->tex, 0, 0) & + PIPE_REFERENCED_FOR_WRITE) || + (pMask && exa->pipe->is_texture_referenced(exa->pipe, pMask->tex, 0, 0) & + PIPE_REFERENCED_FOR_WRITE)) + exa->pipe->flush(exa->pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + if (pSrcPicture && pSrc) { unsigned src_wrap = render_repeat_to_gallium( pSrcPicture->repeatType); @@ -675,7 +658,7 @@ boolean xorg_solid_bind_state(struct exa_context *exa, cso_set_vertex_shader_handle(exa->cso, shader.vs); cso_set_fragment_shader_handle(exa->cso, shader.fs); - return FALSE; + return TRUE; } void xorg_solid(struct exa_context *exa, @@ -684,28 +667,26 @@ void xorg_solid(struct exa_context *exa, { struct pipe_context *pipe = exa->pipe; struct pipe_buffer *buf = 0; - float vertices[4][2][4]; /* 1st vertex */ - setup_vertex0(vertices[0], x0, y0, + setup_vertex0(exa->vertices2[0], x0, y0, exa->solid_color); /* 2nd vertex */ - setup_vertex0(vertices[1], x1, y0, + setup_vertex0(exa->vertices2[1], x1, y0, exa->solid_color); /* 3rd vertex */ - setup_vertex0(vertices[2], x1, y1, + setup_vertex0(exa->vertices2[2], x1, y1, exa->solid_color); /* 4th vertex */ - setup_vertex0(vertices[3], x0, y1, + setup_vertex0(exa->vertices2[3], x0, y1, exa->solid_color); buf = pipe_user_buffer_create(exa->pipe->screen, - vertices, - sizeof(vertices)); + exa->vertices2, + sizeof(exa->vertices2)); if (buf) { - debug_printf("Drawing buf is %p\n", buf); util_draw_vertex_buffer(pipe, buf, 0, PIPE_PRIM_TRIANGLE_FAN, 4, /* verts */ @@ -848,7 +829,7 @@ static void renderer_copy_texture(struct exa_context *exa, t1 = 1; #endif -#if 1 +#if 0 debug_printf("copy texture src=[%f, %f, %f, %f], dst=[%f, %f, %f, %f], tex=[%f, %f, %f, %f]\n", sx1, sy1, sx2, sy2, dx1, dy1, dx2, dy2, s0, t0, s1, t1); @@ -902,6 +883,8 @@ static void renderer_copy_texture(struct exa_context *exa, /* texture */ cso_set_sampler_textures(exa->cso, 1, &src); + bind_rasterizer_state(exa); + /* shaders */ shader = xorg_shaders_get(exa->shaders, VS_COMPOSITE, @@ -1018,7 +1001,9 @@ void xorg_copy_pixmap(struct exa_context *ctx, struct pipe_texture *dst = dst_priv->tex; struct pipe_texture *src = src_priv->tex; - xorg_exa_finish(ctx); + if (ctx->pipe->is_texture_referenced(ctx->pipe, src, 0, 0) & + PIPE_REFERENCED_FOR_WRITE) + ctx->pipe->flush(ctx->pipe, PIPE_FLUSH_RENDER_CACHE, NULL); dst_loc[0] = dx; dst_loc[1] = dy; diff --git a/src/gallium/state_trackers/xorg/xorg_driver.c b/src/gallium/state_trackers/xorg/xorg_driver.c index 643b6b3b9e4..4bc87aa613d 100644 --- a/src/gallium/state_trackers/xorg/xorg_driver.c +++ b/src/gallium/state_trackers/xorg/xorg_driver.c @@ -558,6 +558,7 @@ ScreenInit(int scrnIndex, ScreenPtr pScreen, int argc, char **argv) xf86SetBlackWhitePixels(pScreen); ms->exa = xorg_exa_init(pScrn); + ms->debug_fallback = debug_get_bool_option("XORG_DEBUG_FALLBACK", TRUE); miInitializeBackingStore(pScreen); xf86SetBackingStore(pScreen); @@ -725,8 +726,10 @@ CloseScreen(int scrnIndex, ScreenPtr pScreen) if (ms->exa) xorg_exa_close(pScrn); + if (ms->api->destroy) ms->api->destroy(ms->api); - ms->api = NULL; + ms->api = NULL; + drmClose(ms->fd); ms->fd = -1; diff --git a/src/gallium/state_trackers/xorg/xorg_exa.c b/src/gallium/state_trackers/xorg/xorg_exa.c index dea9f4c2bc3..f7949bafaae 100644 --- a/src/gallium/state_trackers/xorg/xorg_exa.c +++ b/src/gallium/state_trackers/xorg/xorg_exa.c @@ -47,7 +47,9 @@ #include "util/u_rect.h" +#define DEBUG_PRINT 0 #define DEBUG_SOLID 0 +#define DISABLE_ACCEL 0 /* * Helper functions @@ -74,7 +76,7 @@ exa_get_pipe_format(int depth, enum pipe_format *format, int *bbp) assert(*bbp == 16); break; case 8: - *format = PIPE_FORMAT_A8_UNORM; + *format = PIPE_FORMAT_I8_UNORM; assert(*bbp == 8); break; case 4: @@ -200,7 +202,7 @@ ExaPrepareAccess(PixmapPtr pPix, int index) if (!priv->tex) return FALSE; - if (priv->map_count++ == 0) + if (priv->map_count == 0) { if (exa->pipe->is_texture_referenced(exa->pipe, priv->tex, 0, 0) & PIPE_REFERENCED_FOR_WRITE) @@ -208,14 +210,21 @@ ExaPrepareAccess(PixmapPtr pPix, int index) priv->map_transfer = exa->scrn->get_tex_transfer(exa->scrn, priv->tex, 0, 0, 0, +#ifdef EXA_MIXED_PIXMAPS + PIPE_TRANSFER_MAP_DIRECTLY | +#endif PIPE_TRANSFER_READ_WRITE, 0, 0, priv->tex->width[0], priv->tex->height[0]); + if (!priv->map_transfer) + return FALSE; pPix->devPrivate.ptr = exa->scrn->transfer_map(exa->scrn, priv->map_transfer); pPix->devKind = priv->map_transfer->stride; } + priv->map_count++; + return TRUE; } @@ -281,36 +290,36 @@ ExaPrepareSolid(PixmapPtr pPixmap, int alu, Pixel planeMask, Pixel fg) struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pPixmap); struct exa_context *exa = ms->exa; -#if 0 - debug_printf("ExaPrepareSolid - test\n"); +#if DEBUG_PRINT + debug_printf("ExaPrepareSolid(0x%x)\n", fg); #endif - if (!EXA_PM_IS_SOLID(&pPixmap->drawable, planeMask)) - return FALSE; + if (!exa->pipe) + XORG_FALLBACK("solid accle not enabled"); if (!priv || !priv->tex) - return FALSE; + XORG_FALLBACK("solid !priv || !priv->tex"); - if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format, - priv->tex->target, - PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) - return FALSE; + if (!EXA_PM_IS_SOLID(&pPixmap->drawable, planeMask)) + XORG_FALLBACK("solid planeMask is not solid"); if (alu != GXcopy) - return FALSE; - - if (!exa->pipe) - return FALSE; + XORG_FALLBACK("solid not GXcopy"); + if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) { + XORG_FALLBACK("solid bad format %s", pf_name(priv->tex->format)); + } #if DEBUG_SOLID fg = 0xffff0000; #endif -#if 1 - debug_printf(" ExaPrepareSolid(0x%x)\n", fg); -#endif - +#if DISABLE_ACCEL + return FALSE; +#else return xorg_solid_bind_state(exa, priv, fg); +#endif } static void @@ -321,7 +330,9 @@ ExaSolid(PixmapPtr pPixmap, int x0, int y0, int x1, int y1) struct exa_context *exa = ms->exa; struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pPixmap); +#if DEBUG_PRINT debug_printf("\tExaSolid(%d, %d, %d, %d)\n", x0, y0, x1, y1); +#endif #if 0 if (x0 == 0 && y0 == 0 && @@ -375,36 +386,42 @@ ExaPrepareCopy(PixmapPtr pSrcPixmap, PixmapPtr pDstPixmap, int xdir, struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pDstPixmap); struct exa_pixmap_priv *src_priv = exaGetPixmapDriverPrivate(pSrcPixmap); +#if DEBUG_PRINT debug_printf("ExaPrepareCopy\n"); +#endif + if (!exa->pipe) + XORG_FALLBACK("copy accle not enabled"); - if (alu != GXcopy) - return FALSE; + if (!priv || !src_priv) + XORG_FALLBACK("copy !priv || !src_priv"); + + if (!priv->tex || !src_priv->tex) + XORG_FALLBACK("copy !priv->tex || !src_priv->tex"); if (!EXA_PM_IS_SOLID(&pSrcPixmap->drawable, planeMask)) - return FALSE; + XORG_FALLBACK("copy planeMask is not solid"); - if (!priv || !src_priv) - return FALSE; + if (alu != GXcopy) + XORG_FALLBACK("copy alu not GXcopy"); if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format, priv->tex->target, - PIPE_TEXTURE_USAGE_RENDER_TARGET, 0) || - !exa->scrn->is_format_supported(exa->scrn, src_priv->tex->format, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) + XORG_FALLBACK("copy pDst format %s", pf_name(priv->tex->format)); + + if (!exa->scrn->is_format_supported(exa->scrn, src_priv->tex->format, src_priv->tex->target, PIPE_TEXTURE_USAGE_SAMPLER, 0)) - return FALSE; - - if (!priv->tex || !src_priv->tex) - return FALSE; - - if (!exa->pipe) - return FALSE; + XORG_FALLBACK("copy pSrc format %s", pf_name(src_priv->tex->format)); exa->copy.src = src_priv; exa->copy.dst = priv; - /*XXX disabled until some issues with syncing are fixed */ +#if DISABLE_ACCEL return FALSE; +#else + return TRUE; +#endif } static void @@ -416,8 +433,10 @@ ExaCopy(PixmapPtr pDstPixmap, int srcX, int srcY, int dstX, int dstY, struct exa_context *exa = ms->exa; struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pDstPixmap); +#if DEBUG_PRINT debug_printf("\tExaCopy(srcx=%d, srcy=%d, dstX=%d, dstY=%d, w=%d, h=%d)\n", srcX, srcY, dstX, dstY, width, height); +#endif debug_assert(priv == exa->copy.dst); @@ -434,14 +453,55 @@ ExaPrepareComposite(int op, PicturePtr pSrcPicture, ScrnInfoPtr pScrn = xf86Screens[pDst->drawable.pScreen->myNum]; modesettingPtr ms = modesettingPTR(pScrn); struct exa_context *exa = ms->exa; + struct exa_pixmap_priv *priv; +#if DEBUG_PRINT debug_printf("ExaPrepareComposite\n"); +#endif + if (!exa->pipe) + XORG_FALLBACK("comp accle not enabled"); + + priv = exaGetPixmapDriverPrivate(pDst); + if (!priv || !priv->tex) + XORG_FALLBACK("comp pDst %s", !priv ? "!priv" : "!priv->tex"); + + if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) + XORG_FALLBACK("copy pDst format: %s", pf_name(priv->tex->format)); + + if (pSrc) { + priv = exaGetPixmapDriverPrivate(pSrc); + if (!priv || !priv->tex) + XORG_FALLBACK("comp pSrc %s", !priv ? "!priv" : "!priv->tex"); + + if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_SAMPLER, 0)) + XORG_FALLBACK("copy pSrc format: %s", pf_name(priv->tex->format)); + } + + if (pMask) { + priv = exaGetPixmapDriverPrivate(pMask); + if (!priv || !priv->tex) + XORG_FALLBACK("comp pMask %s", !priv ? "!priv" : "!priv->tex"); + + if (!exa->scrn->is_format_supported(exa->scrn, priv->tex->format, + priv->tex->target, + PIPE_TEXTURE_USAGE_SAMPLER, 0)) + XORG_FALLBACK("copy pMask format: %s", pf_name(priv->tex->format)); + } +#if DISABLE_ACCEL + (void) exa; + return FALSE; +#else return xorg_composite_bind_state(exa, op, pSrcPicture, pMaskPicture, pDstPicture, - exaGetPixmapDriverPrivate(pSrc), - exaGetPixmapDriverPrivate(pMask), + pSrc ? exaGetPixmapDriverPrivate(pSrc) : NULL, + pMask ? exaGetPixmapDriverPrivate(pMask) : NULL, exaGetPixmapDriverPrivate(pDst)); +#endif } static void @@ -453,7 +513,9 @@ ExaComposite(PixmapPtr pDst, int srcX, int srcY, int maskX, int maskY, struct exa_context *exa = ms->exa; struct exa_pixmap_priv *priv = exaGetPixmapDriverPrivate(pDst); +#if DEBUG_PRINT debug_printf("\tExaComposite\n"); +#endif xorg_composite(exa, priv, srcX, srcY, maskX, maskY, dstX, dstY, width, height); @@ -464,10 +526,15 @@ ExaCheckComposite(int op, PicturePtr pSrcPicture, PicturePtr pMaskPicture, PicturePtr pDstPicture) { - return xorg_composite_accelerated(op, - pSrcPicture, - pMaskPicture, - pDstPicture); + boolean accelerated = xorg_composite_accelerated(op, + pSrcPicture, + pMaskPicture, + pDstPicture); +#if DEBUG_PRINT + debug_printf("ExaCheckComposite(%d, %p, %p, %p) = %d\n", + op, pSrcPicture, pMaskPicture, pDstPicture, accelerated); +#endif + return accelerated; } static void * @@ -610,67 +677,37 @@ ExaModifyPixmapHeader(PixmapPtr pPixmap, int width, int height, priv->tex->height[0] != height || priv->tex_flags != priv->flags)) { struct pipe_texture *texture = NULL; - -#ifdef DRM_MODE_FEATURE_DIRTYFB - if (priv->flags) -#endif - { - struct pipe_texture template; - - memset(&template, 0, sizeof(template)); - template.target = PIPE_TEXTURE_2D; - exa_get_pipe_format(depth, &template.format, &bitsPerPixel); - pf_get_block(template.format, &template.block); - template.width[0] = width; - template.height[0] = height; - template.depth[0] = 1; - template.last_level = 0; - template.tex_usage = PIPE_TEXTURE_USAGE_RENDER_TARGET | priv->flags; - priv->tex_flags = priv->flags; - texture = exa->scrn->texture_create(exa->scrn, &template); - - if (priv->tex) { - struct pipe_surface *dst_surf; - struct pipe_surface *src_surf; - - dst_surf = exa->scrn->get_tex_surface( - exa->scrn, texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); - src_surf = exa_gpu_surface(exa, priv); - exa->pipe->surface_copy(exa->pipe, dst_surf, 0, 0, src_surf, - 0, 0, min(width, texture->width[0]), - min(height, texture->height[0])); - exa->scrn->tex_surface_destroy(dst_surf); - exa->scrn->tex_surface_destroy(src_surf); - } else if (pPixmap->devPrivate.ptr) { - struct pipe_transfer *transfer; - - if (priv->map_count != 0) - FatalError("doing ExaModifyPixmapHeader on mapped buffer\n"); - - transfer = - exa->scrn->get_tex_transfer(exa->scrn, texture, 0, 0, 0, - PIPE_TRANSFER_WRITE, - 0, 0, width, height); - util_copy_rect(exa->scrn->transfer_map(exa->scrn, transfer), - &texture->block, transfer->stride, 0, 0, - width, height, pPixmap->devPrivate.ptr, - pPixmap->devKind, 0, 0); - exa->scrn->transfer_unmap(exa->scrn, transfer); - exa->scrn->tex_transfer_destroy(transfer); - - xfree(pPixmap->devPrivate.ptr); - pPixmap->devPrivate.ptr = NULL; - } + struct pipe_texture template; + + memset(&template, 0, sizeof(template)); + template.target = PIPE_TEXTURE_2D; + exa_get_pipe_format(depth, &template.format, &bitsPerPixel); + pf_get_block(template.format, &template.block); + template.width[0] = width; + template.height[0] = height; + template.depth[0] = 1; + template.last_level = 0; + template.tex_usage = PIPE_TEXTURE_USAGE_RENDER_TARGET | priv->flags; + priv->tex_flags = priv->flags; + texture = exa->scrn->texture_create(exa->scrn, &template); + + if (priv->tex) { + struct pipe_surface *dst_surf; + struct pipe_surface *src_surf; + + dst_surf = exa->scrn->get_tex_surface( + exa->scrn, texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); + src_surf = exa_gpu_surface(exa, priv); + exa->pipe->surface_copy(exa->pipe, dst_surf, 0, 0, src_surf, + 0, 0, min(width, texture->width[0]), + min(height, texture->height[0])); + exa->scrn->tex_surface_destroy(dst_surf); + exa->scrn->tex_surface_destroy(src_surf); } -#ifdef DRM_MODE_FEATURE_DIRTYFB - else { - xfree(pPixmap->devPrivate.ptr); - pPixmap->devPrivate.ptr = xalloc(pPixmap->drawable.height * - pPixmap->devKind); - } -#endif pipe_texture_reference(&priv->tex, texture); + /* the texture we create has one reference */ + pipe_texture_reference(&texture, NULL); } return TRUE; diff --git a/src/gallium/state_trackers/xorg/xorg_exa.h b/src/gallium/state_trackers/xorg/xorg_exa.h index 43949b04a44..d3f25ca844b 100644 --- a/src/gallium/state_trackers/xorg/xorg_exa.h +++ b/src/gallium/state_trackers/xorg/xorg_exa.h @@ -32,8 +32,11 @@ struct exa_context struct exa_pixmap_priv *src; struct exa_pixmap_priv *dst; } copy; -}; + /* we should combine these two */ + float vertices2[4][2][4]; + float vertices3[4][3][4]; +}; struct exa_pixmap_priv { @@ -47,6 +50,15 @@ struct exa_pixmap_priv unsigned map_count; }; +#define XORG_FALLBACK(s, arg...) \ +do { \ + if (ms->debug_fallback) { \ + xf86DrvMsg(pScrn->scrnIndex, X_INFO, \ + "fallback: " s "\n", ##arg); \ + } \ + return FALSE; \ +} while(0) + struct pipe_surface * exa_gpu_surface(struct exa_context *exa, struct exa_pixmap_priv *priv); diff --git a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c index 2daa5b5628f..bb5a42af37e 100644 --- a/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c +++ b/src/gallium/state_trackers/xorg/xorg_exa_tgsi.c @@ -271,7 +271,7 @@ create_vs(struct pipe_context *pipe, if (has_mask) { src = ureg_DECL_vs_input(ureg, input_slot++); - dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_POSITION, 2); + dst = ureg_DECL_output(ureg, TGSI_SEMANTIC_GENERIC, 2); ureg_MOV(ureg, dst, src); } @@ -330,7 +330,7 @@ create_fs(struct pipe_context *pipe, if (has_mask) { mask_sampler = ureg_DECL_sampler(ureg, 1); mask_pos = ureg_DECL_fs_input(ureg, - TGSI_SEMANTIC_POSITION, + TGSI_SEMANTIC_GENERIC, 1, TGSI_INTERPOLATE_PERSPECTIVE); } @@ -465,7 +465,7 @@ struct xorg_shader xorg_shaders_get(struct xorg_shaders *sc, unsigned vs_traits, unsigned fs_traits) { - struct xorg_shader shader = {0}; + struct xorg_shader shader = { NULL, NULL }; void *vs, *fs; vs = shader_from_cache(sc->exa->pipe, PIPE_SHADER_VERTEX, diff --git a/src/gallium/state_trackers/xorg/xorg_tracker.h b/src/gallium/state_trackers/xorg/xorg_tracker.h index b1ab783a15a..db2f16f63ea 100644 --- a/src/gallium/state_trackers/xorg/xorg_tracker.h +++ b/src/gallium/state_trackers/xorg/xorg_tracker.h @@ -94,6 +94,7 @@ typedef struct _modesettingRec /* exa */ void *exa; Bool noEvict; + Bool debug_fallback; #ifdef DRM_MODE_FEATURE_DIRTYFB DamagePtr damage; @@ -151,5 +152,11 @@ crtc_cursor_destroy(xf86CrtcPtr crtc); void output_init(ScrnInfoPtr pScrn); +/*********************************************************************** + * xorg_xv.c + */ +void +xorg_init_video(ScreenPtr pScreen); + #endif /* _XORG_TRACKER_H_ */ diff --git a/src/gallium/state_trackers/xorg/xorg_xv.c b/src/gallium/state_trackers/xorg/xorg_xv.c new file mode 100644 index 00000000000..88955d47fd3 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xorg_xv.c @@ -0,0 +1,212 @@ +#include "xorg_tracker.h" + +#include <xf86xv.h> +#include <X11/extensions/Xv.h> +#include <fourcc.h> + +/*XXX get these from pipe's texture limits */ +#define IMAGE_MAX_WIDTH 2048 +#define IMAGE_MAX_HEIGHT 2048 + +#define MAKE_ATOM(a) MakeAtom(a, sizeof(a) - 1, TRUE) + +static Atom xvBrightness, xvContrast; + +#define NUM_TEXTURED_ATTRIBUTES 2 +static XF86AttributeRec TexturedAttributes[NUM_TEXTURED_ATTRIBUTES] = { + {XvSettable | XvGettable, -128, 127, "XV_BRIGHTNESS"}, + {XvSettable | XvGettable, 0, 255, "XV_CONTRAST"} +}; + +#define NUM_FORMATS 3 +static XF86VideoFormatRec Formats[NUM_FORMATS] = { + {15, TrueColor}, {16, TrueColor}, {24, TrueColor} +}; + +static XF86VideoEncodingRec DummyEncoding[1] = { + { + 0, + "XV_IMAGE", + IMAGE_MAX_WIDTH, IMAGE_MAX_HEIGHT, + {1, 1} + } +}; + +#define NUM_IMAGES 2 +static XF86ImageRec Images[NUM_IMAGES] = { + XVIMAGE_UYVY, + XVIMAGE_YUY2, +}; + +struct xorg_xv_port_priv { + RegionRec clip; +}; + + +static void +stop_video(ScrnInfoPtr pScrn, pointer data, Bool shutdown) +{ +} + +static int +set_port_attribute(ScrnInfoPtr pScrn, + Atom attribute, INT32 value, pointer data) +{ + return 0; +} + +static int +get_port_attribute(ScrnInfoPtr pScrn, + Atom attribute, INT32 * value, pointer data) +{ + return 0; +} + +static void +query_best_size(ScrnInfoPtr pScrn, + Bool motion, + short vid_w, short vid_h, + short drw_w, short drw_h, + unsigned int *p_w, unsigned int *p_h, pointer data) +{ +} + +static int +put_image(ScrnInfoPtr pScrn, + short src_x, short src_y, + short drw_x, short drw_y, + short src_w, short src_h, + short drw_w, short drw_h, + int id, unsigned char *buf, + short width, short height, + Bool sync, RegionPtr clipBoxes, pointer data, + DrawablePtr pDraw) +{ + return 0; +} + +static int +query_image_attributes(ScrnInfoPtr pScrn, + int id, + unsigned short *w, unsigned short *h, + int *pitches, int *offsets) +{ + return 0; +} + +static struct xorg_xv_port_priv * +port_priv_create(ScreenPtr pScreen) +{ + /*ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];*/ + /*modesettingPtr ms = modesettingPTR(pScrn);*/ + struct xorg_xv_port_priv *priv = NULL; + + priv = calloc(1, sizeof(struct xorg_xv_port_priv)); + + if (!priv) + return NULL; + + REGION_NULL(pScreen, &priv->clip); + + return priv; +} + +static XF86VideoAdaptorPtr +xorg_setup_textured_adapter(ScreenPtr pScreen) +{ + /*ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum];*/ + /*modesettingPtr ms = modesettingPTR(pScrn);*/ + XF86VideoAdaptorPtr adapt; + XF86AttributePtr attrs; + DevUnion *dev_unions; + int nports = 16, i; + int nattributes; + + nattributes = NUM_TEXTURED_ATTRIBUTES; + + adapt = calloc(1, sizeof(XF86VideoAdaptorRec)); + dev_unions = calloc(nports, sizeof(DevUnion)); + attrs = calloc(nattributes, sizeof(XF86AttributeRec)); + if (adapt == NULL || dev_unions == NULL || attrs == NULL) { + free(adapt); + free(dev_unions); + free(attrs); + return NULL; + } + + adapt->type = XvWindowMask | XvInputMask | XvImageMask; + adapt->flags = 0; + adapt->name = "Gallium3D Textured Video"; + adapt->nEncodings = 1; + adapt->pEncodings = DummyEncoding; + adapt->nFormats = NUM_FORMATS; + adapt->pFormats = Formats; + adapt->nPorts = 0; + adapt->pPortPrivates = dev_unions; + adapt->nAttributes = nattributes; + adapt->pAttributes = attrs; + memcpy(attrs, TexturedAttributes, nattributes * sizeof(XF86AttributeRec)); + adapt->nImages = NUM_IMAGES; + adapt->pImages = Images; + adapt->PutVideo = NULL; + adapt->PutStill = NULL; + adapt->GetVideo = NULL; + adapt->GetStill = NULL; + adapt->StopVideo = stop_video; + adapt->SetPortAttribute = set_port_attribute; + adapt->GetPortAttribute = get_port_attribute; + adapt->QueryBestSize = query_best_size; + adapt->PutImage = put_image; + adapt->QueryImageAttributes = query_image_attributes; + + for (i = 0; i < nports; i++) { + struct xorg_xv_port_priv *priv = + port_priv_create(pScreen); + + adapt->pPortPrivates[i].ptr = (pointer) (priv); + adapt->nPorts++; + } + + return adapt; +} + +void +xorg_init_video(ScreenPtr pScreen) +{ + ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum]; + /*modesettingPtr ms = modesettingPTR(pScrn);*/ + XF86VideoAdaptorPtr *adaptors, *new_adaptors = NULL; + XF86VideoAdaptorPtr textured_adapter; + int num_adaptors; + + num_adaptors = xf86XVListGenericAdaptors(pScrn, &adaptors); + new_adaptors = malloc((num_adaptors + 1) * sizeof(XF86VideoAdaptorPtr *)); + if (new_adaptors == NULL) + return; + + memcpy(new_adaptors, adaptors, num_adaptors * sizeof(XF86VideoAdaptorPtr)); + adaptors = new_adaptors; + + /* Add the adaptors supported by our hardware. First, set up the atoms + * that will be used by both output adaptors. + */ + xvBrightness = MAKE_ATOM("XV_BRIGHTNESS"); + xvContrast = MAKE_ATOM("XV_CONTRAST"); + + textured_adapter = xorg_setup_textured_adapter(pScreen); + + debug_assert(textured_adapter); + + if (textured_adapter) { + adaptors[num_adaptors++] = textured_adapter; + } + + if (num_adaptors) { + xf86XVScreenInit(pScreen, adaptors, num_adaptors); + } else { + xf86DrvMsg(pScrn->scrnIndex, X_WARNING, + "Disabling Xv because no adaptors could be initialized.\n"); + } + + free(adaptors); +} diff --git a/src/gallium/state_trackers/xorg/xvmc/Makefile b/src/gallium/state_trackers/xorg/xvmc/Makefile new file mode 100644 index 00000000000..126dc6d58f1 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/Makefile @@ -0,0 +1,16 @@ +TOP = ../../../../.. +include $(TOP)/configs/current + +LIBNAME = xvmctracker + +LIBRARY_INCLUDES = \ + $(shell pkg-config --cflags-only-I xvmc) \ + -I$(TOP)/src/gallium/winsys/g3dvl + +C_SOURCES = block.c \ + surface.c \ + context.c \ + subpicture.c \ + attributes.c + +include ../../../Makefile.template diff --git a/src/gallium/state_trackers/xorg/xvmc/SConscript b/src/gallium/state_trackers/xorg/xvmc/SConscript new file mode 100644 index 00000000000..cb25d68bd80 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/SConscript @@ -0,0 +1,27 @@ +####################################################################### +# SConscript for xvmc state_tracker + +Import('*') + +if 'xorg/xvmc' in env['statetrackers']: + + env = env.Clone() + + env.Append(CPPPATH = [ + '#/src/gallium/include', + '#/src/gallium/auxiliary', + '#/src/gallium/winsys/g3dvl', + ]) + + env.ParseConfig('pkg-config --cflags --libs xvmc') + + st_xvmc = env.ConvenienceLibrary( + target = 'st_xvmc', + source = [ 'block.c', + 'surface.c', + 'context.c', + 'subpicture.c', + 'attributes.c', + ] + ) + Export('st_xvmc') diff --git a/src/gallium/drivers/softpipe/sp_quad_occlusion.c b/src/gallium/state_trackers/xorg/xvmc/attributes.c index dfa7ff3b1d1..79a67838e6e 100644 --- a/src/gallium/drivers/softpipe/sp_quad_occlusion.c +++ b/src/gallium/state_trackers/xorg/xvmc/attributes.c @@ -1,6 +1,6 @@ /************************************************************************** * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * Copyright 2009 Younes Manton. * All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a @@ -25,61 +25,22 @@ * **************************************************************************/ +#include <assert.h> +#include <X11/Xlib.h> +#include <X11/extensions/Xvlib.h> +#include <X11/extensions/XvMClib.h> -/** - * \brief Quad occlusion counter stage - * \author Brian Paul - */ - - -#include "pipe/p_defines.h" -#include "util/u_memory.h" -#include "sp_context.h" -#include "sp_quad.h" -#include "sp_surface.h" -#include "sp_quad_pipe.h" - -static unsigned count_bits( unsigned val ) -{ - unsigned i; - - for (i = 0; val ; val >>= 1) - i += (val & 1); - - return i; -} - -static void -occlusion_count_quad(struct quad_stage *qs, struct quad_header *quad) -{ - struct softpipe_context *softpipe = qs->softpipe; - - softpipe->occlusion_count += count_bits(quad->inout.mask); - - qs->next->run(qs->next, quad); -} - - -static void occlusion_begin(struct quad_stage *qs) +XvAttribute* XvMCQueryAttributes(Display *dpy, XvMCContext *context, int *number) { - qs->next->begin(qs->next); + return NULL; } - -static void occlusion_destroy(struct quad_stage *qs) +Status XvMCSetAttribute(Display *dpy, XvMCContext *context, Atom attribute, int value) { - FREE( qs ); + return BadImplementation; } - -struct quad_stage *sp_quad_occlusion_stage( struct softpipe_context *softpipe ) +Status XvMCGetAttribute(Display *dpy, XvMCContext *context, Atom attribute, int *value) { - struct quad_stage *stage = CALLOC_STRUCT(quad_stage); - - stage->softpipe = softpipe; - stage->begin = occlusion_begin; - stage->run = occlusion_count_quad; - stage->destroy = occlusion_destroy; - - return stage; + return BadImplementation; } diff --git a/src/gallium/state_trackers/xorg/xvmc/block.c b/src/gallium/state_trackers/xorg/xvmc/block.c new file mode 100644 index 00000000000..5102375fcf8 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/block.c @@ -0,0 +1,88 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include <assert.h> +#include <X11/Xlib.h> +#include <X11/extensions/XvMClib.h> +#include <util/u_memory.h> +#include "xvmc_private.h" + +Status XvMCCreateBlocks(Display *dpy, XvMCContext *context, unsigned int num_blocks, XvMCBlockArray *blocks) +{ + assert(dpy); + + if (!context) + return XvMCBadContext; + if (num_blocks == 0) + return BadValue; + + assert(blocks); + + blocks->context_id = context->context_id; + blocks->num_blocks = num_blocks; + blocks->blocks = MALLOC(BLOCK_SIZE_BYTES * num_blocks); + blocks->privData = NULL; + + return Success; +} + +Status XvMCDestroyBlocks(Display *dpy, XvMCBlockArray *blocks) +{ + assert(dpy); + assert(blocks); + FREE(blocks->blocks); + + return Success; +} + +Status XvMCCreateMacroBlocks(Display *dpy, XvMCContext *context, unsigned int num_blocks, XvMCMacroBlockArray *blocks) +{ + assert(dpy); + + if (!context) + return XvMCBadContext; + if (num_blocks == 0) + return BadValue; + + assert(blocks); + + blocks->context_id = context->context_id; + blocks->num_blocks = num_blocks; + blocks->macro_blocks = MALLOC(sizeof(XvMCMacroBlock) * num_blocks); + blocks->privData = NULL; + + return Success; +} + +Status XvMCDestroyMacroBlocks(Display *dpy, XvMCMacroBlockArray *blocks) +{ + assert(dpy); + assert(blocks); + FREE(blocks->macro_blocks); + + return Success; +} diff --git a/src/gallium/state_trackers/xorg/xvmc/context.c b/src/gallium/state_trackers/xorg/xvmc/context.c new file mode 100644 index 00000000000..942692d1bb9 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/context.c @@ -0,0 +1,252 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include <assert.h> +#include <X11/Xlibint.h> +#include <X11/extensions/XvMClib.h> +#include <pipe/p_screen.h> +#include <pipe/p_video_context.h> +#include <pipe/p_video_state.h> +#include <pipe/p_state.h> +#include <vl_winsys.h> +#include <util/u_memory.h> +#include <util/u_debug.h> +#include <vl/vl_csc.h> +#include "xvmc_private.h" + +static Status Validate(Display *dpy, XvPortID port, int surface_type_id, + unsigned int width, unsigned int height, int flags, + bool *found_port, int *screen, int *chroma_format, + int *mc_type, int *surface_flags) +{ + bool found_surface = false; + XvAdaptorInfo *adaptor_info; + unsigned int num_adaptors; + int num_types; + unsigned int max_width, max_height; + Status ret; + + assert(dpy); + assert(found_port); + assert(screen); + assert(chroma_format); + assert(mc_type); + assert(surface_flags); + + *found_port = false; + + for (unsigned int i = 0; i < XScreenCount(dpy); ++i) { + ret = XvQueryAdaptors(dpy, XRootWindow(dpy, i), &num_adaptors, &adaptor_info); + if (ret != Success) + return ret; + + for (unsigned int j = 0; j < num_adaptors && !*found_port; ++j) { + for (unsigned int k = 0; k < adaptor_info[j].num_ports && !*found_port; ++k) { + XvMCSurfaceInfo *surface_info; + + if (adaptor_info[j].base_id + k != port) + continue; + + *found_port = true; + + surface_info = XvMCListSurfaceTypes(dpy, adaptor_info[j].base_id, &num_types); + if (!surface_info) { + XvFreeAdaptorInfo(adaptor_info); + return BadAlloc; + } + + for (unsigned int l = 0; l < num_types && !found_surface; ++l) { + if (surface_info[l].surface_type_id != surface_type_id) + continue; + + found_surface = true; + max_width = surface_info[l].max_width; + max_height = surface_info[l].max_height; + *chroma_format = surface_info[l].chroma_format; + *mc_type = surface_info[l].mc_type; + *surface_flags = surface_info[l].flags; + *screen = i; + } + + XFree(surface_info); + } + } + + XvFreeAdaptorInfo(adaptor_info); + } + + if (!*found_port) + return XvBadPort; + if (!found_surface) + return BadMatch; + if (width > max_width || height > max_height) + return BadValue; + if (flags != XVMC_DIRECT && flags != 0) + return BadValue; + + return Success; +} + +static enum pipe_video_profile ProfileToPipe(int xvmc_profile) +{ + if (xvmc_profile & XVMC_MPEG_1) + assert(0); + if (xvmc_profile & XVMC_MPEG_2) + return PIPE_VIDEO_PROFILE_MPEG2_MAIN; + if (xvmc_profile & XVMC_H263) + assert(0); + if (xvmc_profile & XVMC_MPEG_4) + assert(0); + + assert(0); + + return -1; +} + +static enum pipe_video_chroma_format FormatToPipe(int xvmc_format) +{ + switch (xvmc_format) { + case XVMC_CHROMA_FORMAT_420: + return PIPE_VIDEO_CHROMA_FORMAT_420; + case XVMC_CHROMA_FORMAT_422: + return PIPE_VIDEO_CHROMA_FORMAT_422; + case XVMC_CHROMA_FORMAT_444: + return PIPE_VIDEO_CHROMA_FORMAT_444; + default: + assert(0); + } + + return -1; +} + +Status XvMCCreateContext(Display *dpy, XvPortID port, int surface_type_id, + int width, int height, int flags, XvMCContext *context) +{ + bool found_port; + int scrn; + int chroma_format; + int mc_type; + int surface_flags; + Status ret; + struct pipe_screen *screen; + struct pipe_video_context *vpipe; + XvMCContextPrivate *context_priv; + float csc[16]; + + assert(dpy); + + if (!context) + return XvMCBadContext; + + ret = Validate(dpy, port, surface_type_id, width, height, flags, + &found_port, &scrn, &chroma_format, &mc_type, &surface_flags); + + /* Success and XvBadPort have the same value */ + if (ret != Success || !found_port) + return ret; + + /* XXX: Current limits */ + if (chroma_format != XVMC_CHROMA_FORMAT_420) { + debug_printf("[XvMCg3dvl] Cannot decode requested surface type. Unsupported chroma format.\n"); + return BadImplementation; + } + if (mc_type != (XVMC_MOCOMP | XVMC_MPEG_2)) { + debug_printf("[XvMCg3dvl] Cannot decode requested surface type. Non-MPEG2/Mocomp acceleration unsupported.\n"); + return BadImplementation; + } + if (!(surface_flags & XVMC_INTRA_UNSIGNED)) { + debug_printf("[XvMCg3dvl] Cannot decode requested surface type. Signed intra unsupported.\n"); + return BadImplementation; + } + + context_priv = CALLOC(1, sizeof(XvMCContextPrivate)); + if (!context_priv) + return BadAlloc; + + /* TODO: Reuse screen if process creates another context */ + screen = vl_screen_create(dpy, scrn); + + if (!screen) { + FREE(context_priv); + return BadAlloc; + } + + vpipe = vl_video_create(screen, ProfileToPipe(mc_type), + FormatToPipe(chroma_format), width, height); + + if (!vpipe) { + screen->destroy(screen); + FREE(context_priv); + return BadAlloc; + } + + /* TODO: Define some Xv attribs to allow users to specify color standard, procamp */ + vl_csc_get_matrix + ( + debug_get_bool_option("G3DVL_NO_CSC", FALSE) ? + VL_CSC_COLOR_STANDARD_IDENTITY : VL_CSC_COLOR_STANDARD_BT_601, + NULL, true, csc + ); + vpipe->set_csc_matrix(vpipe, csc); + + context_priv->vpipe = vpipe; + + context->context_id = XAllocID(dpy); + context->surface_type_id = surface_type_id; + context->width = width; + context->height = height; + context->flags = flags; + context->port = port; + context->privData = context_priv; + + SyncHandle(); + + return Success; +} + +Status XvMCDestroyContext(Display *dpy, XvMCContext *context) +{ + struct pipe_screen *screen; + struct pipe_video_context *vpipe; + XvMCContextPrivate *context_priv; + + assert(dpy); + + if (!context || !context->privData) + return XvMCBadContext; + + context_priv = context->privData; + vpipe = context_priv->vpipe; + pipe_surface_reference(&context_priv->backbuffer, NULL); + screen = vpipe->screen; + vpipe->destroy(vpipe); + screen->destroy(screen); + FREE(context_priv); + context->privData = NULL; + + return Success; +} diff --git a/src/gallium/state_trackers/xorg/xvmc/subpicture.c b/src/gallium/state_trackers/xorg/xvmc/subpicture.c new file mode 100644 index 00000000000..69898d5fcd3 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/subpicture.c @@ -0,0 +1,195 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include <assert.h> +#include <X11/Xlibint.h> +#include <X11/extensions/XvMClib.h> + +Status XvMCCreateSubpicture(Display *dpy, XvMCContext *context, XvMCSubpicture *subpicture, + unsigned short width, unsigned short height, int xvimage_id) +{ + assert(dpy); + + if (!context) + return XvMCBadContext; + + assert(subpicture); + + /*if (width > || height > ) + return BadValue;*/ + + /*if (xvimage_id != ) + return BadMatch;*/ + + subpicture->subpicture_id = XAllocID(dpy); + subpicture->context_id = context->context_id; + subpicture->xvimage_id = xvimage_id; + subpicture->width = width; + subpicture->height = height; + subpicture->num_palette_entries = 0; + subpicture->entry_bytes = 0; + subpicture->component_order[0] = 0; + subpicture->component_order[1] = 0; + subpicture->component_order[2] = 0; + subpicture->component_order[3] = 0; + /* TODO: subpicture->privData = ;*/ + + SyncHandle(); + + return Success; +} + +Status XvMCClearSubpicture(Display *dpy, XvMCSubpicture *subpicture, short x, short y, + unsigned short width, unsigned short height, unsigned int color) +{ + assert(dpy); + + if (!subpicture) + return XvMCBadSubpicture; + + /* TODO: Assert clear rect is within bounds? Or clip? */ + + return Success; +} + +Status XvMCCompositeSubpicture(Display *dpy, XvMCSubpicture *subpicture, XvImage *image, + short srcx, short srcy, unsigned short width, unsigned short height, + short dstx, short dsty) +{ + assert(dpy); + + if (!subpicture) + return XvMCBadSubpicture; + + assert(image); + + if (subpicture->xvimage_id != image->id) + return BadMatch; + + /* TODO: Assert rects are within bounds? Or clip? */ + + return Success; +} + +Status XvMCDestroySubpicture(Display *dpy, XvMCSubpicture *subpicture) +{ + assert(dpy); + + if (!subpicture) + return XvMCBadSubpicture; + + return BadImplementation; +} + +Status XvMCSetSubpicturePalette(Display *dpy, XvMCSubpicture *subpicture, unsigned char *palette) +{ + assert(dpy); + + if (!subpicture) + return XvMCBadSubpicture; + + assert(palette); + + /* We don't support paletted subpictures */ + return BadMatch; +} + +Status XvMCBlendSubpicture(Display *dpy, XvMCSurface *target_surface, XvMCSubpicture *subpicture, + short subx, short suby, unsigned short subw, unsigned short subh, + short surfx, short surfy, unsigned short surfw, unsigned short surfh) +{ + assert(dpy); + + if (!target_surface) + return XvMCBadSurface; + + if (!subpicture) + return XvMCBadSubpicture; + + if (target_surface->context_id != subpicture->context_id) + return BadMatch; + + /* TODO: Assert rects are within bounds? Or clip? */ + return Success; +} + +Status XvMCBlendSubpicture2(Display *dpy, XvMCSurface *source_surface, XvMCSurface *target_surface, + XvMCSubpicture *subpicture, short subx, short suby, unsigned short subw, unsigned short subh, + short surfx, short surfy, unsigned short surfw, unsigned short surfh) +{ + assert(dpy); + + if (!source_surface || !target_surface) + return XvMCBadSurface; + + if (!subpicture) + return XvMCBadSubpicture; + + if (source_surface->context_id != subpicture->context_id) + return BadMatch; + + if (source_surface->context_id != subpicture->context_id) + return BadMatch; + + /* TODO: Assert rects are within bounds? Or clip? */ + return Success; +} + +Status XvMCSyncSubpicture(Display *dpy, XvMCSubpicture *subpicture) +{ + assert(dpy); + + if (!subpicture) + return XvMCBadSubpicture; + + return Success; +} + +Status XvMCFlushSubpicture(Display *dpy, XvMCSubpicture *subpicture) +{ + assert(dpy); + + if (!subpicture) + return XvMCBadSubpicture; + + return Success; +} + +Status XvMCGetSubpictureStatus(Display *dpy, XvMCSubpicture *subpicture, int *status) +{ + assert(dpy); + + if (!subpicture) + return XvMCBadSubpicture; + + assert(status); + + /* TODO */ + *status = 0; + + return Success; +} diff --git a/src/gallium/state_trackers/xorg/xvmc/surface.c b/src/gallium/state_trackers/xorg/xvmc/surface.c new file mode 100644 index 00000000000..bf9038f356e --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/surface.c @@ -0,0 +1,409 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include <assert.h> +#include <X11/Xlibint.h> +#include <pipe/p_video_context.h> +#include <pipe/p_video_state.h> +#include <pipe/p_state.h> +#include <util/u_memory.h> +#include "xvmc_private.h" + +static enum pipe_mpeg12_macroblock_type TypeToPipe(int xvmc_mb_type) +{ + if (xvmc_mb_type & XVMC_MB_TYPE_INTRA) + return PIPE_MPEG12_MACROBLOCK_TYPE_INTRA; + if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_FORWARD) + return PIPE_MPEG12_MACROBLOCK_TYPE_FWD; + if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_BACKWARD) + return PIPE_MPEG12_MACROBLOCK_TYPE_BKWD; + if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) + return PIPE_MPEG12_MACROBLOCK_TYPE_BI; + + assert(0); + + return -1; +} + +static enum pipe_mpeg12_picture_type PictureToPipe(int xvmc_pic) +{ + switch (xvmc_pic) { + case XVMC_TOP_FIELD: + return PIPE_MPEG12_PICTURE_TYPE_FIELD_TOP; + case XVMC_BOTTOM_FIELD: + return PIPE_MPEG12_PICTURE_TYPE_FIELD_BOTTOM; + case XVMC_FRAME_PICTURE: + return PIPE_MPEG12_PICTURE_TYPE_FRAME; + default: + assert(0); + } + + return -1; +} + +static enum pipe_mpeg12_motion_type MotionToPipe(int xvmc_motion_type, int xvmc_dct_type) +{ + switch (xvmc_motion_type) { + case XVMC_PREDICTION_FRAME: + return xvmc_dct_type == XVMC_DCT_TYPE_FIELD ? + PIPE_MPEG12_MOTION_TYPE_16x8 : PIPE_MPEG12_MOTION_TYPE_FRAME; + case XVMC_PREDICTION_FIELD: + return PIPE_MPEG12_MOTION_TYPE_FIELD; + case XVMC_PREDICTION_DUAL_PRIME: + return PIPE_MPEG12_MOTION_TYPE_DUALPRIME; + default: + assert(0); + } + + return -1; +} + +static bool +CreateOrResizeBackBuffer(struct pipe_video_context *vpipe, unsigned int width, unsigned int height, + struct pipe_surface **backbuffer) +{ + struct pipe_texture template; + struct pipe_texture *tex; + + assert(vpipe); + + if (*backbuffer) { + if ((*backbuffer)->width != width || (*backbuffer)->height != height) + pipe_surface_reference(backbuffer, NULL); + else + return true; + } + + memset(&template, 0, sizeof(struct pipe_texture)); + template.target = PIPE_TEXTURE_2D; + /* XXX: Needs to match the drawable's format? */ + template.format = PIPE_FORMAT_X8R8G8B8_UNORM; + template.last_level = 0; + template.width[0] = width; + template.height[0] = height; + template.depth[0] = 1; + pf_get_block(template.format, &template.block); + template.tex_usage = PIPE_TEXTURE_USAGE_DISPLAY_TARGET; + + tex = vpipe->screen->texture_create(vpipe->screen, &template); + if (!tex) + return false; + + *backbuffer = vpipe->screen->get_tex_surface(vpipe->screen, tex, 0, 0, 0, + PIPE_BUFFER_USAGE_GPU_READ | + PIPE_BUFFER_USAGE_GPU_WRITE); + pipe_texture_reference(&tex, NULL); + + if (!*backbuffer) + return false; + + /* Clear the backbuffer in case the video doesn't cover the whole window */ + /* FIXME: Need to clear every time a frame moves and leaves dirty rects */ + vpipe->clear_surface(vpipe, 0, 0, width, height, 0, *backbuffer); + + return true; +} + +static void +MacroBlocksToPipe(const XvMCMacroBlockArray *xvmc_macroblocks, + const XvMCBlockArray *xvmc_blocks, + unsigned int first_macroblock, + unsigned int num_macroblocks, + struct pipe_mpeg12_macroblock *pipe_macroblocks) +{ + unsigned int i, j, k, l; + XvMCMacroBlock *xvmc_mb; + + assert(xvmc_macroblocks); + assert(xvmc_blocks); + assert(pipe_macroblocks); + assert(num_macroblocks); + + xvmc_mb = xvmc_macroblocks->macro_blocks + first_macroblock; + + for (i = 0; i < num_macroblocks; ++i) { + pipe_macroblocks->base.codec = PIPE_VIDEO_CODEC_MPEG12; + pipe_macroblocks->mbx = xvmc_mb->x; + pipe_macroblocks->mby = xvmc_mb->y; + pipe_macroblocks->mb_type = TypeToPipe(xvmc_mb->macroblock_type); + if (pipe_macroblocks->mb_type != PIPE_MPEG12_MACROBLOCK_TYPE_INTRA) + pipe_macroblocks->mo_type = MotionToPipe(xvmc_mb->motion_type, xvmc_mb->dct_type); + /* Get rid of Valgrind 'undefined' warnings */ + else + pipe_macroblocks->mo_type = -1; + pipe_macroblocks->dct_type = xvmc_mb->dct_type == XVMC_DCT_TYPE_FIELD ? + PIPE_MPEG12_DCT_TYPE_FIELD : PIPE_MPEG12_DCT_TYPE_FRAME; + + for (j = 0; j < 2; ++j) + for (k = 0; k < 2; ++k) + for (l = 0; l < 2; ++l) + pipe_macroblocks->pmv[j][k][l] = xvmc_mb->PMV[j][k][l]; + + pipe_macroblocks->cbp = xvmc_mb->coded_block_pattern; + pipe_macroblocks->blocks = xvmc_blocks->blocks + xvmc_mb->index * BLOCK_SIZE_SAMPLES; + + ++pipe_macroblocks; + ++xvmc_mb; + } +} + +Status XvMCCreateSurface(Display *dpy, XvMCContext *context, XvMCSurface *surface) +{ + XvMCContextPrivate *context_priv; + struct pipe_video_context *vpipe; + XvMCSurfacePrivate *surface_priv; + struct pipe_video_surface *vsfc; + + assert(dpy); + + if (!context) + return XvMCBadContext; + if (!surface) + return XvMCBadSurface; + + context_priv = context->privData; + vpipe = context_priv->vpipe; + + surface_priv = CALLOC(1, sizeof(XvMCSurfacePrivate)); + if (!surface_priv) + return BadAlloc; + + vsfc = vpipe->screen->video_surface_create(vpipe->screen, vpipe->chroma_format, + vpipe->width, vpipe->height); + if (!vsfc) { + FREE(surface_priv); + return BadAlloc; + } + + surface_priv->pipe_vsfc = vsfc; + surface_priv->context = context; + + surface->surface_id = XAllocID(dpy); + surface->context_id = context->context_id; + surface->surface_type_id = context->surface_type_id; + surface->width = context->width; + surface->height = context->height; + surface->privData = surface_priv; + + SyncHandle(); + + return Success; +} + +Status XvMCRenderSurface(Display *dpy, XvMCContext *context, unsigned int picture_structure, + XvMCSurface *target_surface, XvMCSurface *past_surface, XvMCSurface *future_surface, + unsigned int flags, unsigned int num_macroblocks, unsigned int first_macroblock, + XvMCMacroBlockArray *macroblocks, XvMCBlockArray *blocks +) +{ + struct pipe_video_context *vpipe; + struct pipe_surface *t_vsfc; + struct pipe_surface *p_vsfc; + struct pipe_surface *f_vsfc; + XvMCContextPrivate *context_priv; + XvMCSurfacePrivate *target_surface_priv; + XvMCSurfacePrivate *past_surface_priv; + XvMCSurfacePrivate *future_surface_priv; + struct pipe_mpeg12_macroblock pipe_macroblocks[num_macroblocks]; + + assert(dpy); + + if (!context || !context->privData) + return XvMCBadContext; + if (!target_surface || !target_surface->privData) + return XvMCBadSurface; + + if (picture_structure != XVMC_TOP_FIELD && + picture_structure != XVMC_BOTTOM_FIELD && + picture_structure != XVMC_FRAME_PICTURE) + return BadValue; + /* Bkwd pred equivalent to fwd (past && !future) */ + if (future_surface && !past_surface) + return BadMatch; + + assert(context->context_id == target_surface->context_id); + assert(!past_surface || context->context_id == past_surface->context_id); + assert(!future_surface || context->context_id == future_surface->context_id); + + assert(macroblocks); + assert(blocks); + + assert(macroblocks->context_id == context->context_id); + assert(blocks->context_id == context->context_id); + + assert(flags == 0 || flags == XVMC_SECOND_FIELD); + + target_surface_priv = target_surface->privData; + past_surface_priv = past_surface ? past_surface->privData : NULL; + future_surface_priv = future_surface ? future_surface->privData : NULL; + + assert(target_surface_priv->context == context); + assert(!past_surface || past_surface_priv->context == context); + assert(!future_surface || future_surface_priv->context == context); + + context_priv = context->privData; + vpipe = context_priv->vpipe; + + t_vsfc = target_surface_priv->pipe_vsfc; + p_vsfc = past_surface ? past_surface_priv->pipe_vsfc : NULL; + f_vsfc = future_surface ? future_surface_priv->pipe_vsfc : NULL; + + MacroBlocksToPipe(macroblocks, blocks, first_macroblock, + num_macroblocks, pipe_macroblocks); + + vpipe->set_decode_target(vpipe, t_vsfc); + vpipe->decode_macroblocks(vpipe, p_vsfc, f_vsfc, num_macroblocks, + &pipe_macroblocks->base, target_surface_priv->render_fence); + + return Success; +} + +Status XvMCFlushSurface(Display *dpy, XvMCSurface *surface) +{ + assert(dpy); + + if (!surface) + return XvMCBadSurface; + + return Success; +} + +Status XvMCSyncSurface(Display *dpy, XvMCSurface *surface) +{ + assert(dpy); + + if (!surface) + return XvMCBadSurface; + + return Success; +} + +Status XvMCPutSurface(Display *dpy, XvMCSurface *surface, Drawable drawable, + short srcx, short srcy, unsigned short srcw, unsigned short srch, + short destx, short desty, unsigned short destw, unsigned short desth, + int flags) +{ + Window root; + int x, y; + unsigned int width, height; + unsigned int border_width; + unsigned int depth; + struct pipe_video_context *vpipe; + XvMCSurfacePrivate *surface_priv; + XvMCContextPrivate *context_priv; + XvMCContext *context; + struct pipe_video_rect src_rect = {srcx, srcy, srcw, srch}; + struct pipe_video_rect dst_rect = {destx, desty, destw, desth}; + + assert(dpy); + + if (!surface || !surface->privData) + return XvMCBadSurface; + + if (XGetGeometry(dpy, drawable, &root, &x, &y, &width, &height, &border_width, &depth) == BadDrawable) + return BadDrawable; + + assert(flags == XVMC_TOP_FIELD || flags == XVMC_BOTTOM_FIELD || flags == XVMC_FRAME_PICTURE); + assert(srcx + srcw - 1 < surface->width); + assert(srcy + srch - 1 < surface->height); + /* + * Some apps (mplayer) hit these asserts because they call + * this function after the window has been resized by the WM + * but before they've handled the corresponding XEvent and + * know about the new dimensions. The output should be clipped + * until the app updates destw and desth. + */ + /* + assert(destx + destw - 1 < width); + assert(desty + desth - 1 < height); + */ + + surface_priv = surface->privData; + context = surface_priv->context; + context_priv = context->privData; + vpipe = context_priv->vpipe; + + if (!CreateOrResizeBackBuffer(vpipe, width, height, &context_priv->backbuffer)) + return BadAlloc; + + vpipe->render_picture(vpipe, surface_priv->pipe_vsfc, PictureToPipe(flags), &src_rect, + context_priv->backbuffer, &dst_rect, surface_priv->disp_fence); + + vl_video_bind_drawable(vpipe, drawable); + + vpipe->screen->flush_frontbuffer + ( + vpipe->screen, + context_priv->backbuffer, + vpipe->priv + ); + + return Success; +} + +Status XvMCGetSurfaceStatus(Display *dpy, XvMCSurface *surface, int *status) +{ + assert(dpy); + + if (!surface) + return XvMCBadSurface; + + assert(status); + + *status = 0; + + return Success; +} + +Status XvMCDestroySurface(Display *dpy, XvMCSurface *surface) +{ + XvMCSurfacePrivate *surface_priv; + + assert(dpy); + + if (!surface || !surface->privData) + return XvMCBadSurface; + + surface_priv = surface->privData; + pipe_video_surface_reference(&surface_priv->pipe_vsfc, NULL); + FREE(surface_priv); + surface->privData = NULL; + + return Success; +} + +Status XvMCHideSurface(Display *dpy, XvMCSurface *surface) +{ + assert(dpy); + + if (!surface || !surface->privData) + return XvMCBadSurface; + + /* No op, only for overlaid rendering */ + + return Success; +} diff --git a/src/xvmc/tests/.gitignore b/src/gallium/state_trackers/xorg/xvmc/tests/.gitignore index e1d2f9023df..e1d2f9023df 100644 --- a/src/xvmc/tests/.gitignore +++ b/src/gallium/state_trackers/xorg/xvmc/tests/.gitignore diff --git a/src/gallium/state_trackers/xorg/xvmc/tests/Makefile b/src/gallium/state_trackers/xorg/xvmc/tests/Makefile new file mode 100644 index 00000000000..c875dd76058 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/tests/Makefile @@ -0,0 +1,28 @@ +TOP = ../../../../../.. +include $(TOP)/configs/current + +LIBS = -lXvMCW -lXvMC -lXv -lX11 + +############################################# + +.PHONY: default clean + +default: test_context test_surface test_blocks test_rendering xvmc_bench + +test_context: test_context.o testlib.o + $(CC) $(LDFLAGS) -o $@ $^ $(LIBS) + +test_surface: test_surface.o testlib.o + $(CC) $(LDFLAGS) -o $@ $^ $(LIBS) + +test_blocks: test_blocks.o testlib.o + $(CC) $(LDFLAGS) -o $@ $^ $(LIBS) + +test_rendering: test_rendering.o testlib.o + $(CC) $(LDFLAGS) -o $@ $^ $(LIBS) + +xvmc_bench: xvmc_bench.o testlib.o + $(CC) $(LDFLAGS) -o $@ $^ $(LIBS) + +clean: + $(RM) -rf *.o test_context test_surface test_blocks test_rendering xvmc_bench diff --git a/src/xvmc/tests/test_blocks.c b/src/gallium/state_trackers/xorg/xvmc/tests/test_blocks.c index dc80adfa652..994e3ca4d14 100644 --- a/src/xvmc/tests/test_blocks.c +++ b/src/gallium/state_trackers/xorg/xvmc/tests/test_blocks.c @@ -1,3 +1,30 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + #include <assert.h> #include <error.h> #include "testlib.h" diff --git a/src/xvmc/tests/test_context.c b/src/gallium/state_trackers/xorg/xvmc/tests/test_context.c index 53f7449cd0b..3da957c9330 100644 --- a/src/xvmc/tests/test_context.c +++ b/src/gallium/state_trackers/xorg/xvmc/tests/test_context.c @@ -1,3 +1,30 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + #include <assert.h> #include <error.h> #include "testlib.h" diff --git a/src/xvmc/tests/test_rendering.c b/src/gallium/state_trackers/xorg/xvmc/tests/test_rendering.c index 1e7467a3aa6..6058783a798 100644 --- a/src/xvmc/tests/test_rendering.c +++ b/src/gallium/state_trackers/xorg/xvmc/tests/test_rendering.c @@ -1,3 +1,30 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + #include <assert.h> #include <stdio.h> #include <string.h> @@ -23,6 +50,9 @@ #define DEFAULT_OUTPUT_HEIGHT INPUT_HEIGHT #define DEFAULT_ACCEPTABLE_ERR 0.01 +void ParseArgs(int argc, char **argv, unsigned int *output_width, unsigned int *output_height, double *acceptable_error, int *prompt); +void Gradient(short *block, unsigned int start, unsigned int stop, int horizontal); + void ParseArgs(int argc, char **argv, unsigned int *output_width, unsigned int *output_height, double *acceptable_error, int *prompt) { int fail = 0; diff --git a/src/xvmc/tests/test_surface.c b/src/gallium/state_trackers/xorg/xvmc/tests/test_surface.c index 25ebdcc4fcf..b65eb265c0a 100644 --- a/src/xvmc/tests/test_surface.c +++ b/src/gallium/state_trackers/xorg/xvmc/tests/test_surface.c @@ -1,3 +1,30 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + #include <assert.h> #include <error.h> #include "testlib.h" @@ -6,7 +33,7 @@ int main(int argc, char **argv) { const unsigned int width = 16, height = 16; const unsigned int mc_types[2] = {XVMC_MOCOMP | XVMC_MPEG_2, XVMC_IDCT | XVMC_MPEG_2}; - + Display *display; XvPortID port_num; int surface_type_id; @@ -14,9 +41,9 @@ int main(int argc, char **argv) int colorkey; XvMCContext context; XvMCSurface surface = {0}; - + display = XOpenDisplay(NULL); - + if (!GetPort ( display, @@ -34,15 +61,15 @@ int main(int argc, char **argv) XCloseDisplay(display); error(1, 0, "Error, unable to find a good port.\n"); } - + if (is_overlay) { Atom xv_colorkey = XInternAtom(display, "XV_COLORKEY", 0); XvGetPortAttribute(display, port_num, xv_colorkey, &colorkey); } - + assert(XvMCCreateContext(display, port_num, surface_type_id, width, height, XVMC_DIRECT, &context) == Success); - + /* Test NULL context */ assert(XvMCCreateSurface(display, NULL, &surface) == XvMCBadContext); /* Test NULL surface */ @@ -61,12 +88,11 @@ int main(int argc, char **argv) assert(XvMCDestroySurface(display, &surface) == Success); /* Test NULL surface */ assert(XvMCDestroySurface(display, NULL) == XvMCBadSurface); - + assert(XvMCDestroyContext(display, &context) == Success); - + XvUngrabPort(display, port_num, CurrentTime); XCloseDisplay(display); - + return 0; } - diff --git a/src/xvmc/tests/testlib.c b/src/gallium/state_trackers/xorg/xvmc/tests/testlib.c index 59a03ca813f..142c09bb590 100644 --- a/src/xvmc/tests/testlib.c +++ b/src/gallium/state_trackers/xorg/xvmc/tests/testlib.c @@ -1,3 +1,30 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + #include "testlib.h" #include <stdio.h> diff --git a/src/gallium/state_trackers/xorg/xvmc/tests/testlib.h b/src/gallium/state_trackers/xorg/xvmc/tests/testlib.h new file mode 100644 index 00000000000..0438e52928b --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/tests/testlib.h @@ -0,0 +1,69 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef testlib_h +#define testlib_h + +/* +#define TEST(pred, doc) test(pred, #pred, doc, __FILE__, __LINE__) + +void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line); +*/ + +#include <sys/time.h> +#include <X11/Xlib.h> +#include <X11/extensions/XvMClib.h> + +/* + * display: IN A valid X display + * width, height: IN Surface size that the port must display + * chroma_format: IN Chroma format that the port must display + * mc_types, num_mc_types: IN List of MC types that the port must support, first port that matches the first mc_type will be returned + * port_id: OUT Your port's ID + * surface_type_id: OUT Your port's surface ID + * is_overlay: OUT If 1, port uses overlay surfaces, you need to set a colorkey + * intra_unsigned: OUT If 1, port uses unsigned values for intra-coded blocks + */ +int GetPort +( + Display *display, + unsigned int width, + unsigned int height, + unsigned int chroma_format, + const unsigned int *mc_types, + unsigned int num_mc_types, + XvPortID *port_id, + int *surface_type_id, + unsigned int *is_overlay, + unsigned int *intra_unsigned +); + +unsigned int align(unsigned int value, unsigned int alignment); + +int timeval_subtract(struct timeval *result, struct timeval *x, struct timeval *y); + +#endif diff --git a/src/xvmc/tests/xvmc_bench.c b/src/gallium/state_trackers/xorg/xvmc/tests/xvmc_bench.c index d5a39ecf17d..bf94d856234 100644 --- a/src/xvmc/tests/xvmc_bench.c +++ b/src/gallium/state_trackers/xorg/xvmc/tests/xvmc_bench.c @@ -1,3 +1,30 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + #include <assert.h> #include <stdio.h> #include <string.h> @@ -32,6 +59,8 @@ struct Config unsigned int reps; }; +void ParseArgs(int argc, char **argv, struct Config *config); + void ParseArgs(int argc, char **argv, struct Config *config) { int fail = 0; diff --git a/src/gallium/state_trackers/xorg/xvmc/xvmc_private.h b/src/gallium/state_trackers/xorg/xvmc/xvmc_private.h new file mode 100644 index 00000000000..42337631ca1 --- /dev/null +++ b/src/gallium/state_trackers/xorg/xvmc/xvmc_private.h @@ -0,0 +1,58 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#ifndef xvmc_private_h +#define xvmc_private_h + +#include <X11/Xlib.h> +#include <X11/extensions/XvMClib.h> + +#define BLOCK_SIZE_SAMPLES 64 +#define BLOCK_SIZE_BYTES (BLOCK_SIZE_SAMPLES * 2) + +struct pipe_video_context; +struct pipe_surface; +struct pipe_fence_handle; + +typedef struct +{ + struct pipe_video_context *vpipe; + struct pipe_surface *backbuffer; +} XvMCContextPrivate; + +typedef struct +{ + struct pipe_video_surface *pipe_vsfc; + struct pipe_fence_handle *render_fence; + struct pipe_fence_handle *disp_fence; + + /* Some XvMC functions take a surface but not a context, + so we keep track of which context each surface belongs to. */ + XvMCContext *context; +} XvMCSurfacePrivate; + +#endif /* xvmc_private_h */ diff --git a/src/gallium/winsys/drm/intel/gem/intel_drm_api.c b/src/gallium/winsys/drm/intel/gem/intel_drm_api.c index 4c5a1d2ea89..8b647a769b5 100644 --- a/src/gallium/winsys/drm/intel/gem/intel_drm_api.c +++ b/src/gallium/winsys/drm/intel/gem/intel_drm_api.c @@ -7,6 +7,7 @@ #include "i915simple/i915_context.h" #include "i915simple/i915_screen.h" +#include "trace/tr_drm.h" /* * Helper functions @@ -40,6 +41,7 @@ intel_drm_buffer_from_handle(struct intel_drm_winsys *idws, const char* name, unsigned handle) { struct intel_drm_buffer *buf = CALLOC_STRUCT(intel_drm_buffer); + uint32_t tile = 0, swizzle = 0; if (!buf) return NULL; @@ -52,6 +54,10 @@ intel_drm_buffer_from_handle(struct intel_drm_winsys *idws, if (!buf->bo) goto err; + drm_intel_bo_get_tiling(buf->bo, &tile, &swizzle); + if (tile != INTEL_TILE_NONE) + buf->map_gtt = TRUE; + return (struct intel_buffer *)buf; err: @@ -166,6 +172,7 @@ intel_drm_create_screen(struct drm_api *api, int drmFD, idws->base.destroy = intel_drm_winsys_destroy; idws->pools.gem = drm_intel_bufmgr_gem_init(idws->fd, idws->max_batch_size); + drm_intel_bufmgr_gem_enable_reuse(idws->pools.gem); idws->softpipe = FALSE; idws->dump_cmd = debug_get_bool_option("INTEL_DUMP_CMD", FALSE); @@ -198,5 +205,5 @@ struct drm_api intel_drm_api = struct drm_api * drm_api_create() { - return &intel_drm_api; + return trace_drm_create(&intel_drm_api); } diff --git a/src/gallium/winsys/drm/intel/gem/intel_drm_buffer.c b/src/gallium/winsys/drm/intel/gem/intel_drm_buffer.c index 0030f915a36..327e19fcd68 100644 --- a/src/gallium/winsys/drm/intel/gem/intel_drm_buffer.c +++ b/src/gallium/winsys/drm/intel/gem/intel_drm_buffer.c @@ -58,11 +58,17 @@ intel_drm_buffer_set_fence_reg(struct intel_winsys *iws, unsigned stride, enum intel_buffer_tile tile) { + struct intel_drm_buffer *buf = intel_drm_buffer(buffer); assert(I915_TILING_NONE == INTEL_TILE_NONE); assert(I915_TILING_X == INTEL_TILE_X); assert(I915_TILING_Y == INTEL_TILE_Y); - return drm_intel_bo_set_tiling(intel_bo(buffer), &tile, stride); + if (tile != INTEL_TILE_NONE) { + assert(buf->map_count == 0); + buf->map_gtt = TRUE; + } + + return drm_intel_bo_set_tiling(buf->bo, &tile, stride); } static void * diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm.c b/src/gallium/winsys/drm/radeon/core/radeon_drm.c index a4011db0b87..caab33de1cc 100644 --- a/src/gallium/winsys/drm/radeon/core/radeon_drm.c +++ b/src/gallium/winsys/drm/radeon/core/radeon_drm.c @@ -107,13 +107,18 @@ radeon_texture_from_shared_handle(struct drm_api *api, unsigned handle) { struct pipe_buffer *buffer; + struct pipe_texture *blanket; buffer = radeon_buffer_from_handle(api, screen, name, handle); if (!buffer) { return NULL; } - return screen->texture_blanket(screen, templ, &stride, buffer); + blanket = screen->texture_blanket(screen, templ, &stride, buffer); + + pipe_buffer_reference(&buffer, NULL); + + return blanket; } static boolean radeon_shared_handle_from_texture(struct drm_api *api, diff --git a/src/gallium/winsys/drm/radeon/xorg/Makefile b/src/gallium/winsys/drm/radeon/xorg/Makefile index 0241625f69b..9fa16dab24c 100644 --- a/src/gallium/winsys/drm/radeon/xorg/Makefile +++ b/src/gallium/winsys/drm/radeon/xorg/Makefile @@ -20,6 +20,8 @@ LIBS = \ $(GALLIUMDIR)/state_trackers/xorg/libxorgtracker.a \ $(GALLIUMDIR)/winsys/drm/radeon/core/libradeonwinsys.a \ $(TOP)/src/gallium/drivers/r300/libr300.a \ + $(TOP)/src/gallium/drivers/trace/libtrace.a \ + $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \ $(GALLIUM_AUXILIARIES) ############################################# diff --git a/src/gallium/winsys/g3dvl/Makefile b/src/gallium/winsys/g3dvl/Makefile new file mode 100644 index 00000000000..424ddea87ad --- /dev/null +++ b/src/gallium/winsys/g3dvl/Makefile @@ -0,0 +1,11 @@ +TOP = ../../../.. +include $(TOP)/configs/current + +SUBDIRS = $(GALLIUM_WINSYS_DIRS) + +default install clean: + @for dir in $(SUBDIRS) ; do \ + if [ -d $$dir ] ; then \ + (cd $$dir && $(MAKE) $@) || exit 1; \ + fi \ + done diff --git a/src/gallium/winsys/g3dvl/vl_winsys.h b/src/gallium/winsys/g3dvl/vl_winsys.h index c83db28dd98..22119f95595 100644 --- a/src/gallium/winsys/g3dvl/vl_winsys.h +++ b/src/gallium/winsys/g3dvl/vl_winsys.h @@ -1,14 +1,50 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + #ifndef vl_winsys_h #define vl_winsys_h #include <X11/Xlib.h> +#include <pipe/p_defines.h> +#include <pipe/p_format.h> -struct pipe_context; +struct pipe_screen; +struct pipe_video_context; -struct pipe_context* create_pipe_context(Display *display, int screen); -int destroy_pipe_context(struct pipe_context *pipe); -int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable); -int unbind_pipe_drawable(struct pipe_context *pipe); +struct pipe_screen* +vl_screen_create(Display *display, int screen); -#endif +struct pipe_video_context* +vl_video_create(struct pipe_screen *screen, + enum pipe_video_profile profile, + enum pipe_video_chroma_format chroma_format, + unsigned width, unsigned height); +Drawable +vl_video_bind_drawable(struct pipe_video_context *vpipe, Drawable drawable); + +#endif diff --git a/src/gallium/winsys/g3dvl/xlib/Makefile b/src/gallium/winsys/g3dvl/xlib/Makefile new file mode 100644 index 00000000000..cf765ef51a5 --- /dev/null +++ b/src/gallium/winsys/g3dvl/xlib/Makefile @@ -0,0 +1,74 @@ +# This makefile produces a "stand-alone" libXvMCg3dvl.so which is +# based on Xlib (no DRI HW acceleration) + +TOP = ../../../../.. +include $(TOP)/configs/current + +XVMC_MAJOR = 1 +XVMC_MINOR = 0 +XVMC_LIB = XvMCg3dvl +XVMC_LIB_NAME = lib$(XVMC_LIB).so +XVMC_LIB_DEPS = $(EXTRA_LIB_PATH) -lXvMC -lXv -lX11 -lm + +INCLUDES = -I$(TOP)/src/gallium/include \ + -I$(TOP)/src/gallium/auxiliary \ + -I$(TOP)/src/gallium/drivers \ + -I$(TOP)/src/gallium/winsys/g3dvl + +DEFINES += -DGALLIUM_SOFTPIPE \ + -DGALLIUM_TRACE + +SOURCES = xsp_winsys.c + +# XXX: Hack, if we include libxvmctracker.a in LIBS none of the symbols are +# pulled in by the linker because xsp_winsys.c doesn't refer to them +OBJECTS = $(SOURCES:.c=.o) $(TOP)/src/gallium/state_trackers/xorg/xvmc/*.o + +LIBS = $(TOP)/src/gallium/drivers/softpipe/libsoftpipe.a \ + $(TOP)/src/gallium/auxiliary/vl/libvl.a \ + $(TOP)/src/gallium/auxiliary/tgsi/libtgsi.a \ + $(TOP)/src/gallium/auxiliary/draw/libdraw.a \ + $(TOP)/src/gallium/auxiliary/translate/libtranslate.a \ + $(TOP)/src/gallium/auxiliary/cso_cache/libcso_cache.a \ + $(TOP)/src/gallium/auxiliary/rtasm/librtasm.a \ + $(TOP)/src/gallium/auxiliary/util/libutil.a + +.c.o: + $(CC) -c $(INCLUDES) $(DEFINES) $(CFLAGS) $< -o $@ + +.S.o: + $(CC) -c $(INCLUDES) $(DEFINES) $(CFLAGS) $< -o $@ + +.PHONY: default $(TOP)/$(LIB_DIR)/gallium clean + +default: depend $(TOP)/$(LIB_DIR)/gallium $(TOP)/$(LIB_DIR)/gallium/$(XVMC_LIB_NAME) + +$(TOP)/$(LIB_DIR)/gallium: + @mkdir -p $(TOP)/$(LIB_DIR)/gallium + +# Make the libXvMCg3dvl.so library +$(TOP)/$(LIB_DIR)/gallium/$(XVMC_LIB_NAME): $(OBJECTS) $(LIBS) Makefile + $(MKLIB) -o $(XVMC_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \ + -major $(XVMC_MAJOR) -minor $(XVMC_MINOR) $(MKLIB_OPTIONS) \ + -install $(TOP)/$(LIB_DIR)/gallium -id $(INSTALL_LIB_DIR)/lib$(XVMC_LIB).1.dylib \ + $(XVMC_LIB_DEPS) $(OBJECTS) $(LIBS) + +depend: $(SOURCES) Makefile + $(RM) depend + touch depend + $(MKDEP) $(MKDEP_OPTIONS) $(DEFINES) $(INCLUDES) $(SOURCES) + +#install: default +# $(INSTALL) -d $(INSTALL_DIR)/include/GL +# $(INSTALL) -d $(INSTALL_DIR)/$(LIB_DIR) +# $(INSTALL) -m 644 $(TOP)/include/GL/*.h $(INSTALL_DIR)/include/GL +# @if [ -e $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME) ]; then \ +# $(INSTALL) $(TOP)/$(LIB_DIR)/libGL* $(INSTALL_DIR)/$(LIB_DIR); \ +# fi + +clean: Makefile + $(RM) $(TOP)/$(LIB_DIR)/gallium/$(XVMC_LIB_NAME) + $(RM) *.o *~ + $(RM) depend depend.bak + +-include depend diff --git a/src/gallium/winsys/g3dvl/xlib/xsp_winsys.c b/src/gallium/winsys/g3dvl/xlib/xsp_winsys.c new file mode 100644 index 00000000000..0e5f5a587b4 --- /dev/null +++ b/src/gallium/winsys/g3dvl/xlib/xsp_winsys.c @@ -0,0 +1,331 @@ +/************************************************************************** + * + * Copyright 2009 Younes Manton. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +#include <vl_winsys.h> +#include <X11/Xutil.h> +#include <pipe/internal/p_winsys_screen.h> +#include <pipe/p_state.h> +#include <pipe/p_inlines.h> +#include <util/u_memory.h> +#include <util/u_math.h> +#include <softpipe/sp_winsys.h> +#include <softpipe/sp_video_context.h> +#include <softpipe/sp_texture.h> + +/* pipe_winsys implementation */ + +struct xsp_pipe_winsys +{ + struct pipe_winsys base; + Display *display; + int screen; + XImage *fbimage; +}; + +struct xsp_context +{ + Drawable drawable; + + void (*pipe_destroy)(struct pipe_video_context *vpipe); +}; + +struct xsp_buffer +{ + struct pipe_buffer base; + boolean is_user_buffer; + void *data; + void *mapped_data; +}; + +static struct pipe_buffer* xsp_buffer_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size) +{ + struct xsp_buffer *buffer; + + assert(pws); + + buffer = calloc(1, sizeof(struct xsp_buffer)); + pipe_reference_init(&buffer->base.reference, 1); + buffer->base.alignment = alignment; + buffer->base.usage = usage; + buffer->base.size = size; + buffer->data = align_malloc(size, alignment); + + return (struct pipe_buffer*)buffer; +} + +static struct pipe_buffer* xsp_user_buffer_create(struct pipe_winsys *pws, void *data, unsigned size) +{ + struct xsp_buffer *buffer; + + assert(pws); + + buffer = calloc(1, sizeof(struct xsp_buffer)); + pipe_reference_init(&buffer->base.reference, 1); + buffer->base.size = size; + buffer->is_user_buffer = TRUE; + buffer->data = data; + + return (struct pipe_buffer*)buffer; +} + +static void* xsp_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buffer, unsigned flags) +{ + struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; + + assert(pws); + assert(buffer); + + xsp_buf->mapped_data = xsp_buf->data; + + return xsp_buf->mapped_data; +} + +static void xsp_buffer_unmap(struct pipe_winsys *pws, struct pipe_buffer *buffer) +{ + struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; + + assert(pws); + assert(buffer); + + xsp_buf->mapped_data = NULL; +} + +static void xsp_buffer_destroy(struct pipe_buffer *buffer) +{ + struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; + + assert(buffer); + + if (!xsp_buf->is_user_buffer) + align_free(xsp_buf->data); + + free(xsp_buf); +} + +static struct pipe_buffer* xsp_surface_buffer_create +( + struct pipe_winsys *pws, + unsigned width, + unsigned height, + enum pipe_format format, + unsigned usage, + unsigned tex_usage, + unsigned *stride +) +{ + const unsigned int ALIGNMENT = 1; + struct pipe_format_block block; + unsigned nblocksx, nblocksy; + + pf_get_block(format, &block); + nblocksx = pf_get_nblocksx(&block, width); + nblocksy = pf_get_nblocksy(&block, height); + *stride = align(nblocksx * block.size, ALIGNMENT); + + return pws->buffer_create(pws, ALIGNMENT, usage, + *stride * nblocksy); +} + +static void xsp_fence_reference(struct pipe_winsys *pws, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence) +{ + assert(pws); + assert(ptr); + assert(fence); +} + +static int xsp_fence_signalled(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag) +{ + assert(pws); + assert(fence); + + return 0; +} + +static int xsp_fence_finish(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag) +{ + assert(pws); + assert(fence); + + return 0; +} + +static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private) +{ + struct xsp_pipe_winsys *xsp_winsys; + struct xsp_context *xsp_context; + + assert(pws); + assert(surface); + assert(context_private); + + xsp_winsys = (struct xsp_pipe_winsys*)pws; + xsp_context = (struct xsp_context*)context_private; + xsp_winsys->fbimage->width = surface->width; + xsp_winsys->fbimage->height = surface->height; + xsp_winsys->fbimage->bytes_per_line = surface->width * (xsp_winsys->fbimage->bits_per_pixel >> 3); + xsp_winsys->fbimage->data = (char*)((struct xsp_buffer *)softpipe_texture(surface->texture)->buffer)->data + surface->offset; + + XPutImage + ( + xsp_winsys->display, xsp_context->drawable, + XDefaultGC(xsp_winsys->display, xsp_winsys->screen), + xsp_winsys->fbimage, 0, 0, 0, 0, + surface->width, surface->height + ); + XFlush(xsp_winsys->display); +} + +static const char* xsp_get_name(struct pipe_winsys *pws) +{ + assert(pws); + return "X11 SoftPipe"; +} + +static void xsp_destroy(struct pipe_winsys *pws) +{ + struct xsp_pipe_winsys *xsp_winsys = (struct xsp_pipe_winsys*)pws; + + assert(pws); + + /* XDestroyImage() wants to free the data as well */ + xsp_winsys->fbimage->data = NULL; + + XDestroyImage(xsp_winsys->fbimage); + FREE(xsp_winsys); +} + +/* Called through pipe_video_context::destroy() */ +static void xsp_pipe_destroy(struct pipe_video_context *vpipe) +{ + struct xsp_context *xsp_context; + + assert(vpipe); + + xsp_context = vpipe->priv; + + /* Call the original destroy */ + xsp_context->pipe_destroy(vpipe); + + FREE(xsp_context); +} + +/* Show starts here */ + +Drawable +vl_video_bind_drawable(struct pipe_video_context *vpipe, Drawable drawable) +{ + struct xsp_context *xsp_context; + Drawable old_drawable; + + assert(vpipe); + + xsp_context = vpipe->priv; + old_drawable = xsp_context->drawable; + xsp_context->drawable = drawable; + + return old_drawable; +} + +struct pipe_screen* +vl_screen_create(Display *display, int screen) +{ + struct xsp_pipe_winsys *xsp_winsys; + + assert(display); + + xsp_winsys = CALLOC_STRUCT(xsp_pipe_winsys); + if (!xsp_winsys) + return NULL; + + xsp_winsys->base.buffer_create = xsp_buffer_create; + xsp_winsys->base.user_buffer_create = xsp_user_buffer_create; + xsp_winsys->base.buffer_map = xsp_buffer_map; + xsp_winsys->base.buffer_unmap = xsp_buffer_unmap; + xsp_winsys->base.buffer_destroy = xsp_buffer_destroy; + xsp_winsys->base.surface_buffer_create = xsp_surface_buffer_create; + xsp_winsys->base.fence_reference = xsp_fence_reference; + xsp_winsys->base.fence_signalled = xsp_fence_signalled; + xsp_winsys->base.fence_finish = xsp_fence_finish; + xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer; + xsp_winsys->base.get_name = xsp_get_name; + xsp_winsys->base.destroy = xsp_destroy; + xsp_winsys->display = display; + xsp_winsys->screen = screen; + xsp_winsys->fbimage = XCreateImage + ( + display, + XDefaultVisual(display, screen), + XDefaultDepth(display, screen), + ZPixmap, + 0, + NULL, + 0, /* Don't know the width and height until flush_frontbuffer */ + 0, + 32, + 0 + ); + + if (!xsp_winsys->fbimage) { + FREE(xsp_winsys); + return NULL; + } + + XInitImage(xsp_winsys->fbimage); + + return softpipe_create_screen(&xsp_winsys->base); +} + +struct pipe_video_context* +vl_video_create(struct pipe_screen *screen, + enum pipe_video_profile profile, + enum pipe_video_chroma_format chroma_format, + unsigned width, unsigned height) +{ + struct pipe_video_context *vpipe; + struct xsp_context *xsp_context; + + assert(screen); + assert(width && height); + + vpipe = sp_video_create(screen, profile, chroma_format, width, height); + if (!vpipe) + return NULL; + + xsp_context = CALLOC_STRUCT(xsp_context); + if (!xsp_context) { + vpipe->destroy(vpipe); + return NULL; + } + + /* Override this so we can free our xsp_context when the pipe is freed */ + xsp_context->pipe_destroy = vpipe->destroy; + vpipe->destroy = xsp_pipe_destroy; + + vpipe->priv = xsp_context; + + return vpipe; +} diff --git a/src/gallium/winsys/g3dvl/xsp_winsys.c b/src/gallium/winsys/g3dvl/xsp_winsys.c deleted file mode 100644 index 37d60ce5406..00000000000 --- a/src/gallium/winsys/g3dvl/xsp_winsys.c +++ /dev/null @@ -1,291 +0,0 @@ -#include "vl_winsys.h" -#include <X11/Xutil.h> -#include <pipe/internal/p_winsys_screen.h> -#include <pipe/p_state.h> -#include <pipe/p_inlines.h> -#include <util/u_memory.h> -#include <util/u_math.h> -#include <softpipe/sp_winsys.h> -#include <softpipe/sp_texture.h> - -/* pipe_winsys implementation */ - -struct xsp_pipe_winsys -{ - struct pipe_winsys base; - XImage fbimage; -}; - -struct xsp_context -{ - Display *display; - int screen; - Drawable drawable; - int drawable_bound; -}; - -struct xsp_buffer -{ - struct pipe_buffer base; - boolean is_user_buffer; - void *data; - void *mapped_data; -}; - -static struct pipe_buffer* xsp_buffer_create(struct pipe_winsys *pws, unsigned alignment, unsigned usage, unsigned size) -{ - struct xsp_buffer *buffer; - - assert(pws); - - buffer = calloc(1, sizeof(struct xsp_buffer)); - pipe_reference_init(&buffer->base.reference, 1); - buffer->base.alignment = alignment; - buffer->base.usage = usage; - buffer->base.size = size; - buffer->data = align_malloc(size, alignment); - - return (struct pipe_buffer*)buffer; -} - -static struct pipe_buffer* xsp_user_buffer_create(struct pipe_winsys *pws, void *data, unsigned size) -{ - struct xsp_buffer *buffer; - - assert(pws); - - buffer = calloc(1, sizeof(struct xsp_buffer)); - pipe_reference_init(&buffer->base.reference, 1); - buffer->base.size = size; - buffer->is_user_buffer = TRUE; - buffer->data = data; - - return (struct pipe_buffer*)buffer; -} - -static void* xsp_buffer_map(struct pipe_winsys *pws, struct pipe_buffer *buffer, unsigned flags) -{ - struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; - - assert(pws); - assert(buffer); - - xsp_buf->mapped_data = xsp_buf->data; - - return xsp_buf->mapped_data; -} - -static void xsp_buffer_unmap(struct pipe_winsys *pws, struct pipe_buffer *buffer) -{ - struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; - - assert(pws); - assert(buffer); - - xsp_buf->mapped_data = NULL; -} - -static void xsp_buffer_destroy(struct pipe_winsys *pws, struct pipe_buffer *buffer) -{ - struct xsp_buffer *xsp_buf = (struct xsp_buffer*)buffer; - - assert(pws); - assert(buffer); - - if (!xsp_buf->is_user_buffer) - align_free(xsp_buf->data); - - free(xsp_buf); -} - -static struct pipe_buffer* xsp_surface_buffer_create -( - struct pipe_winsys *pws, - unsigned width, - unsigned height, - enum pipe_format format, - unsigned usage, - unsigned tex_usage, - unsigned *stride -) -{ - const unsigned int ALIGNMENT = 1; - struct pipe_format_block block; - unsigned nblocksx, nblocksy; - - pf_get_block(format, &block); - nblocksx = pf_get_nblocksx(&block, width); - nblocksy = pf_get_nblocksy(&block, height); - *stride = align(nblocksx * block.size, ALIGNMENT); - - return pws->buffer_create(pws, ALIGNMENT, - usage, - *stride * nblocksy); -} - -static void xsp_fence_reference(struct pipe_winsys *pws, struct pipe_fence_handle **ptr, struct pipe_fence_handle *fence) -{ - assert(pws); - assert(ptr); - assert(fence); -} - -static int xsp_fence_signalled(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag) -{ - assert(pws); - assert(fence); - - return 0; -} - -static int xsp_fence_finish(struct pipe_winsys *pws, struct pipe_fence_handle *fence, unsigned flag) -{ - assert(pws); - assert(fence); - - return 0; -} - -static void xsp_flush_frontbuffer(struct pipe_winsys *pws, struct pipe_surface *surface, void *context_private) -{ - struct xsp_pipe_winsys *xsp_winsys; - struct xsp_context *xsp_context; - - assert(pws); - assert(surface); - assert(context_private); - - xsp_winsys = (struct xsp_pipe_winsys*)pws; - xsp_context = (struct xsp_context*)context_private; - - if (!xsp_context->drawable_bound) - return; - - xsp_winsys->fbimage.width = surface->width; - xsp_winsys->fbimage.height = surface->height; - xsp_winsys->fbimage.bytes_per_line = surface->width * (xsp_winsys->fbimage.bits_per_pixel >> 3); - xsp_winsys->fbimage.data = ((struct xsp_buffer *)softpipe_texture(surface->texture)->buffer)->data + surface->offset; - - XPutImage - ( - xsp_context->display, - xsp_context->drawable, - XDefaultGC(xsp_context->display, xsp_context->screen), - &xsp_winsys->fbimage, - 0, - 0, - 0, - 0, - surface->width, - surface->height - ); - XFlush(xsp_context->display); -} - -static const char* xsp_get_name(struct pipe_winsys *pws) -{ - assert(pws); - return "X11 SoftPipe"; -} - -/* Show starts here */ - -int bind_pipe_drawable(struct pipe_context *pipe, Drawable drawable) -{ - struct xsp_context *xsp_context; - - assert(pipe); - - xsp_context = pipe->priv; - xsp_context->drawable = drawable; - xsp_context->drawable_bound = 1; - - return 0; -} - -int unbind_pipe_drawable(struct pipe_context *pipe) -{ - struct xsp_context *xsp_context; - - assert(pipe); - - xsp_context = pipe->priv; - xsp_context->drawable_bound = 0; - - return 0; -} - -struct pipe_context* create_pipe_context(Display *display, int screen) -{ - struct xsp_pipe_winsys *xsp_winsys; - struct xsp_context *xsp_context; - struct pipe_screen *sp_screen; - struct pipe_context *sp_pipe; - - assert(display); - - xsp_winsys = calloc(1, sizeof(struct xsp_pipe_winsys)); - xsp_winsys->base.buffer_create = xsp_buffer_create; - xsp_winsys->base.user_buffer_create = xsp_user_buffer_create; - xsp_winsys->base.buffer_map = xsp_buffer_map; - xsp_winsys->base.buffer_unmap = xsp_buffer_unmap; - xsp_winsys->base.buffer_destroy = xsp_buffer_destroy; - xsp_winsys->base.surface_buffer_create = xsp_surface_buffer_create; - xsp_winsys->base.fence_reference = xsp_fence_reference; - xsp_winsys->base.fence_signalled = xsp_fence_signalled; - xsp_winsys->base.fence_finish = xsp_fence_finish; - xsp_winsys->base.flush_frontbuffer = xsp_flush_frontbuffer; - xsp_winsys->base.get_name = xsp_get_name; - - { - /* XXX: Can't use the returned XImage* directly, - since we don't have control over winsys destruction - and we wouldn't be able to free it */ - XImage *template = XCreateImage - ( - display, - XDefaultVisual(display, XDefaultScreen(display)), - XDefaultDepth(display, XDefaultScreen(display)), - ZPixmap, - 0, - NULL, - 0, /* Don't know the width and height until flush_frontbuffer */ - 0, - 32, - 0 - ); - - memcpy(&xsp_winsys->fbimage, template, sizeof(XImage)); - XInitImage(&xsp_winsys->fbimage); - - XDestroyImage(template); - } - - sp_screen = softpipe_create_screen((struct pipe_winsys*)xsp_winsys); - sp_pipe = softpipe_create(sp_screen); - - xsp_context = calloc(1, sizeof(struct xsp_context)); - xsp_context->display = display; - xsp_context->screen = screen; - - sp_pipe->priv = xsp_context; - - return sp_pipe; -} - -int destroy_pipe_context(struct pipe_context *pipe) -{ - struct pipe_screen *screen; - struct pipe_winsys *winsys; - - assert(pipe); - - screen = pipe->screen; - winsys = pipe->winsys; - free(pipe->priv); - pipe->destroy(pipe); - screen->destroy(screen); - free(winsys); - - return 0; -} diff --git a/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c b/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c index c0c33b45d59..e8bc0f55ac4 100644 --- a/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c +++ b/src/gallium/winsys/gdi/gdi_llvmpipe_winsys.c @@ -43,7 +43,7 @@ #include "util/u_memory.h" #include "llvmpipe/lp_winsys.h" #include "llvmpipe/lp_texture.h" -#include "shared/stw_winsys.h" +#include "stw_winsys.h" struct gdi_llvmpipe_displaytarget @@ -234,9 +234,9 @@ gdi_llvmpipe_context_create(struct pipe_screen *screen) static void -gdi_llvmpipe_flush_frontbuffer(struct pipe_screen *screen, - struct pipe_surface *surface, - HDC hDC) +gdi_llvmpipe_present(struct pipe_screen *screen, + struct pipe_surface *surface, + HDC hDC) { struct llvmpipe_texture *texture; struct gdi_llvmpipe_displaytarget *gdt; @@ -254,7 +254,11 @@ gdi_llvmpipe_flush_frontbuffer(struct pipe_screen *screen, static const struct stw_winsys stw_winsys = { &gdi_llvmpipe_screen_create, &gdi_llvmpipe_context_create, - &gdi_llvmpipe_flush_frontbuffer + &gdi_llvmpipe_present, + NULL, /* get_adapter_luid */ + NULL, /* shared_surface_open */ + NULL, /* shared_surface_close */ + NULL /* compose */ }; diff --git a/src/gallium/winsys/gdi/gdi_softpipe_winsys.c b/src/gallium/winsys/gdi/gdi_softpipe_winsys.c index 66120a6a983..5e0ccf32f48 100644 --- a/src/gallium/winsys/gdi/gdi_softpipe_winsys.c +++ b/src/gallium/winsys/gdi/gdi_softpipe_winsys.c @@ -46,7 +46,7 @@ #include "util/u_memory.h" #include "softpipe/sp_winsys.h" #include "softpipe/sp_texture.h" -#include "shared/stw_winsys.h" +#include "stw_winsys.h" struct gdi_softpipe_buffer @@ -269,9 +269,9 @@ gdi_softpipe_context_create(struct pipe_screen *screen) static void -gdi_softpipe_flush_frontbuffer(struct pipe_screen *screen, - struct pipe_surface *surface, - HDC hDC) +gdi_softpipe_present(struct pipe_screen *screen, + struct pipe_surface *surface, + HDC hDC) { struct softpipe_texture *texture; struct gdi_softpipe_buffer *buffer; @@ -304,7 +304,11 @@ gdi_softpipe_flush_frontbuffer(struct pipe_screen *screen, static const struct stw_winsys stw_winsys = { &gdi_softpipe_screen_create, &gdi_softpipe_context_create, - &gdi_softpipe_flush_frontbuffer + &gdi_softpipe_present, + NULL, /* get_adapter_luid */ + NULL, /* shared_surface_open */ + NULL, /* shared_surface_close */ + NULL /* compose */ }; diff --git a/src/gallium/winsys/xlib/xlib_softpipe.c b/src/gallium/winsys/xlib/xlib_softpipe.c index 67fea023a3b..260b39e2a0f 100644 --- a/src/gallium/winsys/xlib/xlib_softpipe.c +++ b/src/gallium/winsys/xlib/xlib_softpipe.c @@ -75,9 +75,6 @@ struct xmesa_pipe_winsys { struct pipe_winsys base; /* struct xmesa_visual *xm_visual; */ -#ifdef USE_XSHM - int shm; -#endif }; @@ -93,11 +90,6 @@ xm_buffer( struct pipe_buffer *buf ) /** * X Shared Memory Image extension code */ -#ifdef USE_XSHM -#define XSHM_ENABLED(b) ((b)->shm) -#else -#define XSHM_ENABLED(b) 0 -#endif #ifdef USE_XSHM @@ -116,23 +108,23 @@ mesaHandleXError(Display *dpy, XErrorEvent *event) } -static GLboolean alloc_shm(struct xm_buffer *buf, unsigned size) +static char *alloc_shm(struct xm_buffer *buf, unsigned size) { XShmSegmentInfo *const shminfo = & buf->shminfo; shminfo->shmid = shmget(IPC_PRIVATE, size, IPC_CREAT|0777); if (shminfo->shmid < 0) { - return GL_FALSE; + return NULL; } shminfo->shmaddr = (char *) shmat(shminfo->shmid, 0, 0); if (shminfo->shmaddr == (char *) -1) { shmctl(shminfo->shmid, IPC_RMID, 0); - return GL_FALSE; + return NULL; } shminfo->readOnly = False; - return GL_TRUE; + return shminfo->shmaddr; } @@ -258,25 +250,30 @@ xlib_softpipe_display_surface(struct xmesa_buffer *b, return; #ifdef USE_XSHM - if (XSHM_ENABLED(xm_buf) && (xm_buf->tempImage == NULL)) { - assert(surf->texture->block.width == 1); - assert(surf->texture->block.height == 1); - alloc_shm_ximage(xm_buf, b, spt->stride[surf->level] / - surf->texture->block.size, surf->height); - } -#endif + if (xm_buf->shm) + { + if (xm_buf->tempImage == NULL) + { + assert(surf->texture->block.width == 1); + assert(surf->texture->block.height == 1); + alloc_shm_ximage(xm_buf, b, spt->stride[surf->level] / + surf->texture->block.size, surf->height); + } - ximage = (XSHM_ENABLED(xm_buf)) ? xm_buf->tempImage : b->tempImage; - ximage->data = xm_buf->data; + ximage = xm_buf->tempImage; + ximage->data = xm_buf->data; - /* display image in Window */ -#ifdef USE_XSHM - if (XSHM_ENABLED(xm_buf)) { + /* _debug_printf("XSHM\n"); */ XShmPutImage(b->xm_visual->display, b->drawable, b->gc, ximage, 0, 0, 0, 0, surf->width, surf->height, False); - } else + } + else #endif { + /* display image in Window */ + ximage = b->tempImage; + ximage->data = xm_buf->data; + /* check that the XImage has been previously initialized */ assert(ximage->format); assert(ximage->bitmap_unit); @@ -286,6 +283,7 @@ xlib_softpipe_display_surface(struct xmesa_buffer *b, ximage->height = surf->height; ximage->bytes_per_line = spt->stride[surf->level]; + /* _debug_printf("XPUT\n"); */ XPutImage(b->xm_visual->display, b->drawable, b->gc, ximage, 0, 0, 0, 0, surf->width, surf->height); } @@ -322,21 +320,6 @@ xm_buffer_create(struct pipe_winsys *pws, unsigned size) { struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer); -#ifdef USE_XSHM - struct xmesa_pipe_winsys *xpws = (struct xmesa_pipe_winsys *) pws; - - buffer->shminfo.shmid = -1; - buffer->shminfo.shmaddr = (char *) -1; - - if (xpws->shm && (usage & PIPE_BUFFER_USAGE_PIXEL) != 0) { - buffer->shm = xpws->shm; - - if (alloc_shm(buffer, size)) { - buffer->data = buffer->shminfo.shmaddr; - buffer->shm = 1; - } - } -#endif pipe_reference_init(&buffer->base.reference, 1); buffer->base.alignment = alignment; @@ -363,9 +346,6 @@ xm_user_buffer_create(struct pipe_winsys *pws, void *ptr, unsigned bytes) buffer->base.size = bytes; buffer->userBuffer = TRUE; buffer->data = ptr; -#ifdef USE_XSHM - buffer->shm = 0; -#endif return &buffer->base; } @@ -381,16 +361,44 @@ xm_surface_buffer_create(struct pipe_winsys *winsys, { const unsigned alignment = 64; struct pipe_format_block block; - unsigned nblocksx, nblocksy; + unsigned nblocksx, nblocksy, size; pf_get_block(format, &block); nblocksx = pf_get_nblocksx(&block, width); nblocksy = pf_get_nblocksy(&block, height); *stride = align(nblocksx * block.size, alignment); + size = *stride * nblocksy; + +#ifdef USE_XSHM + if (!debug_get_bool_option("XLIB_NO_SHM", FALSE)) + { + struct xm_buffer *buffer = CALLOC_STRUCT(xm_buffer); + + pipe_reference_init(&buffer->base.reference, 1); + buffer->base.alignment = alignment; + buffer->base.usage = usage; + buffer->base.size = size; + buffer->userBuffer = FALSE; + buffer->shminfo.shmid = -1; + buffer->shminfo.shmaddr = (char *) -1; + buffer->shm = TRUE; + + buffer->data = alloc_shm(buffer, size); + if (!buffer->data) + goto out; + + return &buffer->base; + + out: + if (buffer) + FREE(buffer); + } +#endif + return winsys->buffer_create(winsys, alignment, usage, - *stride * nblocksy); + size); } diff --git a/src/glx/x11/glx_pbuffer.c b/src/glx/x11/glx_pbuffer.c index 01569213f2d..501500afc3d 100644 --- a/src/glx/x11/glx_pbuffer.c +++ b/src/glx/x11/glx_pbuffer.c @@ -39,6 +39,13 @@ #include "glxextensions.h" #include "glcontextmodes.h" +#define WARN_ONCE_GLX_1_3(a, b) { \ + static int warned=1; \ + if(warned) { \ + warn_GLX_1_3((a), b ); \ + warned=0; \ + } \ + } /** * Emit a warning when clients use GLX 1.3 functions on pre-1.3 systems. @@ -576,7 +583,7 @@ glXCreatePbuffer(Display * dpy, GLXFBConfig config, const int *attrib_list) width = 0; height = 0; - warn_GLX_1_3(dpy, __func__); + WARN_ONCE_GLX_1_3(dpy, __func__); for (i = 0; attrib_list[i * 2]; i++) { switch (attrib_list[i * 2]) { @@ -611,7 +618,7 @@ PUBLIC void glXQueryDrawable(Display * dpy, GLXDrawable drawable, int attribute, unsigned int *value) { - warn_GLX_1_3(dpy, __func__); + WARN_ONCE_GLX_1_3(dpy, __func__); GetDrawableAttribute(dpy, drawable, attribute, value); } @@ -665,7 +672,7 @@ PUBLIC GLXPixmap glXCreatePixmap(Display * dpy, GLXFBConfig config, Pixmap pixmap, const int *attrib_list) { - warn_GLX_1_3(dpy, __func__); + WARN_ONCE_GLX_1_3(dpy, __func__); return CreateDrawable(dpy, (__GLcontextModes *) config, (Drawable) pixmap, attrib_list, X_GLXCreatePixmap); @@ -676,7 +683,7 @@ PUBLIC GLXWindow glXCreateWindow(Display * dpy, GLXFBConfig config, Window win, const int *attrib_list) { - warn_GLX_1_3(dpy, __func__); + WARN_ONCE_GLX_1_3(dpy, __func__); return CreateDrawable(dpy, (__GLcontextModes *) config, (Drawable) win, attrib_list, X_GLXCreateWindow); @@ -686,7 +693,7 @@ glXCreateWindow(Display * dpy, GLXFBConfig config, Window win, PUBLIC void glXDestroyPixmap(Display * dpy, GLXPixmap pixmap) { - warn_GLX_1_3(dpy, __func__); + WARN_ONCE_GLX_1_3(dpy, __func__); DestroyDrawable(dpy, (GLXDrawable) pixmap, X_GLXDestroyPixmap); } @@ -695,7 +702,7 @@ glXDestroyPixmap(Display * dpy, GLXPixmap pixmap) PUBLIC void glXDestroyWindow(Display * dpy, GLXWindow win) { - warn_GLX_1_3(dpy, __func__); + WARN_ONCE_GLX_1_3(dpy, __func__); DestroyDrawable(dpy, (GLXDrawable) win, X_GLXDestroyWindow); } @@ -716,3 +723,4 @@ GLX_ALIAS_VOID(glXGetSelectedEventSGIX, (Display * dpy, GLXDrawable drawable, unsigned long *mask), (dpy, drawable, mask), glXGetSelectedEvent) + diff --git a/src/glx/x11/glxcmds.c b/src/glx/x11/glxcmds.c index ea55cc49efe..cd4aede74e4 100644 --- a/src/glx/x11/glxcmds.c +++ b/src/glx/x11/glxcmds.c @@ -398,6 +398,10 @@ CreateContext(Display * dpy, XVisualInfo * vis, _XError(dpy, &error); return None; } + if (renderType == 0) { + /* Initialize renderType now */ + renderType = mode->rgbMode ? GLX_RGBA_TYPE : GLX_COLOR_INDEX_TYPE; + } } else { mode = fbconfig; @@ -484,6 +488,8 @@ CreateContext(Display * dpy, XVisualInfo * vis, gc->imported = GL_TRUE; } + gc->renderType = renderType; + return gc; } @@ -1575,7 +1581,7 @@ GLX_ALIAS(Display *, glXGetCurrentDisplayEXT, (void), (), * This function dynamically determines whether to use the EXT_import_context * version of the protocol or the GLX 1.3 version of the protocol. */ - static int __glXQueryContextInfo(Display * dpy, GLXContext ctx) +static int __glXQueryContextInfo(Display * dpy, GLXContext ctx) { __GLXdisplayPrivate *priv = __glXInitialize(dpy); xGLXQueryContextReply reply; @@ -1713,7 +1719,7 @@ GLX_ALIAS(int, glXQueryContextInfoEXT, (Display * dpy, GLXContext ctx, int attribute, int *value), (dpy, ctx, attribute, value), glXQueryContext) - PUBLIC GLXContextID glXGetContextIDEXT(const GLXContext ctx) +PUBLIC GLXContextID glXGetContextIDEXT(const GLXContext ctx) { return ctx->xid; } @@ -2159,18 +2165,19 @@ GLX_ALIAS(int, glXGetFBConfigAttribSGIX, (Display * dpy, GLXFBConfigSGIX config, int attribute, int *value), (dpy, config, attribute, value), glXGetFBConfigAttrib) - PUBLIC GLX_ALIAS(GLXFBConfigSGIX *, glXChooseFBConfigSGIX, - (Display * dpy, int screen, int *attrib_list, - int *nelements), (dpy, screen, attrib_list, nelements), - glXChooseFBConfig) +PUBLIC GLX_ALIAS(GLXFBConfigSGIX *, glXChooseFBConfigSGIX, + (Display * dpy, int screen, int *attrib_list, + int *nelements), (dpy, screen, attrib_list, nelements), + glXChooseFBConfig) - PUBLIC GLX_ALIAS(XVisualInfo *, glXGetVisualFromFBConfigSGIX, - (Display * dpy, GLXFBConfigSGIX config), - (dpy, config), glXGetVisualFromFBConfig) +PUBLIC GLX_ALIAS(XVisualInfo *, glXGetVisualFromFBConfigSGIX, + (Display * dpy, GLXFBConfigSGIX config), + (dpy, config), glXGetVisualFromFBConfig) - PUBLIC GLXPixmap glXCreateGLXPixmapWithConfigSGIX(Display * dpy, - GLXFBConfigSGIX config, - Pixmap pixmap) +PUBLIC GLXPixmap +glXCreateGLXPixmapWithConfigSGIX(Display * dpy, + GLXFBConfigSGIX config, + Pixmap pixmap) { xGLXVendorPrivateWithReplyReq *vpreq; xGLXCreateGLXPixmapWithConfigSGIXReq *req; diff --git a/src/glx/x11/glxhash.c b/src/glx/x11/glxhash.c index 7d28ada49ca..b76ec323459 100644 --- a/src/glx/x11/glxhash.c +++ b/src/glx/x11/glxhash.c @@ -77,6 +77,7 @@ #include <stdio.h> #include <stdlib.h> +#include <string.h> #define HASH_MAGIC 0xdeadbeef #define HASH_DEBUG 0 @@ -87,6 +88,12 @@ #define HASH_ALLOC malloc #define HASH_FREE free +#ifndef __GLIBC__ +#define HASH_RANDOM_DECL char *ps, rs[256] +#define HASH_RANDOM_INIT(seed) ps = initstate(seed, rs, sizeof(rs)) +#define HASH_RANDOM random() +#define HASH_RANDOM_DESTROY setstate(ps) +#else #define HASH_RANDOM_DECL struct random_data rd; int32_t rv; char rs[256] #define HASH_RANDOM_INIT(seed) \ do { \ @@ -95,6 +102,7 @@ } while(0) #define HASH_RANDOM ((void) random_r(&rd, &rv), rv) #define HASH_RANDOM_DESTROY +#endif typedef struct __glxHashBucket { diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c index f09106b77c3..0f8447cb70f 100644 --- a/src/mesa/drivers/common/driverfuncs.c +++ b/src/mesa/drivers/common/driverfuncs.c @@ -106,7 +106,7 @@ _mesa_init_driver_functions(struct dd_function_table *driver) driver->CopyTexSubImage1D = _mesa_meta_CopyTexSubImage1D; driver->CopyTexSubImage2D = _mesa_meta_CopyTexSubImage2D; driver->CopyTexSubImage3D = _mesa_meta_CopyTexSubImage3D; - driver->GenerateMipmap = _mesa_generate_mipmap; + driver->GenerateMipmap = _mesa_meta_GenerateMipmap; driver->TestProxyTexImage = _mesa_test_proxy_teximage; driver->CompressedTexImage1D = _mesa_store_compressed_teximage1d; driver->CompressedTexImage2D = _mesa_store_compressed_teximage2d; diff --git a/src/mesa/drivers/common/meta.c b/src/mesa/drivers/common/meta.c index 532b2c92635..e1732241b3a 100644 --- a/src/mesa/drivers/common/meta.c +++ b/src/mesa/drivers/common/meta.c @@ -55,6 +55,7 @@ #include "main/stencil.h" #include "main/texobj.h" #include "main/texenv.h" +#include "main/texformat.h" #include "main/teximage.h" #include "main/texparam.h" #include "main/texstate.h" @@ -236,7 +237,6 @@ struct copypix_state struct drawpix_state { GLuint ArrayObj; - GLuint VBO; GLuint StencilFP; /**< Fragment program for drawing stencil images */ GLuint DepthFP; /**< Fragment program for drawing depth images */ @@ -327,7 +327,6 @@ _mesa_meta_free(GLcontext *ctx) _mesa_DeleteVertexArraysAPPLE(1, &meta->CopyPix.ArrayObj); /* glDrawPixels */ - _mesa_DeleteBuffersARB(1, & meta->DrawPix.VBO); _mesa_DeleteVertexArraysAPPLE(1, &meta->DrawPix.ArrayObj); _mesa_DeletePrograms(1, &meta->DrawPix.DepthFP); _mesa_DeletePrograms(1, &meta->DrawPix.StencilFP); @@ -976,7 +975,8 @@ setup_copypix_texture(struct temp_texture *tex, * Setup/load texture for glDrawPixels. */ static void -setup_drawpix_texture(struct temp_texture *tex, +setup_drawpix_texture(GLcontext *ctx, + struct temp_texture *tex, GLboolean newTex, GLenum texIntFormat, GLsizei width, GLsizei height, @@ -997,9 +997,17 @@ setup_drawpix_texture(struct temp_texture *tex, tex->Width, tex->Height, 0, format, type, pixels); } else { + struct gl_buffer_object *save_unpack_obj = NULL; + + _mesa_reference_buffer_object(ctx, &save_unpack_obj, + ctx->Unpack.BufferObj); + _mesa_BindBufferARB(GL_PIXEL_UNPACK_BUFFER_ARB, 0); /* create empty texture */ _mesa_TexImage2D(tex->Target, 0, tex->IntFormat, tex->Width, tex->Height, 0, format, type, NULL); + if (save_unpack_obj != NULL) + _mesa_BindBufferARB(GL_PIXEL_UNPACK_BUFFER_ARB, + save_unpack_obj->Name); /* load image */ _mesa_TexSubImage2D(tex->Target, 0, 0, 0, width, height, format, type, pixels); @@ -1047,6 +1055,124 @@ init_blit_depth_pixels(GLcontext *ctx) /** + * Try to do a glBiltFramebuffer using no-copy texturing. + * We can do this when the src renderbuffer is actually a texture. + * But if the src buffer == dst buffer we cannot do this. + * + * \return new buffer mask indicating the buffers left to blit using the + * normal path. + */ +static GLbitfield +blitframebuffer_texture(GLcontext *ctx, + GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, + GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, + GLbitfield mask, GLenum filter) +{ + if (mask & GL_COLOR_BUFFER_BIT) { + const struct gl_framebuffer *drawFb = ctx->DrawBuffer; + const struct gl_framebuffer *readFb = ctx->ReadBuffer; + const struct gl_renderbuffer_attachment *drawAtt = + &drawFb->Attachment[drawFb->_ColorDrawBufferIndexes[0]]; + const struct gl_renderbuffer_attachment *readAtt = + &readFb->Attachment[readFb->_ColorReadBufferIndex]; + + if (readAtt && readAtt->Texture) { + const struct gl_texture_object *texObj = readAtt->Texture; + const GLenum minFilterSave = texObj->MinFilter; + const GLenum magFilterSave = texObj->MagFilter; + const GLenum target = texObj->Target; + + if (drawAtt->Texture == readAtt->Texture) { + /* Can't use same texture as both the source and dest. We need + * to handle overlapping blits and besides, some hw may not + * support this. + */ + return mask; + } + + if (target != GL_TEXTURE_2D && target != GL_TEXTURE_RECTANGLE_ARB) { + /* Can't handle other texture types at this time */ + return mask; + } + + /* + printf("Blit from texture!\n"); + printf(" srcAtt %p dstAtt %p\n", readAtt, drawAtt); + printf(" srcTex %p dstText %p\n", texObj, drawAtt->Texture); + */ + + /* Prepare src texture state */ + _mesa_BindTexture(target, texObj->Name); + _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, filter); + _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, filter); + _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE); + /*_mesa_set_enable(ctx, GL_TEXTURE_RECTANGLE, GL_FALSE);*/ + _mesa_set_enable(ctx, target, GL_TRUE); + + /* Prepare vertex data (the VBO was previously created and bound) */ + { + struct vertex { + GLfloat x, y, s, t; + }; + struct vertex verts[4]; + GLfloat s0, t0, s1, t1; + + if (target == GL_TEXTURE_2D) { + const struct gl_texture_image *texImage + = _mesa_select_tex_image(ctx, texObj, target, + readAtt->TextureLevel); + s0 = srcX0 / (float) texImage->Width; + s1 = srcX1 / (float) texImage->Width; + t0 = srcY0 / (float) texImage->Height; + t1 = srcY1 / (float) texImage->Height; + } + else { + assert(target == GL_TEXTURE_RECTANGLE_ARB); + s0 = srcX0; + s1 = srcX1; + t0 = srcY0; + t1 = srcY1; + } + + verts[0].x = (GLfloat) dstX0; + verts[0].y = (GLfloat) dstY0; + verts[1].x = (GLfloat) dstX1; + verts[1].y = (GLfloat) dstY0; + verts[2].x = (GLfloat) dstX1; + verts[2].y = (GLfloat) dstY1; + verts[3].x = (GLfloat) dstX0; + verts[3].y = (GLfloat) dstY1; + + verts[0].s = s0; + verts[0].t = t0; + verts[1].s = s1; + verts[1].t = t0; + verts[2].s = s1; + verts[2].t = t1; + verts[3].s = s0; + verts[3].t = t1; + + _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); + } + + _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); + + /* Restore texture's filter state, the texture binding will + * be restored by _mesa_meta_end(). + */ + _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, minFilterSave); + _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, magFilterSave); + + /* Done with color buffer */ + mask &= ~GL_COLOR_BUFFER_BIT; + } + } + + return mask; +} + + +/** * Meta implementation of ctx->Driver.BlitFramebuffer() in terms * of texture mapping and polygon rendering. */ @@ -1117,6 +1243,18 @@ _mesa_meta_BlitFramebuffer(GLcontext *ctx, _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, blit->VBO); } + /* Try faster, direct texture approach first */ + mask = blitframebuffer_texture(ctx, srcX0, srcY0, srcX1, srcY1, + dstX0, dstY0, dstX1, dstY1, mask, filter); + if (mask == 0x0) { + _mesa_meta_end(ctx); + return; + } + + /* Continue with "normal" approach which involves copying the src rect + * into a temporary texture and is "blitted" by drawing a textured quad. + */ + newTex = alloc_texture(tex, srcW, srcH, GL_RGBA); /* vertex positions/texcoords (after texture allocation!) */ @@ -1164,7 +1302,7 @@ _mesa_meta_BlitFramebuffer(GLcontext *ctx, _mesa_ReadPixels(srcX, srcY, srcW, srcH, GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, tmp); - setup_drawpix_texture(tex, newTex, GL_DEPTH_COMPONENT, srcW, srcH, + setup_drawpix_texture(ctx, tex, newTex, GL_DEPTH_COMPONENT, srcW, srcH, GL_DEPTH_COMPONENT, GL_UNSIGNED_INT, tmp); _mesa_BindProgram(GL_FRAGMENT_PROGRAM_ARB, blit->DepthFP); @@ -1585,6 +1723,7 @@ _mesa_meta_DrawPixels(GLcontext *ctx, GLenum texIntFormat; GLboolean fallback, newTex; GLbitfield metaExtraSave = 0x0; + GLuint vbo; /* * Determine if we can do the glDrawPixels with texture mapping. @@ -1664,30 +1803,6 @@ _mesa_meta_DrawPixels(GLcontext *ctx, META_VIEWPORT | metaExtraSave)); - if (drawpix->ArrayObj == 0) { - /* one-time setup */ - - /* create vertex array object */ - _mesa_GenVertexArrays(1, &drawpix->ArrayObj); - _mesa_BindVertexArray(drawpix->ArrayObj); - - /* create vertex array buffer */ - _mesa_GenBuffersARB(1, &drawpix->VBO); - _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, drawpix->VBO); - _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts), - NULL, GL_DYNAMIC_DRAW_ARB); - - /* setup vertex arrays */ - _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); - _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); - _mesa_EnableClientState(GL_VERTEX_ARRAY); - _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); - } - else { - _mesa_BindVertexArray(drawpix->ArrayObj); - _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, drawpix->VBO); - } - newTex = alloc_texture(tex, width, height, texIntFormat); /* vertex positions, texcoords (after texture allocation!) */ @@ -1718,10 +1833,25 @@ _mesa_meta_DrawPixels(GLcontext *ctx, verts[3].z = z; verts[3].s = 0.0F; verts[3].t = tex->Ttop; + } - /* upload new vertex data */ - _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); + if (drawpix->ArrayObj == 0) { + /* one-time setup: create vertex array object */ + _mesa_GenVertexArrays(1, &drawpix->ArrayObj); } + _mesa_BindVertexArray(drawpix->ArrayObj); + + /* create vertex array buffer */ + _mesa_GenBuffersARB(1, &vbo); + _mesa_BindBufferARB(GL_ARRAY_BUFFER_ARB, vbo); + _mesa_BufferDataARB(GL_ARRAY_BUFFER_ARB, sizeof(verts), + verts, GL_DYNAMIC_DRAW_ARB); + + /* setup vertex arrays */ + _mesa_VertexPointer(3, GL_FLOAT, sizeof(struct vertex), OFFSET(x)); + _mesa_TexCoordPointer(2, GL_FLOAT, sizeof(struct vertex), OFFSET(s)); + _mesa_EnableClientState(GL_VERTEX_ARRAY); + _mesa_EnableClientState(GL_TEXTURE_COORD_ARRAY); /* set given unpack params */ ctx->Unpack = *unpack; @@ -1735,7 +1865,7 @@ _mesa_meta_DrawPixels(GLcontext *ctx, if (!drawpix->StencilFP) init_draw_stencil_pixels(ctx); - setup_drawpix_texture(tex, newTex, texIntFormat, width, height, + setup_drawpix_texture(ctx, tex, newTex, texIntFormat, width, height, GL_ALPHA, type, pixels); _mesa_ColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE); @@ -1778,20 +1908,22 @@ _mesa_meta_DrawPixels(GLcontext *ctx, _mesa_ProgramLocalParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, 0, ctx->Current.RasterColor); - setup_drawpix_texture(tex, newTex, texIntFormat, width, height, + setup_drawpix_texture(ctx, tex, newTex, texIntFormat, width, height, format, type, pixels); _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); } else { /* Drawing RGBA */ - setup_drawpix_texture(tex, newTex, texIntFormat, width, height, + setup_drawpix_texture(ctx, tex, newTex, texIntFormat, width, height, format, type, pixels); _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); } _mesa_set_enable(ctx, tex->Target, GL_FALSE); + _mesa_DeleteBuffersARB(1, &vbo); + /* restore unpack params */ ctx->Unpack = unpackSave; @@ -1931,7 +2063,7 @@ _mesa_meta_Bitmap(GLcontext *ctx, _mesa_set_enable(ctx, GL_ALPHA_TEST, GL_TRUE); _mesa_AlphaFunc(GL_GREATER, 0.0); - setup_drawpix_texture(tex, newTex, texIntFormat, width, height, + setup_drawpix_texture(ctx, tex, newTex, texIntFormat, width, height, GL_ALPHA, GL_UNSIGNED_BYTE, bitmap8); _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); @@ -1947,6 +2079,33 @@ _mesa_meta_Bitmap(GLcontext *ctx, } +/** + * Check if the call to _mesa_meta_GenerateMipmap() will require a + * software fallback. The fallback path will require that the texture + * images are mapped. + */ +GLboolean +_mesa_meta_check_generate_mipmap_fallback(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj) +{ + struct gl_texture_image *baseImage = + _mesa_select_tex_image(ctx, texObj, target, texObj->BaseLevel); + + /* check for fallbacks */ + if (!ctx->Extensions.EXT_framebuffer_object || + target == GL_TEXTURE_3D || + !baseImage || + baseImage->IsCompressed) { + return GL_TRUE; + } + return GL_FALSE; +} + + +/** + * Called via ctx->Driver.GenerateMipmap() + * Note: texture borders and 3D texture support not yet complete. + */ void _mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, struct gl_texture_object *texObj) @@ -1960,13 +2119,18 @@ _mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, const GLuint maxLevel = texObj->MaxLevel; const GLenum minFilterSave = texObj->MinFilter; const GLenum magFilterSave = texObj->MagFilter; + const GLint baseLevelSave = texObj->BaseLevel; + const GLint maxLevelSave = texObj->MaxLevel; + const GLboolean genMipmapSave = texObj->GenerateMipmap; + const GLenum wrapSSave = texObj->WrapS; + const GLenum wrapTSave = texObj->WrapT; + const GLenum wrapRSave = texObj->WrapR; const GLuint fboSave = ctx->DrawBuffer->Name; GLenum faceTarget; - GLuint level; + GLuint dstLevel; GLuint border = 0; - /* check for fallbacks */ - if (!ctx->Extensions.EXT_framebuffer_object) { + if (_mesa_meta_check_generate_mipmap_fallback(ctx, target, texObj)) { _mesa_generate_mipmap(ctx, target, texObj); return; } @@ -2015,12 +2179,16 @@ _mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, GL_LINEAR); _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, GL_LINEAR); + _mesa_TexParameteri(target, GL_GENERATE_MIPMAP, GL_FALSE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, GL_CLAMP_TO_EDGE); + _mesa_set_enable(ctx, target, GL_TRUE); /* setup texcoords once (XXX what about border?) */ switch (faceTarget) { - case GL_TEXTURE_CUBE_MAP_POSITIVE_X: - break; + case GL_TEXTURE_1D: case GL_TEXTURE_2D: verts[0].s = 0.0F; verts[0].t = 0.0F; @@ -2035,63 +2203,180 @@ _mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, verts[3].t = 1.0F; verts[3].r = 0.0F; break; + case GL_TEXTURE_3D: + abort(); + break; + default: + /* cube face */ + { + static const GLfloat st[4][2] = { + {0.0f, 0.0f}, {1.0f, 0.0f}, {1.0f, 1.0f}, {0.0f, 1.0f} + }; + GLuint i; + + /* loop over quad verts */ + for (i = 0; i < 4; i++) { + /* Compute sc = +/-scale and tc = +/-scale. + * Not +/-1 to avoid cube face selection ambiguity near the edges, + * though that can still sometimes happen with this scale factor... + */ + const GLfloat scale = 0.9999f; + const GLfloat sc = (2.0f * st[i][0] - 1.0f) * scale; + const GLfloat tc = (2.0f * st[i][1] - 1.0f) * scale; + + switch (faceTarget) { + case GL_TEXTURE_CUBE_MAP_POSITIVE_X: + verts[i].s = 1.0f; + verts[i].t = -tc; + verts[i].r = -sc; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_X: + verts[i].s = -1.0f; + verts[i].t = -tc; + verts[i].r = sc; + break; + case GL_TEXTURE_CUBE_MAP_POSITIVE_Y: + verts[i].s = sc; + verts[i].t = 1.0f; + verts[i].r = tc; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y: + verts[i].s = sc; + verts[i].t = -1.0f; + verts[i].r = -tc; + break; + case GL_TEXTURE_CUBE_MAP_POSITIVE_Z: + verts[i].s = sc; + verts[i].t = -tc; + verts[i].r = 1.0f; + break; + case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z: + verts[i].s = -sc; + verts[i].t = -tc; + verts[i].r = -1.0f; + break; + default: + assert(0); + } + } + } } + _mesa_set_enable(ctx, target, GL_TRUE); + + /* texture is already locked, unlock now */ + _mesa_unlock_texture(ctx, texObj); - for (level = baseLevel + 1; level <= maxLevel; level++) { + for (dstLevel = baseLevel + 1; dstLevel <= maxLevel; dstLevel++) { const struct gl_texture_image *srcImage; - const GLuint srcLevel = level - 1; - GLsizei srcWidth, srcHeight; - GLsizei newWidth, newHeight; + const GLuint srcLevel = dstLevel - 1; + GLsizei srcWidth, srcHeight, srcDepth; + GLsizei dstWidth, dstHeight, dstDepth; GLenum status; - srcImage = _mesa_select_tex_image(ctx, texObj, target, srcLevel); + srcImage = _mesa_select_tex_image(ctx, texObj, faceTarget, srcLevel); assert(srcImage->Border == 0); /* XXX we can fix this */ + /* src size w/out border */ srcWidth = srcImage->Width - 2 * border; srcHeight = srcImage->Height - 2 * border; + srcDepth = srcImage->Depth - 2 * border; - newWidth = MAX2(1, srcWidth / 2) + 2 * border; - newHeight = MAX2(1, srcHeight / 2) + 2 * border; + /* new dst size w/ border */ + dstWidth = MAX2(1, srcWidth / 2) + 2 * border; + dstHeight = MAX2(1, srcHeight / 2) + 2 * border; + dstDepth = MAX2(1, srcDepth / 2) + 2 * border; - if (newWidth == srcImage->Width && newHeight == srcImage->Height) { - break; + if (dstWidth == srcImage->Width && + dstHeight == srcImage->Height && + dstDepth == srcImage->Depth) { + /* all done */ + break; } - /* Create empty image */ - _mesa_TexImage2D(GL_TEXTURE_2D, level, srcImage->InternalFormat, - newWidth, newHeight, border, - GL_RGBA, GL_UNSIGNED_BYTE, NULL); + /* Create empty dest image */ + if (target == GL_TEXTURE_1D) { + _mesa_TexImage1D(target, dstLevel, srcImage->InternalFormat, + dstWidth, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + else if (target == GL_TEXTURE_3D) { + _mesa_TexImage3D(target, dstLevel, srcImage->InternalFormat, + dstWidth, dstHeight, dstDepth, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + else { + /* 2D or cube */ + _mesa_TexImage2D(faceTarget, dstLevel, srcImage->InternalFormat, + dstWidth, dstHeight, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + + if (target == GL_TEXTURE_CUBE_MAP) { + /* If texturing from a cube, we need to make sure all src faces + * have been defined (even if we're not sampling from them.) + * Otherwise the texture object will be 'incomplete' and + * texturing from it will not be allowed. + */ + GLuint face; + for (face = 0; face < 6; face++) { + if (!texObj->Image[face][srcLevel] || + texObj->Image[face][srcLevel]->Width != srcWidth) { + _mesa_TexImage2D(GL_TEXTURE_CUBE_MAP_POSITIVE_X + face, + srcLevel, srcImage->InternalFormat, + srcWidth, srcHeight, border, + GL_RGBA, GL_UNSIGNED_BYTE, NULL); + } + } + } + } - /* vertex positions */ + /* setup vertex positions */ { verts[0].x = 0.0F; verts[0].y = 0.0F; - verts[1].x = (GLfloat) newWidth; + verts[1].x = (GLfloat) dstWidth; verts[1].y = 0.0F; - verts[2].x = (GLfloat) newWidth; - verts[2].y = (GLfloat) newHeight; + verts[2].x = (GLfloat) dstWidth; + verts[2].y = (GLfloat) dstHeight; verts[3].x = 0.0F; - verts[3].y = (GLfloat) newHeight; + verts[3].y = (GLfloat) dstHeight; /* upload new vertex data */ _mesa_BufferSubDataARB(GL_ARRAY_BUFFER_ARB, 0, sizeof(verts), verts); } /* limit sampling to src level */ - _mesa_TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, srcLevel); - _mesa_TexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, srcLevel); - - /* Set to draw into the current level */ - _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, - GL_COLOR_ATTACHMENT0_EXT, - target, - texObj->Name, - level); + _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, srcLevel); + _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, srcLevel); + + /* Set to draw into the current dstLevel */ + if (target == GL_TEXTURE_1D) { + _mesa_FramebufferTexture1DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + target, + texObj->Name, + dstLevel); + } + else if (target == GL_TEXTURE_3D) { + GLint zoffset = 0; /* XXX unfinished */ + _mesa_FramebufferTexture3DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + target, + texObj->Name, + dstLevel, zoffset); + } + else { + /* 2D / cube */ + _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, + GL_COLOR_ATTACHMENT0_EXT, + faceTarget, + texObj->Name, + dstLevel); + } - /* Choose to render to the color attachment. */ _mesa_DrawBuffer(GL_COLOR_ATTACHMENT0_EXT); + /* sanity check */ status = _mesa_CheckFramebufferStatusEXT (GL_FRAMEBUFFER_EXT); if (status != GL_FRAMEBUFFER_COMPLETE_EXT) { abort(); @@ -2101,12 +2386,19 @@ _mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); } + _mesa_lock_texture(ctx, texObj); /* relock */ + _mesa_meta_end(ctx); _mesa_TexParameteri(target, GL_TEXTURE_MIN_FILTER, minFilterSave); _mesa_TexParameteri(target, GL_TEXTURE_MAG_FILTER, magFilterSave); + _mesa_TexParameteri(target, GL_TEXTURE_BASE_LEVEL, baseLevelSave); + _mesa_TexParameteri(target, GL_TEXTURE_MAX_LEVEL, maxLevelSave); + _mesa_TexParameteri(target, GL_GENERATE_MIPMAP, genMipmapSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_S, wrapSSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_T, wrapTSave); + _mesa_TexParameteri(target, GL_TEXTURE_WRAP_R, wrapRSave); - /* restore (XXX add to meta_begin/end()? */ _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, fboSave); } @@ -2180,6 +2472,12 @@ copy_tex_image(GLcontext *ctx, GLuint dims, GLenum target, GLint level, return; } + if (texImage->TexFormat == &_mesa_null_texformat) + texImage->TexFormat = ctx->Driver.ChooseTextureFormat(ctx, + internalFormat, + format, + type); + _mesa_unlock_texture(ctx, texObj); /* need to unlock first */ /* diff --git a/src/mesa/drivers/common/meta.h b/src/mesa/drivers/common/meta.h index 7f659528dc0..6225b941893 100644 --- a/src/mesa/drivers/common/meta.h +++ b/src/mesa/drivers/common/meta.h @@ -60,6 +60,10 @@ _mesa_meta_Bitmap(GLcontext *ctx, const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap); +extern GLboolean +_mesa_meta_check_generate_mipmap_fallback(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj); + extern void _mesa_meta_GenerateMipmap(GLcontext *ctx, GLenum target, struct gl_texture_object *texObj); diff --git a/src/mesa/drivers/dri/glcore/Makefile b/src/mesa/drivers/dri/glcore/Makefile deleted file mode 100644 index ac7e1de9285..00000000000 --- a/src/mesa/drivers/dri/glcore/Makefile +++ /dev/null @@ -1,84 +0,0 @@ -# src/mesa/drivers/dri/glcore/Makefile - -TOP = ../../../../.. -include $(TOP)/configs/current - -LIBNAME = glcore_dri.so - -DRIVER_SOURCES = glcore_driver.c \ - $(TOP)/src/mesa/drivers/common/driverfuncs.c \ - ../common/dri_util.c - -C_SOURCES = \ - $(DRIVER_SOURCES) \ - $(DRI_SOURCES) - - -# Include directories -INCLUDE_DIRS = \ - -I. \ - -I../common \ - -I../dri_client \ - -I../dri_client/imports \ - -Iserver \ - -I$(TOP)/include \ - -I$(DRM_SOURCE_PATH)/shared-core \ - -I$(TOP)/src/mesa \ - -I$(TOP)/src/mesa/main \ - -I$(TOP)/src/mesa/glapi \ - -I$(TOP)/src/mesa/math \ - -I$(TOP)/src/mesa/transform \ - -I$(TOP)/src/mesa/shader \ - -I$(TOP)/src/mesa/swrast \ - -I$(TOP)/src/mesa/swrast_setup - -# Core Mesa objects -MESA_MODULES = $(TOP)/src/mesa/libmesa.a - -# Libraries that the driver shared lib depends on -LIB_DEPS = -lm -lpthread -lc -# LIB_DEPS = -lGL -lm -lpthread -lc - - -ASM_SOURCES = - -OBJECTS = $(C_SOURCES:.c=.o) \ - $(ASM_SOURCES:.S=.o) - - -##### RULES ##### - -.c.o: - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES) $< -o $@ - -.S.o: - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES) $< -o $@ - - -##### TARGETS ##### - -default: depend $(TOP)/$(LIB_DIR)/$(LIBNAME) - - -$(TOP)/$(LIB_DIR)/$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(WINOBJ) Makefile - CC="$(CC)" CXX="$(CXX)" $(TOP)/bin/mklib -o $(LIBNAME) -noprefix -install $(TOP)/$(LIB_DIR) \ - $(OBJECTS) $(WINLIB) $(LIB_DEPS) $(WINOBJ) $(MESA_MODULES) - - -depend: $(C_SOURCES) $(ASM_SOURCES) - rm -f depend - touch depend - $(MKDEP) $(MKDEP_OPTIONS) $(INCLUDE_DIRS) $(C_SOURCES) $(ASM_SOURCES) \ - > /dev/null - - -# Emacs tags -tags: - etags `find . -name \*.[ch]` `find ../include` - - -clean: - -rm -f *.o server/*.o - - -include depend diff --git a/src/mesa/drivers/dri/i915/Makefile b/src/mesa/drivers/dri/i915/Makefile index 9d049dea8fe..393312e7328 100644 --- a/src/mesa/drivers/dri/i915/Makefile +++ b/src/mesa/drivers/dri/i915/Makefile @@ -19,7 +19,6 @@ DRIVER_SOURCES = \ intel_batchbuffer.c \ intel_clear.c \ intel_extensions.c \ - intel_generatemipmap.c \ intel_mipmap_tree.c \ intel_tex_layout.c \ intel_tex_image.c \ diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c index 3ab7d682ee7..7d4c7cfbabb 100644 --- a/src/mesa/drivers/dri/i915/i915_context.c +++ b/src/mesa/drivers/dri/i915/i915_context.c @@ -40,6 +40,7 @@ #include "utils.h" #include "i915_reg.h" +#include "i915_program.h" #include "intel_regions.h" #include "intel_batchbuffer.h" @@ -80,6 +81,8 @@ i915InvalidateState(GLcontext * ctx, GLuint new_state) i915_update_stencil(ctx); if (new_state & (_NEW_LIGHT)) i915_update_provoking_vertex(ctx); + if (new_state & (_NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS)) + i915_update_program(ctx); } @@ -139,7 +142,7 @@ i915CreateContext(const __GLcontextModes * mesaVis, ctx->Const.MaxTextureUnits = I915_TEX_UNITS; ctx->Const.MaxTextureImageUnits = I915_TEX_UNITS; ctx->Const.MaxTextureCoordUnits = I915_TEX_UNITS; - + ctx->Const.MaxVarying = I915_TEX_UNITS; /* Advertise the full hardware capabilities. The new memory * manager should cope much better with overload situations: diff --git a/src/mesa/drivers/dri/i915/i915_context.h b/src/mesa/drivers/dri/i915/i915_context.h index 8de4a9d0d36..082d6144425 100644 --- a/src/mesa/drivers/dri/i915/i915_context.h +++ b/src/mesa/drivers/dri/i915/i915_context.h @@ -121,10 +121,14 @@ enum { #define I915_MAX_CONSTANT 32 #define I915_CONSTANT_SIZE (2+(4*I915_MAX_CONSTANT)) +#define I915_MAX_INSN (I915_MAX_DECL_INSN + \ + I915_MAX_TEX_INSN + \ + I915_MAX_ALU_INSN) -#define I915_PROGRAM_SIZE 192 - -#define I915_MAX_INSN (I915_MAX_TEX_INSN+I915_MAX_ALU_INSN) +/* Maximum size of the program packet, which matches the limits on + * decl, tex, and ALU instructions. + */ +#define I915_PROGRAM_SIZE (I915_MAX_INSN * 3 + 1) /* Hardware version of a parsed fragment program. "Derived" from the * mesa fragment_program struct. @@ -154,8 +158,9 @@ struct i915_fragment_program */ GLcontext *ctx; - GLuint declarations[I915_PROGRAM_SIZE]; - GLuint program[I915_PROGRAM_SIZE]; + /* declarations contains the packet header. */ + GLuint declarations[I915_MAX_DECL_INSN * 3 + 1]; + GLuint program[(I915_MAX_TEX_INSN + I915_MAX_ALU_INSN) * 3]; GLfloat constant[I915_MAX_CONSTANT][4]; GLuint constant_flags[I915_MAX_CONSTANT]; diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c index 2db10c60e99..d9c61446f52 100644 --- a/src/mesa/drivers/dri/i915/i915_fragprog.c +++ b/src/mesa/drivers/dri/i915/i915_fragprog.c @@ -89,7 +89,8 @@ src_vector(struct i915_fragment_program *p, */ case PROGRAM_TEMPORARY: if (source->Index >= I915_MAX_TEMPORARY) { - i915_program_error(p, "Exceeded max temporary reg"); + i915_program_error(p, "Exceeded max temporary reg: %d/%d", + source->Index, I915_MAX_TEMPORARY); return 0; } src = UREG(REG_TYPE_R, source->Index); @@ -121,10 +122,23 @@ src_vector(struct i915_fragment_program *p, src = i915_emit_decl(p, REG_TYPE_T, T_TEX0 + (source->Index - FRAG_ATTRIB_TEX0), D0_CHANNEL_ALL); + break; + + case FRAG_ATTRIB_VAR0: + case FRAG_ATTRIB_VAR0 + 1: + case FRAG_ATTRIB_VAR0 + 2: + case FRAG_ATTRIB_VAR0 + 3: + case FRAG_ATTRIB_VAR0 + 4: + case FRAG_ATTRIB_VAR0 + 5: + case FRAG_ATTRIB_VAR0 + 6: + case FRAG_ATTRIB_VAR0 + 7: + src = i915_emit_decl(p, REG_TYPE_T, + T_TEX0 + (source->Index - FRAG_ATTRIB_VAR0), + D0_CHANNEL_ALL); break; default: - i915_program_error(p, "Bad source->Index"); + i915_program_error(p, "Bad source->Index: %d", source->Index); return 0; } break; @@ -146,6 +160,7 @@ src_vector(struct i915_fragment_program *p, case PROGRAM_CONSTANT: case PROGRAM_STATE_VAR: case PROGRAM_NAMED_PARAM: + case PROGRAM_UNIFORM: src = i915_emit_param4fv(p, program->Base.Parameters->ParameterValues[source-> @@ -153,7 +168,7 @@ src_vector(struct i915_fragment_program *p, break; default: - i915_program_error(p, "Bad source->File"); + i915_program_error(p, "Bad source->File: %d", source->File); return 0; } @@ -186,13 +201,14 @@ get_result_vector(struct i915_fragment_program *p, p->depth_written = 1; return UREG(REG_TYPE_OD, 0); default: - i915_program_error(p, "Bad inst->DstReg.Index"); + i915_program_error(p, "Bad inst->DstReg.Index: %d", + inst->DstReg.Index); return 0; } case PROGRAM_TEMPORARY: return UREG(REG_TYPE_R, inst->DstReg.Index); default: - i915_program_error(p, "Bad inst->DstReg.File"); + i915_program_error(p, "Bad inst->DstReg.File: %d", inst->DstReg.File); return 0; } } @@ -231,7 +247,7 @@ translate_tex_src_target(struct i915_fragment_program *p, GLubyte bit) case TEXTURE_CUBE_INDEX: return D0_SAMPLE_TYPE_CUBE; default: - i915_program_error(p, "TexSrcBit"); + i915_program_error(p, "TexSrcBit: %d", bit); return 0; } } @@ -351,7 +367,7 @@ upload_program(struct i915_fragment_program *p) while (1) { GLuint src0, src1, src2, flags; - GLuint tmp = 0, consts0 = 0, consts1 = 0; + GLuint tmp = 0, dst, consts0 = 0, consts1 = 0; switch (inst->Opcode) { case OPCODE_ABS: @@ -503,6 +519,10 @@ upload_program(struct i915_fragment_program *p) EMIT_1ARG_ARITH(A0_FLR); break; + case OPCODE_TRUNC: + EMIT_1ARG_ARITH(A0_TRC); + break; + case OPCODE_FRC: EMIT_1ARG_ARITH(A0_FRC); break; @@ -516,6 +536,22 @@ upload_program(struct i915_fragment_program *p) 0, src0, T0_TEXKILL); break; + case OPCODE_KIL_NV: + if (inst->DstReg.CondMask == COND_TR) { + tmp = i915_get_utemp(p); + + i915_emit_texld(p, get_live_regs(p, inst), + tmp, A0_DEST_CHANNEL_ALL, + 0, /* use a dummy dest reg */ + swizzle(tmp, ONE, ONE, ONE, ONE), /* always */ + T0_TEXKILL); + } else { + p->error = 1; + i915_program_error(p, "Unsupported KIL_NV condition code: %d", + inst->DstReg.CondMask); + } + break; + case OPCODE_LG2: src0 = src_vector(p, &inst->SrcReg[0], program); @@ -615,6 +651,20 @@ upload_program(struct i915_fragment_program *p) EMIT_2ARG_ARITH(A0_MUL); break; + case OPCODE_NOISE1: + case OPCODE_NOISE2: + case OPCODE_NOISE3: + case OPCODE_NOISE4: + /* Don't implement noise because we just don't have the instructions + * to spare. We aren't the first vendor to do so. + */ + i915_program_error(p, "Stubbed-out noise functions"); + i915_emit_arith(p, + A0_MOV, + get_result_vector(p, inst), + get_result_flags(inst), 0, + swizzle(src0, ZERO, ZERO, ZERO, ZERO), 0, 0); + case OPCODE_POW: src0 = src_vector(p, &inst->SrcReg[0], program); src1 = src_vector(p, &inst->SrcReg[1], program); @@ -721,9 +771,38 @@ upload_program(struct i915_fragment_program *p) } break; - case OPCODE_SGE: - EMIT_2ARG_ARITH(A0_SGE); - break; + case OPCODE_SEQ: + tmp = i915_get_utemp(p); + flags = get_result_flags(inst); + dst = get_result_vector(p, inst); + + /* dst = src1 >= src2 */ + i915_emit_arith(p, + A0_SGE, + dst, + flags, 0, + src_vector(p, &inst->SrcReg[0], program), + src_vector(p, &inst->SrcReg[1], program), + 0); + /* tmp = src1 <= src2 */ + i915_emit_arith(p, + A0_SGE, + tmp, + flags, 0, + negate(src_vector(p, &inst->SrcReg[0], program), + 1, 1, 1, 1), + negate(src_vector(p, &inst->SrcReg[1], program), + 1, 1, 1, 1), + 0); + /* dst = tmp && dst */ + i915_emit_arith(p, + A0_MUL, + dst, + flags, 0, + dst, + tmp, + 0); + break; case OPCODE_SIN: src0 = src_vector(p, &inst->SrcReg[0], program); @@ -809,10 +888,71 @@ upload_program(struct i915_fragment_program *p) break; + case OPCODE_SGE: + EMIT_2ARG_ARITH(A0_SGE); + break; + + case OPCODE_SGT: + i915_emit_arith(p, + A0_SLT, + get_result_vector( p, inst ), + get_result_flags( inst ), 0, + negate(src_vector( p, &inst->SrcReg[0], program), + 1, 1, 1, 1), + negate(src_vector( p, &inst->SrcReg[1], program), + 1, 1, 1, 1), + 0); + break; + + case OPCODE_SLE: + i915_emit_arith(p, + A0_SGE, + get_result_vector( p, inst ), + get_result_flags( inst ), 0, + negate(src_vector( p, &inst->SrcReg[0], program), + 1, 1, 1, 1), + negate(src_vector( p, &inst->SrcReg[1], program), + 1, 1, 1, 1), + 0); + break; + case OPCODE_SLT: EMIT_2ARG_ARITH(A0_SLT); break; + case OPCODE_SNE: + tmp = i915_get_utemp(p); + flags = get_result_flags(inst); + dst = get_result_vector(p, inst); + + /* dst = src1 < src2 */ + i915_emit_arith(p, + A0_SLT, + dst, + flags, 0, + src_vector(p, &inst->SrcReg[0], program), + src_vector(p, &inst->SrcReg[1], program), + 0); + /* tmp = src1 > src2 */ + i915_emit_arith(p, + A0_SLT, + tmp, + flags, 0, + negate(src_vector(p, &inst->SrcReg[0], program), + 1, 1, 1, 1), + negate(src_vector(p, &inst->SrcReg[1], program), + 1, 1, 1, 1), + 0); + /* dst = tmp || dst */ + i915_emit_arith(p, + A0_ADD, + dst, + flags | A0_DEST_SATURATE, 0, + dst, + tmp, + 0); + break; + case OPCODE_SUB: src0 = src_vector(p, &inst->SrcReg[0], program); src1 = src_vector(p, &inst->SrcReg[1], program); @@ -869,8 +1009,39 @@ upload_program(struct i915_fragment_program *p) case OPCODE_END: return; + case OPCODE_BGNLOOP: + case OPCODE_BGNSUB: + case OPCODE_BRA: + case OPCODE_BRK: + case OPCODE_CAL: + case OPCODE_CONT: + case OPCODE_DDX: + case OPCODE_DDY: + case OPCODE_ELSE: + case OPCODE_ENDIF: + case OPCODE_ENDLOOP: + case OPCODE_ENDSUB: + case OPCODE_IF: + case OPCODE_RET: + p->error = 1; + i915_program_error(p, "Unsupported opcode: %s", + _mesa_opcode_string(inst->Opcode)); + return; + + case OPCODE_EXP: + case OPCODE_LOG: + /* These opcodes are claimed as GLSL, NV_vp, and ARB_vp in + * prog_instruction.h, but apparently GLSL doesn't ever emit them. + * Instead, it translates to EX2 or LG2. + */ + case OPCODE_TXD: + case OPCODE_TXL: + /* These opcodes are claimed by GLSL in prog_instruction.h, but + * only NV_vp/fp appears to emit them. + */ default: - i915_program_error(p, "bad opcode"); + i915_program_error(p, "bad opcode: %s", + _mesa_opcode_string(inst->Opcode)); return; } @@ -906,7 +1077,7 @@ check_wpos(struct i915_fragment_program *p) p->wpos_tex = -1; for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) { - if (inputs & FRAG_BIT_TEX(i)) + if (inputs & (FRAG_BIT_TEX(i) | FRAG_BIT_VAR(i))) continue; else if (inputs & FRAG_BIT_WPOS) { p->wpos_tex = i; @@ -1055,6 +1226,28 @@ i915ProgramStringNotify(GLcontext * ctx, _tnl_program_string(ctx, target, prog); } +void +i915_update_program(GLcontext *ctx) +{ + struct intel_context *intel = intel_context(ctx); + struct i915_context *i915 = i915_context(&intel->ctx); + struct i915_fragment_program *fp = + (struct i915_fragment_program *) ctx->FragmentProgram._Current; + + if (i915->current_program != fp) { + if (i915->current_program) { + i915->current_program->on_hardware = 0; + i915->current_program->params_uptodate = 0; + } + + i915->current_program = fp; + } + + if (!fp->translated) + translate_program(fp); + + FALLBACK(&i915->intel, I915_FALLBACK_PROGRAM, fp->error); +} void i915ValidateFragmentProgram(struct i915_context *i915) @@ -1072,16 +1265,6 @@ i915ValidateFragmentProgram(struct i915_context *i915) GLuint s2 = S2_TEXCOORD_NONE; int i, offset = 0; - if (i915->current_program != p) { - if (i915->current_program) { - i915->current_program->on_hardware = 0; - i915->current_program->params_uptodate = 0; - } - - i915->current_program = p; - } - - /* Important: */ VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr; @@ -1125,6 +1308,14 @@ i915ValidateFragmentProgram(struct i915_context *i915) EMIT_ATTR(_TNL_ATTRIB_TEX0 + i, EMIT_SZ(sz), 0, sz * 4); } + else if (inputsRead & FRAG_BIT_VAR(i)) { + int sz = VB->AttribPtr[_TNL_ATTRIB_GENERIC0 + i]->size; + + s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK); + s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz)); + + EMIT_ATTR(_TNL_ATTRIB_GENERIC0 + i, EMIT_SZ(sz), 0, sz * 4); + } else if (i == p->wpos_tex) { /* If WPOS is required, duplicate the XYZ position data in an diff --git a/src/mesa/drivers/dri/i915/i915_program.c b/src/mesa/drivers/dri/i915/i915_program.c index e87700f8e0a..e7908bd48fc 100644 --- a/src/mesa/drivers/dri/i915/i915_program.c +++ b/src/mesa/drivers/dri/i915/i915_program.c @@ -130,6 +130,7 @@ i915_emit_decl(struct i915_fragment_program *p, *(p->decl++) = (D0_DCL | D0_DEST(reg) | d0_flags); *(p->decl++) = D1_MBZ; *(p->decl++) = D2_MBZ; + assert(p->decl <= p->declarations + ARRAY_SIZE(p->declarations)); p->nr_decl_insn++; return reg; @@ -186,6 +187,11 @@ i915_emit_arith(struct i915_fragment_program * p, p->utemp_flag = old_utemp_flag; /* restore */ } + if (p->csr >= p->program + ARRAY_SIZE(p->program)) { + i915_program_error(p, "Program contains too many instructions"); + return UREG_BAD; + } + *(p->csr++) = (op | A0_DEST(dest) | mask | saturate | A0_SRC0(src0)); *(p->csr++) = (A1_SRC0(src0) | A1_SRC1(src1)); *(p->csr++) = (A2_SRC1(src1) | A2_SRC2(src2)); @@ -270,6 +276,11 @@ GLuint i915_emit_texld( struct i915_fragment_program *p, p->register_phases[GET_UREG_NR(coord)] == p->nr_tex_indirect) p->nr_tex_indirect++; + if (p->csr >= p->program + ARRAY_SIZE(p->program)) { + i915_program_error(p, "Program contains too many instructions"); + return UREG_BAD; + } + *(p->csr++) = (op | T0_DEST( dest ) | T0_SAMPLER( sampler )); @@ -424,12 +435,21 @@ i915_emit_param4fv(struct i915_fragment_program * p, const GLfloat * values) return 0; } - - +/* Warning the user about program errors seems to be quite valuable, from + * our bug reports. It unfortunately means piglit reporting errors + * when we fall back to software due to an unsupportable program, though. + */ void -i915_program_error(struct i915_fragment_program *p, const char *msg) +i915_program_error(struct i915_fragment_program *p, const char *fmt, ...) { - _mesa_problem(NULL, "i915_program_error: %s", msg); + va_list args; + + fprintf(stderr, "i915_program_error: "); + va_start(args, fmt); + vfprintf(stderr, fmt, args); + va_end(args); + + fprintf(stderr, "\n"); p->error = 1; } @@ -511,7 +531,8 @@ i915_upload_program(struct i915_context *i915, GLuint program_size = p->csr - p->program; GLuint decl_size = p->decl - p->declarations; - FALLBACK(&i915->intel, I915_FALLBACK_PROGRAM, p->error); + if (p->error) + return; /* Could just go straight to the batchbuffer from here: */ diff --git a/src/mesa/drivers/dri/i915/i915_program.h b/src/mesa/drivers/dri/i915/i915_program.h index 14a3f08801f..0d17d048653 100644 --- a/src/mesa/drivers/dri/i915/i915_program.h +++ b/src/mesa/drivers/dri/i915/i915_program.h @@ -145,7 +145,7 @@ extern GLuint i915_emit_param4fv(struct i915_fragment_program *p, const GLfloat * values); extern void i915_program_error(struct i915_fragment_program *p, - const char *msg); + const char *fmt, ...); extern void i915_init_program(struct i915_context *i915, struct i915_fragment_program *p); @@ -155,7 +155,6 @@ extern void i915_upload_program(struct i915_context *i915, extern void i915_fini_program(struct i915_fragment_program *p); - - +extern void i915_update_program(GLcontext *ctx); #endif diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c index 9a723d3cd73..9e2523932f1 100644 --- a/src/mesa/drivers/dri/i915/i915_vtbl.c +++ b/src/mesa/drivers/dri/i915/i915_vtbl.c @@ -54,8 +54,7 @@ i915_render_prevalidate(struct intel_context *intel) { struct i915_context *i915 = i915_context(&intel->ctx); - if (!intel->Fallback) - i915ValidateFragmentProgram(i915); + i915ValidateFragmentProgram(i915); } static void diff --git a/src/mesa/drivers/dri/i915/intel_generatemipmap.c b/src/mesa/drivers/dri/i915/intel_generatemipmap.c deleted file mode 120000 index 4c6b37ada01..00000000000 --- a/src/mesa/drivers/dri/i915/intel_generatemipmap.c +++ /dev/null @@ -1 +0,0 @@ -../intel/intel_generatemipmap.c
\ No newline at end of file diff --git a/src/mesa/drivers/dri/i915/intel_tris.c b/src/mesa/drivers/dri/i915/intel_tris.c index a905455342d..0641e6df9d7 100644 --- a/src/mesa/drivers/dri/i915/intel_tris.c +++ b/src/mesa/drivers/dri/i915/intel_tris.c @@ -1076,7 +1076,9 @@ intelRunPipeline(GLcontext * ctx) intel->NewGLState = 0; } + intel_map_vertex_shader_textures(ctx); _tnl_run_pipeline(ctx); + intel_unmap_vertex_shader_textures(ctx); _mesa_unlock_context_textures(ctx); } diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile index 6e9a9a29a39..57dcc915869 100644 --- a/src/mesa/drivers/dri/i965/Makefile +++ b/src/mesa/drivers/dri/i965/Makefile @@ -14,7 +14,6 @@ DRIVER_SOURCES = \ intel_decode.c \ intel_extensions.c \ intel_fbo.c \ - intel_generatemipmap.c \ intel_mipmap_tree.c \ intel_regions.c \ intel_screen.c \ diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c index 3c5b8483197..c300c33adce 100644 --- a/src/mesa/drivers/dri/i965/brw_context.c +++ b/src/mesa/drivers/dri/i965/brw_context.c @@ -125,6 +125,7 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis, /* We want the GLSL compiler to emit code that uses condition codes */ ctx->Shader.EmitCondCodes = GL_TRUE; + ctx->Shader.EmitNVTempInitialization = GL_TRUE; ctx->Const.VertexProgram.MaxNativeInstructions = (16 * 1024); ctx->Const.VertexProgram.MaxAluInstructions = 0; diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h index a5209ac41be..fa3e32c7ff1 100644 --- a/src/mesa/drivers/dri/i965/brw_context.h +++ b/src/mesa/drivers/dri/i965/brw_context.h @@ -705,10 +705,6 @@ void brw_debug_batch(struct intel_context *intel); /*====================================================================== * brw_tex.c */ -void brwUpdateTextureState( struct intel_context *intel ); -void brw_FrameBufferTexInit( struct brw_context *brw, - struct intel_region *region ); -void brw_FrameBufferTexDestroy( struct brw_context *brw ); void brw_validate_textures( struct brw_context *brw ); diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c index 0b0e6931a06..4be6c77aa1e 100644 --- a/src/mesa/drivers/dri/i965/brw_curbe.c +++ b/src/mesa/drivers/dri/i965/brw_curbe.c @@ -248,6 +248,9 @@ static void prepare_constant_buffer(struct brw_context *brw) GLuint offset = brw->curbe.vs_start * 16; GLuint nr = brw->vs.prog_data->nr_params / 4; + if (brw->vertex_program->IsNVProgram) + _mesa_load_tracked_matrices(ctx); + /* Updates the ParamaterValues[i] pointers for all parameters of the * basic type of PROGRAM_STATE_VAR. */ diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h index 78572356a3d..d639656b9d4 100644 --- a/src/mesa/drivers/dri/i965/brw_state.h +++ b/src/mesa/drivers/dri/i965/brw_state.h @@ -86,9 +86,6 @@ const struct brw_tracked_state brw_psp_urb_cbs; const struct brw_tracked_state brw_pipe_control; -const struct brw_tracked_state brw_clear_surface_cache; -const struct brw_tracked_state brw_clear_batch_cache; - const struct brw_tracked_state brw_drawing_rect; const struct brw_tracked_state brw_indices; const struct brw_tracked_state brw_vertices; @@ -154,6 +151,7 @@ void brw_state_cache_check_size( struct brw_context *brw ); void brw_init_caches( struct brw_context *brw ); void brw_destroy_caches( struct brw_context *brw ); +void brw_state_cache_bo_delete(struct brw_cache *cache, dri_bo *bo); /*********************************************************************** * brw_state_batch.c @@ -165,7 +163,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw, const void *data, GLuint sz ); void brw_destroy_batch_cache( struct brw_context *brw ); -void brw_clear_batch_cache_flush( struct brw_context *brw ); +void brw_clear_batch_cache( struct brw_context *brw ); /* brw_wm_surface_state.c */ dri_bo * diff --git a/src/mesa/drivers/dri/i965/brw_state_batch.c b/src/mesa/drivers/dri/i965/brw_state_batch.c index 811940edc05..7821898cf9b 100644 --- a/src/mesa/drivers/dri/i965/brw_state_batch.c +++ b/src/mesa/drivers/dri/i965/brw_state_batch.c @@ -79,7 +79,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw, return GL_TRUE; } -static void clear_batch_cache( struct brw_context *brw ) +void brw_clear_batch_cache( struct brw_context *brw ) { struct brw_cached_batch_item *item = brw->cached_batch_items; @@ -93,18 +93,7 @@ static void clear_batch_cache( struct brw_context *brw ) brw->cached_batch_items = NULL; } -void brw_clear_batch_cache_flush( struct brw_context *brw ) -{ - clear_batch_cache(brw); - - brw->state.dirty.mesa |= ~0; - brw->state.dirty.brw |= ~0; - brw->state.dirty.cache |= ~0; -} - - - void brw_destroy_batch_cache( struct brw_context *brw ) { - clear_batch_cache(brw); + brw_clear_batch_cache(brw); } diff --git a/src/mesa/drivers/dri/i965/brw_state_cache.c b/src/mesa/drivers/dri/i965/brw_state_cache.c index e40d7a04164..f8e46aacf71 100644 --- a/src/mesa/drivers/dri/i965/brw_state_cache.c +++ b/src/mesa/drivers/dri/i965/brw_state_cache.c @@ -517,6 +517,55 @@ brw_clear_cache(struct brw_context *brw, struct brw_cache *cache) brw->state.dirty.cache |= ~0; } +/* Clear all entries from the cache that point to the given bo. + * + * This lets us release memory for reuse earlier for known-dead buffers, + * at the cost of walking the entire hash table. + */ +void +brw_state_cache_bo_delete(struct brw_cache *cache, dri_bo *bo) +{ + struct brw_cache_item **prev; + GLuint i; + + if (INTEL_DEBUG & DEBUG_STATE) + _mesa_printf("%s\n", __FUNCTION__); + + for (i = 0; i < cache->size; i++) { + for (prev = &cache->items[i]; *prev;) { + struct brw_cache_item *c = *prev; + int j; + + for (j = 0; j < c->nr_reloc_bufs; j++) { + if (c->reloc_bufs[j] == bo) + break; + } + + if (j != c->nr_reloc_bufs) { + + *prev = c->next; + + for (j = 0; j < c->nr_reloc_bufs; j++) + dri_bo_unreference(c->reloc_bufs[j]); + dri_bo_unreference(c->bo); + free((void *)c->key); + free(c); + cache->n_items--; + + /* Delete up the tree. Notably we're trying to get from + * a request to delete the surface, to deleting the surface state + * object, to deleting the binding table. We're slack and restart + * the deletion process when we do this because the other delete + * may kill our *prev. + */ + brw_state_cache_bo_delete(cache, c->bo); + prev = &cache->items[i]; + } else { + prev = &(*prev)->next; + } + } + } +} void brw_state_cache_check_size(struct brw_context *brw) diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c index 414620d0b39..b817b741e77 100644 --- a/src/mesa/drivers/dri/i965/brw_state_upload.c +++ b/src/mesa/drivers/dri/i965/brw_state_upload.c @@ -287,6 +287,7 @@ void brw_validate_state( struct brw_context *brw ) if (brw->emit_state_always) { state->mesa |= ~0; state->brw |= ~0; + state->cache |= ~0; } if (brw->fragment_program != ctx->FragmentProgram._Current) { @@ -305,7 +306,7 @@ void brw_validate_state( struct brw_context *brw ) return; if (brw->state.dirty.brw & BRW_NEW_CONTEXT) - brw_clear_batch_cache_flush(brw); + brw_clear_batch_cache(brw); brw->intel.Fallback = 0; diff --git a/src/mesa/drivers/dri/i965/brw_tex.c b/src/mesa/drivers/dri/i965/brw_tex.c index 71bff166dda..e911b105b23 100644 --- a/src/mesa/drivers/dri/i965/brw_tex.c +++ b/src/mesa/drivers/dri/i965/brw_tex.c @@ -39,38 +39,6 @@ #include "intel_tex.h" #include "brw_context.h" - -void brw_FrameBufferTexInit( struct brw_context *brw, - struct intel_region *region ) -{ - struct intel_context *intel = &brw->intel; - GLcontext *ctx = &intel->ctx; - struct gl_texture_object *obj; - struct gl_texture_image *img; - - intel->frame_buffer_texobj = obj = - ctx->Driver.NewTextureObject( ctx, (GLuint) -1, GL_TEXTURE_2D ); - - obj->MinFilter = GL_NEAREST; - obj->MagFilter = GL_NEAREST; - - img = ctx->Driver.NewTextureImage( ctx ); - - _mesa_init_teximage_fields( ctx, GL_TEXTURE_2D, img, - region->pitch, region->height, 1, 0, - region->cpp == 4 ? GL_RGBA : GL_RGB ); - - _mesa_set_tex_image( obj, GL_TEXTURE_2D, 0, img ); -} - -void brw_FrameBufferTexDestroy( struct brw_context *brw ) -{ - if (brw->intel.frame_buffer_texobj != NULL) - brw->intel.ctx.Driver.DeleteTexture( &brw->intel.ctx, - brw->intel.frame_buffer_texobj ); - brw->intel.frame_buffer_texobj = NULL; -} - /** * Finalizes all textures, completing any rendering that needs to be done * to prepare them. diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c index e3111c66800..f0c79efbd96 100644 --- a/src/mesa/drivers/dri/i965/brw_vs.c +++ b/src/mesa/drivers/dri/i965/brw_vs.c @@ -90,8 +90,6 @@ static void brw_upload_vs_prog(struct brw_context *brw) struct brw_vertex_program *vp = (struct brw_vertex_program *)brw->vertex_program; - assert (vp && !vp->program.IsNVProgram); - memset(&key, 0, sizeof(key)); /* Just upload the program verbatim for now. Always send it all diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c index ac117901515..124fde25fe0 100644 --- a/src/mesa/drivers/dri/i965/brw_vtbl.c +++ b/src/mesa/drivers/dri/i965/brw_vtbl.c @@ -69,8 +69,6 @@ static void brw_destroy_context( struct intel_context *intel ) _mesa_free(brw->wm.compile_data); - brw_FrameBufferTexDestroy( brw ); - for (i = 0; i < brw->state.nr_color_regions; i++) intel_region_release(&brw->state.color_regions[i]); brw->state.nr_color_regions = 0; diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c index 51539ac1e73..9c28a22a298 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c +++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c @@ -660,7 +660,7 @@ brw_wm_get_binding_table(struct brw_context *brw) if (bind_bo == NULL) { GLuint data_size = brw->wm.nr_surfaces * sizeof(GLuint); - uint32_t *data = malloc(data_size); + uint32_t data[BRW_WM_MAX_SURF]; int i; for (i = 0; i < brw->wm.nr_surfaces; i++) @@ -685,8 +685,6 @@ brw_wm_get_binding_table(struct brw_context *brw) brw->wm.surf_bo[i]); } } - - free(data); } return bind_bo; @@ -724,17 +722,8 @@ static void prepare_wm_surfaces(struct brw_context *brw ) /* _NEW_TEXTURE, BRW_NEW_TEXDATA */ if (texUnit->_ReallyEnabled) { - if (texUnit->_Current == intel->frame_buffer_texobj) { - /* render to texture */ - dri_bo_unreference(brw->wm.surf_bo[surf]); - brw->wm.surf_bo[surf] = brw->wm.surf_bo[0]; - dri_bo_reference(brw->wm.surf_bo[surf]); - brw->wm.nr_surfaces = surf + 1; - } else { - /* regular texture */ - brw_update_texture_surface(ctx, i); - brw->wm.nr_surfaces = surf + 1; - } + brw_update_texture_surface(ctx, i); + brw->wm.nr_surfaces = surf + 1; } else { dri_bo_unreference(brw->wm.surf_bo[surf]); brw->wm.surf_bo[surf] = NULL; diff --git a/src/mesa/drivers/dri/i965/intel_generatemipmap.c b/src/mesa/drivers/dri/i965/intel_generatemipmap.c deleted file mode 120000 index 4c6b37ada01..00000000000 --- a/src/mesa/drivers/dri/i965/intel_generatemipmap.c +++ /dev/null @@ -1 +0,0 @@ -../intel/intel_generatemipmap.c
\ No newline at end of file diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c index 43141c509c7..9e114db6c78 100644 --- a/src/mesa/drivers/dri/intel/intel_blit.c +++ b/src/mesa/drivers/dri/intel/intel_blit.c @@ -371,8 +371,6 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask) skipBuffers = BUFFER_BIT_STENCIL; } - /* XXX Move this flush/lock into the following conditional? */ - intelFlush(&intel->ctx); LOCK_HARDWARE(intel); intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off); diff --git a/src/mesa/drivers/dri/intel/intel_buffer_objects.c b/src/mesa/drivers/dri/intel/intel_buffer_objects.c index c55c5c426e0..a0225936c81 100644 --- a/src/mesa/drivers/dri/intel/intel_buffer_objects.c +++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.c @@ -207,8 +207,12 @@ intel_bufferobj_subdata(GLcontext * ctx, if (intel_obj->sys_buffer) memcpy((char *)intel_obj->sys_buffer + offset, data, size); - else + else { + /* Flush any existing batchbuffer that might reference this data. */ + intelFlush(ctx); + dri_bo_subdata(intel_obj->buffer, offset, size, data); + } } @@ -225,7 +229,10 @@ intel_bufferobj_get_subdata(GLcontext * ctx, struct intel_buffer_object *intel_obj = intel_buffer_object(obj); assert(intel_obj); - dri_bo_get_subdata(intel_obj->buffer, offset, size, data); + if (intel_obj->sys_buffer) + memcpy(data, (char *)intel_obj->sys_buffer + offset, size); + else + dri_bo_get_subdata(intel_obj->buffer, offset, size, data); } diff --git a/src/mesa/drivers/dri/intel/intel_clear.c b/src/mesa/drivers/dri/intel/intel_clear.c index fef977f4655..fb62f0f430a 100644 --- a/src/mesa/drivers/dri/intel/intel_clear.c +++ b/src/mesa/drivers/dri/intel/intel_clear.c @@ -173,7 +173,6 @@ intelClear(GLcontext *ctx, GLbitfield mask) } _mesa_meta_Clear(&intel->ctx, tri_mask); - intel_batchbuffer_flush(intel->batch); } if (swrast_mask) { diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h index 03e7cf39d68..b104096912c 100644 --- a/src/mesa/drivers/dri/intel/intel_context.h +++ b/src/mesa/drivers/dri/intel/intel_context.h @@ -254,9 +254,6 @@ struct intel_context intel_line_func draw_line; intel_tri_func draw_tri; - /* These refer to the current drawing buffer: - */ - struct gl_texture_object *frame_buffer_texobj; /** * Set to true if a single constant cliprect should be used in the * batchbuffer. Otherwise, cliprects must be calculated at batchbuffer @@ -296,7 +293,6 @@ struct intel_context GLboolean use_texture_tiling; GLboolean use_early_z; - drm_clip_rect_t fboRect; /**< cliprect for FBO rendering */ int perf_boxes; diff --git a/src/mesa/drivers/dri/intel/intel_extensions.c b/src/mesa/drivers/dri/intel/intel_extensions.c index 5431cf90a1b..2eb08a8f057 100644 --- a/src/mesa/drivers/dri/intel/intel_extensions.c +++ b/src/mesa/drivers/dri/intel/intel_extensions.c @@ -80,6 +80,9 @@ static const struct dri_extension card_extensions[] = { { "GL_ARB_multitexture", NULL }, { "GL_ARB_point_parameters", GL_ARB_point_parameters_functions }, { "GL_ARB_point_sprite", NULL }, + { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, + { "GL_ARB_shading_language_100", GL_VERSION_2_0_functions }, + { "GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, { "GL_ARB_sync", GL_ARB_sync_functions }, { "GL_ARB_texture_border_clamp", NULL }, { "GL_ARB_texture_cube_map", NULL }, @@ -91,6 +94,7 @@ static const struct dri_extension card_extensions[] = { { "GL_ARB_texture_rectangle", NULL }, { "GL_ARB_vertex_array_object", GL_ARB_vertex_array_object_functions}, { "GL_ARB_vertex_program", GL_ARB_vertex_program_functions }, + { "GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions }, { "GL_ARB_window_pos", GL_ARB_window_pos_functions }, { "GL_EXT_blend_color", GL_EXT_blend_color_functions }, { "GL_EXT_blend_equation_separate", GL_EXT_blend_equation_separate_functions }, @@ -150,13 +154,9 @@ static const struct dri_extension brw_extensions[] = { { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, { "GL_ARB_point_sprite", NULL }, { "GL_ARB_seamless_cube_map", NULL }, - { "GL_ARB_shader_objects", GL_ARB_shader_objects_functions }, - { "GL_ARB_shading_language_100", GL_VERSION_2_0_functions }, - { "GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, { "GL_ARB_shadow", NULL }, { "GL_MESA_texture_signed_rgba", NULL }, { "GL_ARB_texture_non_power_of_two", NULL }, - { "GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions }, { "GL_EXT_shadow_funcs", NULL }, { "GL_EXT_stencil_two_side", GL_EXT_stencil_two_side_functions }, { "GL_EXT_texture_sRGB", NULL }, @@ -171,6 +171,7 @@ static const struct dri_extension brw_extensions[] = { static const struct dri_extension arb_oq_extensions[] = { + { "GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions }, { NULL, NULL } }; @@ -182,6 +183,10 @@ static const struct dri_extension ttm_extensions[] = { { NULL, NULL } }; +static const struct dri_extension fragment_shader_extensions[] = { + { "GL_ARB_fragment_shader", NULL }, + { NULL, NULL } +}; /** * Initializes potential list of extensions if ctx == NULL, or actually enables @@ -205,6 +210,14 @@ intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging) driInitExtensions(ctx, brw_extensions, GL_FALSE); if (intel == NULL || IS_915(intel->intelScreen->deviceID) - || IS_945(intel->intelScreen->deviceID)) + || IS_945(intel->intelScreen->deviceID)) { driInitExtensions(ctx, i915_extensions, GL_FALSE); + + if (intel == NULL || driQueryOptionb(&intel->optionCache, "fragment_shader")) + driInitExtensions(ctx, fragment_shader_extensions, GL_FALSE); + + if (intel == NULL || driQueryOptionb(&intel->optionCache, + "stub_occlusion_query")) + driInitExtensions(ctx, arb_oq_extensions, GL_FALSE); + } } diff --git a/src/mesa/drivers/dri/intel/intel_generatemipmap.c b/src/mesa/drivers/dri/intel/intel_generatemipmap.c deleted file mode 100644 index 237754d4695..00000000000 --- a/src/mesa/drivers/dri/intel/intel_generatemipmap.c +++ /dev/null @@ -1,279 +0,0 @@ -/* - * Copyright (C) 1999-2007 Brian Paul All Rights Reserved. - * Copyright © 2009 Intel Corporation - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS - * IN THE SOFTWARE. - * - * Authors: - * Eric Anholt <[email protected]> - * - */ - -#include "main/glheader.h" -#include "main/enums.h" -#include "main/image.h" -#include "main/mtypes.h" -#include "main/macros.h" -#include "main/bufferobj.h" -#include "main/teximage.h" -#include "main/texenv.h" -#include "main/texobj.h" -#include "main/texstate.h" -#include "main/texparam.h" -#include "main/varray.h" -#include "main/attrib.h" -#include "main/enable.h" -#include "main/buffers.h" -#include "main/fbobject.h" -#include "main/framebuffer.h" -#include "main/renderbuffer.h" -#include "main/depth.h" -#include "main/hash.h" -#include "main/mipmap.h" -#include "main/blend.h" -#include "glapi/dispatch.h" -#include "swrast/swrast.h" - -#include "intel_screen.h" -#include "intel_context.h" -#include "intel_batchbuffer.h" -#include "intel_pixel.h" -#include "intel_tex.h" -#include "intel_mipmap_tree.h" - -static const char *intel_fp_tex2d = - "!!ARBfp1.0\n" - "TEX result.color, fragment.texcoord[0], texture[0], 2D;\n" - "END\n"; - -static GLboolean -intel_generate_mipmap_level(GLcontext *ctx, GLuint tex_name, - int level, int width, int height) -{ - struct intel_context *intel = intel_context(ctx); - GLfloat vertices[4][2]; - GLint status; - - /* Set to source from the previous level */ - _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, level - 1); - _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, level - 1); - - /* Set to draw into the current level */ - _mesa_FramebufferTexture2DEXT(GL_FRAMEBUFFER_EXT, - GL_COLOR_ATTACHMENT0_EXT, - GL_TEXTURE_2D, - tex_name, - level); - /* Choose to render to the color attachment. */ - _mesa_DrawBuffer(GL_COLOR_ATTACHMENT0_EXT); - - status = _mesa_CheckFramebufferStatusEXT (GL_FRAMEBUFFER_EXT); - if (status != GL_FRAMEBUFFER_COMPLETE_EXT) - return GL_FALSE; - - meta_set_passthrough_transform(&intel->meta); - - /* XXX: Doing it right would involve setting up the transformation to do - * 0-1 mapping or something, and not changing the vertex data. - */ - vertices[0][0] = 0; - vertices[0][1] = 0; - vertices[1][0] = width; - vertices[1][1] = 0; - vertices[2][0] = width; - vertices[2][1] = height; - vertices[3][0] = 0; - vertices[3][1] = height; - - _mesa_VertexPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &vertices); - _mesa_Enable(GL_VERTEX_ARRAY); - meta_set_default_texrect(&intel->meta); - - _mesa_DrawArrays(GL_TRIANGLE_FAN, 0, 4); - - meta_restore_texcoords(&intel->meta); - meta_restore_transform(&intel->meta); - - return GL_TRUE; -} - -static GLboolean -intel_generate_mipmap_2d(GLcontext *ctx, - GLenum target, - struct gl_texture_object *texObj) -{ - struct intel_context *intel = intel_context(ctx); - GLint old_active_texture; - int level, max_levels, start_level, end_level; - GLuint fb_name; - GLboolean success = GL_FALSE; - struct gl_framebuffer *saved_fbo = NULL; - - _mesa_PushAttrib(GL_ENABLE_BIT | GL_TEXTURE_BIT | - GL_CURRENT_BIT | GL_COLOR_BUFFER_BIT | - GL_DEPTH_BUFFER_BIT); - _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT); - old_active_texture = ctx->Texture.CurrentUnit; - _mesa_reference_framebuffer(&saved_fbo, ctx->DrawBuffer); - - _mesa_Disable(GL_POLYGON_STIPPLE); - _mesa_Disable(GL_DEPTH_TEST); - _mesa_Disable(GL_STENCIL_TEST); - _mesa_ColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE); - _mesa_DepthMask(GL_FALSE); - - /* Bind the given texture to GL_TEXTURE_2D with linear filtering for our - * minification. - */ - _mesa_ActiveTextureARB(GL_TEXTURE0_ARB); - _mesa_Enable(GL_TEXTURE_2D); - _mesa_BindTexture(GL_TEXTURE_2D, texObj->Name); - _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, - GL_LINEAR_MIPMAP_NEAREST); - _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); - - /* Bind the new renderbuffer to the color attachment point. */ - _mesa_GenFramebuffersEXT(1, &fb_name); - _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, fb_name); - - meta_set_fragment_program(&intel->meta, &intel->meta.tex2d_fp, - intel_fp_tex2d); - meta_set_passthrough_vertex_program(&intel->meta); - - max_levels = _mesa_max_texture_levels(ctx, texObj->Target); - start_level = texObj->BaseLevel; - end_level = texObj->MaxLevel; - - /* Loop generating level+1 from level. */ - for (level = start_level; level < end_level && level < max_levels - 1; level++) { - const struct gl_texture_image *srcImage; - int width, height; - - srcImage = _mesa_select_tex_image(ctx, texObj, target, level); - if (srcImage->Border != 0) - goto fail; - - width = srcImage->Width / 2; - if (width < 1) - width = 1; - height = srcImage->Height / 2; - if (height < 1) - height = 1; - - if (width == srcImage->Width && - height == srcImage->Height) { - /* Neither _mesa_max_texture_levels nor texObj->MaxLevel are the - * maximum texture level for the object, so break out when we've gone - * over the edge. - */ - break; - } - - /* Make sure that there's space allocated for the target level. - * We could skip this if there's already space allocated and save some - * time. - */ - _mesa_TexImage2D(GL_TEXTURE_2D, level + 1, srcImage->InternalFormat, - width, height, 0, - GL_RGBA, GL_UNSIGNED_INT, NULL); - - if (!intel_generate_mipmap_level(ctx, texObj->Name, level + 1, - width, height)) - goto fail; - } - - success = GL_TRUE; - -fail: - meta_restore_fragment_program(&intel->meta); - meta_restore_vertex_program(&intel->meta); - - _mesa_DeleteFramebuffersEXT(1, &fb_name); - _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture); - if (saved_fbo) - _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, saved_fbo->Name); - _mesa_reference_framebuffer(&saved_fbo, NULL); - _mesa_PopClientAttrib(); - _mesa_PopAttrib(); - - return success; -} - - -/** - * Generate new mipmap data from BASE+1 to BASE+p (the minimally-sized mipmap - * level). - * - * The texture object's miptree must be mapped. - * - * This function should also include an accelerated path. - */ -void -intel_generate_mipmap(GLcontext *ctx, GLenum target, - struct gl_texture_object *texObj) -{ - struct intel_context *intel = intel_context(ctx); - struct intel_texture_object *intelObj = intel_texture_object(texObj); - GLuint nr_faces = (intelObj->base.Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1; - int face, i; - - /* HW path */ - if (target == GL_TEXTURE_2D && - ctx->Extensions.EXT_framebuffer_object && - ctx->Extensions.ARB_fragment_program && - ctx->Extensions.ARB_vertex_program) { - GLboolean success; - - /* We'll be accessing this texture using GL entrypoints, which should - * be resilient against other access to this texture. - */ - _mesa_unlock_texture(ctx, texObj); - success = intel_generate_mipmap_2d(ctx, target, texObj); - _mesa_lock_texture(ctx, texObj); - - if (success) - return; - } - - /* SW path */ - intel_tex_map_level_images(intel, intelObj, texObj->BaseLevel); - _mesa_generate_mipmap(ctx, target, texObj); - intel_tex_unmap_level_images(intel, intelObj, texObj->BaseLevel); - - /* Update the level information in our private data in the new images, since - * it didn't get set as part of a normal TexImage path. - */ - for (face = 0; face < nr_faces; face++) { - for (i = texObj->BaseLevel + 1; i < texObj->MaxLevel; i++) { - struct intel_texture_image *intelImage; - - intelImage = intel_texture_image(texObj->Image[face][i]); - if (intelImage == NULL) - break; - - intelImage->level = i; - intelImage->face = face; - /* Unreference the miptree to signal that the new Data is a bare - * pointer from mesa. - */ - intel_miptree_release(intel, &intelImage->mt); - } - } -} diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c index c985da5aa25..4f5101a3128 100644 --- a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c +++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c @@ -29,6 +29,9 @@ #include "intel_mipmap_tree.h" #include "intel_regions.h" #include "intel_chipset.h" +#ifndef I915 +#include "brw_state.h" +#endif #include "main/enums.h" #define FILE_DEBUG_FLAG DEBUG_MIPTREE @@ -269,6 +272,19 @@ intel_miptree_release(struct intel_context *intel, DBG("%s deleting %p\n", __FUNCTION__, *mt); +#ifndef I915 + /* Free up cached binding tables holding a reference on our buffer, to + * avoid excessive memory consumption. + * + * This isn't as aggressive as we could be, as we'd like to do + * it from any time we free the last ref on a region. But intel_region.c + * is context-agnostic. Perhaps our constant state cache should be, as + * well. + */ + brw_state_cache_bo_delete(&brw_context(&intel->ctx)->surface_cache, + (*mt)->region->buffer); +#endif + intel_region_release(&((*mt)->region)); for (i = 0; i < MAX_TEXTURE_LEVELS; i++) diff --git a/src/mesa/drivers/dri/intel/intel_pixel.c b/src/mesa/drivers/dri/intel/intel_pixel.c index a3001416559..993e427a992 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel.c +++ b/src/mesa/drivers/dri/intel/intel_pixel.c @@ -129,20 +129,6 @@ intel_check_blit_fragment_ops(GLcontext * ctx, GLboolean src_alpha_is_one) return GL_TRUE; } - -GLboolean -intel_check_meta_tex_fragment_ops(GLcontext * ctx) -{ - if (ctx->NewState) - _mesa_update_state(ctx); - - /* Some of _ImageTransferState (scale, bias) could be done with - * fragment programs on i915. - */ - return !(ctx->_ImageTransferState || ctx->Fog.Enabled || /* not done yet */ - ctx->Texture._EnabledUnits || ctx->FragmentProgram._Enabled); -} - /* The intel_region struct doesn't really do enough to capture the * format of the pixels in the region. For now this code assumes that * the region is a display surface and hence is either ARGB8888 or diff --git a/src/mesa/drivers/dri/intel/intel_pixel.h b/src/mesa/drivers/dri/intel/intel_pixel.h index 96a6dd17b25..743b6497c52 100644 --- a/src/mesa/drivers/dri/intel/intel_pixel.h +++ b/src/mesa/drivers/dri/intel/intel_pixel.h @@ -34,8 +34,6 @@ void intelInitPixelFuncs(struct dd_function_table *functions); GLboolean intel_check_blit_fragment_ops(GLcontext * ctx, GLboolean src_alpha_is_one); -GLboolean intel_check_meta_tex_fragment_ops(GLcontext * ctx); - GLboolean intel_check_blit_format(struct intel_region *region, GLenum format, GLenum type); diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c index 1b8c56e68d6..24f7fbc9922 100644 --- a/src/mesa/drivers/dri/intel/intel_screen.c +++ b/src/mesa/drivers/dri/intel/intel_screen.c @@ -79,6 +79,10 @@ PUBLIC const char __driConfigOptions[] = DRI_CONF_DESC(en, "Enable early Z in classic mode (unstable, 945-only).") DRI_CONF_OPT_END + DRI_CONF_OPT_BEGIN(fragment_shader, bool, false) + DRI_CONF_DESC(en, "Enable limited ARB_fragment_shader support on 915/945.") + DRI_CONF_OPT_END + DRI_CONF_SECTION_END DRI_CONF_SECTION_QUALITY DRI_CONF_FORCE_S3TC_ENABLE(false) @@ -88,10 +92,14 @@ PUBLIC const char __driConfigOptions[] = DRI_CONF_NO_RAST(false) DRI_CONF_ALWAYS_FLUSH_BATCH(false) DRI_CONF_ALWAYS_FLUSH_CACHE(false) + + DRI_CONF_OPT_BEGIN(stub_occlusion_query, bool, false) + DRI_CONF_DESC(en, "Enable stub ARB_occlusion_query support on 915/945.") + DRI_CONF_OPT_END DRI_CONF_SECTION_END DRI_CONF_END; -const GLuint __driNConfigOptions = 10; +const GLuint __driNConfigOptions = 12; #ifdef USE_NEW_INTERFACE static PFNGLXCREATECONTEXTMODES create_context_modes = NULL; diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c index 28eabbc0054..dcfcad1d952 100644 --- a/src/mesa/drivers/dri/intel/intel_span.c +++ b/src/mesa/drivers/dri/intel/intel_span.c @@ -558,6 +558,43 @@ intelInitSpanFuncs(GLcontext * ctx) swdd->SpanRenderFinish = intelSpanRenderFinish; } +void +intel_map_vertex_shader_textures(GLcontext *ctx) +{ + struct intel_context *intel = intel_context(ctx); + int i; + + if (ctx->VertexProgram._Current == NULL) + return; + + for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) { + if (ctx->Texture.Unit[i]._ReallyEnabled && + ctx->VertexProgram._Current->Base.TexturesUsed[i] != 0) { + struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current; + + intel_tex_map_images(intel, intel_texture_object(texObj)); + } + } +} + +void +intel_unmap_vertex_shader_textures(GLcontext *ctx) +{ + struct intel_context *intel = intel_context(ctx); + int i; + + if (ctx->VertexProgram._Current == NULL) + return; + + for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) { + if (ctx->Texture.Unit[i]._ReallyEnabled && + ctx->VertexProgram._Current->Base.TexturesUsed[i] != 0) { + struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current; + + intel_tex_unmap_images(intel, intel_texture_object(texObj)); + } + } +} /** * Plug in appropriate span read/write functions for the given renderbuffer. diff --git a/src/mesa/drivers/dri/intel/intel_span.h b/src/mesa/drivers/dri/intel/intel_span.h index acbeb4abe1c..bffe109aa5b 100644 --- a/src/mesa/drivers/dri/intel/intel_span.h +++ b/src/mesa/drivers/dri/intel/intel_span.h @@ -36,5 +36,7 @@ void intel_renderbuffer_map(struct intel_context *intel, struct gl_renderbuffer *rb); void intel_renderbuffer_unmap(struct intel_context *intel, struct gl_renderbuffer *rb); +void intel_map_vertex_shader_textures(GLcontext *ctx); +void intel_unmap_vertex_shader_textures(GLcontext *ctx); #endif diff --git a/src/mesa/drivers/dri/intel/intel_tex.c b/src/mesa/drivers/dri/intel/intel_tex.c index df63f29a42c..3cbc379dbd3 100644 --- a/src/mesa/drivers/dri/intel/intel_tex.c +++ b/src/mesa/drivers/dri/intel/intel_tex.c @@ -2,6 +2,7 @@ #include "main/texobj.h" #include "main/teximage.h" #include "main/mipmap.h" +#include "drivers/common/meta.h" #include "intel_context.h" #include "intel_mipmap_tree.h" #include "intel_tex.h" @@ -158,11 +159,36 @@ timed_memcpy(void *dest, const void *src, size_t n) } #endif /* DO_DEBUG */ + +/** + * Called via ctx->Driver.GenerateMipmap() + * This is basically a wrapper for _mesa_meta_GenerateMipmap() which checks + * if we'll be using software mipmap generation. In that case, we need to + * map/unmap the base level texture image. + */ +static void +intelGenerateMipmap(GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj) +{ + if (_mesa_meta_check_generate_mipmap_fallback(ctx, target, texObj)) { + /* sw path: need to map texture images */ + struct intel_context *intel = intel_context(ctx); + struct intel_texture_object *intelObj = intel_texture_object(texObj); + intel_tex_map_level_images(intel, intelObj, texObj->BaseLevel); + _mesa_generate_mipmap(ctx, target, texObj); + intel_tex_unmap_level_images(intel, intelObj, texObj->BaseLevel); + } + else { + _mesa_meta_GenerateMipmap(ctx, target, texObj); + } +} + + void intelInitTextureFuncs(struct dd_function_table *functions) { functions->ChooseTextureFormat = intelChooseTextureFormat; - functions->GenerateMipmap = intel_generate_mipmap; + functions->GenerateMipmap = intelGenerateMipmap; functions->NewTextureObject = intelNewTextureObject; functions->NewTextureImage = intelNewTextureImage; diff --git a/src/mesa/drivers/dri/intel/intel_tex.h b/src/mesa/drivers/dri/intel/intel_tex.h index 471aa2a240b..57ed0b1aabd 100644 --- a/src/mesa/drivers/dri/intel/intel_tex.h +++ b/src/mesa/drivers/dri/intel/intel_tex.h @@ -71,7 +71,4 @@ void intel_tex_unmap_images(struct intel_context *intel, int intel_compressed_num_bytes(GLuint mesaFormat); -void intel_generate_mipmap(GLcontext *ctx, GLenum target, - struct gl_texture_object *texObj); - #endif diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c index ac557a92005..8f467e65144 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_copy.c +++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c @@ -75,6 +75,7 @@ get_teximage_source(struct intel_context *intel, GLenum internalFormat) case GL_RGBA: case GL_RGBA8: case GL_RGB: + case GL_RGB8: return intel_readbuf_region(intel); default: return NULL; diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_common.c b/src/mesa/drivers/dri/r300/r300_fragprog_common.c index 7f8a0174595..39618b78dc6 100644 --- a/src/mesa/drivers/dri/r300/r300_fragprog_common.c +++ b/src/mesa/drivers/dri/r300/r300_fragprog_common.c @@ -100,8 +100,8 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler, { int i; + fp->wpos_attr = FRAG_ATTRIB_MAX; if (!(compiler->Base.Program.InputsRead & FRAG_BIT_WPOS)) { - fp->wpos_attr = FRAG_ATTRIB_MAX; return; } @@ -113,6 +113,13 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler, } } + /* No free texcoord found, fall-back to software rendering */ + if (fp->wpos_attr == FRAG_ATTRIB_MAX) + { + compiler->Base.Error = 1; + return; + } + rc_transform_fragment_wpos(&compiler->Base, FRAG_ATTRIB_WPOS, fp->wpos_attr); } @@ -128,8 +135,8 @@ static void rewriteFog(struct r300_fragment_program_compiler *compiler, struct r struct rc_src_register src; int i; + fp->fog_attr = FRAG_ATTRIB_MAX; if (!(compiler->Base.Program.InputsRead & FRAG_BIT_FOGC)) { - fp->fog_attr = FRAG_ATTRIB_MAX; return; } @@ -141,6 +148,13 @@ static void rewriteFog(struct r300_fragment_program_compiler *compiler, struct r } } + /* No free texcoord found, fall-back to software rendering */ + if (fp->fog_attr == FRAG_ATTRIB_MAX) + { + compiler->Base.Error = 1; + return; + } + memset(&src, 0, sizeof(src)); src.File = RC_FILE_INPUT; src.Index = fp->fog_attr; diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c index b5ddfdc9f82..3cd38753b8a 100644 --- a/src/mesa/drivers/dri/r300/r300_render.c +++ b/src/mesa/drivers/dri/r300/r300_render.c @@ -475,7 +475,7 @@ void r300SwitchFallback(GLcontext *ctx, uint32_t bit, GLboolean mode) /* update only if we have disabled all tcl fallbacks */ if (rmesa->options.hw_tcl_enabled) { - if ((old_fallback & R300_RASTER_FALLBACK_MASK) == bit) { + if ((old_fallback & R300_TCL_FALLBACK_MASK) == bit) { R300_STATECHANGE(rmesa, vap_cntl_status); rmesa->hw.vap_cntl_status.cmd[1] &= ~R300_VAP_TCL_BYPASS; } diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c index 8ad75fb5c06..fb8d6bceda8 100644 --- a/src/mesa/drivers/dri/r300/r300_vertprog.c +++ b/src/mesa/drivers/dri/r300/r300_vertprog.c @@ -43,6 +43,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. #include "compiler/radeon_compiler.h" #include "radeon_mesa_to_rc.h" #include "r300_context.h" +#include "r300_fragprog_common.h" #include "r300_state.h" /** @@ -298,6 +299,20 @@ struct r300_vertex_program * r300SelectAndTranslateVertexShader(GLcontext *ctx) struct r300_vertex_program *vp; vpc = (struct r300_vertex_program_cont *)ctx->VertexProgram._Current; + + if (!r300->selected_fp) { + /* This can happen when GetProgramiv is called to check + * whether the program runs natively. + * + * To be honest, this is not a very good solution, + * but solving the problem of reporting good values + * for those queries is tough anyway considering that + * we recompile vertex programs based on the precise + * fragment program that is in use. + */ + r300SelectAndTranslateFragmentShader(ctx); + } + wanted_key.FpReads = r300->selected_fp->InputsRead; wanted_key.FogAttr = r300->selected_fp->fog_attr; wanted_key.WPosAttr = r300->selected_fp->wpos_attr; diff --git a/src/mesa/drivers/dri/r600/Makefile b/src/mesa/drivers/dri/r600/Makefile index 36bf773c054..7d5a7b1ab6f 100644 --- a/src/mesa/drivers/dri/r600/Makefile +++ b/src/mesa/drivers/dri/r600/Makefile @@ -29,6 +29,7 @@ COMMON_SOURCES = \ RADEON_COMMON_SOURCES = \ radeon_bo_legacy.c \ radeon_common_context.c \ + radeon_buffer_objects.c \ radeon_common.c \ radeon_cs_legacy.c \ radeon_dma.c \ diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c index f8fd9c13d77..969144ba123 100644 --- a/src/mesa/drivers/dri/r600/r600_context.c +++ b/src/mesa/drivers/dri/r600/r600_context.c @@ -59,6 +59,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "radeon_debug.h" #include "r600_context.h" #include "radeon_common_context.h" +#include "radeon_buffer_objects.h" #include "radeon_span.h" #include "r600_cmdbuf.h" #include "r600_emit.h" @@ -257,6 +258,7 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, r600InitTextureFuncs(&functions); r700InitShaderFuncs(&functions); r700InitIoctlFuncs(&functions); + radeonInitBufferObjectFuncs(&functions); if (!radeonInitContext(&r600->radeon, &functions, glVisual, driContextPriv, @@ -284,8 +286,8 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, ctx->Const.MaxTextureMaxAnisotropy = 16.0; ctx->Const.MaxTextureLodBias = 16.0; - ctx->Const.MaxTextureLevels = 13; - ctx->Const.MaxTextureRectSize = 4096; + ctx->Const.MaxTextureLevels = 13; /* hw support 14 */ + ctx->Const.MaxTextureRectSize = 4096; /* hw support 8192 */ ctx->Const.MinPointSize = 0x0001 / 8.0; ctx->Const.MinPointSizeAA = 0x0001 / 8.0; @@ -330,26 +332,27 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, _tnl_allow_pixel_fog(ctx, GL_FALSE); _tnl_allow_vertex_fog(ctx, GL_TRUE); - /* currently bogus data */ - ctx->Const.VertexProgram.MaxInstructions = VSF_MAX_FRAGMENT_LENGTH / 4; - ctx->Const.VertexProgram.MaxNativeInstructions = - VSF_MAX_FRAGMENT_LENGTH / 4; - ctx->Const.VertexProgram.MaxNativeAttribs = 16; /* r420 */ - ctx->Const.VertexProgram.MaxTemps = 32; - ctx->Const.VertexProgram.MaxNativeTemps = - /*VSF_MAX_FRAGMENT_TEMPS */ 32; - ctx->Const.VertexProgram.MaxNativeParameters = 256; /* r420 */ - ctx->Const.VertexProgram.MaxNativeAddressRegs = 1; - - ctx->Const.FragmentProgram.MaxNativeTemps = PFS_NUM_TEMP_REGS; - ctx->Const.FragmentProgram.MaxNativeAttribs = 11; /* copy i915... */ - ctx->Const.FragmentProgram.MaxNativeParameters = PFS_NUM_CONST_REGS; - ctx->Const.FragmentProgram.MaxNativeAluInstructions = PFS_MAX_ALU_INST; - ctx->Const.FragmentProgram.MaxNativeTexInstructions = PFS_MAX_TEX_INST; - ctx->Const.FragmentProgram.MaxNativeInstructions = - PFS_MAX_ALU_INST + PFS_MAX_TEX_INST; - ctx->Const.FragmentProgram.MaxNativeTexIndirections = - PFS_MAX_TEX_INDIRECT; + /* 256 for reg-based consts, inline consts also supported */ + ctx->Const.VertexProgram.MaxInstructions = 8192; /* in theory no limit */ + ctx->Const.VertexProgram.MaxNativeInstructions = 8192; + ctx->Const.VertexProgram.MaxNativeAttribs = 160; + ctx->Const.VertexProgram.MaxTemps = 128; + ctx->Const.VertexProgram.MaxNativeTemps = 128; + ctx->Const.VertexProgram.MaxNativeParameters = 256; + ctx->Const.VertexProgram.MaxNativeAddressRegs = 1; /* ??? */ + + ctx->Const.FragmentProgram.MaxNativeTemps = 128; + ctx->Const.FragmentProgram.MaxNativeAttribs = 32; + ctx->Const.FragmentProgram.MaxNativeParameters = 256; + ctx->Const.FragmentProgram.MaxNativeAluInstructions = 8192; + /* 8 per clause on r6xx, 16 on rv670/r7xx */ + if ((screen->chip_family == CHIP_FAMILY_RV670) || + (screen->chip_family >= CHIP_FAMILY_RV770)) + ctx->Const.FragmentProgram.MaxNativeTexInstructions = 16; + else + ctx->Const.FragmentProgram.MaxNativeTexInstructions = 8; + ctx->Const.FragmentProgram.MaxNativeInstructions = 8192; + ctx->Const.FragmentProgram.MaxNativeTexIndirections = 8; /* ??? */ ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0; /* and these are?? */ ctx->VertexProgram._MaintainTnlProgram = GL_TRUE; ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE; @@ -374,6 +377,8 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc"); } + r700InitDraw(ctx); + radeon_fbo_init(&r600->radeon); radeonInitSpanFuncs( ctx ); diff --git a/src/mesa/drivers/dri/r600/r600_context.h b/src/mesa/drivers/dri/r600/r600_context.h index c59df7505af..7f68820fda1 100644 --- a/src/mesa/drivers/dri/r600/r600_context.h +++ b/src/mesa/drivers/dri/r600/r600_context.h @@ -86,29 +86,10 @@ extern int hw_tcl_on; #include "tnl_dd/t_dd_vertex.h" #undef TAG -#define PFS_MAX_ALU_INST 64 -#define PFS_MAX_TEX_INST 64 -#define PFS_MAX_TEX_INDIRECT 4 -#define PFS_NUM_TEMP_REGS 32 -#define PFS_NUM_CONST_REGS 16 - -#define R600_MAX_AOS_ARRAYS 16 - -#define REG_COORDS 0 -#define REG_COLOR0 1 -#define REG_TEX0 2 - #define R600_FALLBACK_NONE 0 #define R600_FALLBACK_TCL 1 #define R600_FALLBACK_RAST 2 -enum -{ - NO_SHIFT = 0, - LEFT_SHIFT = 1, - RIGHT_SHIFT = 2, -}; - struct r600_hw_state { struct radeon_state_atom sq; struct radeon_state_atom db; @@ -145,6 +126,32 @@ struct r600_hw_state { struct radeon_state_atom tx_brdr_clr; }; +typedef struct StreamDesc +{ + GLint size; //number of data element + GLenum type; //data element type + GLsizei stride; + + struct radeon_bo *bo; + GLint bo_offset; + + GLuint dwords; + GLuint dst_loc; + GLuint _signed; + GLboolean normalize; + GLboolean is_named_bo; + GLubyte element; +} StreamDesc; + +typedef struct r700_index_buffer +{ + struct radeon_bo *bo; + int bo_offset; + + GLboolean is_32bit; + GLuint count; +} r700_index_buffer; + /** * \brief R600 context structure. */ @@ -163,6 +170,9 @@ struct r600_context { GLvector4f dummy_attrib[_TNL_ATTRIB_MAX]; GLvector4f *temp_attrib[_TNL_ATTRIB_MAX]; + GLint nNumActiveAos; + StreamDesc stream_desc[VERT_ATTRIB_MAX]; + struct r700_index_buffer ind_buf; }; #define R700_CONTEXT(ctx) ((context_t *)(ctx->DriverCtx)) @@ -196,6 +206,7 @@ extern GLboolean r700SyncSurf(context_t *context, extern void r700SetupStreams(GLcontext * ctx); extern void r700Start3D(context_t *context); extern void r600InitAtoms(context_t *context); +extern void r700InitDraw(GLcontext *ctx); #define RADEON_D_CAPTURE 0 #define RADEON_D_PLAYBACK 1 diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c index efeccb25f1e..903b6968be1 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.c +++ b/src/mesa/drivers/dri/r600/r700_assembler.c @@ -213,7 +213,7 @@ GLboolean is_reduction_opcode(PVSDWORD* dest) { if (dest->dst.op3 == 0) { - if ( (dest->dst.opcode == SQ_OP2_INST_DOT4 || dest->dst.opcode == SQ_OP2_INST_DOT4_IEEE) ) + if ( (dest->dst.opcode == SQ_OP2_INST_DOT4 || dest->dst.opcode == SQ_OP2_INST_DOT4_IEEE || dest->dst.opcode == SQ_OP2_INST_CUBE) ) { return GL_TRUE; } @@ -350,6 +350,7 @@ unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm) case SQ_OP2_INST_PRED_SETNE: case SQ_OP2_INST_DOT4: case SQ_OP2_INST_DOT4_IEEE: + case SQ_OP2_INST_CUBE: return 2; case SQ_OP2_INST_MOV: @@ -469,6 +470,9 @@ int Init_r700_AssemblerBase(SHADER_PIPE_TYPE spt, r700_AssemblerBase* pAsm, R700 pAsm->number_of_inputs = 0; + pAsm->is_tex = GL_FALSE; + pAsm->need_tex_barrier = GL_FALSE; + return 0; } @@ -682,7 +686,7 @@ GLboolean add_tex_instruction(r700_AssemblerBase* pAsm, // If this clause constains any TEX instruction that is dependent on a previous instruction, // set the barrier bit - if( pAsm->pInstDeps[pAsm->uiCurInst].nDstDep > (-1) ) + if( pAsm->pInstDeps[pAsm->uiCurInst].nDstDep > (-1) || pAsm->need_tex_barrier == GL_TRUE ) { pAsm->cf_current_tex_clause_ptr->m_Word1.f.barrier = 0x1; } @@ -786,6 +790,133 @@ GLboolean assemble_vfetch_instruction(r700_AssemblerBase* pAsm, return GL_TRUE; } +GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm, + GLuint destination_register, + GLenum type, + GLint size, + GLubyte element, + GLuint _signed, + GLboolean normalize, + VTX_FETCH_METHOD * pFetchMethod) +{ + GLuint client_size_inbyte; + GLuint data_format; + GLuint mega_fetch_count; + GLuint is_mega_fetch_flag; + + R700VertexGenericFetch* vfetch_instruction_ptr; + R700VertexGenericFetch* assembled_vfetch_instruction_ptr + = pAsm->vfetch_instruction_ptr_array[element]; + + if (assembled_vfetch_instruction_ptr == NULL) + { + vfetch_instruction_ptr = (R700VertexGenericFetch*) CALLOC_STRUCT(R700VertexGenericFetch); + if (vfetch_instruction_ptr == NULL) + { + return GL_FALSE; + } + Init_R700VertexGenericFetch(vfetch_instruction_ptr); + } + else + { + vfetch_instruction_ptr = assembled_vfetch_instruction_ptr; + } + + data_format = GetSurfaceFormat(type, size, &client_size_inbyte); + + if(GL_TRUE == pFetchMethod->bEnableMini) //More conditions here + { + //TODO : mini fetch + } + else + { + mega_fetch_count = MEGA_FETCH_BYTES - 1; + is_mega_fetch_flag = 0x1; + pFetchMethod->mega_fetch_remainder = MEGA_FETCH_BYTES - client_size_inbyte; + } + + vfetch_instruction_ptr->m_Word0.f.vtx_inst = SQ_VTX_INST_FETCH; + vfetch_instruction_ptr->m_Word0.f.fetch_type = SQ_VTX_FETCH_VERTEX_DATA; + vfetch_instruction_ptr->m_Word0.f.fetch_whole_quad = 0x0; + + vfetch_instruction_ptr->m_Word0.f.buffer_id = element; + vfetch_instruction_ptr->m_Word0.f.src_gpr = 0x0; + vfetch_instruction_ptr->m_Word0.f.src_rel = SQ_ABSOLUTE; + vfetch_instruction_ptr->m_Word0.f.src_sel_x = SQ_SEL_X; + vfetch_instruction_ptr->m_Word0.f.mega_fetch_count = mega_fetch_count; + + vfetch_instruction_ptr->m_Word1.f.dst_sel_x = (size < 1) ? SQ_SEL_0 : SQ_SEL_X; + vfetch_instruction_ptr->m_Word1.f.dst_sel_y = (size < 2) ? SQ_SEL_0 : SQ_SEL_Y; + vfetch_instruction_ptr->m_Word1.f.dst_sel_z = (size < 3) ? SQ_SEL_0 : SQ_SEL_Z; + vfetch_instruction_ptr->m_Word1.f.dst_sel_w = (size < 4) ? SQ_SEL_1 : SQ_SEL_W; + + vfetch_instruction_ptr->m_Word1.f.use_const_fields = 1; + vfetch_instruction_ptr->m_Word1.f.data_format = data_format; + vfetch_instruction_ptr->m_Word2.f.endian_swap = SQ_ENDIAN_NONE; + + if(1 == _signed) + { + vfetch_instruction_ptr->m_Word1.f.format_comp_all = SQ_FORMAT_COMP_SIGNED; + } + else + { + vfetch_instruction_ptr->m_Word1.f.format_comp_all = SQ_FORMAT_COMP_UNSIGNED; + } + + if(GL_TRUE == normalize) + { + vfetch_instruction_ptr->m_Word1.f.num_format_all = SQ_NUM_FORMAT_NORM; + } + else + { + vfetch_instruction_ptr->m_Word1.f.num_format_all = SQ_NUM_FORMAT_INT; + } + + // Destination register + vfetch_instruction_ptr->m_Word1_GPR.f.dst_gpr = destination_register; + vfetch_instruction_ptr->m_Word1_GPR.f.dst_rel = SQ_ABSOLUTE; + + vfetch_instruction_ptr->m_Word2.f.offset = 0; + vfetch_instruction_ptr->m_Word2.f.const_buf_no_stride = 0x0; + + vfetch_instruction_ptr->m_Word2.f.mega_fetch = is_mega_fetch_flag; + + if (assembled_vfetch_instruction_ptr == NULL) + { + if ( GL_FALSE == add_vfetch_instruction(pAsm, (R700VertexInstruction *)vfetch_instruction_ptr) ) + { + return GL_FALSE; + } + + if (pAsm->vfetch_instruction_ptr_array[element] != NULL) + { + return GL_FALSE; + } + else + { + pAsm->vfetch_instruction_ptr_array[element] = vfetch_instruction_ptr; + } + } + + return GL_TRUE; +} + +GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm) +{ + GLint i; + pAsm->cf_current_clause_type = CF_EMPTY_CLAUSE; + pAsm->cf_current_vtx_clause_ptr = NULL; + + for (i=0; i<VERT_ATTRIB_MAX; i++) + { + pAsm->vfetch_instruction_ptr_array[ i ] = NULL; + } + + cleanup_vfetch_shaderinst(pAsm->pR700Shader); + + return GL_TRUE; +} + GLuint gethelpr(r700_AssemblerBase* pAsm) { GLuint r = pAsm->uHelpReg; @@ -1152,42 +1283,48 @@ GLboolean tex_src(r700_AssemblerBase *pAsm) GLboolean bValidTexCoord = GL_FALSE; + if(pAsm->aArgSubst[1] >= 0) + { + bValidTexCoord = GL_TRUE; + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = pAsm->aArgSubst[1]; + } + else + { switch (pILInst->SrcReg[0].File) { - case PROGRAM_CONSTANT: - case PROGRAM_LOCAL_PARAM: - case PROGRAM_ENV_PARAM: - case PROGRAM_STATE_VAR: - bValidTexCoord = GL_TRUE; - setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; - pAsm->S[0].src.reg = pAsm->aArgSubst[1]; - break; - case PROGRAM_TEMPORARY: - bValidTexCoord = GL_TRUE; - pAsm->S[0].src.reg = pILInst->SrcReg[0].Index + - pAsm->starting_temp_register_number; - pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; - break; - case PROGRAM_INPUT: - switch (pILInst->SrcReg[0].Index) - { - case FRAG_ATTRIB_COL0: - case FRAG_ATTRIB_COL1: - case FRAG_ATTRIB_TEX0: - case FRAG_ATTRIB_TEX1: - case FRAG_ATTRIB_TEX2: - case FRAG_ATTRIB_TEX3: - case FRAG_ATTRIB_TEX4: - case FRAG_ATTRIB_TEX5: - case FRAG_ATTRIB_TEX6: - case FRAG_ATTRIB_TEX7: - bValidTexCoord = GL_TRUE; - pAsm->S[0].src.reg = - pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index]; - pAsm->S[0].src.rtype = SRC_REG_INPUT; - break; - } - break; + case PROGRAM_CONSTANT: + case PROGRAM_LOCAL_PARAM: + case PROGRAM_ENV_PARAM: + case PROGRAM_STATE_VAR: + break; + case PROGRAM_TEMPORARY: + bValidTexCoord = GL_TRUE; + pAsm->S[0].src.reg = pILInst->SrcReg[0].Index + + pAsm->starting_temp_register_number; + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + break; + case PROGRAM_INPUT: + switch (pILInst->SrcReg[0].Index) + { + case FRAG_ATTRIB_COL0: + case FRAG_ATTRIB_COL1: + case FRAG_ATTRIB_TEX0: + case FRAG_ATTRIB_TEX1: + case FRAG_ATTRIB_TEX2: + case FRAG_ATTRIB_TEX3: + case FRAG_ATTRIB_TEX4: + case FRAG_ATTRIB_TEX5: + case FRAG_ATTRIB_TEX6: + case FRAG_ATTRIB_TEX7: + bValidTexCoord = GL_TRUE; + pAsm->S[0].src.reg = + pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index]; + pAsm->S[0].src.rtype = SRC_REG_INPUT; + break; + } + break; + } } if(GL_TRUE == bValidTexCoord) @@ -1941,9 +2078,9 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) GLuint contiguous_slots_needed; GLuint uNumSrc = r700GetNumOperands(pAsm); - GLuint channel_swizzle, j; - GLuint chan_counter[4] = {0, 0, 0, 0}; - PVSSRC * pSource[3]; + //GLuint channel_swizzle, j; + //GLuint chan_counter[4] = {0, 0, 0, 0}; + //PVSSRC * pSource[3]; GLboolean bSplitInst = GL_FALSE; if (1 == pAsm->D.dst.math) @@ -1955,7 +2092,9 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) { is_single_scalar_operation = GL_FALSE; number_of_scalar_operations = 4; - + +/* current assembler doesn't do more than 1 register per source */ +#if 0 /* check read port, only very preliminary algorithm, not count in src0/1 same comp case and prev slot repeat case; also not count relative addressing. TODO: improve performance. */ @@ -1990,6 +2129,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) { bSplitInst = GL_TRUE; } +#endif } contiguous_slots_needed = 0; @@ -2024,7 +2164,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm) return GL_FALSE; } - if (pAsm->D.dst.math == 0) + if (uNumSrc > 1) { // Process source 1 current_source_index = 1; @@ -2210,9 +2350,7 @@ GLboolean next_ins(r700_AssemblerBase *pAsm) { struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]); - if( GL_TRUE == IsTex(pILInst->Opcode) && - /* handle const moves to temp register */ - !(pAsm->D.dst.opcode == SQ_OP2_INST_MOV) ) + if( GL_TRUE == pAsm->is_tex ) { if (pILInst->TexSrcTarget == TEXTURE_RECT_INDEX) { if( GL_FALSE == assemble_tex_instruction(pAsm, GL_FALSE) ) @@ -2256,7 +2394,8 @@ GLboolean next_ins(r700_AssemblerBase *pAsm) pAsm->S[0].bits = 0; pAsm->S[1].bits = 0; pAsm->S[2].bits = 0; - + pAsm->is_tex = GL_FALSE; + pAsm->need_tex_barrier = GL_FALSE; return GL_TRUE; } @@ -2880,6 +3019,11 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) return GL_FALSE; } + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + /* dst.y = max(src.x, 0.0) */ pAsm->D.dst.opcode = SQ_OP2_INST_MAX; pAsm->D.dst.rtype = dstType; @@ -2891,11 +3035,6 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_X; - pAsm->S[0].src.swizzley = SQ_SEL_X; - pAsm->S[0].src.swizzlez = SQ_SEL_X; - pAsm->S[0].src.swizzlew = SQ_SEL_X; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; pAsm->S[1].src.reg = tmp; setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); @@ -2909,34 +3048,47 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) return GL_FALSE; } - /* before: dst.w = log(src.y) - * after : dst.x = log(src.y) - * why change dest register is that dst.w has been initialized as 1 before - */ + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_Y, SQ_SEL_Y, SQ_SEL_Y, SQ_SEL_Y); + + /* dst.z = log(src.y) */ pAsm->D.dst.opcode = SQ_OP2_INST_LOG_CLAMPED; pAsm->D.dst.math = 1; pAsm->D.dst.rtype = dstType; pAsm->D.dst.reg = dstReg; - pAsm->D.dst.writex = 1; + pAsm->D.dst.writex = 0; pAsm->D.dst.writey = 0; - pAsm->D.dst.writez = 0; + pAsm->D.dst.writez = 1; pAsm->D.dst.writew = 0; pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_Y; - pAsm->S[0].src.swizzley = SQ_SEL_Y; - pAsm->S[0].src.swizzlez = SQ_SEL_Y; - pAsm->S[0].src.swizzlew = SQ_SEL_Y; if( GL_FALSE == next_ins(pAsm) ) { return GL_FALSE; } - /* before: tmp.x = amd MUL_LIT(src.w, dst.w, src.x ) */ - /* after : tmp.x = amd MUL_LIT(src.w, dst.x, src.x ) */ + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + if( GL_FALSE == assemble_src(pAsm, 0, 2) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_W, SQ_SEL_W, SQ_SEL_W, SQ_SEL_W); + + swizzleagain_PVSSRC(&(pAsm->S[2].src), SQ_SEL_X, SQ_SEL_X, SQ_SEL_X, SQ_SEL_X); + + /* tmp.x = amd MUL_LIT(src.w, dst.z, src.x ) */ pAsm->D.dst.opcode = SQ_OP3_INST_MUL_LIT; + pAsm->D.dst.math = 1; pAsm->D.dst.op3 = 1; pAsm->D.dst.rtype = DST_REG_TEMPORARY; pAsm->D.dst.reg = tmp; @@ -2948,29 +3100,19 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm) pAsm->S[0].src.rtype = srcType; pAsm->S[0].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[0].src)); - pAsm->S[0].src.swizzlex = SQ_SEL_W; - pAsm->S[0].src.swizzley = SQ_SEL_W; - pAsm->S[0].src.swizzlez = SQ_SEL_W; - pAsm->S[0].src.swizzlew = SQ_SEL_W; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; pAsm->S[1].src.reg = dstReg; setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); noneg_PVSSRC(&(pAsm->S[1].src)); - pAsm->S[1].src.swizzlex = SQ_SEL_X; - pAsm->S[1].src.swizzley = SQ_SEL_X; - pAsm->S[1].src.swizzlez = SQ_SEL_X; - pAsm->S[1].src.swizzlew = SQ_SEL_X; + pAsm->S[1].src.swizzlex = SQ_SEL_Z; + pAsm->S[1].src.swizzley = SQ_SEL_Z; + pAsm->S[1].src.swizzlez = SQ_SEL_Z; + pAsm->S[1].src.swizzlew = SQ_SEL_Z; pAsm->S[2].src.rtype = srcType; pAsm->S[2].src.reg = srcReg; setaddrmode_PVSSRC(&(pAsm->S[2].src), ADDR_ABSOLUTE); - noneg_PVSSRC(&(pAsm->S[2].src)); - pAsm->S[2].src.swizzlex = SQ_SEL_X; - pAsm->S[2].src.swizzley = SQ_SEL_X; - pAsm->S[2].src.swizzlez = SQ_SEL_X; - pAsm->S[2].src.swizzlew = SQ_SEL_X; if( GL_FALSE == next_ins(pAsm) ) { @@ -3376,7 +3518,10 @@ GLboolean assemble_STP(r700_AssemblerBase *pAsm) GLboolean assemble_TEX(r700_AssemblerBase *pAsm) { GLboolean src_const; + GLboolean need_barrier = GL_FALSE; + checkop1(pAsm); + switch (pAsm->pILInst[pAsm->uiCurInst].SrcReg[0].File) { case PROGRAM_CONSTANT: @@ -3396,20 +3541,18 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) { if ( GL_FALSE == mov_temp(pAsm, 0) ) return GL_FALSE; + need_barrier = GL_TRUE; } switch (pAsm->pILInst[pAsm->uiCurInst].Opcode) { case OPCODE_TEX: - pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; break; case OPCODE_TXB: radeon_error("do not support TXB yet\n"); return GL_FALSE; break; case OPCODE_TXP: - /* TODO : tex proj version : divid first 3 components by 4th */ - pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; break; default: radeon_error("Internal error: bad texture op (not TEX)\n"); @@ -3417,6 +3560,190 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) break; } + if (pAsm->pILInst[pAsm->uiCurInst].Opcode == OPCODE_TXP) + { + GLuint tmp = gethelpr(pAsm); + pAsm->D.dst.opcode = SQ_OP2_INST_RECIP_IEEE; + pAsm->D.dst.math = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp; + pAsm->D.dst.writew = 1; + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_W, SQ_SEL_W, SQ_SEL_W, SQ_SEL_W); + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + pAsm->D.dst.opcode = SQ_OP2_INST_MUL; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 1; + pAsm->D.dst.writew = 0; + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = tmp; + setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_W); + + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + pAsm->aArgSubst[1] = tmp; + need_barrier = GL_TRUE; + } + + if (pAsm->pILInst[pAsm->uiCurInst].TexSrcTarget == TEXTURE_CUBE_INDEX ) + { + GLuint tmp1 = gethelpr(pAsm); + GLuint tmp2 = gethelpr(pAsm); + + /* tmp1.xyzw = CUBE(R0.zzxy, R0.yxzz) */ + pAsm->D.dst.opcode = SQ_OP2_INST_CUBE; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + nomask_PVSDST(&(pAsm->D.dst)); + + if( GL_FALSE == assemble_src(pAsm, 0, -1) ) + { + return GL_FALSE; + } + + if( GL_FALSE == assemble_src(pAsm, 0, 1) ) + { + return GL_FALSE; + } + + swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_Z, SQ_SEL_Z, SQ_SEL_X, SQ_SEL_Y); + swizzleagain_PVSSRC(&(pAsm->S[1].src), SQ_SEL_Y, SQ_SEL_X, SQ_SEL_Z, SQ_SEL_Z); + + if( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } + + /* tmp1.z = ABS(tmp1.z) dont have abs support in assembler currently + * have to do explicit instruction + */ + pAsm->D.dst.opcode = SQ_OP2_INST_MAX; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writez = 1; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + pAsm->S[1].bits = pAsm->S[0].bits; + flipneg_PVSSRC(&(pAsm->S[1].src)); + + next_ins(pAsm); + + /* tmp1.z = RCP_e(|tmp1.z|) */ + pAsm->D.dst.opcode = SQ_OP2_INST_RECIP_IEEE; + pAsm->D.dst.math = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writez = 1; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + pAsm->S[0].src.swizzlex = SQ_SEL_Z; + + next_ins(pAsm); + + /* MULADD R0.x, R0.x, PS1, (0x3FC00000, 1.5f).x + * MULADD R0.y, R0.y, PS1, (0x3FC00000, 1.5f).x + * muladd has no writemask, have to use another temp + * also no support for imm constants, so add 1 here + */ + pAsm->D.dst.opcode = SQ_OP3_INST_MULADD; + pAsm->D.dst.op3 = 1; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp2; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp1; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = tmp1; + setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_Z); + setaddrmode_PVSSRC(&(pAsm->S[2].src), ADDR_ABSOLUTE); + pAsm->S[2].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[2].src.reg = tmp1; + setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_1); + + next_ins(pAsm); + + /* ADD the remaining .5 */ + pAsm->D.dst.opcode = SQ_OP2_INST_ADD; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp2; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 0; + pAsm->D.dst.writew = 0; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp2; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE); + pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[1].src.reg = 252; // SQ_ALU_SRC_0_5 + noswizzle_PVSSRC(&(pAsm->S[1].src)); + + next_ins(pAsm); + + /* tmp1.xy = temp2.xy */ + pAsm->D.dst.opcode = SQ_OP2_INST_MOV; + setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE); + pAsm->D.dst.rtype = DST_REG_TEMPORARY; + pAsm->D.dst.reg = tmp1; + pAsm->D.dst.writex = 1; + pAsm->D.dst.writey = 1; + pAsm->D.dst.writez = 0; + pAsm->D.dst.writew = 0; + + setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE); + pAsm->S[0].src.rtype = SRC_REG_TEMPORARY; + pAsm->S[0].src.reg = tmp2; + noswizzle_PVSSRC(&(pAsm->S[0].src)); + + next_ins(pAsm); + pAsm->aArgSubst[1] = tmp1; + need_barrier = GL_TRUE; + + } + + pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE; + pAsm->is_tex = GL_TRUE; + if ( GL_TRUE == need_barrier ) + { + pAsm->need_tex_barrier = GL_TRUE; + } // Set src1 to tex unit id pAsm->S[1].src.reg = pAsm->pILInst[pAsm->uiCurInst].TexSrcUnit; pAsm->S[1].src.rtype = SRC_REG_TEMPORARY; @@ -3437,10 +3764,25 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm) return GL_FALSE; } - if ( GL_FALSE == next_ins(pAsm) ) + if(pAsm->pILInst[pAsm->uiCurInst].Opcode == OPCODE_TXP) { - return GL_FALSE; + /* hopefully did swizzles before */ + noswizzle_PVSSRC(&(pAsm->S[0].src)); } + + if(pAsm->pILInst[pAsm->uiCurInst].TexSrcTarget == TEXTURE_CUBE_INDEX) + { + /* SAMPLE dst, tmp.yxwy, CUBE */ + pAsm->S[0].src.swizzlex = SQ_SEL_Y; + pAsm->S[0].src.swizzley = SQ_SEL_X; + pAsm->S[0].src.swizzlez = SQ_SEL_W; + pAsm->S[0].src.swizzlew = SQ_SEL_Y; + } + + if ( GL_FALSE == next_ins(pAsm) ) + { + return GL_FALSE; + } return GL_TRUE; } diff --git a/src/mesa/drivers/dri/r600/r700_assembler.h b/src/mesa/drivers/dri/r600/r700_assembler.h index f9c4d849c65..0d4283e4bad 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.h +++ b/src/mesa/drivers/dri/r600/r700_assembler.h @@ -374,6 +374,10 @@ typedef struct r700_AssemblerBase struct prog_instruction * pILInst; GLuint uiCurInst; GLboolean bR6xx; + /* helper to decide which type of instruction to assemble */ + GLboolean is_tex; + /* we inserted helper intructions and need barrier on next TEX ins */ + GLboolean need_tex_barrier; } r700_AssemblerBase; //Internal use @@ -411,6 +415,15 @@ GLboolean assemble_vfetch_instruction(r700_AssemblerBase* pAsm, GLuint number_of_elements, GLenum dataElementType, VTX_FETCH_METHOD* pFetchMethod); +GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm, + GLuint destination_register, + GLenum type, + GLint size, + GLubyte element, + GLuint _signed, + GLboolean normalize, + VTX_FETCH_METHOD * pFetchMethod); +GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm); GLuint gethelpr(r700_AssemblerBase* pAsm); void resethelpr(r700_AssemblerBase* pAsm); void checkop_init(r700_AssemblerBase* pAsm); diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c index 06d7e9c9ab1..3b7f6fffe03 100644 --- a/src/mesa/drivers/dri/r600/r700_chip.c +++ b/src/mesa/drivers/dri/r600/r700_chip.c @@ -208,6 +208,93 @@ static void r700SetupVTXConstants(GLcontext * ctx, } +extern int getTypeSize(GLenum type); +static void r700SetupVTXConstants2(GLcontext * ctx, + void * pAos, + StreamDesc * pStreamDesc) +{ + context_t *context = R700_CONTEXT(ctx); + struct radeon_aos * paos = (struct radeon_aos *)pAos; + unsigned int nVBsize; + BATCH_LOCALS(&context->radeon); + + unsigned int uSQ_VTX_CONSTANT_WORD0_0; + unsigned int uSQ_VTX_CONSTANT_WORD1_0; + unsigned int uSQ_VTX_CONSTANT_WORD2_0 = 0; + unsigned int uSQ_VTX_CONSTANT_WORD3_0 = 0; + unsigned int uSQ_VTX_CONSTANT_WORD6_0 = 0; + + if (!paos->bo) + return; + + if ((context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV610) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV620) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS780) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RS880) || + (context->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV710)) + r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, TC_ACTION_ENA_bit); + else + r700SyncSurf(context, paos->bo, RADEON_GEM_DOMAIN_GTT, 0, VC_ACTION_ENA_bit); + + if(0 == pStreamDesc->stride) + { + nVBsize = paos->count * pStreamDesc->size * getTypeSize(pStreamDesc->type); + } + else + { + nVBsize = paos->count * pStreamDesc->stride; + } + + uSQ_VTX_CONSTANT_WORD0_0 = paos->offset; + uSQ_VTX_CONSTANT_WORD1_0 = nVBsize - 1; + + SETfield(uSQ_VTX_CONSTANT_WORD2_0, 0, BASE_ADDRESS_HI_shift, BASE_ADDRESS_HI_mask); /* TODO */ + SETfield(uSQ_VTX_CONSTANT_WORD2_0, pStreamDesc->stride, SQ_VTX_CONSTANT_WORD2_0__STRIDE_shift, + SQ_VTX_CONSTANT_WORD2_0__STRIDE_mask); + SETfield(uSQ_VTX_CONSTANT_WORD2_0, GetSurfaceFormat(pStreamDesc->type, pStreamDesc->size, NULL), + SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_shift, + SQ_VTX_CONSTANT_WORD2_0__DATA_FORMAT_mask); /* TODO : trace back api for initial data type, not only GL_FLOAT */ + + if(GL_TRUE == pStreamDesc->normalize) + { + SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_NORM, + SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask); + } + //else + //{ + // SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_INT, + // SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask); + //} + + if(1 == pStreamDesc->_signed) + { + SETbit(uSQ_VTX_CONSTANT_WORD2_0, SQ_VTX_CONSTANT_WORD2_0__FORMAT_COMP_ALL_bit); + } + + SETfield(uSQ_VTX_CONSTANT_WORD3_0, 1, MEM_REQUEST_SIZE_shift, MEM_REQUEST_SIZE_mask); + SETfield(uSQ_VTX_CONSTANT_WORD6_0, SQ_TEX_VTX_VALID_BUFFER, + SQ_TEX_RESOURCE_WORD6_0__TYPE_shift, SQ_TEX_RESOURCE_WORD6_0__TYPE_mask); + + BEGIN_BATCH_NO_AUTOSTATE(9 + 2); + + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7)); + R600_OUT_BATCH((pStreamDesc->element + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD0_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD1_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD2_0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD3_0); + R600_OUT_BATCH(0); + R600_OUT_BATCH(0); + R600_OUT_BATCH(uSQ_VTX_CONSTANT_WORD6_0); + R600_OUT_BATCH_RELOC(uSQ_VTX_CONSTANT_WORD0_0, + paos->bo, + uSQ_VTX_CONSTANT_WORD0_0, + RADEON_GEM_DOMAIN_GTT, 0, 0); + END_BATCH(); + COMMIT_BATCH(); + +} + void r700SetupStreams(GLcontext *ctx) { context_t *context = R700_CONTEXT(ctx); @@ -256,14 +343,24 @@ static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom) COMMIT_BATCH(); for(i=0; i<VERT_ATTRIB_MAX; i++) { - if(vp->mesa_program->Base.InputsRead & (1 << i)) { - /* currently aos are packed */ - r700SetupVTXConstants(ctx, - i, - (void*)(&context->radeon.tcl.aos[j]), - (unsigned int)context->radeon.tcl.aos[j].components, - (unsigned int)context->radeon.tcl.aos[j].stride * 4, - (unsigned int)context->radeon.tcl.aos[j].count); + if(vp->mesa_program->Base.InputsRead & (1 << i)) + { + if(1 == context->selected_vp->uiVersion) + { + /* currently aos are packed */ + r700SetupVTXConstants(ctx, + i, + (void*)(&context->radeon.tcl.aos[j]), + (unsigned int)context->radeon.tcl.aos[j].components, + (unsigned int)context->radeon.tcl.aos[j].stride * 4, + (unsigned int)context->radeon.tcl.aos[j].count); + } + else + { /* context->selected_vp->uiVersion == 2 : aos not always packed */ + r700SetupVTXConstants2(ctx, + (void*)(&context->radeon.tcl.aos[j]), + &(context->stream_desc[j])); + } j++; } } diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.c b/src/mesa/drivers/dri/r600/r700_fragprog.c index 78ce3ae4366..62a1ea1a22a 100644 --- a/src/mesa/drivers/dri/r600/r700_fragprog.c +++ b/src/mesa/drivers/dri/r600/r700_fragprog.c @@ -341,6 +341,11 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx) SETbit(r700->SPI_PS_IN_CONTROL_0.u32All, POSITION_ENA_bit); SETbit(r700->SPI_INPUT_Z.u32All, PROVIDE_Z_TO_SPI_bit); } + else + { + CLEARbit(r700->SPI_PS_IN_CONTROL_0.u32All, POSITION_ENA_bit); + CLEARbit(r700->SPI_INPUT_Z.u32All, PROVIDE_Z_TO_SPI_bit); + } ui = (unNumOfReg < ui) ? ui : unNumOfReg; diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c index b1c3648ca56..0aef0b7ea1f 100644 --- a/src/mesa/drivers/dri/r600/r700_render.c +++ b/src/mesa/drivers/dri/r600/r700_render.c @@ -43,6 +43,7 @@ #include "tnl/t_context.h" #include "tnl/t_vertex.h" #include "tnl/t_pipeline.h" +#include "vbo/vbo_context.h" #include "r600_context.h" #include "r600_cmdbuf.h" @@ -53,6 +54,7 @@ #include "r700_fragprog.h" #include "r700_state.h" +#include "radeon_buffer_objects.h" #include "radeon_common_context.h" void r700WaitForIdle(context_t *context); @@ -249,78 +251,134 @@ static int r700NumVerts(int num_verts, int prim) static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim) { - context_t *context = R700_CONTEXT(ctx); - BATCH_LOCALS(&context->radeon); - int type, i, total_emit; - int num_indices; - uint32_t vgt_draw_initiator = 0; - uint32_t vgt_index_type = 0; - uint32_t vgt_primitive_type = 0; - uint32_t vgt_num_indices = 0; - TNLcontext *tnl = TNL_CONTEXT(ctx); - struct vertex_buffer *vb = &tnl->vb; - - type = r700PrimitiveType(prim); - num_indices = r700NumVerts(end - start, prim); - - radeon_print(RADEON_RENDER, RADEON_TRACE, - "%s type %x num_indices %d\n", - __func__, type, num_indices); - - if (type < 0 || num_indices <= 0) - return; + context_t *context = R700_CONTEXT(ctx); + BATCH_LOCALS(&context->radeon); + int type, i, total_emit; + int num_indices; + uint32_t vgt_draw_initiator = 0; + uint32_t vgt_index_type = 0; + uint32_t vgt_primitive_type = 0; + uint32_t vgt_num_indices = 0; + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *vb = &tnl->vb; + GLboolean bUseDrawIndex; + + if(NULL != context->ind_buf.bo) + { + bUseDrawIndex = GL_TRUE; + } + else + { + bUseDrawIndex = GL_FALSE; + } + + type = r700PrimitiveType(prim); + num_indices = r700NumVerts(end - start, prim); + + radeon_print(RADEON_RENDER, RADEON_TRACE, + "%s type %x num_indices %d\n", + __func__, type, num_indices); + + if (type < 0 || num_indices <= 0) + return; + if(GL_TRUE == bUseDrawIndex) + { + total_emit = 3 /* VGT_PRIMITIVE_TYPE */ + + 2 /* VGT_INDEX_TYPE */ + + 2 /* NUM_INSTANCES */ + + 5 + 2; /* DRAW_INDEX */ + } + else + { total_emit = 3 /* VGT_PRIMITIVE_TYPE */ + 2 /* VGT_INDEX_TYPE */ - + 2 /* NUM_INSTANCES */ + + 2 /* NUM_INSTANCES */ + num_indices + 3; /* DRAW_INDEX_IMMD */ + } - BEGIN_BATCH_NO_AUTOSTATE(total_emit); - // prim - SETfield(vgt_primitive_type, type, - VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask); - R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); - R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX); - R600_OUT_BATCH(vgt_primitive_type); + BEGIN_BATCH_NO_AUTOSTATE(total_emit); + // prim + SETfield(vgt_primitive_type, type, + VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask); + R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CONFIG_REG, 1)); + R600_OUT_BATCH(mmVGT_PRIMITIVE_TYPE - ASIC_CONFIG_BASE_INDEX); + R600_OUT_BATCH(vgt_primitive_type); // index type - SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); - R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0)); - R600_OUT_BATCH(vgt_index_type); + SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); - // num instances - R600_OUT_BATCH(CP_PACKET3(R600_IT_NUM_INSTANCES, 0)); - R600_OUT_BATCH(1); + if(GL_TRUE == bUseDrawIndex) + { + if(GL_TRUE != context->ind_buf.is_32bit) + { + SETfield(vgt_index_type, DI_INDEX_SIZE_16_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask); + } + } + + R600_OUT_BATCH(CP_PACKET3(R600_IT_INDEX_TYPE, 0)); + R600_OUT_BATCH(vgt_index_type); + + // num instances + R600_OUT_BATCH(CP_PACKET3(R600_IT_NUM_INSTANCES, 0)); + R600_OUT_BATCH(1); + + // draw packet + vgt_num_indices = num_indices; - // draw packet - vgt_num_indices = num_indices; + if(GL_TRUE == bUseDrawIndex) + { + SETfield(vgt_draw_initiator, DI_SRC_SEL_DMA, SOURCE_SELECT_shift, SOURCE_SELECT_mask); + } + else + { SETfield(vgt_draw_initiator, DI_SRC_SEL_IMMEDIATE, SOURCE_SELECT_shift, SOURCE_SELECT_mask); + } + SETfield(vgt_draw_initiator, DI_MAJOR_MODE_0, MAJOR_MODE_shift, MAJOR_MODE_mask); + if(GL_TRUE == bUseDrawIndex) + { + R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX, 3)); + R600_OUT_BATCH(context->ind_buf.bo_offset); + R600_OUT_BATCH(0); + R600_OUT_BATCH(vgt_num_indices); + R600_OUT_BATCH(vgt_draw_initiator); + R600_OUT_BATCH_RELOC(context->ind_buf.bo_offset, + context->ind_buf.bo, + context->ind_buf.bo_offset, + RADEON_GEM_DOMAIN_GTT, 0, 0); + } + else + { R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (num_indices + 1))); R600_OUT_BATCH(vgt_num_indices); R600_OUT_BATCH(vgt_draw_initiator); - for (i = start; i < (start + num_indices); i++) { - if(vb->Elts) - R600_OUT_BATCH(vb->Elts[i]); - else - R600_OUT_BATCH(i); + for (i = start; i < (start + num_indices); i++) + { + if(vb->Elts) + { + R600_OUT_BATCH(vb->Elts[i]); + } + else + { + R600_OUT_BATCH(i); + } } - END_BATCH(); - COMMIT_BATCH(); + } + END_BATCH(); + COMMIT_BATCH(); } /* start 3d, idle, cb/db flush */ #define PRE_EMIT_STATE_BUFSZ 10 + 5 + 14 -static GLuint r700PredictRenderSize(GLcontext* ctx) +static GLuint r700PredictRenderSize(GLcontext* ctx, GLuint nr_prims) { context_t *context = R700_CONTEXT(ctx); - TNLcontext *tnl = TNL_CONTEXT(ctx); struct r700_vertex_program *vp = context->selected_vp; - struct vertex_buffer *vb = &tnl->vb; GLboolean flushed; GLuint dwords, i; GLuint state_size; @@ -328,8 +386,15 @@ static GLuint r700PredictRenderSize(GLcontext* ctx) context->radeon.tcl.aos_count = _mesa_bitcount(vp->mesa_program->Base.InputsRead); dwords = PRE_EMIT_STATE_BUFSZ; - for (i = 0; i < vb->PrimitiveCount; i++) - dwords += vb->Primitive[i].count + 10; + if (nr_prims) + dwords += nr_prims * 14; + else { + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *vb = &tnl->vb; + + for (i = 0; i < vb->PrimitiveCount; i++) + dwords += vb->Primitive[i].count + 10; + } state_size = radeonCountStateEmitSize(&context->radeon); flushed = rcommonEnsureCmdBufSpace(&context->radeon, dwords + state_size, __FUNCTION__); @@ -369,7 +434,7 @@ static GLboolean r700RunRender(GLcontext * ctx, r700SetupFragmentProgram(ctx); r600UpdateTextureState(ctx); - GLuint emit_end = r700PredictRenderSize(ctx) + GLuint emit_end = r700PredictRenderSize(ctx, 0) + context->radeon.cmdbuf.cs->cdw; r700SetupStreams(ctx); @@ -477,4 +542,544 @@ const struct tnl_pipeline_stage *r700_pipeline[] = 0, }; +#define CONVERT( TYPE, MACRO ) do { \ + GLuint i, j, sz; \ + sz = input->Size; \ + if (input->Normalized) { \ + for (i = 0; i < count; i++) { \ + const TYPE *in = (TYPE *)src_ptr; \ + for (j = 0; j < sz; j++) { \ + *dst_ptr++ = MACRO(*in); \ + in++; \ + } \ + src_ptr += stride; \ + } \ + } else { \ + for (i = 0; i < count; i++) { \ + const TYPE *in = (TYPE *)src_ptr; \ + for (j = 0; j < sz; j++) { \ + *dst_ptr++ = (GLfloat)(*in); \ + in++; \ + } \ + src_ptr += stride; \ + } \ + } \ +} while (0) + +/** + * Convert attribute data type to float + * If the attribute uses named buffer object replace the bo with newly allocated bo + */ +static void r700ConvertAttrib(GLcontext *ctx, int count, + const struct gl_client_array *input, + struct StreamDesc *attr) +{ + context_t *context = R700_CONTEXT(ctx); + const GLvoid *src_ptr; + GLboolean mapped_named_bo = GL_FALSE; + GLfloat *dst_ptr; + GLuint stride; + + stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size : input->StrideB; + + /* Convert value for first element only */ + if (input->StrideB == 0) + { + count = 1; + } + + if (input->BufferObj->Name) + { + if (!input->BufferObj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj); + mapped_named_bo = GL_TRUE; + } + + src_ptr = ADD_POINTERS(input->BufferObj->Pointer, input->Ptr); + } + else + { + src_ptr = input->Ptr; + } + + radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, + sizeof(GLfloat) * input->Size * count, 32); + dst_ptr = (GLfloat *)ADD_POINTERS(attr->bo->ptr, attr->bo_offset); + + assert(src_ptr != NULL); + + switch (input->Type) + { + case GL_DOUBLE: + CONVERT(GLdouble, (GLfloat)); + break; + case GL_UNSIGNED_INT: + CONVERT(GLuint, UINT_TO_FLOAT); + break; + case GL_INT: + CONVERT(GLint, INT_TO_FLOAT); + break; + case GL_UNSIGNED_SHORT: + CONVERT(GLushort, USHORT_TO_FLOAT); + break; + case GL_SHORT: + CONVERT(GLshort, SHORT_TO_FLOAT); + break; + case GL_UNSIGNED_BYTE: + assert(input->Format != GL_BGRA); + CONVERT(GLubyte, UBYTE_TO_FLOAT); + break; + case GL_BYTE: + CONVERT(GLbyte, BYTE_TO_FLOAT); + break; + default: + assert(0); + break; + } + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj); + } +} + +static void r700AlignDataToDword(GLcontext *ctx, + const struct gl_client_array *input, + int count, + struct StreamDesc *attr) +{ + context_t *context = R700_CONTEXT(ctx); + const int dst_stride = (input->StrideB + 3) & ~3; + const int size = getTypeSize(input->Type) * input->Size * count; + GLboolean mapped_named_bo = GL_FALSE; + + radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, size, 32); + + if (!input->BufferObj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj); + mapped_named_bo = GL_TRUE; + } + + { + GLvoid *src_ptr = ADD_POINTERS(input->BufferObj->Pointer, input->Ptr); + GLvoid *dst_ptr = ADD_POINTERS(attr->bo->ptr, attr->bo_offset); + int i; + + for (i = 0; i < count; ++i) + { + _mesa_memcpy(dst_ptr, src_ptr, input->StrideB); + src_ptr += input->StrideB; + dst_ptr += dst_stride; + } + } + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj); + } + + attr->stride = dst_stride; +} + +static void r700SetupStreams2(GLcontext *ctx, const struct gl_client_array *input[], int count) +{ + context_t *context = R700_CONTEXT(ctx); + GLuint stride; + int ret; + int i, index; + + R600_STATECHANGE(context, vtx); + + for(index = 0; index < context->nNumActiveAos; index++) + { + struct radeon_aos *aos = &context->radeon.tcl.aos[index]; + i = context->stream_desc[index].element; + + stride = (input[i]->StrideB == 0) ? getTypeSize(input[i]->Type) * input[i]->Size : input[i]->StrideB; + + if (input[i]->Type == GL_DOUBLE || input[i]->Type == GL_UNSIGNED_INT || input[i]->Type == GL_INT || +#if MESA_BIG_ENDIAN + getTypeSize(input[i]->Type) != 4 || +#endif + stride < 4) + { + r700ConvertAttrib(ctx, count, input[i], &context->stream_desc[index]); + } + else + { + if (input[i]->BufferObj->Name) + { + if (stride % 4 != 0) + { + assert(((intptr_t) input[i]->Ptr) % input[i]->StrideB == 0); + r700AlignDataToDword(ctx, input[i], count, &context->stream_desc[index]); + context->stream_desc[index].is_named_bo = GL_FALSE; + } + else + { + context->stream_desc[index].stride = input[i]->StrideB; + context->stream_desc[index].bo_offset = (intptr_t) input[i]->Ptr; + context->stream_desc[index].bo = get_radeon_buffer_object(input[i]->BufferObj)->bo; + context->stream_desc[index].is_named_bo = GL_TRUE; + } + } + else + { + int size; + int local_count = count; + uint32_t *dst; + + if (input[i]->StrideB == 0) + { + size = getTypeSize(input[i]->Type) * input[i]->Size; + local_count = 1; + } + else + { + size = getTypeSize(input[i]->Type) * input[i]->Size * local_count; + } + + radeonAllocDmaRegion(&context->radeon, &context->stream_desc[index].bo, + &context->stream_desc[index].bo_offset, size, 32); + assert(context->stream_desc[index].bo->ptr != NULL); + dst = (uint32_t *)ADD_POINTERS(context->stream_desc[index].bo->ptr, + context->stream_desc[index].bo_offset); + + switch (context->stream_desc[index].dwords) + { + case 1: + radeonEmitVec4(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 4; + break; + case 2: + radeonEmitVec8(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 8; + break; + case 3: + radeonEmitVec12(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 12; + break; + case 4: + radeonEmitVec16(dst, input[i]->Ptr, input[i]->StrideB, local_count); + context->stream_desc[index].stride = 16; + break; + default: + assert(0); + break; + } + } + } + + aos->count = context->stream_desc[index].stride == 0 ? 1 : count; + aos->stride = context->stream_desc[index].stride / sizeof(float); + aos->components = context->stream_desc[index].dwords; + aos->bo = context->stream_desc[index].bo; + aos->offset = context->stream_desc[index].bo_offset; + + if(context->stream_desc[index].is_named_bo) + { + radeon_cs_space_add_persistent_bo(context->radeon.cmdbuf.cs, + context->stream_desc[index].bo, + RADEON_GEM_DOMAIN_GTT, 0); + } + } + + context->radeon.tcl.aos_count = context->nNumActiveAos; + ret = radeon_cs_space_check_with_bo(context->radeon.cmdbuf.cs, + first_elem(&context->radeon.dma.reserved)->bo, + RADEON_GEM_DOMAIN_GTT, 0); +} + +static void r700FreeData(GLcontext *ctx) +{ + /* Need to zero tcl.aos[n].bo and tcl.elt_dma_bo + * to prevent double unref in radeonReleaseArrays + * called during context destroy + */ + context_t *context = R700_CONTEXT(ctx); + + int i; + + for (i = 0; i < context->nNumActiveAos; i++) + { + if (!context->stream_desc[i].is_named_bo) + { + radeon_bo_unref(context->stream_desc[i].bo); + } + context->radeon.tcl.aos[i].bo = NULL; + } + + if (context->ind_buf.bo != NULL) + { + radeon_bo_unref(context->ind_buf.bo); + } +} + +static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer *mesa_ind_buf) +{ + context_t *context = R700_CONTEXT(ctx); + GLvoid *src_ptr; + GLuint *out; + int i; + GLboolean mapped_named_bo = GL_FALSE; + + if (mesa_ind_buf->obj->Name && !mesa_ind_buf->obj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY_ARB, mesa_ind_buf->obj); + mapped_named_bo = GL_TRUE; + assert(mesa_ind_buf->obj->Pointer != NULL); + } + src_ptr = ADD_POINTERS(mesa_ind_buf->obj->Pointer, mesa_ind_buf->ptr); + + if (mesa_ind_buf->type == GL_UNSIGNED_BYTE) + { + GLuint size = sizeof(GLushort) * ((mesa_ind_buf->count + 1) & ~1); + GLubyte *in = (GLubyte *)src_ptr; + + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + + assert(context->ind_buf.bo->ptr != NULL); + out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + + for (i = 0; i + 1 < mesa_ind_buf->count; i += 2) + { + *out++ = in[i] | in[i + 1] << 16; + } + + if (i < mesa_ind_buf->count) + { + *out++ = in[i]; + } + +#if MESA_BIG_ENDIAN + } + else + { /* if (mesa_ind_buf->type == GL_UNSIGNED_SHORT) */ + GLushort *in = (GLushort *)src_ptr; + GLuint size = sizeof(GLushort) * ((mesa_ind_buf->count + 1) & ~1); + + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + + assert(context->ind_buf.bo->ptr != NULL); + out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + + for (i = 0; i + 1 < mesa_ind_buf->count; i += 2) + { + *out++ = in[i] | in[i + 1] << 16; + } + + if (i < mesa_ind_buf->count) + { + *out++ = in[i]; + } +#endif + } + + context->ind_buf.is_32bit = GL_FALSE; + context->ind_buf.count = mesa_ind_buf->count; + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, mesa_ind_buf->obj); + } +} + +static void r700SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer *mesa_ind_buf) +{ + context_t *context = R700_CONTEXT(ctx); + + if (!mesa_ind_buf) { + context->ind_buf.bo = NULL; + return; + } + +#if MESA_BIG_ENDIAN + if (mesa_ind_buf->type == GL_UNSIGNED_INT) + { +#else + if (mesa_ind_buf->type != GL_UNSIGNED_BYTE) + { +#endif + const GLvoid *src_ptr; + GLvoid *dst_ptr; + GLboolean mapped_named_bo = GL_FALSE; + + if (mesa_ind_buf->obj->Name && !mesa_ind_buf->obj->Pointer) + { + ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, GL_READ_ONLY_ARB, mesa_ind_buf->obj); + assert(mesa_ind_buf->obj->Pointer != NULL); + mapped_named_bo = GL_TRUE; + } + + src_ptr = ADD_POINTERS(mesa_ind_buf->obj->Pointer, mesa_ind_buf->ptr); + + const GLuint size = mesa_ind_buf->count * getTypeSize(mesa_ind_buf->type); + + radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo, + &context->ind_buf.bo_offset, size, 4); + assert(context->ind_buf.bo->ptr != NULL); + dst_ptr = ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset); + + _mesa_memcpy(dst_ptr, src_ptr, size); + + context->ind_buf.is_32bit = (mesa_ind_buf->type == GL_UNSIGNED_INT); + context->ind_buf.count = mesa_ind_buf->count; + + if (mapped_named_bo) + { + ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, mesa_ind_buf->obj); + } + } + else + { + r700FixupIndexBuffer(ctx, mesa_ind_buf); + } +} + +static GLboolean r700TryDrawPrims(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLuint min_index, + GLuint max_index ) +{ + context_t *context = R700_CONTEXT(ctx); + radeonContextPtr radeon = &context->radeon; + GLuint i, id = 0; + struct radeon_renderbuffer *rrb; + + if (ctx->NewState) + { + _mesa_update_state( ctx ); + } + + _tnl_UpdateFixedFunctionProgram(ctx); + r700SetVertexFormat(ctx, arrays, max_index + 1); + r700SetupIndexBuffer(ctx, ib); + /* shaders need to be updated before buffers are validated */ + r700UpdateShaders2(ctx); + if (!r600ValidateBuffers(ctx)) + return GL_FALSE; + + /* always emit CB base to prevent + * lock ups on some chips. + */ + R600_STATECHANGE(context, cb_target); + /* mark vtx as dirty since it changes per-draw */ + R600_STATECHANGE(context, vtx); + + r700SetScissor(context); + r700SetupVertexProgram(ctx); + r700SetupFragmentProgram(ctx); + r600UpdateTextureState(ctx); + + GLuint emit_end = r700PredictRenderSize(ctx, nr_prims) + + context->radeon.cmdbuf.cs->cdw; + + r700SetupStreams2(ctx, arrays, max_index + 1); + + radeonEmitState(radeon); + + radeon_debug_add_indent(); + for (i = 0; i < nr_prims; ++i) + { + r700RunRenderPrimitive(ctx, + prim[i].start, + prim[i].start + prim[i].count, + prim[i].mode); + } + radeon_debug_remove_indent(); + + /* Flush render op cached for last several quads. */ + r700WaitForIdleClean(context); + + rrb = radeon_get_colorbuffer(&context->radeon); + if (rrb && rrb->bo) + r700SyncSurf(context, rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM, + CB_ACTION_ENA_bit | (1 << (id + 6))); + + rrb = radeon_get_depthbuffer(&context->radeon); + if (rrb && rrb->bo) + r700SyncSurf(context, rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM, + DB_ACTION_ENA_bit | DB_DEST_BASE_ENA_bit); + + r700FreeData(ctx); + + if (emit_end < context->radeon.cmdbuf.cs->cdw) + { + WARN_ONCE("Rendering was %d commands larger than predicted size." + " We might overflow command buffer.\n", context->radeon.cmdbuf.cs->cdw - emit_end); + } + + return GL_TRUE; +} + +static void r700DrawPrimsRe(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index) +{ + GLboolean retval = GL_FALSE; + + /* This check should get folded into just the places that + * min/max index are really needed. + */ + if (!index_bounds_valid) { + vbo_get_minmax_index(ctx, prim, ib, &min_index, &max_index); + } + + if (min_index) { + vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, r700DrawPrimsRe ); + return; + } + + /* Make an attempt at drawing */ + retval = r700TryDrawPrims(ctx, arrays, prim, nr_prims, ib, min_index, max_index); + + /* If failed run tnl pipeline - it should take care of fallbacks */ + if (!retval) + _tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index); +} + +static void r700DrawPrims(GLcontext *ctx, + const struct gl_client_array *arrays[], + const struct _mesa_prim *prim, + GLuint nr_prims, + const struct _mesa_index_buffer *ib, + GLboolean index_bounds_valid, + GLuint min_index, + GLuint max_index) +{ + context_t *context = R700_CONTEXT(ctx); + + /* For non indexed drawing, using tnl pipe. */ + if(!ib) + { + context->ind_buf.bo = NULL; + + _tnl_vbo_draw_prims(ctx, arrays, prim, nr_prims, ib, + index_bounds_valid, min_index, max_index); + return; + } + + r700DrawPrimsRe(ctx, arrays, prim, nr_prims, ib, index_bounds_valid, min_index, max_index); +} + +void r700InitDraw(GLcontext *ctx) +{ + struct vbo_context *vbo = vbo_context(ctx); + + /* to be enabled */ + vbo->draw_prims = r700DrawPrims; +} + diff --git a/src/mesa/drivers/dri/r600/r700_shader.c b/src/mesa/drivers/dri/r600/r700_shader.c index b4fd51c1370..955ea4e4e1d 100644 --- a/src/mesa/drivers/dri/r600/r700_shader.c +++ b/src/mesa/drivers/dri/r600/r700_shader.c @@ -60,6 +60,55 @@ void AddInstToList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * plstCFInstructions->uNumOfNode++; } +void TakeInstOutFromList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * pInst) +{ + GLuint ulIndex = 0; + GLboolean bFound = GL_FALSE; + R700ShaderInstruction * pPrevInst = NULL; + R700ShaderInstruction * pCurInst = plstCFInstructions->pHead; + + /* Need go thro list to make sure pInst is there. */ + while(NULL != pCurInst) + { + if(pCurInst == pInst) + { + bFound = GL_TRUE; + break; + } + + pPrevInst = pCurInst; + pCurInst = pCurInst->pNextInst; + } + if(GL_TRUE == bFound) + { + plstCFInstructions->uNumOfNode--; + + pCurInst = pInst->pNextInst; + ulIndex = pInst->m_uIndex; + while(NULL != pCurInst) + { + pCurInst->m_uIndex = ulIndex; + ulIndex++; + pCurInst = pCurInst->pNextInst; + } + + if(plstCFInstructions->pHead == pInst) + { + plstCFInstructions->pHead = pInst->pNextInst; + } + if(plstCFInstructions->pTail == pInst) + { + plstCFInstructions->pTail = pPrevInst; + } + if(NULL != pPrevInst) + { + pPrevInst->pNextInst = pInst->pNextInst; + } + + FREE(pInst); + } +} + void Init_R700_Shader(R700_Shader * pShader) { pShader->Type = R700_SHADER_INVALID; @@ -488,6 +537,47 @@ void DebugPrint(void) { } +void cleanup_vfetch_shaderinst(R700_Shader *pShader) +{ + R700ShaderInstruction *pInst; + R700ShaderInstruction *pInstToFree; + R700VertexInstruction *pVTXInst; + R700ControlFlowInstruction *pCFInst; + + pInst = pShader->lstVTXInstructions.pHead; + while(NULL != pInst) + { + pVTXInst = (R700VertexInstruction *)pInst; + pShader->uShaderBinaryDWORDSize -= GetInstructionSize(pVTXInst->m_ShaderInstType); + + if(NULL != pVTXInst->m_pLinkedGenericClause) + { + pCFInst = (R700ControlFlowInstruction*)(pVTXInst->m_pLinkedGenericClause); + + TakeInstOutFromList(&(pShader->lstCFInstructions), + (R700ShaderInstruction*)pCFInst); + + pShader->uShaderBinaryDWORDSize -= GetInstructionSize(pCFInst->m_ShaderInstType); + } + + pInst = pInst->pNextInst; + }; + + //destroy each item in pShader->lstVTXInstructions; + pInst = pShader->lstVTXInstructions.pHead; + while(NULL != pInst) + { + pInstToFree = pInst; + pInst = pInst->pNextInst; + FREE(pInstToFree); + }; + + //set NULL pShader->lstVTXInstructions + pShader->lstVTXInstructions.pHead=NULL; + pShader->lstVTXInstructions.pTail=NULL; + pShader->lstVTXInstructions.uNumOfNode=0; +} + void Clean_Up_Shader(R700_Shader *pShader) { FREE(pShader->pProgram); diff --git a/src/mesa/drivers/dri/r600/r700_shader.h b/src/mesa/drivers/dri/r600/r700_shader.h index bfd01e1a93a..997cb05aaf8 100644 --- a/src/mesa/drivers/dri/r600/r700_shader.h +++ b/src/mesa/drivers/dri/r600/r700_shader.h @@ -143,6 +143,7 @@ void LoadProgram(R700_Shader *pShader); void UpdateShaderRegisters(R700_Shader *pShader); void DeleteInstructions(R700_Shader *pShader); void DebugPrint(void); +void cleanup_vfetch_shaderinst(R700_Shader *pShader); void Clean_Up_Shader(R700_Shader *pShader); diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c index fc0b5116843..7e8b48f91eb 100644 --- a/src/mesa/drivers/dri/r600/r700_state.c +++ b/src/mesa/drivers/dri/r600/r700_state.c @@ -92,7 +92,25 @@ void r700UpdateShaders (GLcontext * ctx) //---------------------------------- } } - r700SelectVertexShader(ctx); + r700SelectVertexShader(ctx, 1); + r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS); + context->radeon.NewGLState = 0; +} + +void r700UpdateShaders2(GLcontext * ctx) +{ + context_t *context = R700_CONTEXT(ctx); + + /* should only happenen once, just after context is created */ + /* TODO: shouldn't we fallback to sw here? */ + if (!ctx->FragmentProgram._Current) { + _mesa_fprintf(stderr, "No ctx->FragmentProgram._Current!!\n"); + return; + } + + r700SelectFragmentShader(ctx); + + r700SelectVertexShader(ctx, 2); r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS); context->radeon.NewGLState = 0; } @@ -475,10 +493,10 @@ static void r700SetBlendState(GLcontext * ctx) eqn, COLOR_COMB_FCN_shift, COLOR_COMB_FCN_mask); SETfield(blend_reg, - blend_factor(ctx->Color.BlendSrcRGB, GL_TRUE), + blend_factor(ctx->Color.BlendSrcA, GL_TRUE), ALPHA_SRCBLEND_shift, ALPHA_SRCBLEND_mask); SETfield(blend_reg, - blend_factor(ctx->Color.BlendDstRGB, GL_FALSE), + blend_factor(ctx->Color.BlendDstA, GL_FALSE), ALPHA_DESTBLEND_shift, ALPHA_DESTBLEND_mask); switch (ctx->Color.BlendEquationA) { @@ -753,9 +771,9 @@ static void r700ColorMask(GLcontext * ctx, (b ? 4 : 0) | (a ? 8 : 0)); - if (mask != r700->CB_SHADER_MASK.u32All) { + if (mask != r700->CB_TARGET_MASK.u32All) { R600_STATECHANGE(context, cb); - SETfield(r700->CB_SHADER_MASK.u32All, mask, OUTPUT0_ENABLE_shift, OUTPUT0_ENABLE_mask); + SETfield(r700->CB_TARGET_MASK.u32All, mask, TARGET0_ENABLE_shift, TARGET0_ENABLE_mask); } } @@ -1277,10 +1295,11 @@ void r700SetScissor(context_t *context) //--------------- return; } if (context->radeon.state.scissor.enabled) { + /* r600 has exclusive scissors */ x1 = context->radeon.state.scissor.rect.x1; y1 = context->radeon.state.scissor.rect.y1; - x2 = context->radeon.state.scissor.rect.x2; - y2 = context->radeon.state.scissor.rect.y2; + x2 = context->radeon.state.scissor.rect.x2 + 1; + y2 = context->radeon.state.scissor.rect.y2 + 1; } else { if (context->radeon.radeonScreen->driScreen->dri2.enabled) { x1 = 0; @@ -1761,7 +1780,7 @@ void r700InitState(GLcontext * ctx) //------------------- r700->CB_CLRCMP_MSK.u32All = 0xFFFFFFFF; /* screen/window/view */ - SETfield(r700->CB_TARGET_MASK.u32All, 0xF, (4 * id), TARGET0_ENABLE_mask); + SETfield(r700->CB_SHADER_MASK.u32All, 0xF, (4 * id), OUTPUT0_ENABLE_mask); context->radeon.hw.all_dirty = GL_TRUE; diff --git a/src/mesa/drivers/dri/r600/r700_state.h b/src/mesa/drivers/dri/r600/r700_state.h index 0f53d5b4c59..209189d8d72 100644 --- a/src/mesa/drivers/dri/r600/r700_state.h +++ b/src/mesa/drivers/dri/r600/r700_state.h @@ -35,6 +35,7 @@ extern void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state); extern void r700UpdateShaders (GLcontext * ctx); +extern void r700UpdateShaders2(GLcontext * ctx); extern void r700UpdateViewportOffset(GLcontext * ctx); diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c index 9ee26286d9b..e7a209be9d8 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.c +++ b/src/mesa/drivers/dri/r600/r700_vertprog.c @@ -159,7 +159,35 @@ GLboolean Process_Vertex_Program_Vfetch_Instructions( return GL_TRUE; } -void Map_Vertex_Program(struct r700_vertex_program *vp, +GLboolean Process_Vertex_Program_Vfetch_Instructions2( + GLcontext *ctx, + struct r700_vertex_program *vp, + struct gl_vertex_program *mesa_vp) +{ + int i; + context_t *context = R700_CONTEXT(ctx); + + VTX_FETCH_METHOD vtxFetchMethod; + vtxFetchMethod.bEnableMini = GL_FALSE; + vtxFetchMethod.mega_fetch_remainder = 0; + + for(i=0; i<context->nNumActiveAos; i++) + { + assemble_vfetch_instruction2(&vp->r700AsmCode, + vp->r700AsmCode.ucVP_AttributeMap[context->stream_desc[i].element], + context->stream_desc[i].type, + context->stream_desc[i].size, + context->stream_desc[i].element, + context->stream_desc[i]._signed, + context->stream_desc[i].normalize, + &vtxFetchMethod); + } + + return GL_TRUE; +} + +void Map_Vertex_Program(GLcontext *ctx, + struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp) { GLuint ui; @@ -175,11 +203,22 @@ void Map_Vertex_Program(struct r700_vertex_program *vp, pAsm->number_used_registers += num_inputs; // Create VFETCH instructions for inputs - if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions(vp, mesa_vp) ) - { - radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions return error. \n"); - return; //error - } + if(1 == vp->uiVersion) + { + if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions(vp, mesa_vp) ) + { + radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions return error. \n"); + return; + } + } + else + { + if (GL_TRUE != Process_Vertex_Program_Vfetch_Instructions2(ctx, vp, mesa_vp) ) + { + radeon_error("Calling Process_Vertex_Program_Vfetch_Instructions2 return error. \n"); + return; + } + } // Map Outputs pAsm->number_of_exports = Map_Vertex_Output(pAsm, mesa_vp, pAsm->number_used_registers); @@ -261,7 +300,8 @@ GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp, } struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, - struct gl_vertex_program *mesa_vp) + struct gl_vertex_program *mesa_vp, + GLint nVer) { context_t *context = R700_CONTEXT(ctx); struct r700_vertex_program *vp; @@ -271,6 +311,7 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, unsigned int i; vp = _mesa_calloc(sizeof(*vp)); + vp->uiVersion = nVer; vp->mesa_program = (struct gl_vertex_program *)_mesa_clone_program(ctx, &mesa_vp->Base); if (mesa_vp->IsPositionInvariant) @@ -296,7 +337,7 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, //Init_Program Init_r700_AssemblerBase(SPT_VP, &(vp->r700AsmCode), &(vp->r700Shader) ); - Map_Vertex_Program( vp, vp->mesa_program ); + Map_Vertex_Program(ctx, vp, vp->mesa_program ); if(GL_FALSE == Find_Instruction_Dependencies_vp(vp, vp->mesa_program)) { @@ -325,7 +366,7 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, return vp; } -void r700SelectVertexShader(GLcontext *ctx) +void r700SelectVertexShader(GLcontext *ctx, GLint nVersion) { context_t *context = R700_CONTEXT(ctx); struct r700_vertex_program_cont *vpc; @@ -365,7 +406,7 @@ void r700SelectVertexShader(GLcontext *ctx) } } - vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program) ); + vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program), nVersion); if(!vp) { radeon_error("Failed to translate vertex shader. \n"); @@ -377,6 +418,140 @@ void r700SelectVertexShader(GLcontext *ctx) return; } +int getTypeSize(GLenum type) +{ + switch (type) + { + case GL_DOUBLE: + return sizeof(GLdouble); + case GL_FLOAT: + return sizeof(GLfloat); + case GL_INT: + return sizeof(GLint); + case GL_UNSIGNED_INT: + return sizeof(GLuint); + case GL_SHORT: + return sizeof(GLshort); + case GL_UNSIGNED_SHORT: + return sizeof(GLushort); + case GL_BYTE: + return sizeof(GLbyte); + case GL_UNSIGNED_BYTE: + return sizeof(GLubyte); + default: + assert(0); + return 0; + } +} + +static void r700TranslateAttrib(GLcontext *ctx, GLuint unLoc, int count, const struct gl_client_array *input) +{ + context_t *context = R700_CONTEXT(ctx); + + StreamDesc * pStreamDesc = &(context->stream_desc[context->nNumActiveAos]); + + GLuint stride; + + stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size + : input->StrideB; + + if (input->Type == GL_DOUBLE || input->Type == GL_UNSIGNED_INT || input->Type == GL_INT || +#if MESA_BIG_ENDIAN + getTypeSize(input->Type) != 4 || +#endif + stride < 4) + { + pStreamDesc->type = GL_FLOAT; + + if (input->StrideB == 0) + { + pStreamDesc->stride = 0; + } + else + { + pStreamDesc->stride = sizeof(GLfloat) * input->Size; + } + pStreamDesc->dwords = input->Size; + pStreamDesc->is_named_bo = GL_FALSE; + } + else + { + pStreamDesc->type = input->Type; + pStreamDesc->dwords = (getTypeSize(input->Type) * input->Size + 3)/ 4; + if (!input->BufferObj->Name) + { + if (input->StrideB == 0) + { + pStreamDesc->stride = 0; + } + else + { + pStreamDesc->stride = (getTypeSize(pStreamDesc->type) * input->Size + 3) & ~3; + } + + pStreamDesc->is_named_bo = GL_FALSE; + } + } + + pStreamDesc->size = input->Size; + pStreamDesc->dst_loc = context->nNumActiveAos; + pStreamDesc->element = unLoc; + + switch (pStreamDesc->type) + { //GetSurfaceFormat + case GL_FLOAT: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = GL_FALSE; + break; + case GL_SHORT: + pStreamDesc->_signed = 1; + pStreamDesc->normalize = input->Normalized; + break; + case GL_BYTE: + pStreamDesc->_signed = 1; + pStreamDesc->normalize = input->Normalized; + break; + case GL_UNSIGNED_SHORT: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = input->Normalized; + break; + case GL_UNSIGNED_BYTE: + pStreamDesc->_signed = 0; + pStreamDesc->normalize = input->Normalized; + break; + default: + case GL_INT: + case GL_UNSIGNED_INT: + case GL_DOUBLE: + assert(0); + break; + } + context->nNumActiveAos++; +} + +void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count) +{ + context_t *context = R700_CONTEXT(ctx); + struct r700_vertex_program *vpc + = (struct r700_vertex_program *)ctx->VertexProgram._Current; + + struct gl_vertex_program * mesa_vp = (struct gl_vertex_program *)&(vpc->mesa_program); + unsigned int unLoc = 0; + unsigned int unBit = mesa_vp->Base.InputsRead; + context->nNumActiveAos = 0; + + while(unBit) + { + if(unBit & 1) + { + r700TranslateAttrib(ctx, unLoc, count, arrays[unLoc]); + } + + unBit >>= 1; + ++unLoc; + } +} + void * r700GetActiveVpShaderBo(GLcontext * ctx) { context_t *context = R700_CONTEXT(ctx); diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.h b/src/mesa/drivers/dri/r600/r700_vertprog.h index c48764c43ba..f9a3e395ee9 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.h +++ b/src/mesa/drivers/dri/r600/r700_vertprog.h @@ -52,7 +52,7 @@ struct r700_vertex_program GLboolean translated; GLboolean loaded; - GLboolean needUpdateVF; + GLint uiVersion; void * shaderbo; @@ -76,19 +76,28 @@ unsigned int Map_Vertex_Input(r700_AssemblerBase *pAsm, GLboolean Process_Vertex_Program_Vfetch_Instructions( struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp); -void Map_Vertex_Program(struct r700_vertex_program *vp, +GLboolean Process_Vertex_Program_Vfetch_Instructions2( + GLcontext *ctx, + struct r700_vertex_program *vp, + struct gl_vertex_program *mesa_vp); +void Map_Vertex_Program(GLcontext *ctx, + struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp); GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp, struct gl_vertex_program *mesa_vp); struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx, - struct gl_vertex_program *mesa_vp); + struct gl_vertex_program *mesa_vp, + GLint nVer); /* Interface */ -extern void r700SelectVertexShader(GLcontext *ctx); +extern void r700SelectVertexShader(GLcontext *ctx, GLint nVersion); +extern void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count); extern GLboolean r700SetupVertexProgram(GLcontext * ctx); extern void * r700GetActiveVpShaderBo(GLcontext * ctx); +extern int getTypeSize(GLenum type); + #endif /* _R700_VERTPROG_H_ */ diff --git a/src/mesa/drivers/dri/r600/radeon_buffer_objects.c b/src/mesa/drivers/dri/r600/radeon_buffer_objects.c new file mode 120000 index 00000000000..f6a5f664701 --- /dev/null +++ b/src/mesa/drivers/dri/r600/radeon_buffer_objects.c @@ -0,0 +1 @@ +../radeon/radeon_buffer_objects.c
\ No newline at end of file diff --git a/src/mesa/drivers/dri/r600/radeon_buffer_objects.h b/src/mesa/drivers/dri/r600/radeon_buffer_objects.h new file mode 120000 index 00000000000..2f134fd17b8 --- /dev/null +++ b/src/mesa/drivers/dri/r600/radeon_buffer_objects.h @@ -0,0 +1 @@ +../radeon/radeon_buffer_objects.h
\ No newline at end of file diff --git a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c index a24b6dac265..8fac5c6c512 100644 --- a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c +++ b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c @@ -187,7 +187,11 @@ radeonMapBuffer(GLcontext * ctx, radeon_bo_map(radeon_obj->bo, access == GL_WRITE_ONLY_ARB); - return obj->Pointer = radeon_obj->bo->ptr; + obj->Pointer = radeon_obj->bo->ptr; + obj->Length = obj->Size; + obj->Offset = 0; + + return obj->Pointer; } @@ -203,9 +207,12 @@ radeonUnmapBuffer(GLcontext * ctx, if (radeon_obj->bo != NULL) { radeon_bo_unmap(radeon_obj->bo); - obj->Pointer = NULL; } + obj->Pointer = NULL; + obj->Offset = 0; + obj->Length = 0; + return GL_TRUE; } diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c index 264392b3270..f8a4cdb4954 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common.c +++ b/src/mesa/drivers/dri/radeon/radeon_common.c @@ -232,13 +232,13 @@ void radeonUpdateScissor( GLcontext *ctx ) __DRIdrawablePrivate *dPriv = radeon_get_drawable(rmesa); x1 += dPriv->x; - x2 += dPriv->x + 1; + x2 += dPriv->x; min_x += dPriv->x; - max_x += dPriv->x + 1; + max_x += dPriv->x; y1 += dPriv->y; - y2 += dPriv->y + 1; + y2 += dPriv->y; min_y += dPriv->y; - max_y += dPriv->y + 1; + max_y += dPriv->y; } rmesa->state.scissor.rect.x1 = CLAMP(x1, min_x, max_x); diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c index 1c53c04da77..6b9b1e3c5e4 100644 --- a/src/mesa/drivers/dri/radeon/radeon_common_context.c +++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c @@ -227,11 +227,8 @@ GLboolean radeonInitContext(radeonContextPtr radeon, fthrottle_mode = driQueryOptioni(&radeon->optionCache, "fthrottle_mode"); radeon->iw.irq_seq = -1; radeon->irqsEmitted = 0; - if (IS_R600_CLASS(radeon->radeonScreen)) - radeon->do_irqs = 0; - else - radeon->do_irqs = (fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS && - radeon->radeonScreen->irq); + radeon->do_irqs = (fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS && + radeon->radeonScreen->irq); radeon->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS); diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.c b/src/mesa/drivers/dri/radeon/radeon_texture.c index 049284ef8c5..7b7392b2179 100644 --- a/src/mesa/drivers/dri/radeon/radeon_texture.c +++ b/src/mesa/drivers/dri/radeon/radeon_texture.c @@ -101,7 +101,12 @@ void radeonFreeTexImageData(GLcontext *ctx, struct gl_texture_image *timage) /* Set Data pointer and additional data for mapped texture image */ static void teximage_set_map_data(radeon_texture_image *image) { - radeon_mipmap_level *lvl = &image->mt->levels[image->mtlevel]; + radeon_mipmap_level *lvl; + + if (!image->mt) + return; + + lvl = &image->mt->levels[image->mtlevel]; image->base.Data = image->mt->bo->ptr + lvl->faces[image->mtface].offset; image->base.RowStride = lvl->rowstride / image->mt->bpp; @@ -969,7 +974,7 @@ int radeon_validate_texture_miptree(GLcontext * ctx, struct gl_texture_object *t radeon_texture_image *image = get_radeon_texture_image(texObj->Image[face][level]); if (RADEON_DEBUG & RADEON_TEXTURE) fprintf(stderr, " face %i, level %i... %p vs %p ", face, level, t->mt, image->mt); - if (t->mt == image->mt) { + if (t->mt == image->mt || (!image->mt && !image->base.Data)) { if (RADEON_DEBUG & RADEON_TEXTURE) fprintf(stderr, "OK\n"); diff --git a/src/mesa/drivers/dri/swrast/swrast.c b/src/mesa/drivers/dri/swrast/swrast.c index cd499cd5d2e..f4947daa063 100644 --- a/src/mesa/drivers/dri/swrast/swrast.c +++ b/src/mesa/drivers/dri/swrast/swrast.c @@ -116,6 +116,7 @@ const struct dri_extension card_extensions[] = { "GL_NV_depth_clamp", NULL }, { "GL_NV_vertex_program", GL_NV_vertex_program_functions }, { "GL_NV_fragment_program", GL_NV_fragment_program_functions }, + { "GL_NV_fragment_program_option", NULL }, { NULL, NULL } }; diff --git a/src/mesa/drivers/x11/fakeglx.c b/src/mesa/drivers/x11/fakeglx.c index 34e0b8bc8d7..eb7c4f6417c 100644 --- a/src/mesa/drivers/x11/fakeglx.c +++ b/src/mesa/drivers/x11/fakeglx.c @@ -2477,9 +2477,9 @@ Fake_glXQueryContext( Display *dpy, GLXContext ctx, int attribute, int *value ) break; case GLX_RENDER_TYPE: if (xmctx->xm_visual->mesa_visual.rgbMode) - *value = GLX_RGBA_BIT; + *value = GLX_RGBA_TYPE; else - *value = GLX_COLOR_INDEX_BIT; + *value = GLX_COLOR_INDEX_TYPE; break; case GLX_SCREEN: *value = 0; diff --git a/src/mesa/main/accum.c b/src/mesa/main/accum.c index 2345695f3c6..032e13b96eb 100644 --- a/src/mesa/main/accum.c +++ b/src/mesa/main/accum.c @@ -29,6 +29,10 @@ #include "macros.h" #include "state.h" #include "mtypes.h" +#include "glapi/dispatch.h" + + +#if FEATURE_accum void GLAPIENTRY @@ -51,7 +55,7 @@ _mesa_ClearAccum( GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_Accum( GLenum op, GLfloat value ) { GET_CURRENT_CONTEXT(ctx); @@ -99,6 +103,16 @@ _mesa_Accum( GLenum op, GLfloat value ) } +void +_mesa_init_accum_dispatch(struct _glapi_table *disp) +{ + SET_Accum(disp, _mesa_Accum); + SET_ClearAccum(disp, _mesa_ClearAccum); +} + + +#endif /* FEATURE_accum */ + void _mesa_init_accum( GLcontext *ctx ) diff --git a/src/mesa/main/accum.h b/src/mesa/main/accum.h index ce92688a5b9..63740f07edc 100644 --- a/src/mesa/main/accum.h +++ b/src/mesa/main/accum.h @@ -38,25 +38,40 @@ #define ACCUM_H -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL - -extern void GLAPIENTRY -_mesa_Accum( GLenum op, GLfloat value ); +#if FEATURE_accum +#define _MESA_INIT_ACCUM_FUNCTIONS(driver, impl) \ + do { \ + (driver)->Accum = impl ## Accum; \ + } while (0) extern void GLAPIENTRY _mesa_ClearAccum( GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha ); -extern void -_mesa_init_accum( GLcontext *ctx ); +extern void +_mesa_init_accum_dispatch(struct _glapi_table *disp); + +#else /* FEATURE_accum */ -#else +#define _MESA_INIT_ACCUM_FUNCTIONS(driver, impl) do { } while (0) -/** No-op */ -#define _mesa_init_accum( c ) ((void)0) +static INLINE void +_mesa_ClearAccum( GLfloat red, GLfloat green, GLfloat blue, GLfloat alpha ) +{ + /* this is used in _mesa_PopAttrib */ + ASSERT_NO_FEATURE(); +} -#endif +static INLINE void +_mesa_init_accum_dispatch(struct _glapi_table *disp) +{ +} + +#endif /* FEATURE_accum */ + +extern void +_mesa_init_accum( GLcontext *ctx ); -#endif +#endif /* ACCUM_H */ diff --git a/src/mesa/main/api_arrayelt.c b/src/mesa/main/api_arrayelt.c index 2462a1b0037..a058227110d 100644 --- a/src/mesa/main/api_arrayelt.c +++ b/src/mesa/main/api_arrayelt.c @@ -71,6 +71,10 @@ typedef struct { */ #define TYPE_IDX(t) ( (t) == GL_DOUBLE ? 7 : (t) & 7 ) + +#if FEATURE_arrayelt + + static const int ColorFuncs[2][8] = { { _gloffset_Color3bv, @@ -1160,7 +1164,7 @@ static void _ae_update_state( GLcontext *ctx ) at->array = attribArray; /* Note: we can't grab the _glapi_Dispatch->VertexAttrib1fvNV * function pointer here (for float arrays) since the pointer may - * change from one execution of _ae_loopback_array_elt() to + * change from one execution of _ae_ArrayElement() to * the next. Doing so caused UT to break. */ if (ctx->VertexProgram._Enabled @@ -1254,7 +1258,7 @@ void _ae_unmap_vbos( GLcontext *ctx ) * for all enabled vertex arrays (for elt-th element). * Note: this may be called during display list construction. */ -void GLAPIENTRY _ae_loopback_array_elt( GLint elt ) +void GLAPIENTRY _ae_ArrayElement( GLint elt ) { GET_CURRENT_CONTEXT(ctx); const AEcontext *actx = AE_CONTEXT(ctx); @@ -1317,3 +1321,13 @@ void _ae_invalidate_state( GLcontext *ctx, GLuint new_state ) actx->NewState |= new_state; } } + + +void _mesa_install_arrayelt_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt) +{ + SET_ArrayElement(disp, vfmt->ArrayElement); +} + + +#endif /* FEATURE_arrayelt */ diff --git a/src/mesa/main/api_arrayelt.h b/src/mesa/main/api_arrayelt.h index e621724fb29..d18c0792c35 100644 --- a/src/mesa/main/api_arrayelt.h +++ b/src/mesa/main/api_arrayelt.h @@ -27,16 +27,57 @@ #ifndef API_ARRAYELT_H #define API_ARRAYELT_H -#include "mtypes.h" + +#include "main/mtypes.h" + +#if FEATURE_arrayelt + +#define _MESA_INIT_ARRAYELT_VTXFMT(vfmt, impl) \ + do { \ + (vfmt)->ArrayElement = impl ## ArrayElement; \ + } while (0) extern GLboolean _ae_create_context( GLcontext *ctx ); extern void _ae_destroy_context( GLcontext *ctx ); extern void _ae_invalidate_state( GLcontext *ctx, GLuint new_state ); -extern void GLAPIENTRY _ae_loopback_array_elt( GLint elt ); +extern void GLAPIENTRY _ae_ArrayElement( GLint elt ); /* May optionally be called before a batch of element calls: */ extern void _ae_map_vbos( GLcontext *ctx ); extern void _ae_unmap_vbos( GLcontext *ctx ); -#endif +extern void +_mesa_install_arrayelt_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt); + +#else /* FEATURE_arrayelt */ + +#define _MESA_INIT_ARRAYELT_VTXFMT(vfmt, impl) do { } while (0) + +static INLINE GLboolean +_ae_create_context( GLcontext *ctx ) +{ + return GL_TRUE; +} + +static INLINE void +_ae_destroy_context( GLcontext *ctx ) +{ +} + +static INLINE void +_ae_invalidate_state( GLcontext *ctx, GLuint new_state ) +{ +} + +static INLINE void +_mesa_install_arrayelt_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt) +{ +} + +#endif /* FEATURE_arrayelt */ + + +#endif /* API_ARRAYELT_H */ diff --git a/src/mesa/main/api_exec.c b/src/mesa/main/api_exec.c index 02550ae1088..1559984f43a 100644 --- a/src/mesa/main/api_exec.c +++ b/src/mesa/main/api_exec.c @@ -30,9 +30,7 @@ #include "mfeatures.h" -#if FEATURE_accum #include "accum.h" -#endif #include "api_loopback.h" #include "api_exec.h" #if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program @@ -41,9 +39,7 @@ #if FEATURE_ATI_fragment_shader #include "shader/atifragshader.h" #endif -#if FEATURE_attrib_stack #include "attrib.h" -#endif #include "blend.h" #if FEATURE_ARB_vertex_buffer_object #include "bufferobj.h" @@ -54,29 +50,17 @@ #endif #include "clear.h" #include "clip.h" -#if FEATURE_colortable #include "colortab.h" -#endif #include "context.h" -#if FEATURE_convolve #include "convolve.h" -#endif #include "depth.h" -#if FEATURE_dlist #include "dlist.h" -#endif -#if FEATURE_drawpix #include "drawpix.h" #include "rastpos.h" -#endif #include "enable.h" -#if FEATURE_evaluators #include "eval.h" -#endif #include "get.h" -#if FEATURE_feedback #include "feedback.h" -#endif #include "fog.h" #if FEATURE_EXT_framebuffer_object #include "fbobject.h" @@ -84,24 +68,18 @@ #include "ffvertex_prog.h" #include "framebuffer.h" #include "hint.h" -#if FEATURE_histogram #include "histogram.h" -#endif #include "imports.h" #include "light.h" #include "lines.h" #include "macros.h" #include "matrix.h" #include "multisample.h" -#if FEATURE_pixel_transfer #include "pixel.h" -#endif #include "pixelstore.h" #include "points.h" #include "polygon.h" -#if FEATURE_ARB_occlusion_query || FEATURE_EXT_timer_query #include "queryobj.h" -#endif #include "readpix.h" #include "scissor.h" #include "state.h" @@ -109,9 +87,7 @@ #include "texenv.h" #include "texgetimage.h" #include "teximage.h" -#if FEATURE_texgen #include "texgen.h" -#endif #include "texobj.h" #include "texparam.h" #include "texstate.h" @@ -199,20 +175,10 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_TexParameteri(exec, _mesa_TexParameteri); SET_Translatef(exec, _mesa_Translatef); SET_Viewport(exec, _mesa_Viewport); -#if FEATURE_accum - SET_Accum(exec, _mesa_Accum); - SET_ClearAccum(exec, _mesa_ClearAccum); -#endif -#if FEATURE_dlist - SET_CallList(exec, _mesa_CallList); - SET_CallLists(exec, _mesa_CallLists); - SET_DeleteLists(exec, _mesa_DeleteLists); - SET_EndList(exec, _mesa_EndList); - SET_GenLists(exec, _mesa_GenLists); - SET_IsList(exec, _mesa_IsList); - SET_ListBase(exec, _mesa_ListBase); - SET_NewList(exec, _mesa_NewList); -#endif + + _mesa_init_accum_dispatch(exec); + _mesa_init_dlist_dispatch(exec); + SET_ClearDepth(exec, _mesa_ClearDepth); SET_ClearIndex(exec, _mesa_ClearIndex); SET_ClipPlane(exec, _mesa_ClipPlane); @@ -222,21 +188,10 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_DepthFunc(exec, _mesa_DepthFunc); SET_DepthMask(exec, _mesa_DepthMask); SET_DepthRange(exec, _mesa_DepthRange); -#if FEATURE_drawpix - SET_Bitmap(exec, _mesa_Bitmap); - SET_CopyPixels(exec, _mesa_CopyPixels); - SET_DrawPixels(exec, _mesa_DrawPixels); -#endif -#if FEATURE_feedback - SET_InitNames(exec, _mesa_InitNames); - SET_FeedbackBuffer(exec, _mesa_FeedbackBuffer); - SET_LoadName(exec, _mesa_LoadName); - SET_PassThrough(exec, _mesa_PassThrough); - SET_PopName(exec, _mesa_PopName); - SET_PushName(exec, _mesa_PushName); - SET_SelectBuffer(exec, _mesa_SelectBuffer); - SET_RenderMode(exec, _mesa_RenderMode); -#endif + + _mesa_init_drawpix_dispatch(exec); + _mesa_init_feedback_dispatch(exec); + SET_FogCoordPointerEXT(exec, _mesa_FogCoordPointerEXT); SET_Fogf(exec, _mesa_Fogf); SET_Fogfv(exec, _mesa_Fogfv); @@ -270,68 +225,22 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_Lighti(exec, _mesa_Lighti); SET_Lightiv(exec, _mesa_Lightiv); SET_LoadMatrixd(exec, _mesa_LoadMatrixd); -#if FEATURE_evaluators - SET_GetMapdv(exec, _mesa_GetMapdv); - SET_GetMapfv(exec, _mesa_GetMapfv); - SET_GetMapiv(exec, _mesa_GetMapiv); - SET_Map1d(exec, _mesa_Map1d); - SET_Map1f(exec, _mesa_Map1f); - SET_Map2d(exec, _mesa_Map2d); - SET_Map2f(exec, _mesa_Map2f); - SET_MapGrid1d(exec, _mesa_MapGrid1d); - SET_MapGrid1f(exec, _mesa_MapGrid1f); - SET_MapGrid2d(exec, _mesa_MapGrid2d); - SET_MapGrid2f(exec, _mesa_MapGrid2f); -#endif + + _mesa_init_eval_dispatch(exec); + SET_MultMatrixd(exec, _mesa_MultMatrixd); -#if FEATURE_pixel_transfer - SET_GetPixelMapfv(exec, _mesa_GetPixelMapfv); - SET_GetPixelMapuiv(exec, _mesa_GetPixelMapuiv); - SET_GetPixelMapusv(exec, _mesa_GetPixelMapusv); - SET_PixelMapfv(exec, _mesa_PixelMapfv); - SET_PixelMapuiv(exec, _mesa_PixelMapuiv); - SET_PixelMapusv(exec, _mesa_PixelMapusv); - SET_PixelTransferf(exec, _mesa_PixelTransferf); - SET_PixelTransferi(exec, _mesa_PixelTransferi); - SET_PixelZoom(exec, _mesa_PixelZoom); -#endif + + _mesa_init_pixel_dispatch(exec); + SET_PixelStoref(exec, _mesa_PixelStoref); SET_PointSize(exec, _mesa_PointSize); SET_PolygonMode(exec, _mesa_PolygonMode); SET_PolygonOffset(exec, _mesa_PolygonOffset); SET_PolygonStipple(exec, _mesa_PolygonStipple); -#if FEATURE_attrib_stack - SET_PopAttrib(exec, _mesa_PopAttrib); - SET_PushAttrib(exec, _mesa_PushAttrib); - SET_PopClientAttrib(exec, _mesa_PopClientAttrib); - SET_PushClientAttrib(exec, _mesa_PushClientAttrib); -#endif -#if FEATURE_drawpix - SET_RasterPos2f(exec, _mesa_RasterPos2f); - SET_RasterPos2fv(exec, _mesa_RasterPos2fv); - SET_RasterPos2i(exec, _mesa_RasterPos2i); - SET_RasterPos2iv(exec, _mesa_RasterPos2iv); - SET_RasterPos2d(exec, _mesa_RasterPos2d); - SET_RasterPos2dv(exec, _mesa_RasterPos2dv); - SET_RasterPos2s(exec, _mesa_RasterPos2s); - SET_RasterPos2sv(exec, _mesa_RasterPos2sv); - SET_RasterPos3d(exec, _mesa_RasterPos3d); - SET_RasterPos3dv(exec, _mesa_RasterPos3dv); - SET_RasterPos3f(exec, _mesa_RasterPos3f); - SET_RasterPos3fv(exec, _mesa_RasterPos3fv); - SET_RasterPos3i(exec, _mesa_RasterPos3i); - SET_RasterPos3iv(exec, _mesa_RasterPos3iv); - SET_RasterPos3s(exec, _mesa_RasterPos3s); - SET_RasterPos3sv(exec, _mesa_RasterPos3sv); - SET_RasterPos4d(exec, _mesa_RasterPos4d); - SET_RasterPos4dv(exec, _mesa_RasterPos4dv); - SET_RasterPos4f(exec, _mesa_RasterPos4f); - SET_RasterPos4fv(exec, _mesa_RasterPos4fv); - SET_RasterPos4i(exec, _mesa_RasterPos4i); - SET_RasterPos4iv(exec, _mesa_RasterPos4iv); - SET_RasterPos4s(exec, _mesa_RasterPos4s); - SET_RasterPos4sv(exec, _mesa_RasterPos4sv); -#endif + + _mesa_init_attrib_dispatch(exec); + _mesa_init_rastpos_dispatch(exec); + SET_ReadPixels(exec, _mesa_ReadPixels); SET_Rotated(exec, _mesa_Rotated); SET_Scaled(exec, _mesa_Scaled); @@ -339,17 +248,7 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_TexEnvf(exec, _mesa_TexEnvf); SET_TexEnviv(exec, _mesa_TexEnviv); -#if FEATURE_texgen - SET_GetTexGendv(exec, _mesa_GetTexGendv); - SET_GetTexGenfv(exec, _mesa_GetTexGenfv); - SET_GetTexGeniv(exec, _mesa_GetTexGeniv); - SET_TexGend(exec, _mesa_TexGend); - SET_TexGendv(exec, _mesa_TexGendv); - SET_TexGenf(exec, _mesa_TexGenf); - SET_TexGenfv(exec, _mesa_TexGenfv); - SET_TexGeni(exec, _mesa_TexGeni); - SET_TexGeniv(exec, _mesa_TexGeniv); -#endif + _mesa_init_texgen_dispatch(exec); SET_TexImage1D(exec, _mesa_TexImage1D); SET_TexParameterf(exec, _mesa_TexParameterf); @@ -395,45 +294,9 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_BlendEquation(exec, _mesa_BlendEquation); SET_BlendEquationSeparateEXT(exec, _mesa_BlendEquationSeparateEXT); -#if FEATURE_colortable - SET_ColorSubTable(exec, _mesa_ColorSubTable); - SET_ColorTable(exec, _mesa_ColorTable); - SET_ColorTableParameterfv(exec, _mesa_ColorTableParameterfv); - SET_ColorTableParameteriv(exec, _mesa_ColorTableParameteriv); - SET_CopyColorSubTable(exec, _mesa_CopyColorSubTable); - SET_CopyColorTable(exec, _mesa_CopyColorTable); - SET_GetColorTable(exec, _mesa_GetColorTable); - SET_GetColorTableParameterfv(exec, _mesa_GetColorTableParameterfv); - SET_GetColorTableParameteriv(exec, _mesa_GetColorTableParameteriv); -#endif - -#if FEATURE_convolve - SET_ConvolutionFilter1D(exec, _mesa_ConvolutionFilter1D); - SET_ConvolutionFilter2D(exec, _mesa_ConvolutionFilter2D); - SET_ConvolutionParameterf(exec, _mesa_ConvolutionParameterf); - SET_ConvolutionParameterfv(exec, _mesa_ConvolutionParameterfv); - SET_ConvolutionParameteri(exec, _mesa_ConvolutionParameteri); - SET_ConvolutionParameteriv(exec, _mesa_ConvolutionParameteriv); - SET_CopyConvolutionFilter1D(exec, _mesa_CopyConvolutionFilter1D); - SET_CopyConvolutionFilter2D(exec, _mesa_CopyConvolutionFilter2D); - SET_GetConvolutionFilter(exec, _mesa_GetConvolutionFilter); - SET_GetConvolutionParameterfv(exec, _mesa_GetConvolutionParameterfv); - SET_GetConvolutionParameteriv(exec, _mesa_GetConvolutionParameteriv); - SET_SeparableFilter2D(exec, _mesa_SeparableFilter2D); -#endif -#if FEATURE_histogram - SET_GetHistogram(exec, _mesa_GetHistogram); - SET_GetHistogramParameterfv(exec, _mesa_GetHistogramParameterfv); - SET_GetHistogramParameteriv(exec, _mesa_GetHistogramParameteriv); - SET_GetMinmax(exec, _mesa_GetMinmax); - SET_GetMinmaxParameterfv(exec, _mesa_GetMinmaxParameterfv); - SET_GetMinmaxParameteriv(exec, _mesa_GetMinmaxParameteriv); - SET_GetSeparableFilter(exec, _mesa_GetSeparableFilter); - SET_Histogram(exec, _mesa_Histogram); - SET_Minmax(exec, _mesa_Minmax); - SET_ResetHistogram(exec, _mesa_ResetHistogram); - SET_ResetMinmax(exec, _mesa_ResetMinmax); -#endif + _mesa_init_colortable_dispatch(exec); + _mesa_init_convolve_dispatch(exec); + _mesa_init_histogram_dispatch(exec); /* OpenGL 2.0 */ SET_StencilFuncSeparate(exec, _mesa_StencilFuncSeparate); @@ -545,32 +408,7 @@ _mesa_init_exec_table(struct _glapi_table *exec) #endif /* 197. GL_MESA_window_pos */ -#if FEATURE_drawpix - SET_WindowPos2dMESA(exec, _mesa_WindowPos2dMESA); - SET_WindowPos2dvMESA(exec, _mesa_WindowPos2dvMESA); - SET_WindowPos2fMESA(exec, _mesa_WindowPos2fMESA); - SET_WindowPos2fvMESA(exec, _mesa_WindowPos2fvMESA); - SET_WindowPos2iMESA(exec, _mesa_WindowPos2iMESA); - SET_WindowPos2ivMESA(exec, _mesa_WindowPos2ivMESA); - SET_WindowPos2sMESA(exec, _mesa_WindowPos2sMESA); - SET_WindowPos2svMESA(exec, _mesa_WindowPos2svMESA); - SET_WindowPos3dMESA(exec, _mesa_WindowPos3dMESA); - SET_WindowPos3dvMESA(exec, _mesa_WindowPos3dvMESA); - SET_WindowPos3fMESA(exec, _mesa_WindowPos3fMESA); - SET_WindowPos3fvMESA(exec, _mesa_WindowPos3fvMESA); - SET_WindowPos3iMESA(exec, _mesa_WindowPos3iMESA); - SET_WindowPos3ivMESA(exec, _mesa_WindowPos3ivMESA); - SET_WindowPos3sMESA(exec, _mesa_WindowPos3sMESA); - SET_WindowPos3svMESA(exec, _mesa_WindowPos3svMESA); - SET_WindowPos4dMESA(exec, _mesa_WindowPos4dMESA); - SET_WindowPos4dvMESA(exec, _mesa_WindowPos4dvMESA); - SET_WindowPos4fMESA(exec, _mesa_WindowPos4fMESA); - SET_WindowPos4fvMESA(exec, _mesa_WindowPos4fvMESA); - SET_WindowPos4iMESA(exec, _mesa_WindowPos4iMESA); - SET_WindowPos4ivMESA(exec, _mesa_WindowPos4ivMESA); - SET_WindowPos4sMESA(exec, _mesa_WindowPos4sMESA); - SET_WindowPos4svMESA(exec, _mesa_WindowPos4svMESA); -#endif + /* part of _mesa_init_rastpos_dispatch(exec); */ /* 200. GL_IBM_multimode_draw_arrays */ #if _HAVE_FULL_GL @@ -762,16 +600,7 @@ _mesa_init_exec_table(struct _glapi_table *exec) #endif /* ARB 29. GL_ARB_occlusion_query */ -#if FEATURE_ARB_occlusion_query - SET_GenQueriesARB(exec, _mesa_GenQueriesARB); - SET_DeleteQueriesARB(exec, _mesa_DeleteQueriesARB); - SET_IsQueryARB(exec, _mesa_IsQueryARB); - SET_BeginQueryARB(exec, _mesa_BeginQueryARB); - SET_EndQueryARB(exec, _mesa_EndQueryARB); - SET_GetQueryivARB(exec, _mesa_GetQueryivARB); - SET_GetQueryObjectivARB(exec, _mesa_GetQueryObjectivARB); - SET_GetQueryObjectuivARB(exec, _mesa_GetQueryObjectuivARB); -#endif + _mesa_init_queryobj_dispatch(exec); /* ARB 37. GL_ARB_draw_buffers */ #if FEATURE_draw_read_buffer @@ -881,11 +710,6 @@ _mesa_init_exec_table(struct _glapi_table *exec) SET_GenerateMipmapEXT(exec, _mesa_GenerateMipmapEXT); #endif -#if FEATURE_EXT_timer_query - SET_GetQueryObjecti64vEXT(exec, _mesa_GetQueryObjecti64vEXT); - SET_GetQueryObjectui64vEXT(exec, _mesa_GetQueryObjectui64vEXT); -#endif - #if FEATURE_EXT_framebuffer_blit SET_BlitFramebufferEXT(exec, _mesa_BlitFramebufferEXT); #endif diff --git a/src/mesa/main/api_loopback.c b/src/mesa/main/api_loopback.c index 0e3f5ff9570..3d466ac44a8 100644 --- a/src/mesa/main/api_loopback.c +++ b/src/mesa/main/api_loopback.c @@ -77,6 +77,10 @@ #define FOGCOORDF(x) CALL_FogCoordfEXT(GET_DISPATCH(), (x)) #define SECONDARYCOLORF(a,b,c) CALL_SecondaryColor3fEXT(GET_DISPATCH(), (a,b,c)) + +#if FEATURE_beginend + + static void GLAPIENTRY loopback_Color3b_f( GLbyte red, GLbyte green, GLbyte blue ) { @@ -1655,3 +1659,6 @@ _mesa_loopback_init_api_table( struct _glapi_table *dest ) SET_VertexAttrib4NusvARB(dest, loopback_VertexAttrib4NusvARB); SET_VertexAttrib4NuivARB(dest, loopback_VertexAttrib4NuivARB); } + + +#endif /* FEATURE_beginend */ diff --git a/src/mesa/main/api_loopback.h b/src/mesa/main/api_loopback.h index 6f85bbc1d92..3140eb515ee 100644 --- a/src/mesa/main/api_loopback.h +++ b/src/mesa/main/api_loopback.h @@ -27,11 +27,19 @@ #ifndef API_LOOPBACK_H #define API_LOOPBACK_H -#include "glheader.h" +#include "main/mtypes.h" - -struct _glapi_table; +#if FEATURE_beginend extern void _mesa_loopback_init_api_table( struct _glapi_table *dest ); -#endif +#else /* FEATURE_beginend */ + +static INLINE void +_mesa_loopback_init_api_table( struct _glapi_table *dest ) +{ +} + +#endif /* FEATURE_beginend */ + +#endif /* API_LOOPBACK_H */ diff --git a/src/mesa/main/api_noop.c b/src/mesa/main/api_noop.c index 0b669e7e7fc..f72f957300d 100644 --- a/src/mesa/main/api_noop.c +++ b/src/mesa/main/api_noop.c @@ -30,9 +30,8 @@ #include "context.h" #include "light.h" #include "macros.h" -#if FEATURE_dlist #include "dlist.h" -#endif +#include "eval.h" #include "glapi/dispatch.h" @@ -44,6 +43,9 @@ */ +#if FEATURE_beginend + + static void GLAPIENTRY _mesa_noop_EdgeFlag( GLboolean b ) { GET_CURRENT_CONTEXT(ctx); @@ -992,26 +994,21 @@ _mesa_noop_EvalMesh2( GLenum mode, GLint i1, GLint i2, GLint j1, GLint j2 ) void _mesa_noop_vtxfmt_init( GLvertexformat *vfmt ) { - vfmt->ArrayElement = _ae_loopback_array_elt; /* generic helper */ + _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_); + vfmt->Begin = _mesa_noop_Begin; -#if FEATURE_dlist - vfmt->CallList = _mesa_CallList; - vfmt->CallLists = _mesa_CallLists; -#endif + + _MESA_INIT_DLIST_VTXFMT(vfmt, _mesa_); + vfmt->Color3f = _mesa_noop_Color3f; vfmt->Color3fv = _mesa_noop_Color3fv; vfmt->Color4f = _mesa_noop_Color4f; vfmt->Color4fv = _mesa_noop_Color4fv; vfmt->EdgeFlag = _mesa_noop_EdgeFlag; vfmt->End = _mesa_noop_End; -#if FEATURE_evaluators - vfmt->EvalCoord1f = _mesa_noop_EvalCoord1f; - vfmt->EvalCoord1fv = _mesa_noop_EvalCoord1fv; - vfmt->EvalCoord2f = _mesa_noop_EvalCoord2f; - vfmt->EvalCoord2fv = _mesa_noop_EvalCoord2fv; - vfmt->EvalPoint1 = _mesa_noop_EvalPoint1; - vfmt->EvalPoint2 = _mesa_noop_EvalPoint2; -#endif + + _MESA_INIT_EVAL_VTXFMT(vfmt, _mesa_noop_); + vfmt->FogCoordfEXT = _mesa_noop_FogCoordfEXT; vfmt->FogCoordfvEXT = _mesa_noop_FogCoordfvEXT; vfmt->Indexf = _mesa_noop_Indexf; @@ -1069,6 +1066,7 @@ _mesa_noop_vtxfmt_init( GLvertexformat *vfmt ) vfmt->DrawElementsBaseVertex = _mesa_noop_DrawElementsBaseVertex; vfmt->DrawRangeElementsBaseVertex = _mesa_noop_DrawRangeElementsBaseVertex; vfmt->MultiDrawElementsBaseVertex = _mesa_noop_MultiDrawElementsBaseVertex; - vfmt->EvalMesh1 = _mesa_noop_EvalMesh1; - vfmt->EvalMesh2 = _mesa_noop_EvalMesh2; } + + +#endif /* FEATURE_beginend */ diff --git a/src/mesa/main/api_noop.h b/src/mesa/main/api_noop.h index 1150984d64b..e7fd49bafbb 100644 --- a/src/mesa/main/api_noop.h +++ b/src/mesa/main/api_noop.h @@ -25,8 +25,9 @@ #ifndef _API_NOOP_H #define _API_NOOP_H -#include "mtypes.h" -#include "context.h" +#include "main/mtypes.h" + +#if FEATURE_beginend extern void GLAPIENTRY _mesa_noop_Rectf(GLfloat x1, GLfloat y1, GLfloat x2, GLfloat y2); @@ -54,4 +55,6 @@ _mesa_noop_MultiDrawElementsBaseVertex(GLenum mode, const GLsizei *count, extern void _mesa_noop_vtxfmt_init(GLvertexformat *vfmt); -#endif +#endif /* FEATURE_beginend */ + +#endif /* _API_NOOP_H */ diff --git a/src/mesa/main/api_validate.c b/src/mesa/main/api_validate.c index 4a1448deee6..e71e5a6ce86 100644 --- a/src/mesa/main/api_validate.c +++ b/src/mesa/main/api_validate.c @@ -52,6 +52,51 @@ index_bytes(GLenum type, GLsizei count) /** + * Find the max index in the given element/index buffer + */ +GLuint +_mesa_max_buffer_index(GLcontext *ctx, GLuint count, GLenum type, + const void *indices, + struct gl_buffer_object *elementBuf) +{ + const GLubyte *map = NULL; + GLuint max = 0; + GLuint i; + + if (_mesa_is_bufferobj(elementBuf)) { + /* elements are in a user-defined buffer object. need to map it */ + map = ctx->Driver.MapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER, + GL_READ_ONLY, elementBuf); + /* Actual address is the sum of pointers */ + indices = (const GLvoid *) ADD_POINTERS(map, (const GLubyte *) indices); + } + + if (type == GL_UNSIGNED_INT) { + for (i = 0; i < count; i++) + if (((GLuint *) indices)[i] > max) + max = ((GLuint *) indices)[i]; + } + else if (type == GL_UNSIGNED_SHORT) { + for (i = 0; i < count; i++) + if (((GLushort *) indices)[i] > max) + max = ((GLushort *) indices)[i]; + } + else { + ASSERT(type == GL_UNSIGNED_BYTE); + for (i = 0; i < count; i++) + if (((GLubyte *) indices)[i] > max) + max = ((GLubyte *) indices)[i]; + } + + if (map) { + ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, elementBuf); + } + + return max; +} + + +/** * Check if OK to draw arrays/elements. */ static GLboolean diff --git a/src/mesa/main/api_validate.h b/src/mesa/main/api_validate.h index 1d3ae157d73..6064d15fe6c 100644 --- a/src/mesa/main/api_validate.h +++ b/src/mesa/main/api_validate.h @@ -30,6 +30,12 @@ #include "mtypes.h" + +extern GLuint +_mesa_max_buffer_index(GLcontext *ctx, GLuint count, GLenum type, + const void *indices, + struct gl_buffer_object *elementBuf); + extern GLboolean _mesa_validate_DrawArrays(GLcontext *ctx, GLenum mode, GLint start, GLsizei count); diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c index 0fb8fa3bba2..246c5521b7d 100644 --- a/src/mesa/main/attrib.c +++ b/src/mesa/main/attrib.c @@ -57,6 +57,7 @@ #include "varray.h" #include "viewport.h" #include "mtypes.h" +#include "glapi/dispatch.h" /** @@ -174,6 +175,9 @@ struct texture_state }; +#if FEATURE_attrib_stack + + /** * Allocate new attribute node of given type/kind. Attach payload data. * Insert it into the linked list named by 'head'. @@ -1464,6 +1468,19 @@ _mesa_PopClientAttrib(void) } +void +_mesa_init_attrib_dispatch(struct _glapi_table *disp) +{ + SET_PopAttrib(disp, _mesa_PopAttrib); + SET_PushAttrib(disp, _mesa_PushAttrib); + SET_PopClientAttrib(disp, _mesa_PopClientAttrib); + SET_PushClientAttrib(disp, _mesa_PushClientAttrib); +} + + +#endif /* FEATURE_attrib_stack */ + + /** * Free any attribute state data that might be attached to the context. */ diff --git a/src/mesa/main/attrib.h b/src/mesa/main/attrib.h index 2cf8fe69345..6b48a176630 100644 --- a/src/mesa/main/attrib.h +++ b/src/mesa/main/attrib.h @@ -26,10 +26,10 @@ #define ATTRIB_H -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL +#if FEATURE_attrib_stack extern void GLAPIENTRY _mesa_PushAttrib( GLbitfield mask ); @@ -43,18 +43,34 @@ _mesa_PushClientAttrib( GLbitfield mask ); extern void GLAPIENTRY _mesa_PopClientAttrib( void ); +extern void +_mesa_init_attrib_dispatch(struct _glapi_table *disp); + +#else /* FEATURE_attrib_stack */ + +static INLINE void +_mesa_PushClientAttrib( GLbitfield mask ) +{ + ASSERT_NO_FEATURE(); +} + +static INLINE void +_mesa_PopClientAttrib( void ) +{ + ASSERT_NO_FEATURE(); +} + +static INLINE void +_mesa_init_attrib_dispatch(struct _glapi_table *disp) +{ +} + +#endif /* FEATURE_attrib_stack */ + extern void _mesa_init_attrib( GLcontext *ctx ); extern void _mesa_free_attrib_data( GLcontext *ctx ); -#else - -/** No-op */ -#define _mesa_init_attrib( c ) ((void)0) -#define _mesa_free_attrib_data( c ) ((void)0) - -#endif - -#endif +#endif /* ATTRIB_H */ diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c index b95e00af5bb..189b5e16558 100644 --- a/src/mesa/main/bufferobj.c +++ b/src/mesa/main/bufferobj.c @@ -1173,7 +1173,7 @@ _mesa_BufferDataARB(GLenum target, GLsizeiptrARB size, ASSERT(ctx->Driver.BufferData); if (!ctx->Driver.BufferData( ctx, target, size, data, usage, bufObj )) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glBufferDataARB(access)"); + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glBufferDataARB()"); } } @@ -1262,7 +1262,7 @@ _mesa_MapBufferARB(GLenum target, GLenum access) ASSERT(ctx->Driver.MapBuffer); map = ctx->Driver.MapBuffer( ctx, target, access, bufObj ); if (!map) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(access)"); + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(map failed)"); return NULL; } else { @@ -1593,7 +1593,7 @@ _mesa_MapBufferRange(GLenum target, GLintptr offset, GLsizeiptr length, map = ctx->Driver.MapBufferRange(ctx, target, offset, length, access, bufObj); if (!map) { - _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(access)"); + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(map failed)"); } else { /* The driver callback should have set all these fields. diff --git a/src/mesa/main/colortab.c b/src/mesa/main/colortab.c index 5447eb18ef9..5ede76c1fb1 100644 --- a/src/mesa/main/colortab.c +++ b/src/mesa/main/colortab.c @@ -32,6 +32,10 @@ #include "state.h" #include "teximage.h" #include "texstate.h" +#include "glapi/dispatch.h" + + +#if FEATURE_colortable /** @@ -536,7 +540,7 @@ _mesa_ColorSubTable( GLenum target, GLsizei start, -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyColorTable(GLenum target, GLenum internalformat, GLint x, GLint y, GLsizei width) { @@ -552,7 +556,7 @@ _mesa_CopyColorTable(GLenum target, GLenum internalformat, -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyColorSubTable(GLenum target, GLsizei start, GLint x, GLint y, GLsizei width) { @@ -568,7 +572,7 @@ _mesa_CopyColorSubTable(GLenum target, GLsizei start, -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetColorTable( GLenum target, GLenum format, GLenum type, GLvoid *data ) { @@ -702,7 +706,7 @@ _mesa_GetColorTable( GLenum target, GLenum format, -void GLAPIENTRY +static void GLAPIENTRY _mesa_ColorTableParameterfv(GLenum target, GLenum pname, const GLfloat *params) { GLfloat *scale, *bias; @@ -747,7 +751,7 @@ _mesa_ColorTableParameterfv(GLenum target, GLenum pname, const GLfloat *params) -void GLAPIENTRY +static void GLAPIENTRY _mesa_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params) { GLfloat fparams[4]; @@ -770,7 +774,7 @@ _mesa_ColorTableParameteriv(GLenum target, GLenum pname, const GLint *params) -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params ) { GET_CURRENT_CONTEXT(ctx); @@ -897,7 +901,7 @@ _mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params ) -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params ) { GET_CURRENT_CONTEXT(ctx); @@ -1052,6 +1056,25 @@ _mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params ) } } + +void +_mesa_init_colortable_dispatch(struct _glapi_table *disp) +{ + SET_ColorSubTable(disp, _mesa_ColorSubTable); + SET_ColorTable(disp, _mesa_ColorTable); + SET_ColorTableParameterfv(disp, _mesa_ColorTableParameterfv); + SET_ColorTableParameteriv(disp, _mesa_ColorTableParameteriv); + SET_CopyColorSubTable(disp, _mesa_CopyColorSubTable); + SET_CopyColorTable(disp, _mesa_CopyColorTable); + SET_GetColorTable(disp, _mesa_GetColorTable); + SET_GetColorTableParameterfv(disp, _mesa_GetColorTableParameterfv); + SET_GetColorTableParameteriv(disp, _mesa_GetColorTableParameteriv); +} + + +#endif /* FEATURE_colortable */ + + /**********************************************************************/ /***** Initialization *****/ /**********************************************************************/ diff --git a/src/mesa/main/colortab.h b/src/mesa/main/colortab.h index b6ff737a65f..652fb582463 100644 --- a/src/mesa/main/colortab.h +++ b/src/mesa/main/colortab.h @@ -27,9 +27,16 @@ #define COLORTAB_H -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL +#if FEATURE_colortable + +#define _MESA_INIT_COLORTABLE_FUNCTIONS(driver, impl) \ + do { \ + (driver)->CopyColorTable = impl ## CopyColorTable; \ + (driver)->CopyColorSubTable = impl ## CopyColorSubTable; \ + (driver)->UpdateTexturePalette = impl ## UpdateTexturePalette; \ + } while (0) extern void GLAPIENTRY _mesa_ColorTable( GLenum target, GLenum internalformat, @@ -41,32 +48,35 @@ _mesa_ColorSubTable( GLenum target, GLsizei start, GLsizei count, GLenum format, GLenum type, const GLvoid *table ); -extern void GLAPIENTRY -_mesa_CopyColorSubTable(GLenum target, GLsizei start, - GLint x, GLint y, GLsizei width); - -extern void GLAPIENTRY -_mesa_CopyColorTable(GLenum target, GLenum internalformat, - GLint x, GLint y, GLsizei width); +extern void +_mesa_init_colortable_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_GetColorTable( GLenum target, GLenum format, - GLenum type, GLvoid *table ); +#else /* FEATURE_colortable */ -extern void GLAPIENTRY -_mesa_ColorTableParameterfv(GLenum target, GLenum pname, - const GLfloat *params); +#define _MESA_INIT_COLORTABLE_FUNCTIONS(driver, impl) do { } while (0) -extern void GLAPIENTRY -_mesa_ColorTableParameteriv(GLenum target, GLenum pname, - const GLint *params); +static INLINE void GLAPIENTRY +_mesa_ColorTable( GLenum target, GLenum internalformat, + GLsizei width, GLenum format, GLenum type, + const GLvoid *table ) +{ + ASSERT_NO_FEATURE(); +} -extern void GLAPIENTRY -_mesa_GetColorTableParameterfv( GLenum target, GLenum pname, GLfloat *params ); +static INLINE void GLAPIENTRY +_mesa_ColorSubTable( GLenum target, GLsizei start, + GLsizei count, GLenum format, GLenum type, + const GLvoid *table ) +{ + ASSERT_NO_FEATURE(); +} -extern void GLAPIENTRY -_mesa_GetColorTableParameteriv( GLenum target, GLenum pname, GLint *params ); +static INLINE void +_mesa_init_colortable_dispatch(struct _glapi_table *disp) +{ +} +#endif /* FEATURE_colortable */ extern void @@ -81,20 +91,5 @@ _mesa_init_colortables( GLcontext *ctx ); extern void _mesa_free_colortables_data( GLcontext *ctx ); -#else - -/** No-op */ -#define _mesa_init_colortable( p ) ((void) 0) - -/** No-op */ -#define _mesa_free_colortable_data( p ) ((void) 0) - -/** No-op */ -#define _mesa_init_colortables( p ) ((void)0) - -/** No-op */ -#define _mesa_free_colortables_data( p ) ((void)0) - -#endif -#endif +#endif /* COLORTAB_H */ diff --git a/src/mesa/main/compiler.h b/src/mesa/main/compiler.h index 9319505a75d..380663ec971 100644 --- a/src/mesa/main/compiler.h +++ b/src/mesa/main/compiler.h @@ -107,8 +107,7 @@ extern "C" { /** * finite macro. */ -#if defined(_WIN32) && !defined(__WIN32__) && !defined(__CYGWIN__) && !defined(BUILD_FOR_SNAP) -# define __WIN32__ +#if defined(_MSC_VER) # define finite _finite #elif defined(__WATCOMC__) # define finite _finite diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c index f6d4ac45957..11b1d244538 100644 --- a/src/mesa/main/context.c +++ b/src/mesa/main/context.c @@ -79,42 +79,28 @@ #include "glheader.h" #include "mfeatures.h" #include "imports.h" -#if FEATURE_accum #include "accum.h" -#endif #include "api_exec.h" #include "arrayobj.h" -#if FEATURE_attrib_stack #include "attrib.h" -#endif #include "blend.h" #include "buffers.h" #include "bufferobj.h" -#if FEATURE_colortable #include "colortab.h" -#endif #include "context.h" #include "cpuinfo.h" #include "debug.h" #include "depth.h" -#if FEATURE_dlist #include "dlist.h" -#endif -#if FEATURE_evaluators #include "eval.h" -#endif #include "enums.h" #include "extensions.h" #include "fbobject.h" -#if FEATURE_feedback #include "feedback.h" -#endif #include "fog.h" #include "framebuffer.h" #include "get.h" -#if FEATURE_histogram #include "histogram.h" -#endif #include "hint.h" #include "hash.h" #include "light.h" @@ -126,15 +112,11 @@ #include "pixelstore.h" #include "points.h" #include "polygon.h" -#if FEATURE_ARB_occlusion_query #include "queryobj.h" -#endif #if FEATURE_ARB_sync #include "syncobj.h" #endif -#if FEATURE_drawpix #include "rastpos.h" -#endif #include "scissor.h" #include "shared.h" #include "simple_list.h" @@ -678,36 +660,20 @@ init_attrib_groups(GLcontext *ctx) _mesa_init_extensions( ctx ); /* Attribute Groups */ -#if FEATURE_accum _mesa_init_accum( ctx ); -#endif -#if FEATURE_attrib_stack _mesa_init_attrib( ctx ); -#endif _mesa_init_buffer_objects( ctx ); _mesa_init_color( ctx ); -#if FEATURE_colortable _mesa_init_colortables( ctx ); -#endif _mesa_init_current( ctx ); _mesa_init_depth( ctx ); _mesa_init_debug( ctx ); -#if FEATURE_dlist _mesa_init_display_list( ctx ); -#endif -#if FEATURE_evaluators _mesa_init_eval( ctx ); -#endif _mesa_init_fbobjects( ctx ); -#if FEATURE_feedback _mesa_init_feedback( ctx ); -#else - ctx->RenderMode = GL_RENDER; -#endif _mesa_init_fog( ctx ); -#if FEATURE_histogram _mesa_init_histogram( ctx ); -#endif _mesa_init_hint( ctx ); _mesa_init_line( ctx ); _mesa_init_lighting( ctx ); @@ -718,15 +684,11 @@ init_attrib_groups(GLcontext *ctx) _mesa_init_point( ctx ); _mesa_init_polygon( ctx ); _mesa_init_program( ctx ); -#if FEATURE_ARB_occlusion_query - _mesa_init_query( ctx ); -#endif + _mesa_init_queryobj( ctx ); #if FEATURE_ARB_sync _mesa_init_sync( ctx ); #endif -#if FEATURE_drawpix _mesa_init_rastpos( ctx ); -#endif _mesa_init_scissor( ctx ); _mesa_init_shader_state( ctx ); _mesa_init_stencil( ctx ); @@ -903,10 +865,12 @@ _mesa_initialize_context(GLcontext *ctx, _mesa_init_exec_table(ctx->Exec); #endif ctx->CurrentDispatch = ctx->Exec; + #if FEATURE_dlist - _mesa_init_dlist_table(ctx->Save); + _mesa_init_save_table(ctx->Save); _mesa_install_save_vtxfmt( ctx, &ctx->ListState.ListVtxfmt ); #endif + /* Neutral tnl module stuff */ _mesa_init_exec_vtxfmt( ctx ); ctx->TnlModule.Current = NULL; @@ -1005,24 +969,16 @@ _mesa_free_context_data( GLcontext *ctx ) _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, NULL); _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._TexEnvProgram, NULL); -#if FEATURE_attrib_stack _mesa_free_attrib_data(ctx); -#endif _mesa_free_lighting_data( ctx ); -#if FEATURE_evaluators _mesa_free_eval_data( ctx ); -#endif _mesa_free_texture_data( ctx ); _mesa_free_matrix_data( ctx ); _mesa_free_viewport_data( ctx ); -#if FEATURE_colortable _mesa_free_colortables_data( ctx ); -#endif _mesa_free_program_data(ctx); _mesa_free_shader_state(ctx); -#if FEATURE_ARB_occlusion_query - _mesa_free_query_data(ctx); -#endif + _mesa_free_queryobj_data(ctx); #if FEATURE_ARB_sync _mesa_free_sync_data(ctx); #endif diff --git a/src/mesa/main/convolve.c b/src/mesa/main/convolve.c index bf6f6619d4f..8db3e79d384 100644 --- a/src/mesa/main/convolve.c +++ b/src/mesa/main/convolve.c @@ -40,6 +40,10 @@ #include "mtypes.h" #include "pixel.h" #include "state.h" +#include "glapi/dispatch.h" + + +#if FEATURE_convolve /* @@ -256,7 +260,7 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalFormat, GLsizei width, G } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param) { GET_CURRENT_CONTEXT(ctx); @@ -299,7 +303,7 @@ _mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat param) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ConvolutionParameterfv(GLenum target, GLenum pname, const GLfloat *params) { GET_CURRENT_CONTEXT(ctx); @@ -351,7 +355,7 @@ _mesa_ConvolutionParameterfv(GLenum target, GLenum pname, const GLfloat *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint param) { GET_CURRENT_CONTEXT(ctx); @@ -394,7 +398,7 @@ _mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint param) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params) { GET_CURRENT_CONTEXT(ctx); @@ -459,7 +463,7 @@ _mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalFormat, GLint x, GLint y, GLsizei width) { GLint baseFormat; @@ -491,7 +495,7 @@ _mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalFormat, GLint x, GLi } -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalFormat, GLint x, GLint y, GLsizei width, GLsizei height) { GLint baseFormat; @@ -527,7 +531,7 @@ _mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalFormat, GLint x, GLi } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type, GLvoid *image) { @@ -586,7 +590,7 @@ _mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params) { GET_CURRENT_CONTEXT(ctx); @@ -647,7 +651,7 @@ _mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params) { GET_CURRENT_CONTEXT(ctx); @@ -717,7 +721,7 @@ _mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type, GLvoid *row, GLvoid *column, GLvoid *span) { @@ -784,7 +788,7 @@ _mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_SeparableFilter2D(GLenum target, GLenum internalFormat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *row, const GLvoid *column) { const GLint colStart = MAX_CONVOLUTION_WIDTH * 4; @@ -1433,3 +1437,25 @@ _mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions, *height = *height - (MAX2(ctx->Separable2D.Height, 1) - 1); } } + + +void +_mesa_init_convolve_dispatch(struct _glapi_table *disp) +{ + SET_ConvolutionFilter1D(disp, _mesa_ConvolutionFilter1D); + SET_ConvolutionFilter2D(disp, _mesa_ConvolutionFilter2D); + SET_ConvolutionParameterf(disp, _mesa_ConvolutionParameterf); + SET_ConvolutionParameterfv(disp, _mesa_ConvolutionParameterfv); + SET_ConvolutionParameteri(disp, _mesa_ConvolutionParameteri); + SET_ConvolutionParameteriv(disp, _mesa_ConvolutionParameteriv); + SET_CopyConvolutionFilter1D(disp, _mesa_CopyConvolutionFilter1D); + SET_CopyConvolutionFilter2D(disp, _mesa_CopyConvolutionFilter2D); + SET_GetConvolutionFilter(disp, _mesa_GetConvolutionFilter); + SET_GetConvolutionParameterfv(disp, _mesa_GetConvolutionParameterfv); + SET_GetConvolutionParameteriv(disp, _mesa_GetConvolutionParameteriv); + SET_SeparableFilter2D(disp, _mesa_SeparableFilter2D); + SET_GetSeparableFilter(disp, _mesa_GetSeparableFilter); +} + + +#endif /* FEATURE_convolve */ diff --git a/src/mesa/main/convolve.h b/src/mesa/main/convolve.h index 4505cdae01e..59492bc7c54 100644 --- a/src/mesa/main/convolve.h +++ b/src/mesa/main/convolve.h @@ -28,10 +28,17 @@ #define CONVOLVE_H -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL +#if FEATURE_convolve + +#define _MESA_INIT_CONVOLVE_FUNCTIONS(driver, impl) \ + do { \ + (driver)->CopyConvolutionFilter1D = impl ## CopyConvolutionFilter1D; \ + (driver)->CopyConvolutionFilter2D = impl ## CopyConvolutionFilter2D; \ + } while (0) + extern void GLAPIENTRY _mesa_ConvolutionFilter1D(GLenum target, GLenum internalformat, GLsizei width, GLenum format, GLenum type, const GLvoid *image); @@ -41,74 +48,79 @@ _mesa_ConvolutionFilter2D(GLenum target, GLenum internalformat, GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *image); -extern void GLAPIENTRY -_mesa_ConvolutionParameterf(GLenum target, GLenum pname, GLfloat params); - -extern void GLAPIENTRY -_mesa_ConvolutionParameterfv(GLenum target, GLenum pname, - const GLfloat *params); - -extern void GLAPIENTRY -_mesa_ConvolutionParameteri(GLenum target, GLenum pname, GLint params); - -extern void GLAPIENTRY -_mesa_ConvolutionParameteriv(GLenum target, GLenum pname, const GLint *params); - -extern void GLAPIENTRY -_mesa_CopyConvolutionFilter1D(GLenum target, GLenum internalformat, - GLint x, GLint y, GLsizei width); +extern void +_mesa_convolve_1d_image(const GLcontext *ctx, GLsizei *width, + const GLfloat *srcImage, GLfloat *dstImage); -extern void GLAPIENTRY -_mesa_CopyConvolutionFilter2D(GLenum target, GLenum internalformat, - GLint x, GLint y, GLsizei width, GLsizei height); +extern void +_mesa_convolve_2d_image(const GLcontext *ctx, GLsizei *width, GLsizei *height, + const GLfloat *srcImage, GLfloat *dstImage); -extern void GLAPIENTRY -_mesa_GetConvolutionFilter(GLenum target, GLenum format, GLenum type, - GLvoid *image); +extern void +_mesa_convolve_sep_image(const GLcontext *ctx, + GLsizei *width, GLsizei *height, + const GLfloat *srcImage, GLfloat *dstImage); -extern void GLAPIENTRY -_mesa_GetConvolutionParameterfv(GLenum target, GLenum pname, GLfloat *params); +extern void +_mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions, + GLsizei *width, GLsizei *height); -extern void GLAPIENTRY -_mesa_GetConvolutionParameteriv(GLenum target, GLenum pname, GLint *params); +extern void +_mesa_init_convolve_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_GetSeparableFilter(GLenum target, GLenum format, GLenum type, - GLvoid *row, GLvoid *column, GLvoid *span); +#else /* FEATURE_convolve */ -extern void GLAPIENTRY -_mesa_SeparableFilter2D(GLenum target, GLenum internalformat, - GLsizei width, GLsizei height, - GLenum format, GLenum type, - const GLvoid *row, const GLvoid *column); +#define _MESA_INIT_CONVOLVE_FUNCTIONS(driver, impl) do { } while (0) +static INLINE void GLAPIENTRY +_mesa_ConvolutionFilter1D(GLenum target, GLenum internalformat, GLsizei width, + GLenum format, GLenum type, const GLvoid *image) +{ + ASSERT_NO_FEATURE(); +} +static INLINE void GLAPIENTRY +_mesa_ConvolutionFilter2D(GLenum target, GLenum internalformat, GLsizei width, + GLsizei height, GLenum format, GLenum type, + const GLvoid *image) +{ + ASSERT_NO_FEATURE(); +} -extern void +static INLINE void _mesa_convolve_1d_image(const GLcontext *ctx, GLsizei *width, - const GLfloat *srcImage, GLfloat *dstImage); - + const GLfloat *srcImage, GLfloat *dstImage) +{ + ASSERT_NO_FEATURE(); +} -extern void +static INLINE void _mesa_convolve_2d_image(const GLcontext *ctx, GLsizei *width, GLsizei *height, - const GLfloat *srcImage, GLfloat *dstImage); + const GLfloat *srcImage, GLfloat *dstImage) +{ + ASSERT_NO_FEATURE(); +} -extern void +static INLINE void _mesa_convolve_sep_image(const GLcontext *ctx, GLsizei *width, GLsizei *height, - const GLfloat *srcImage, GLfloat *dstImage); - + const GLfloat *srcImage, GLfloat *dstImage) +{ + ASSERT_NO_FEATURE(); +} -extern void +static INLINE void _mesa_adjust_image_for_convolution(const GLcontext *ctx, GLuint dimensions, - GLsizei *width, GLsizei *height); + GLsizei *width, GLsizei *height) +{ +} + +static INLINE void +_mesa_init_convolve_dispatch(struct _glapi_table *disp) +{ +} -#else -#define _mesa_adjust_image_for_convolution(c, d, w, h) ((void)0) -#define _mesa_convolve_1d_image(c,w,s,d) ((void)0) -#define _mesa_convolve_2d_image(c,w,h,s,d) ((void)0) -#define _mesa_convolve_sep_image(c,w,h,s,d) ((void)0) -#endif +#endif /* FEATURE_convolve */ -#endif +#endif /* CONVOLVE_H */ diff --git a/src/mesa/main/dlist.c b/src/mesa/main/dlist.c index 9c25de41871..6354ed74748 100644 --- a/src/mesa/main/dlist.c +++ b/src/mesa/main/dlist.c @@ -73,6 +73,7 @@ #include "texstate.h" #include "mtypes.h" #include "varray.h" +#include "vtxfmt.h" #if FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program #include "shader/arbprogram.h" #include "shader/program.h" @@ -438,6 +439,10 @@ typedef union gl_dlist_node Node; */ static GLuint InstSize[OPCODE_END_OF_LIST + 1]; + +#if FEATURE_dlist + + void mesa_print_display_list(GLuint list); @@ -1047,8 +1052,8 @@ static void invalidate_saved_current_state( GLcontext *ctx ) ctx->Driver.CurrentSavePrimitive = PRIM_UNKNOWN; } -void GLAPIENTRY -_mesa_save_CallList(GLuint list) +static void GLAPIENTRY +save_CallList(GLuint list) { GET_CURRENT_CONTEXT(ctx); Node *n; @@ -1070,8 +1075,8 @@ _mesa_save_CallList(GLuint list) } -void GLAPIENTRY -_mesa_save_CallLists(GLsizei num, GLenum type, const GLvoid * lists) +static void GLAPIENTRY +save_CallLists(GLsizei num, GLenum type, const GLvoid * lists) { GET_CURRENT_CONTEXT(ctx); GLint i; @@ -1876,7 +1881,7 @@ save_Enable(GLenum cap) static void GLAPIENTRY -_mesa_save_EvalMesh1(GLenum mode, GLint i1, GLint i2) +save_EvalMesh1(GLenum mode, GLint i1, GLint i2) { GET_CURRENT_CONTEXT(ctx); Node *n; @@ -1894,7 +1899,7 @@ _mesa_save_EvalMesh1(GLenum mode, GLint i1, GLint i2) static void GLAPIENTRY -_mesa_save_EvalMesh2(GLenum mode, GLint i1, GLint i2, GLint j1, GLint j2) +save_EvalMesh2(GLenum mode, GLint i1, GLint i2, GLint j1, GLint j2) { GET_CURRENT_CONTEXT(ctx); Node *n; @@ -4856,7 +4861,7 @@ save_ProgramStringARB(GLenum target, GLenum format, GLsizei len, #endif /* FEATURE_ARB_vertex_program || FEATURE_ARB_fragment_program */ -#if FEATURE_ARB_occlusion_query +#if FEATURE_queryobj static void GLAPIENTRY save_BeginQueryARB(GLenum target, GLuint id) @@ -4890,7 +4895,7 @@ save_EndQueryARB(GLenum target) } } -#endif /* FEATURE_ARB_occlusion_query */ +#endif /* FEATURE_queryobj */ static void GLAPIENTRY @@ -7155,7 +7160,7 @@ execute_list(GLcontext *ctx, GLuint list) n[6].f)); break; #endif -#if FEATURE_ARB_occlusion_query +#if FEATURE_queryobj case OPCODE_BEGIN_QUERY_ARB: CALL_BeginQueryARB(ctx->Exec, (n[1].e, n[2].ui)); break; @@ -7426,7 +7431,7 @@ execute_list(GLcontext *ctx, GLuint list) /** * Test if a display list number is valid. */ -GLboolean GLAPIENTRY +static GLboolean GLAPIENTRY _mesa_IsList(GLuint list) { GET_CURRENT_CONTEXT(ctx); @@ -7439,7 +7444,7 @@ _mesa_IsList(GLuint list) /** * Delete a sequence of consecutive display lists. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_DeleteLists(GLuint list, GLsizei range) { GET_CURRENT_CONTEXT(ctx); @@ -7461,7 +7466,7 @@ _mesa_DeleteLists(GLuint list, GLsizei range) * Return a display list number, n, such that lists n through n+range-1 * are free. */ -GLuint GLAPIENTRY +static GLuint GLAPIENTRY _mesa_GenLists(GLsizei range) { GET_CURRENT_CONTEXT(ctx); @@ -7501,7 +7506,7 @@ _mesa_GenLists(GLsizei range) /** * Begin a new display list. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_NewList(GLuint name, GLenum mode) { GET_CURRENT_CONTEXT(ctx); @@ -7551,7 +7556,7 @@ _mesa_NewList(GLuint name, GLenum mode) /** * End definition of current display list. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_EndList(void) { GET_CURRENT_CONTEXT(ctx); @@ -7685,7 +7690,7 @@ _mesa_CallLists(GLsizei n, GLenum type, const GLvoid * lists) /** * Set the offset added to list numbers in glCallLists. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_ListBase(GLuint base) { GET_CURRENT_CONTEXT(ctx); @@ -8420,7 +8425,7 @@ exec_MultiModeDrawElementsIBM(const GLenum * mode, * struct. */ void -_mesa_init_dlist_table(struct _glapi_table *table) +_mesa_init_save_table(struct _glapi_table *table) { _mesa_loopback_init_api_table(table); @@ -8429,8 +8434,8 @@ _mesa_init_dlist_table(struct _glapi_table *table) SET_AlphaFunc(table, save_AlphaFunc); SET_Bitmap(table, save_Bitmap); SET_BlendFunc(table, save_BlendFunc); - SET_CallList(table, _mesa_save_CallList); - SET_CallLists(table, _mesa_save_CallLists); + SET_CallList(table, save_CallList); + SET_CallLists(table, save_CallLists); SET_Clear(table, save_Clear); SET_ClearAccum(table, save_ClearAccum); SET_ClearColor(table, save_ClearColor); @@ -8451,8 +8456,8 @@ _mesa_init_dlist_table(struct _glapi_table *table) SET_DrawPixels(table, save_DrawPixels); SET_Enable(table, save_Enable); SET_EndList(table, _mesa_EndList); - SET_EvalMesh1(table, _mesa_save_EvalMesh1); - SET_EvalMesh2(table, _mesa_save_EvalMesh2); + SET_EvalMesh1(table, save_EvalMesh1); + SET_EvalMesh2(table, save_EvalMesh2); SET_Finish(table, exec_Finish); SET_Flush(table, exec_Flush); SET_Fogf(table, save_Fogf); @@ -8924,7 +8929,7 @@ _mesa_init_dlist_table(struct _glapi_table *table) SET_UnmapBufferARB(table, _mesa_UnmapBufferARB); #endif -#if FEATURE_ARB_occlusion_query +#if FEATURE_queryobj SET_BeginQueryARB(table, save_BeginQueryARB); SET_EndQueryARB(table, save_EndQueryARB); SET_GenQueriesARB(table, _mesa_GenQueriesARB); @@ -9291,22 +9296,21 @@ mesa_print_display_list(GLuint list) void _mesa_save_vtxfmt_init(GLvertexformat * vfmt) { - vfmt->ArrayElement = _ae_loopback_array_elt; /* generic helper */ + _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_); + vfmt->Begin = save_Begin; - vfmt->CallList = _mesa_save_CallList; - vfmt->CallLists = _mesa_save_CallLists; + + _MESA_INIT_DLIST_VTXFMT(vfmt, save_); + vfmt->Color3f = save_Color3f; vfmt->Color3fv = save_Color3fv; vfmt->Color4f = save_Color4f; vfmt->Color4fv = save_Color4fv; vfmt->EdgeFlag = save_EdgeFlag; vfmt->End = save_End; - vfmt->EvalCoord1f = save_EvalCoord1f; - vfmt->EvalCoord1fv = save_EvalCoord1fv; - vfmt->EvalCoord2f = save_EvalCoord2f; - vfmt->EvalCoord2fv = save_EvalCoord2fv; - vfmt->EvalPoint1 = save_EvalPoint1; - vfmt->EvalPoint2 = save_EvalPoint2; + + _MESA_INIT_EVAL_VTXFMT(vfmt, save_); + vfmt->FogCoordfEXT = save_FogCoordfEXT; vfmt->FogCoordfvEXT = save_FogCoordfvEXT; vfmt->Indexf = save_Indexf; @@ -9355,8 +9359,6 @@ _mesa_save_vtxfmt_init(GLvertexformat * vfmt) vfmt->VertexAttrib4fARB = save_VertexAttrib4fARB; vfmt->VertexAttrib4fvARB = save_VertexAttrib4fvARB; - vfmt->EvalMesh1 = _mesa_save_EvalMesh1; - vfmt->EvalMesh2 = _mesa_save_EvalMesh2; vfmt->Rectf = save_Rectf; /* The driver is required to implement these as @@ -9377,6 +9379,32 @@ _mesa_save_vtxfmt_init(GLvertexformat * vfmt) } +void +_mesa_install_dlist_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt) +{ + SET_CallList(disp, vfmt->CallList); + SET_CallLists(disp, vfmt->CallLists); +} + + +void _mesa_init_dlist_dispatch(struct _glapi_table *disp) +{ + SET_CallList(disp, _mesa_CallList); + SET_CallLists(disp, _mesa_CallLists); + + SET_DeleteLists(disp, _mesa_DeleteLists); + SET_EndList(disp, _mesa_EndList); + SET_GenLists(disp, _mesa_GenLists); + SET_IsList(disp, _mesa_IsList); + SET_ListBase(disp, _mesa_ListBase); + SET_NewList(disp, _mesa_NewList); +} + + +#endif /* FEATURE_dlist */ + + /** * Initialize display list state for given context. */ @@ -9401,5 +9429,7 @@ _mesa_init_display_list(GLcontext *ctx) /* Display List group */ ctx->List.ListBase = 0; +#if FEATURE_dlist _mesa_save_vtxfmt_init(&ctx->ListState.ListVtxfmt); +#endif } diff --git a/src/mesa/main/dlist.h b/src/mesa/main/dlist.h index ab7ec2c8db3..af36991d5ec 100644 --- a/src/mesa/main/dlist.h +++ b/src/mesa/main/dlist.h @@ -33,41 +33,34 @@ #define DLIST_H -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL +#if FEATURE_dlist -extern void -_mesa_delete_list(GLcontext *ctx, struct gl_display_list *dlist); +#define _MESA_INIT_DLIST_FUNCTIONS(driver, impl) \ + do { \ + (driver)->NewList = impl ## NewList; \ + (driver)->EndList = impl ## EndList; \ + (driver)->BeginCallList = impl ## BeginCallList; \ + (driver)->EndCallList = impl ## EndCallList; \ + (driver)->SaveFlushVertices = impl ## SaveFlushVertices; \ + (driver)->NotifySaveBegin = impl ## NotifyBegin; \ + } while (0) + +#define _MESA_INIT_DLIST_VTXFMT(vfmt, impl) \ + do { \ + (vfmt)->CallList = impl ## CallList; \ + (vfmt)->CallLists = impl ## CallLists; \ + } while (0) extern void GLAPIENTRY _mesa_CallList( GLuint list ); extern void GLAPIENTRY _mesa_CallLists( GLsizei n, GLenum type, const GLvoid *lists ); -extern void GLAPIENTRY _mesa_DeleteLists( GLuint list, GLsizei range ); - -extern void GLAPIENTRY _mesa_EndList( void ); - -extern GLuint GLAPIENTRY _mesa_GenLists( GLsizei range ); - -extern GLboolean GLAPIENTRY _mesa_IsList( GLuint list ); - -extern void GLAPIENTRY _mesa_ListBase( GLuint base ); - -extern void GLAPIENTRY _mesa_NewList( GLuint list, GLenum mode ); - -extern void GLAPIENTRY _mesa_save_CallLists( GLsizei n, GLenum type, const GLvoid *lists ); - -extern void GLAPIENTRY _mesa_save_CallList( GLuint list ); - - - -extern void _mesa_init_dlist_table( struct _glapi_table *table ); extern void _mesa_compile_error( GLcontext *ctx, GLenum error, const char *s ); - extern void *_mesa_alloc_instruction(GLcontext *ctx, GLuint opcode, GLuint sz); extern GLint _mesa_alloc_opcode( GLcontext *ctx, GLuint sz, @@ -75,22 +68,43 @@ extern GLint _mesa_alloc_opcode( GLcontext *ctx, GLuint sz, void (*destroy)( GLcontext *, void * ), void (*print)( GLcontext *, void * ) ); -extern void _mesa_init_display_list( GLcontext * ctx ); +extern void _mesa_delete_list(GLcontext *ctx, struct gl_display_list *dlist); extern void _mesa_save_vtxfmt_init( GLvertexformat *vfmt ); +extern void _mesa_init_save_table( struct _glapi_table *table ); + +extern void _mesa_install_dlist_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt); + +extern void _mesa_init_dlist_dispatch(struct _glapi_table *disp); -#else +#else /* FEATURE_dlist */ -/** No-op */ -#define _mesa_init_dlist_table(t,ts) ((void)0) +#define _MESA_INIT_DLIST_FUNCTIONS(driver, impl) do { } while (0) +#define _MESA_INIT_DLIST_VTXFMT(vfmt, impl) do { } while (0) -/** No-op */ -#define _mesa_init_display_list(c) ((void)0) +static INLINE void +_mesa_delete_list(GLcontext *ctx, struct gl_display_list *dlist) +{ + /* there should be no list to delete */ + ASSERT_NO_FEATURE(); +} -/** No-op */ -#define _mesa_save_vtxfmt_init(v) ((void)0) +static INLINE void +_mesa_install_dlist_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt) +{ +} + +static INLINE void +_mesa_init_dlist_dispatch(struct _glapi_table *disp) +{ +} + +#endif /* FEATURE_dlist */ + +extern void _mesa_init_display_list( GLcontext * ctx ); -#endif -#endif +#endif /* DLIST_H */ diff --git a/src/mesa/main/drawpix.c b/src/mesa/main/drawpix.c index aef6585641e..5d4b53af4cb 100644 --- a/src/mesa/main/drawpix.c +++ b/src/mesa/main/drawpix.c @@ -33,8 +33,11 @@ #include "image.h" #include "readpix.h" #include "state.h" +#include "glapi/dispatch.h" +#if FEATURE_drawpix + /** * If a fragment program is enabled, check that it's valid. @@ -47,12 +50,10 @@ valid_fragment_program(GLcontext *ctx) } -#if _HAVE_FULL_GL - /* * Execute glDrawPixels */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_DrawPixels( GLsizei width, GLsizei height, GLenum format, GLenum type, const GLvoid *pixels ) { @@ -140,7 +141,7 @@ end: } -void GLAPIENTRY +static void GLAPIENTRY _mesa_CopyPixels( GLint srcx, GLint srcy, GLsizei width, GLsizei height, GLenum type ) { @@ -225,11 +226,8 @@ end: _mesa_set_vp_override(ctx, GL_FALSE); } -#endif /* _HAVE_FULL_GL */ - - -void GLAPIENTRY +static void GLAPIENTRY _mesa_Bitmap( GLsizei width, GLsizei height, GLfloat xorig, GLfloat yorig, GLfloat xmove, GLfloat ymove, const GLubyte *bitmap ) @@ -309,3 +307,15 @@ _mesa_Bitmap( GLsizei width, GLsizei height, ctx->Current.RasterPos[0] += xmove; ctx->Current.RasterPos[1] += ymove; } + + +void +_mesa_init_drawpix_dispatch(struct _glapi_table *disp) +{ + SET_Bitmap(disp, _mesa_Bitmap); + SET_CopyPixels(disp, _mesa_CopyPixels); + SET_DrawPixels(disp, _mesa_DrawPixels); +} + + +#endif /* FEATURE_drawpix */ diff --git a/src/mesa/main/drawpix.h b/src/mesa/main/drawpix.h index 6177adad6da..8ffb1a6d884 100644 --- a/src/mesa/main/drawpix.h +++ b/src/mesa/main/drawpix.h @@ -22,28 +22,35 @@ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +#ifndef DRAWPIX_H +#define DRAWPIX_H -#ifndef DRAWPIXELS_H -#define DRAWPIXELS_H +#include "main/mtypes.h" -#include "main/glheader.h" +#if FEATURE_drawpix -extern void GLAPIENTRY -_mesa_DrawPixels( GLsizei width, GLsizei height, - GLenum format, GLenum type, const GLvoid *pixels ); +#define _MESA_INIT_DRAWPIX_FUNCTIONS(driver, impl) \ + do { \ + (driver)->DrawPixels = impl ## DrawPixels; \ + (driver)->CopyPixels = impl ## CopyPixels; \ + (driver)->Bitmap = impl ## Bitmap; \ + } while (0) +extern void +_mesa_init_drawpix_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_CopyPixels( GLint srcx, GLint srcy, GLsizei width, GLsizei height, - GLenum type ); +#else /* FEATURE_drawpix */ +#define _MESA_INIT_DRAWPIX_FUNCTIONS(driver, impl) do { } while (0) -extern void GLAPIENTRY -_mesa_Bitmap( GLsizei width, GLsizei height, - GLfloat xorig, GLfloat yorig, GLfloat xmove, GLfloat ymove, - const GLubyte *bitmap ); +static INLINE void +_mesa_init_drawpix_dispatch(struct _glapi_table *disp) +{ +} +#endif /* FEATURE_drawpix */ -#endif + +#endif /* DRAWPIX_H */ diff --git a/src/mesa/main/enable.c b/src/mesa/main/enable.c index 4c1f46102d1..4383aed6692 100644 --- a/src/mesa/main/enable.c +++ b/src/mesa/main/enable.c @@ -37,6 +37,7 @@ #include "mtypes.h" #include "enums.h" #include "api_arrayelt.h" +#include "texstate.h" @@ -228,12 +229,11 @@ get_texcoord_unit(GLcontext *ctx) static GLboolean enable_texture(GLcontext *ctx, GLboolean state, GLbitfield texBit) { - const GLuint curr = ctx->Texture.CurrentUnit; - struct gl_texture_unit *texUnit = &ctx->Texture.Unit[curr]; + struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx); const GLbitfield newenabled = state ? (texUnit->Enabled | texBit) : (texUnit->Enabled & ~texBit); - if (!ctx->DrawBuffer->Visual.rgbMode || texUnit->Enabled == newenabled) + if (texUnit->Enabled == newenabled) return GL_FALSE; FLUSH_VERTICES(ctx, _NEW_TEXTURE); @@ -935,11 +935,6 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state) /* GL_EXT_depth_bounds_test */ case GL_DEPTH_BOUNDS_TEST_EXT: CHECK_EXTENSION(EXT_depth_bounds_test, cap); - if (state && ctx->DrawBuffer->Visual.depthBits == 0) { - _mesa_warning(ctx, - "glEnable(GL_DEPTH_BOUNDS_TEST_EXT) but no depth buffer"); - return; - } if (ctx->Depth.BoundsTest == state) return; FLUSH_VERTICES(ctx, _NEW_DEPTH); diff --git a/src/mesa/main/eval.c b/src/mesa/main/eval.c index 3f89f9c1eae..95d6e231879 100644 --- a/src/mesa/main/eval.c +++ b/src/mesa/main/eval.c @@ -44,6 +44,10 @@ #include "eval.h" #include "macros.h" #include "mtypes.h" +#include "glapi/dispatch.h" + + +#if FEATURE_evaluators /* @@ -417,7 +421,7 @@ map1(GLenum target, GLfloat u1, GLfloat u2, GLint ustride, -void GLAPIENTRY +static void GLAPIENTRY _mesa_Map1f( GLenum target, GLfloat u1, GLfloat u2, GLint stride, GLint order, const GLfloat *points ) { @@ -425,7 +429,7 @@ _mesa_Map1f( GLenum target, GLfloat u1, GLfloat u2, GLint stride, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_Map1d( GLenum target, GLdouble u1, GLdouble u2, GLint stride, GLint order, const GLdouble *points ) { @@ -516,7 +520,7 @@ map2( GLenum target, GLfloat u1, GLfloat u2, GLint ustride, GLint uorder, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_Map2f( GLenum target, GLfloat u1, GLfloat u2, GLint ustride, GLint uorder, GLfloat v1, GLfloat v2, GLint vstride, GLint vorder, @@ -527,7 +531,7 @@ _mesa_Map2f( GLenum target, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_Map2d( GLenum target, GLdouble u1, GLdouble u2, GLint ustride, GLint uorder, GLdouble v1, GLdouble v2, GLint vstride, GLint vorder, @@ -539,7 +543,7 @@ _mesa_Map2d( GLenum target, -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetMapdv( GLenum target, GLenum query, GLdouble *v ) { GET_CURRENT_CONTEXT(ctx); @@ -604,7 +608,7 @@ _mesa_GetMapdv( GLenum target, GLenum query, GLdouble *v ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetMapfv( GLenum target, GLenum query, GLfloat *v ) { GET_CURRENT_CONTEXT(ctx); @@ -669,7 +673,7 @@ _mesa_GetMapfv( GLenum target, GLenum query, GLfloat *v ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetMapiv( GLenum target, GLenum query, GLint *v ) { GET_CURRENT_CONTEXT(ctx); @@ -735,7 +739,7 @@ _mesa_GetMapiv( GLenum target, GLenum query, GLint *v ) -void GLAPIENTRY +static void GLAPIENTRY _mesa_MapGrid1f( GLint un, GLfloat u1, GLfloat u2 ) { GET_CURRENT_CONTEXT(ctx); @@ -753,14 +757,14 @@ _mesa_MapGrid1f( GLint un, GLfloat u1, GLfloat u2 ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_MapGrid1d( GLint un, GLdouble u1, GLdouble u2 ) { _mesa_MapGrid1f( un, (GLfloat) u1, (GLfloat) u2 ); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_MapGrid2f( GLint un, GLfloat u1, GLfloat u2, GLint vn, GLfloat v1, GLfloat v2 ) { @@ -788,7 +792,7 @@ _mesa_MapGrid2f( GLint un, GLfloat u1, GLfloat u2, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_MapGrid2d( GLint un, GLdouble u1, GLdouble u2, GLint vn, GLdouble v1, GLdouble v2 ) { @@ -797,6 +801,41 @@ _mesa_MapGrid2d( GLint un, GLdouble u1, GLdouble u2, } +void +_mesa_install_eval_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt) +{ + SET_EvalCoord1f(disp, vfmt->EvalCoord1f); + SET_EvalCoord1fv(disp, vfmt->EvalCoord1fv); + SET_EvalCoord2f(disp, vfmt->EvalCoord2f); + SET_EvalCoord2fv(disp, vfmt->EvalCoord2fv); + SET_EvalPoint1(disp, vfmt->EvalPoint1); + SET_EvalPoint2(disp, vfmt->EvalPoint2); + + SET_EvalMesh1(disp, vfmt->EvalMesh1); + SET_EvalMesh2(disp, vfmt->EvalMesh2); +} + + +void +_mesa_init_eval_dispatch(struct _glapi_table *disp) +{ + SET_GetMapdv(disp, _mesa_GetMapdv); + SET_GetMapfv(disp, _mesa_GetMapfv); + SET_GetMapiv(disp, _mesa_GetMapiv); + SET_Map1d(disp, _mesa_Map1d); + SET_Map1f(disp, _mesa_Map1f); + SET_Map2d(disp, _mesa_Map2d); + SET_Map2f(disp, _mesa_Map2f); + SET_MapGrid1d(disp, _mesa_MapGrid1d); + SET_MapGrid1f(disp, _mesa_MapGrid1f); + SET_MapGrid2d(disp, _mesa_MapGrid2d); + SET_MapGrid2f(disp, _mesa_MapGrid2f); +} + + +#endif /* FEATURE_evaluators */ + /**********************************************************************/ /***** Initialization *****/ diff --git a/src/mesa/main/eval.h b/src/mesa/main/eval.h index b3ff0a96f82..ffd1bab76da 100644 --- a/src/mesa/main/eval.h +++ b/src/mesa/main/eval.h @@ -37,13 +37,22 @@ #define EVAL_H -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL -extern void _mesa_init_eval( GLcontext *ctx ); -extern void _mesa_free_eval_data( GLcontext *ctx ); +#if FEATURE_evaluators +#define _MESA_INIT_EVAL_VTXFMT(vfmt, impl) \ + do { \ + (vfmt)->EvalCoord1f = impl ## EvalCoord1f; \ + (vfmt)->EvalCoord1fv = impl ## EvalCoord1fv; \ + (vfmt)->EvalCoord2f = impl ## EvalCoord2f; \ + (vfmt)->EvalCoord2fv = impl ## EvalCoord2fv; \ + (vfmt)->EvalPoint1 = impl ## EvalPoint1; \ + (vfmt)->EvalPoint2 = impl ## EvalPoint2; \ + (vfmt)->EvalMesh1 = impl ## EvalMesh1; \ + (vfmt)->EvalMesh2 = impl ## EvalMesh2; \ + } while (0) extern GLuint _mesa_evaluator_components( GLenum target ); @@ -70,59 +79,32 @@ extern GLfloat *_mesa_copy_map_points2d(GLenum target, GLint vstride, GLint vorder, const GLdouble *points ); +extern void +_mesa_install_eval_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt); +extern void +_mesa_init_eval_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_Map1f( GLenum target, GLfloat u1, GLfloat u2, GLint stride, - GLint order, const GLfloat *points ); - -extern void GLAPIENTRY -_mesa_Map2f( GLenum target, - GLfloat u1, GLfloat u2, GLint ustride, GLint uorder, - GLfloat v1, GLfloat v2, GLint vstride, GLint vorder, - const GLfloat *points ); - -extern void GLAPIENTRY -_mesa_Map1d( GLenum target, GLdouble u1, GLdouble u2, GLint stride, - GLint order, const GLdouble *points ); - -extern void GLAPIENTRY -_mesa_Map2d( GLenum target, - GLdouble u1, GLdouble u2, GLint ustride, GLint uorder, - GLdouble v1, GLdouble v2, GLint vstride, GLint vorder, - const GLdouble *points ); - -extern void GLAPIENTRY -_mesa_MapGrid1f( GLint un, GLfloat u1, GLfloat u2 ); +#else /* FEATURE_evaluators */ -extern void GLAPIENTRY -_mesa_MapGrid1d( GLint un, GLdouble u1, GLdouble u2 ); +#define _MESA_INIT_EVAL_VTXFMT(vfmt, impl) do { } while (0) -extern void GLAPIENTRY -_mesa_MapGrid2f( GLint un, GLfloat u1, GLfloat u2, - GLint vn, GLfloat v1, GLfloat v2 ); +static INLINE void +_mesa_install_eval_vtxfmt(struct _glapi_table *disp, + const GLvertexformat *vfmt) +{ +} -extern void GLAPIENTRY -_mesa_MapGrid2d( GLint un, GLdouble u1, GLdouble u2, - GLint vn, GLdouble v1, GLdouble v2 ); +static INLINE void +_mesa_init_eval_dispatch(struct _glapi_table *disp) +{ +} -extern void GLAPIENTRY -_mesa_GetMapdv( GLenum target, GLenum query, GLdouble *v ); +#endif /* FEATURE_evaluators */ -extern void GLAPIENTRY -_mesa_GetMapfv( GLenum target, GLenum query, GLfloat *v ); - -extern void GLAPIENTRY -_mesa_GetMapiv( GLenum target, GLenum query, GLint *v ); - -#else - -/** No-op */ -#define _mesa_init_eval( c ) ((void)0) - -/** No-op */ -#define _mesa_free_eval_data( c ) ((void)0) +extern void _mesa_init_eval( GLcontext *ctx ); +extern void _mesa_free_eval_data( GLcontext *ctx ); -#endif -#endif +#endif /* EVAL_H */ diff --git a/src/mesa/main/execmem.c b/src/mesa/main/execmem.c index 57c1e117c89..4c6139985fe 100644 --- a/src/mesa/main/execmem.c +++ b/src/mesa/main/execmem.c @@ -80,11 +80,10 @@ init_heap(void) exec_heap = mmInit( 0, EXEC_HEAP_SIZE ); if (!exec_mem) - exec_mem = (unsigned char *) mmap(0, EXEC_HEAP_SIZE, - PROT_EXEC | PROT_READ | PROT_WRITE, - MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); + exec_mem = mmap(NULL, EXEC_HEAP_SIZE, PROT_EXEC | PROT_READ | PROT_WRITE, + MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); - return (exec_mem != NULL); + return (exec_mem != MAP_FAILED); } diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c index 6b11aceb5ce..54cf37c5f48 100644 --- a/src/mesa/main/extensions.c +++ b/src/mesa/main/extensions.c @@ -167,6 +167,7 @@ static const struct { { OFF, "GL_NV_blend_square", F(NV_blend_square) }, { OFF, "GL_NV_depth_clamp", F(ARB_depth_clamp) }, { OFF, "GL_NV_fragment_program", F(NV_fragment_program) }, + { OFF, "GL_NV_fragment_program_option", F(NV_fragment_program_option) }, { ON, "GL_NV_light_max_exponent", F(NV_light_max_exponent) }, { OFF, "GL_NV_point_sprite", F(NV_point_sprite) }, { OFF, "GL_NV_texture_env_combine4", F(NV_texture_env_combine4) }, @@ -214,7 +215,7 @@ _mesa_enable_sw_extensions(GLcontext *ctx) ctx->Extensions.ARB_imaging = GL_TRUE; ctx->Extensions.ARB_map_buffer_range = GL_TRUE; ctx->Extensions.ARB_multitexture = GL_TRUE; -#if FEATURE_ARB_occlusion_query +#if FEATURE_queryobj ctx->Extensions.ARB_occlusion_query = GL_TRUE; #endif ctx->Extensions.ARB_point_sprite = GL_TRUE; @@ -315,6 +316,9 @@ _mesa_enable_sw_extensions(GLcontext *ctx) #if FEATURE_NV_fragment_program ctx->Extensions.NV_fragment_program = GL_TRUE; #endif +#if FEATURE_NV_fragment_program && FEATURE_ARB_fragment_program + ctx->Extensions.NV_fragment_program_option = GL_TRUE; +#endif ctx->Extensions.SGI_color_matrix = GL_TRUE; ctx->Extensions.SGI_color_table = GL_TRUE; ctx->Extensions.SGI_texture_color_table = GL_TRUE; diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c index 13f49da5a78..6e767bb24d1 100644 --- a/src/mesa/main/fbobject.c +++ b/src/mesa/main/fbobject.c @@ -44,6 +44,7 @@ #include "teximage.h" #include "texobj.h" #include "texstore.h" +#include "texstate.h" /** Set this to 1 to help debug FBO incompleteness problems */ @@ -717,12 +718,6 @@ _mesa_BindRenderbufferEXT(GLenum target, GLuint renderbuffer) } FLUSH_CURRENT(ctx, _NEW_BUFFERS); - /* The above doesn't fully flush the drivers in the way that a - * glFlush does, but that is required here: - */ - if (ctx->Driver.Flush) - ctx->Driver.Flush(ctx); - if (renderbuffer) { newRb = _mesa_lookup_renderbuffer(ctx, renderbuffer); @@ -1212,9 +1207,6 @@ _mesa_BindFramebufferEXT(GLenum target, GLuint framebuffer) } FLUSH_CURRENT(ctx, _NEW_BUFFERS); - if (ctx->Driver.Flush) { - ctx->Driver.Flush(ctx); - } if (framebuffer) { /* Binding a user-created framebuffer object */ @@ -1293,11 +1285,6 @@ _mesa_DeleteFramebuffersEXT(GLsizei n, const GLuint *framebuffers) ASSERT_OUTSIDE_BEGIN_END(ctx); FLUSH_CURRENT(ctx, _NEW_BUFFERS); - /* The above doesn't fully flush the drivers in the way that a - * glFlush does, but that is required here: - */ - if (ctx->Driver.Flush) - ctx->Driver.Flush(ctx); for (i = 0; i < n; i++) { if (framebuffers[i] > 0) { @@ -1531,11 +1518,6 @@ framebuffer_texture(GLcontext *ctx, const char *caller, GLenum target, } FLUSH_CURRENT(ctx, _NEW_BUFFERS); - /* The above doesn't fully flush the drivers in the way that a - * glFlush does, but that is required here: - */ - if (ctx->Driver.Flush) - ctx->Driver.Flush(ctx); _glthread_LOCK_MUTEX(fb->Mutex); if (texObj) { @@ -1712,11 +1694,6 @@ _mesa_FramebufferRenderbufferEXT(GLenum target, GLenum attachment, FLUSH_CURRENT(ctx, _NEW_BUFFERS); - /* The above doesn't fully flush the drivers in the way that a - * glFlush does, but that is required here: - */ - if (ctx->Driver.Flush) - ctx->Driver.Flush(ctx); assert(ctx->Driver.FramebufferRenderbuffer); ctx->Driver.FramebufferRenderbuffer(ctx, fb, attachment, rb); @@ -1793,11 +1770,6 @@ _mesa_GetFramebufferAttachmentParameterivEXT(GLenum target, GLenum attachment, } FLUSH_CURRENT(ctx, _NEW_BUFFERS); - /* The above doesn't fully flush the drivers in the way that a - * glFlush does, but that is required here: - */ - if (ctx->Driver.Flush) - ctx->Driver.Flush(ctx); switch (pname) { case GL_FRAMEBUFFER_ATTACHMENT_OBJECT_TYPE_EXT: @@ -1955,7 +1927,7 @@ _mesa_GenerateMipmapEXT(GLenum target) return; } - texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit]; + texUnit = _mesa_get_current_tex_unit(ctx); texObj = _mesa_select_tex_object(ctx, texUnit, target); _mesa_lock_texture(ctx, texObj); diff --git a/src/mesa/main/feedback.c b/src/mesa/main/feedback.c index 818a8045407..fcdbb75fc4f 100644 --- a/src/mesa/main/feedback.c +++ b/src/mesa/main/feedback.c @@ -36,9 +36,10 @@ #include "feedback.h" #include "macros.h" #include "mtypes.h" +#include "glapi/dispatch.h" -#if _HAVE_FULL_GL +#if FEATURE_feedback #define FB_3D 0x01 @@ -49,7 +50,7 @@ -void GLAPIENTRY +static void GLAPIENTRY _mesa_FeedbackBuffer( GLsizei size, GLenum type, GLfloat *buffer ) { GET_CURRENT_CONTEXT(ctx); @@ -103,7 +104,7 @@ _mesa_FeedbackBuffer( GLsizei size, GLenum type, GLfloat *buffer ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PassThrough( GLfloat token ) { GET_CURRENT_CONTEXT(ctx); @@ -153,9 +154,6 @@ _mesa_feedback_vertex(GLcontext *ctx, } -#endif /* _HAVE_FULL_GL */ - - /**********************************************************************/ /** \name Selection */ /*@{*/ @@ -173,7 +171,7 @@ _mesa_feedback_vertex(GLcontext *ctx, * Verifies we're not in selection mode, flushes the vertices and initialize * the fields in __GLcontextRec::Select with the given buffer. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_SelectBuffer( GLsizei size, GLuint *buffer ) { GET_CURRENT_CONTEXT(ctx); @@ -280,7 +278,7 @@ write_hit_record(GLcontext *ctx) * the hit record data in gl_selection. Marks new render mode in * __GLcontextRec::NewState. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_InitNames( void ) { GET_CURRENT_CONTEXT(ctx); @@ -311,7 +309,7 @@ _mesa_InitNames( void ) * * sa __GLcontextRec::Select. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_LoadName( GLuint name ) { GET_CURRENT_CONTEXT(ctx); @@ -350,7 +348,7 @@ _mesa_LoadName( GLuint name ) * * sa __GLcontextRec::Select. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_PushName( GLuint name ) { GET_CURRENT_CONTEXT(ctx); @@ -381,7 +379,7 @@ _mesa_PushName( GLuint name ) * * sa __GLcontextRec::Select. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_PopName( void ) { GET_CURRENT_CONTEXT(ctx); @@ -424,7 +422,7 @@ _mesa_PopName( void ) * __GLcontextRec::RenderMode and notifies the driver via the * dd_function_table::RenderMode callback. */ -GLint GLAPIENTRY +static GLint GLAPIENTRY _mesa_RenderMode( GLenum mode ) { GET_CURRENT_CONTEXT(ctx); @@ -507,6 +505,23 @@ _mesa_RenderMode( GLenum mode ) /*@}*/ +void +_mesa_init_feedback_dispatch(struct _glapi_table *disp) +{ + SET_InitNames(disp, _mesa_InitNames); + SET_FeedbackBuffer(disp, _mesa_FeedbackBuffer); + SET_LoadName(disp, _mesa_LoadName); + SET_PassThrough(disp, _mesa_PassThrough); + SET_PopName(disp, _mesa_PopName); + SET_PushName(disp, _mesa_PushName); + SET_SelectBuffer(disp, _mesa_SelectBuffer); + SET_RenderMode(disp, _mesa_RenderMode); +} + + +#endif /* FEATURE_feedback */ + + /**********************************************************************/ /** \name Initialization */ /*@{*/ diff --git a/src/mesa/main/feedback.h b/src/mesa/main/feedback.h index 72c2acd5edd..7a648f444fc 100644 --- a/src/mesa/main/feedback.h +++ b/src/mesa/main/feedback.h @@ -27,11 +27,15 @@ #define FEEDBACK_H -#include "mtypes.h" +#include "main/mtypes.h" -extern void -_mesa_init_feedback( GLcontext *ctx ); +#if FEATURE_feedback + +#define _MESA_INIT_FEEDBACK_FUNCTIONS(driver, impl) \ + do { \ + (driver)->RenderMode = impl ## RenderMode; \ + } while (0) extern void _mesa_feedback_vertex( GLcontext *ctx, @@ -55,29 +59,47 @@ extern void _mesa_update_hitflag( GLcontext *ctx, GLfloat z ); -extern void GLAPIENTRY -_mesa_PassThrough( GLfloat token ); +extern void +_mesa_init_feedback_dispatch(struct _glapi_table *disp); + +#else /* FEATURE_feedback */ -extern void GLAPIENTRY -_mesa_FeedbackBuffer( GLsizei size, GLenum type, GLfloat *buffer ); +#define _MESA_INIT_FEEDBACK_FUNCTIONS(driver, impl) do { } while (0) -extern void GLAPIENTRY -_mesa_SelectBuffer( GLsizei size, GLuint *buffer ); +static INLINE void +_mesa_feedback_vertex( GLcontext *ctx, + const GLfloat win[4], + const GLfloat color[4], + GLfloat index, + const GLfloat texcoord[4] ) +{ + /* render mode is always GL_RENDER */ + ASSERT_NO_FEATURE(); +} -extern void GLAPIENTRY -_mesa_InitNames( void ); -extern void GLAPIENTRY -_mesa_LoadName( GLuint name ); +static INLINE void +_mesa_feedback_token( GLcontext *ctx, GLfloat token ) +{ + /* render mode is always GL_RENDER */ + ASSERT_NO_FEATURE(); +} -extern void GLAPIENTRY -_mesa_PushName( GLuint name ); +static INLINE void +_mesa_update_hitflag( GLcontext *ctx, GLfloat z ) +{ + /* render mode is always GL_RENDER */ + ASSERT_NO_FEATURE(); +} -extern void GLAPIENTRY -_mesa_PopName( void ); +static INLINE void +_mesa_init_feedback_dispatch(struct _glapi_table *disp) +{ +} -extern GLint GLAPIENTRY -_mesa_RenderMode( GLenum mode ); +#endif /* FEATURE_feedback */ +extern void +_mesa_init_feedback( GLcontext *ctx ); -#endif +#endif /* FEEDBACK_H */ diff --git a/src/mesa/main/histogram.c b/src/mesa/main/histogram.c index ceb0d5a6a86..87816d31329 100644 --- a/src/mesa/main/histogram.c +++ b/src/mesa/main/histogram.c @@ -29,8 +29,11 @@ #include "context.h" #include "image.h" #include "histogram.h" +#include "glapi/dispatch.h" +#if FEATURE_histogram + /* * XXX the packed pixel formats haven't been tested. @@ -614,7 +617,11 @@ base_histogram_format( GLenum format ) */ -void GLAPIENTRY +/* this is defined below */ +static void GLAPIENTRY _mesa_ResetMinmax(GLenum target); + + +static void GLAPIENTRY _mesa_GetMinmax(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values) { GET_CURRENT_CONTEXT(ctx); @@ -677,7 +684,7 @@ _mesa_GetMinmax(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvo } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetHistogram(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values) { GET_CURRENT_CONTEXT(ctx); @@ -737,7 +744,7 @@ _mesa_GetHistogram(GLenum target, GLboolean reset, GLenum format, GLenum type, G } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetHistogramParameterfv(GLenum target, GLenum pname, GLfloat *params) { GET_CURRENT_CONTEXT(ctx); @@ -784,7 +791,7 @@ _mesa_GetHistogramParameterfv(GLenum target, GLenum pname, GLfloat *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetHistogramParameteriv(GLenum target, GLenum pname, GLint *params) { GET_CURRENT_CONTEXT(ctx); @@ -831,7 +838,7 @@ _mesa_GetHistogramParameteriv(GLenum target, GLenum pname, GLint *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetMinmaxParameterfv(GLenum target, GLenum pname, GLfloat *params) { GET_CURRENT_CONTEXT(ctx); @@ -857,7 +864,7 @@ _mesa_GetMinmaxParameterfv(GLenum target, GLenum pname, GLfloat *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetMinmaxParameteriv(GLenum target, GLenum pname, GLint *params) { GET_CURRENT_CONTEXT(ctx); @@ -883,7 +890,7 @@ _mesa_GetMinmaxParameteriv(GLenum target, GLenum pname, GLint *params) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_Histogram(GLenum target, GLsizei width, GLenum internalFormat, GLboolean sink) { GLuint i; @@ -966,7 +973,7 @@ _mesa_Histogram(GLenum target, GLsizei width, GLenum internalFormat, GLboolean s } -void GLAPIENTRY +static void GLAPIENTRY _mesa_Minmax(GLenum target, GLenum internalFormat, GLboolean sink) { GET_CURRENT_CONTEXT(ctx); @@ -994,7 +1001,7 @@ _mesa_Minmax(GLenum target, GLenum internalFormat, GLboolean sink) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ResetHistogram(GLenum target) { GLuint i; @@ -1020,7 +1027,7 @@ _mesa_ResetHistogram(GLenum target) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_ResetMinmax(GLenum target) { GET_CURRENT_CONTEXT(ctx); @@ -1043,6 +1050,24 @@ _mesa_ResetMinmax(GLenum target) } +void +_mesa_init_histogram_dispatch(struct _glapi_table *disp) +{ + SET_GetHistogram(disp, _mesa_GetHistogram); + SET_GetHistogramParameterfv(disp, _mesa_GetHistogramParameterfv); + SET_GetHistogramParameteriv(disp, _mesa_GetHistogramParameteriv); + SET_GetMinmax(disp, _mesa_GetMinmax); + SET_GetMinmaxParameterfv(disp, _mesa_GetMinmaxParameterfv); + SET_GetMinmaxParameteriv(disp, _mesa_GetMinmaxParameteriv); + SET_Histogram(disp, _mesa_Histogram); + SET_Minmax(disp, _mesa_Minmax); + SET_ResetHistogram(disp, _mesa_ResetHistogram); + SET_ResetMinmax(disp, _mesa_ResetMinmax); +} + + +#endif /* FEATURE_histogram */ + /**********************************************************************/ /***** Initialization *****/ diff --git a/src/mesa/main/histogram.h b/src/mesa/main/histogram.h index 367e9b11ba1..dbae1bbd06b 100644 --- a/src/mesa/main/histogram.h +++ b/src/mesa/main/histogram.h @@ -36,48 +36,22 @@ #ifndef HISTOGRAM_H #define HISTOGRAM_H -#include "glheader.h" -#include "mtypes.h" +#include "main/mtypes.h" -#if _HAVE_FULL_GL +#if FEATURE_histogram -extern void GLAPIENTRY -_mesa_GetMinmax(GLenum target, GLboolean reset, GLenum format, GLenum types, GLvoid *values); +extern void +_mesa_init_histogram_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_GetHistogram(GLenum target, GLboolean reset, GLenum format, GLenum type, GLvoid *values); +#else /* FEATURE_histogram */ -extern void GLAPIENTRY -_mesa_GetHistogramParameterfv(GLenum target, GLenum pname, GLfloat *params); +static INLINE void +_mesa_init_histogram_dispatch(struct _glapi_table *disp) +{ +} -extern void GLAPIENTRY -_mesa_GetHistogramParameteriv(GLenum target, GLenum pname, GLint *params); - -extern void GLAPIENTRY -_mesa_GetMinmaxParameterfv(GLenum target, GLenum pname, GLfloat *params); - -extern void GLAPIENTRY -_mesa_GetMinmaxParameteriv(GLenum target, GLenum pname, GLint *params); - -extern void GLAPIENTRY -_mesa_Histogram(GLenum target, GLsizei width, GLenum internalformat, GLboolean sink); - -extern void GLAPIENTRY -_mesa_Minmax(GLenum target, GLenum internalformat, GLboolean sink); - -extern void GLAPIENTRY -_mesa_ResetHistogram(GLenum target); - -extern void GLAPIENTRY -_mesa_ResetMinmax(GLenum target); +#endif /* FEATURE_histogram */ extern void _mesa_init_histogram( GLcontext * ctx ); -#else - -/** No-op */ -#define _mesa_init_histogram( c ) ((void) 0) - -#endif - -#endif +#endif /* HISTOGRAM_H */ diff --git a/src/mesa/main/mfeatures.h b/src/mesa/main/mfeatures.h index e23cdb1f426..c2fb8404b16 100644 --- a/src/mesa/main/mfeatures.h +++ b/src/mesa/main/mfeatures.h @@ -36,12 +36,47 @@ #define _HAVE_FULL_GL 1 #endif +/* assert that a feature is disabled and should never be used */ +#define ASSERT_NO_FEATURE() ASSERT(0) + +/** + * A feature can be anything. But most of them share certain characteristics. + * + * When a feature defines driver entries, they can be initialized by + * _MESA_INIT_<FEATURE>_FUNCTIONS + * + * When a feature defines vtxfmt entries, they can be initialized and + * installed by + * _MESA_INIT_<FEATURE>_VTXFMT + * _mesa_install_<feature>_vtxfmt + * + * When a feature defines dispatch entries, they are initialized by + * _mesa_init_<feature>_dispatch + * + * When a feature has states, they are initialized and freed by + * _mesa_init_<feature> + * _mesa_free_<feature>_data + * + * Except for states, the others compile to no-op when a feature is disabled. + * + * The GLAPIENTRYs and helper functions defined by a feature should also + * compile to no-op when it is disabled. But to save typings and to catch + * bugs, some of them may be unavailable, or compile to ASSERT_NO_FEATURE() + * when the feature is disabled. + * + * A feature following the conventions may be used without knowing if it is + * enabled or not. + */ + #define FEATURE_accum _HAVE_FULL_GL +#define FEATURE_arrayelt _HAVE_FULL_GL #define FEATURE_attrib_stack _HAVE_FULL_GL +/* this disables vtxfmt, api_loopback, and api_noop completely */ +#define FEATURE_beginend _HAVE_FULL_GL #define FEATURE_colortable _HAVE_FULL_GL #define FEATURE_convolve _HAVE_FULL_GL #define FEATURE_dispatch _HAVE_FULL_GL -#define FEATURE_dlist _HAVE_FULL_GL +#define FEATURE_dlist (_HAVE_FULL_GL && FEATURE_arrayelt && FEATURE_beginend) #define FEATURE_draw_read_buffer _HAVE_FULL_GL #define FEATURE_drawpix _HAVE_FULL_GL #define FEATURE_evaluators _HAVE_FULL_GL @@ -50,15 +85,15 @@ #define FEATURE_histogram _HAVE_FULL_GL #define FEATURE_pixel_transfer _HAVE_FULL_GL #define FEATURE_point_size_array 0 +#define FEATURE_queryobj _HAVE_FULL_GL +#define FEATURE_rastpos _HAVE_FULL_GL #define FEATURE_texgen _HAVE_FULL_GL #define FEATURE_texture_fxt1 _HAVE_FULL_GL #define FEATURE_texture_s3tc _HAVE_FULL_GL #define FEATURE_userclip _HAVE_FULL_GL #define FEATURE_vertex_array_byte 0 -#define FEATURE_windowpos _HAVE_FULL_GL #define FEATURE_es2_glsl 0 -#define FEATURE_ARB_occlusion_query _HAVE_FULL_GL #define FEATURE_ARB_fragment_program _HAVE_FULL_GL #define FEATURE_ARB_framebuffer_object _HAVE_FULL_GL #define FEATURE_ARB_map_buffer_range _HAVE_FULL_GL @@ -76,7 +111,6 @@ #define FEATURE_EXT_framebuffer_object _HAVE_FULL_GL #define FEATURE_EXT_pixel_buffer_object _HAVE_FULL_GL #define FEATURE_EXT_texture_sRGB _HAVE_FULL_GL -#define FEATURE_EXT_timer_query _HAVE_FULL_GL #define FEATURE_ATI_fragment_shader _HAVE_FULL_GL #define FEATURE_NV_fence _HAVE_FULL_GL #define FEATURE_NV_fragment_program _HAVE_FULL_GL diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h index ae169fb9db6..d0050646454 100644 --- a/src/mesa/main/mtypes.h +++ b/src/mesa/main/mtypes.h @@ -1834,7 +1834,6 @@ struct gl_vertex_program struct gl_program Base; /**< base class */ GLboolean IsNVProgram; /**< is this a GL_NV_vertex_program program? */ GLboolean IsPositionInvariant; - void *TnlData; /**< should probably use Base.DriverData */ }; @@ -2067,6 +2066,8 @@ struct gl_shader_program #define GLSL_OPT 0x4 /**< Force optimizations (override pragmas) */ #define GLSL_NO_OPT 0x8 /**< Force no optimizations (override pragmas) */ #define GLSL_UNIFORMS 0x10 /**< Print glUniform calls */ +#define GLSL_NOP_VERT 0x20 /**< Force no-op vertex shaders */ +#define GLSL_NOP_FRAG 0x40 /**< Force no-op fragment shaders */ /** @@ -2080,6 +2081,7 @@ struct gl_shader_state GLboolean EmitContReturn; /**< Emit CONT/RET opcodes? */ GLboolean EmitCondCodes; /**< Use condition codes? */ GLboolean EmitComments; /**< Annotated instructions */ + GLboolean EmitNVTempInitialization; /**< 0-fill NV temp registers */ void *MemPool; GLbitfield Flags; /**< Mask of GLSL_x flags */ struct gl_sl_pragmas DefaultPragmas; /**< Default #pragma settings */ @@ -2583,6 +2585,7 @@ struct gl_extensions GLboolean MESA_texture_signed_rgba; GLboolean NV_blend_square; GLboolean NV_fragment_program; + GLboolean NV_fragment_program_option; GLboolean NV_light_max_exponent; GLboolean NV_point_sprite; GLboolean NV_texgen_reflection; diff --git a/src/mesa/main/pixel.c b/src/mesa/main/pixel.c index fcef6dfd422..3820ebd8894 100644 --- a/src/mesa/main/pixel.c +++ b/src/mesa/main/pixel.c @@ -36,13 +36,17 @@ #include "macros.h" #include "pixel.h" #include "mtypes.h" +#include "glapi/dispatch.h" + + +#if FEATURE_pixel_transfer /**********************************************************************/ /***** glPixelZoom *****/ /**********************************************************************/ -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelZoom( GLfloat xfactor, GLfloat yfactor ) { GET_CURRENT_CONTEXT(ctx); @@ -163,7 +167,7 @@ validate_pbo_access(GLcontext *ctx, struct gl_pixelstore_attrib *pack, } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values ) { GET_CURRENT_CONTEXT(ctx); @@ -205,7 +209,7 @@ _mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values ) { GLfloat fvalues[MAX_PIXEL_MAP_TABLE]; @@ -261,7 +265,7 @@ _mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values ) { GLfloat fvalues[MAX_PIXEL_MAP_TABLE]; @@ -317,7 +321,7 @@ _mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetPixelMapfv( GLenum map, GLfloat *values ) { GET_CURRENT_CONTEXT(ctx); @@ -362,7 +366,7 @@ _mesa_GetPixelMapfv( GLenum map, GLfloat *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetPixelMapuiv( GLenum map, GLuint *values ) { GET_CURRENT_CONTEXT(ctx); @@ -406,7 +410,7 @@ _mesa_GetPixelMapuiv( GLenum map, GLuint *values ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetPixelMapusv( GLenum map, GLushort *values ) { GET_CURRENT_CONTEXT(ctx); @@ -468,7 +472,7 @@ _mesa_GetPixelMapusv( GLenum map, GLushort *values ) * Implements glPixelTransfer[fi] whether called immediately or from a * display list. */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelTransferf( GLenum pname, GLfloat param ) { GET_CURRENT_CONTEXT(ctx); @@ -662,7 +666,7 @@ _mesa_PixelTransferf( GLenum pname, GLfloat param ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_PixelTransferi( GLenum pname, GLint param ) { _mesa_PixelTransferf( pname, (GLfloat) param ); @@ -756,6 +760,24 @@ void _mesa_update_pixel( GLcontext *ctx, GLuint new_state ) } +void +_mesa_init_pixel_dispatch(struct _glapi_table *disp) +{ + SET_GetPixelMapfv(disp, _mesa_GetPixelMapfv); + SET_GetPixelMapuiv(disp, _mesa_GetPixelMapuiv); + SET_GetPixelMapusv(disp, _mesa_GetPixelMapusv); + SET_PixelMapfv(disp, _mesa_PixelMapfv); + SET_PixelMapuiv(disp, _mesa_PixelMapuiv); + SET_PixelMapusv(disp, _mesa_PixelMapusv); + SET_PixelTransferf(disp, _mesa_PixelTransferf); + SET_PixelTransferi(disp, _mesa_PixelTransferi); + SET_PixelZoom(disp, _mesa_PixelZoom); +} + + +#endif /* FEATURE_pixel_transfer */ + + /**********************************************************************/ /***** Initialization *****/ /**********************************************************************/ diff --git a/src/mesa/main/pixel.h b/src/mesa/main/pixel.h index cb6c5262a39..f4d3f1efdb0 100644 --- a/src/mesa/main/pixel.h +++ b/src/mesa/main/pixel.h @@ -33,48 +33,35 @@ #define PIXEL_H -#include "mtypes.h" +#include "main/mtypes.h" -/** \name API functions */ -/*@{*/ +#if FEATURE_pixel_transfer -extern void GLAPIENTRY -_mesa_GetPixelMapfv( GLenum map, GLfloat *values ); - -extern void GLAPIENTRY -_mesa_GetPixelMapuiv( GLenum map, GLuint *values ); - -extern void GLAPIENTRY -_mesa_GetPixelMapusv( GLenum map, GLushort *values ); - -extern void GLAPIENTRY -_mesa_PixelMapfv( GLenum map, GLsizei mapsize, const GLfloat *values ); - -extern void GLAPIENTRY -_mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values ); - -extern void GLAPIENTRY -_mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values ); +extern void +_mesa_update_pixel( GLcontext *ctx, GLuint newstate ); -extern void GLAPIENTRY -_mesa_PixelTransferf( GLenum pname, GLfloat param ); +extern void +_mesa_init_pixel_dispatch( struct _glapi_table * disp ); -extern void GLAPIENTRY -_mesa_PixelTransferi( GLenum pname, GLint param ); +#else /* FEATURE_pixel_transfer */ -extern void GLAPIENTRY -_mesa_PixelZoom( GLfloat xfactor, GLfloat yfactor ); +static INLINE void +_mesa_update_pixel(GLcontext *ctx, GLuint newstate) +{ +} -/*@}*/ +static INLINE void +_mesa_init_pixel_dispatch(struct _glapi_table *disp) +{ +} +#endif /* FEATURE_pixel_transfer */ -extern void -_mesa_update_pixel( GLcontext *ctx, GLuint newstate ); extern void _mesa_init_pixel( GLcontext * ctx ); /*@}*/ -#endif +#endif /* PIXEL_H */ diff --git a/src/mesa/main/queryobj.c b/src/mesa/main/queryobj.c index a73c6e05081..f6eb4ee7e1b 100644 --- a/src/mesa/main/queryobj.c +++ b/src/mesa/main/queryobj.c @@ -29,6 +29,10 @@ #include "imports.h" #include "queryobj.h" #include "mtypes.h" +#include "glapi/dispatch.h" + + +#if FEATURE_queryobj /** @@ -216,7 +220,7 @@ _mesa_IsQueryARB(GLuint id) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_BeginQueryARB(GLenum target, GLuint id) { struct gl_query_object *q; @@ -236,7 +240,6 @@ _mesa_BeginQueryARB(GLenum target, GLuint id) return; } break; -#if FEATURE_EXT_timer_query case GL_TIME_ELAPSED_EXT: if (!ctx->Extensions.EXT_timer_query) { _mesa_error(ctx, GL_INVALID_ENUM, "glBeginQueryARB(target)"); @@ -247,7 +250,6 @@ _mesa_BeginQueryARB(GLenum target, GLuint id) return; } break; -#endif default: _mesa_error(ctx, GL_INVALID_ENUM, "glBeginQueryARB(target)"); return; @@ -285,17 +287,15 @@ _mesa_BeginQueryARB(GLenum target, GLuint id) if (target == GL_SAMPLES_PASSED_ARB) { ctx->Query.CurrentOcclusionObject = q; } -#if FEATURE_EXT_timer_query else if (target == GL_TIME_ELAPSED_EXT) { ctx->Query.CurrentTimerObject = q; } -#endif ctx->Driver.BeginQuery(ctx, q); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_EndQueryARB(GLenum target) { struct gl_query_object *q; @@ -313,7 +313,6 @@ _mesa_EndQueryARB(GLenum target) q = ctx->Query.CurrentOcclusionObject; ctx->Query.CurrentOcclusionObject = NULL; break; -#if FEATURE_EXT_timer_query case GL_TIME_ELAPSED_EXT: if (!ctx->Extensions.EXT_timer_query) { _mesa_error(ctx, GL_INVALID_ENUM, "glEndQueryARB(target)"); @@ -322,7 +321,6 @@ _mesa_EndQueryARB(GLenum target) q = ctx->Query.CurrentTimerObject; ctx->Query.CurrentTimerObject = NULL; break; -#endif default: _mesa_error(ctx, GL_INVALID_ENUM, "glEndQueryARB(target)"); return; @@ -354,7 +352,6 @@ _mesa_GetQueryivARB(GLenum target, GLenum pname, GLint *params) } q = ctx->Query.CurrentOcclusionObject; break; -#if FEATURE_EXT_timer_query case GL_TIME_ELAPSED_EXT: if (!ctx->Extensions.EXT_timer_query) { _mesa_error(ctx, GL_INVALID_ENUM, "glEndQueryARB(target)"); @@ -362,7 +359,6 @@ _mesa_GetQueryivARB(GLenum target, GLenum pname, GLint *params) } q = ctx->Query.CurrentTimerObject; break; -#endif default: _mesa_error(ctx, GL_INVALID_ENUM, "glGetQueryivARB(target)"); return; @@ -462,12 +458,10 @@ _mesa_GetQueryObjectuivARB(GLuint id, GLenum pname, GLuint *params) } -#if FEATURE_EXT_timer_query - /** * New with GL_EXT_timer_query */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetQueryObjecti64vEXT(GLuint id, GLenum pname, GLint64EXT *params) { struct gl_query_object *q = NULL; @@ -504,7 +498,7 @@ _mesa_GetQueryObjecti64vEXT(GLuint id, GLenum pname, GLint64EXT *params) /** * New with GL_EXT_timer_query */ -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetQueryObjectui64vEXT(GLuint id, GLenum pname, GLuint64EXT *params) { struct gl_query_object *q = NULL; @@ -537,19 +531,35 @@ _mesa_GetQueryObjectui64vEXT(GLuint id, GLenum pname, GLuint64EXT *params) } } -#endif /* FEATURE_EXT_timer_query */ + +void +_mesa_init_queryobj_dispatch(struct _glapi_table *disp) +{ + SET_GenQueriesARB(disp, _mesa_GenQueriesARB); + SET_DeleteQueriesARB(disp, _mesa_DeleteQueriesARB); + SET_IsQueryARB(disp, _mesa_IsQueryARB); + SET_BeginQueryARB(disp, _mesa_BeginQueryARB); + SET_EndQueryARB(disp, _mesa_EndQueryARB); + SET_GetQueryivARB(disp, _mesa_GetQueryivARB); + SET_GetQueryObjectivARB(disp, _mesa_GetQueryObjectivARB); + SET_GetQueryObjectuivARB(disp, _mesa_GetQueryObjectuivARB); + + SET_GetQueryObjecti64vEXT(disp, _mesa_GetQueryObjecti64vEXT); + SET_GetQueryObjectui64vEXT(disp, _mesa_GetQueryObjectui64vEXT); +} + + +#endif /* FEATURE_queryobj */ /** * Allocate/init the context state related to query objects. */ void -_mesa_init_query(GLcontext *ctx) +_mesa_init_queryobj(GLcontext *ctx) { -#if FEATURE_ARB_occlusion_query ctx->Query.QueryObjects = _mesa_NewHashTable(); ctx->Query.CurrentOcclusionObject = NULL; -#endif } @@ -569,7 +579,7 @@ delete_queryobj_cb(GLuint id, void *data, void *userData) * Free the context state related to query objects. */ void -_mesa_free_query_data(GLcontext *ctx) +_mesa_free_queryobj_data(GLcontext *ctx) { _mesa_HashDeleteAll(ctx->Query.QueryObjects, delete_queryobj_cb, ctx); _mesa_DeleteHashTable(ctx->Query.QueryObjects); diff --git a/src/mesa/main/queryobj.h b/src/mesa/main/queryobj.h index ee775ef959e..6cf3c76d74d 100644 --- a/src/mesa/main/queryobj.h +++ b/src/mesa/main/queryobj.h @@ -23,19 +23,24 @@ */ -#ifndef OCCLUDE_H -#define OCCLUDE_H +#ifndef QUERYOBJ_H +#define QUERYOBJ_H -extern void -_mesa_init_query(GLcontext *ctx); +#include "main/mtypes.h" -extern void -_mesa_free_query_data(GLcontext *ctx); -extern void -_mesa_init_query_object_functions(struct dd_function_table *driver); +#if FEATURE_queryobj +#define _MESA_INIT_QUERYOBJ_FUNCTIONS(driver, impl) \ + do { \ + (driver)->NewQueryObject = impl ## NewQueryObject; \ + (driver)->DeleteQuery = impl ## DeleteQuery; \ + (driver)->BeginQuery = impl ## BeginQuery; \ + (driver)->EndQuery = impl ## EndQuery; \ + (driver)->WaitQuery = impl ## WaitQuery; \ + (driver)->CheckQuery = impl ## CheckQuery; \ + } while (0) extern void GLAPIENTRY _mesa_GenQueriesARB(GLsizei n, GLuint *ids); @@ -47,12 +52,6 @@ extern GLboolean GLAPIENTRY _mesa_IsQueryARB(GLuint id); extern void GLAPIENTRY -_mesa_BeginQueryARB(GLenum target, GLuint id); - -extern void GLAPIENTRY -_mesa_EndQueryARB(GLenum target); - -extern void GLAPIENTRY _mesa_GetQueryivARB(GLenum target, GLenum pname, GLint *params); extern void GLAPIENTRY @@ -61,11 +60,33 @@ _mesa_GetQueryObjectivARB(GLuint id, GLenum pname, GLint *params); extern void GLAPIENTRY _mesa_GetQueryObjectuivARB(GLuint id, GLenum pname, GLuint *params); -extern void GLAPIENTRY -_mesa_GetQueryObjecti64vEXT(GLuint id, GLenum pname, GLint64EXT *params); +extern void +_mesa_init_query_object_functions(struct dd_function_table *driver); -extern void GLAPIENTRY -_mesa_GetQueryObjectui64vEXT(GLuint id, GLenum pname, GLuint64EXT *params); +extern void +_mesa_init_queryobj_dispatch(struct _glapi_table *disp); + +#else /* FEATURE_queryobj */ + +#define _MESA_INIT_QUERYOBJ_FUNCTIONS(driver, impl) do { } while (0) + +static INLINE void +_mesa_init_query_object_functions(struct dd_function_table *driver) +{ +} + +static INLINE void +_mesa_init_queryobj_dispatch(struct _glapi_table *disp) +{ +} + +#endif /* FEATURE_queryobj */ + +extern void +_mesa_init_queryobj(GLcontext *ctx); + +extern void +_mesa_free_queryobj_data(GLcontext *ctx); -#endif /* OCCLUDE_H */ +#endif /* QUERYOBJ_H */ diff --git a/src/mesa/main/rastpos.c b/src/mesa/main/rastpos.c index 9f309d6ab84..703b47ec742 100644 --- a/src/mesa/main/rastpos.c +++ b/src/mesa/main/rastpos.c @@ -34,6 +34,10 @@ #include "macros.h" #include "rastpos.h" #include "state.h" +#include "glapi/dispatch.h" + + +#if FEATURE_rastpos /** @@ -60,147 +64,147 @@ rasterpos(GLfloat x, GLfloat y, GLfloat z, GLfloat w) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos2d(GLdouble x, GLdouble y) { rasterpos((GLfloat)x, (GLfloat)y, (GLfloat)0.0, (GLfloat)1.0); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos2f(GLfloat x, GLfloat y) { rasterpos(x, y, 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos2i(GLint x, GLint y) { rasterpos((GLfloat) x, (GLfloat) y, 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos2s(GLshort x, GLshort y) { rasterpos(x, y, 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos3d(GLdouble x, GLdouble y, GLdouble z) { rasterpos((GLfloat) x, (GLfloat) y, (GLfloat) z, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos3f(GLfloat x, GLfloat y, GLfloat z) { rasterpos(x, y, z, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos3i(GLint x, GLint y, GLint z) { rasterpos((GLfloat) x, (GLfloat) y, (GLfloat) z, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos3s(GLshort x, GLshort y, GLshort z) { rasterpos(x, y, z, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos4d(GLdouble x, GLdouble y, GLdouble z, GLdouble w) { rasterpos((GLfloat) x, (GLfloat) y, (GLfloat) z, (GLfloat) w); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos4f(GLfloat x, GLfloat y, GLfloat z, GLfloat w) { rasterpos(x, y, z, w); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos4i(GLint x, GLint y, GLint z, GLint w) { rasterpos((GLfloat) x, (GLfloat) y, (GLfloat) z, (GLfloat) w); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos4s(GLshort x, GLshort y, GLshort z, GLshort w) { rasterpos(x, y, z, w); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos2dv(const GLdouble *v) { rasterpos((GLfloat) v[0], (GLfloat) v[1], 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos2fv(const GLfloat *v) { rasterpos(v[0], v[1], 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos2iv(const GLint *v) { rasterpos((GLfloat) v[0], (GLfloat) v[1], 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos2sv(const GLshort *v) { rasterpos(v[0], v[1], 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos3dv(const GLdouble *v) { rasterpos((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos3fv(const GLfloat *v) { rasterpos(v[0], v[1], v[2], 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos3iv(const GLint *v) { rasterpos((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos3sv(const GLshort *v) { rasterpos(v[0], v[1], v[2], 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos4dv(const GLdouble *v) { rasterpos((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], (GLfloat) v[3]); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos4fv(const GLfloat *v) { rasterpos(v[0], v[1], v[2], v[3]); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos4iv(const GLint *v) { rasterpos((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], (GLfloat) v[3]); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_RasterPos4sv(const GLshort *v) { rasterpos(v[0], v[1], v[2], v[3]); @@ -211,7 +215,7 @@ _mesa_RasterPos4sv(const GLshort *v) /*** GL_ARB_window_pos / GL_MESA_window_pos ***/ /**********************************************************************/ -#if FEATURE_drawpix + /** * All glWindowPosMESA and glWindowPosARB commands call this function to * update the current raster position. @@ -290,153 +294,152 @@ window_pos4f(GLfloat x, GLfloat y, GLfloat z, GLfloat w) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos2dMESA(GLdouble x, GLdouble y) { window_pos4f((GLfloat) x, (GLfloat) y, 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos2fMESA(GLfloat x, GLfloat y) { window_pos4f(x, y, 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos2iMESA(GLint x, GLint y) { window_pos4f((GLfloat) x, (GLfloat) y, 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos2sMESA(GLshort x, GLshort y) { window_pos4f(x, y, 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos3dMESA(GLdouble x, GLdouble y, GLdouble z) { window_pos4f((GLfloat) x, (GLfloat) y, (GLfloat) z, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos3fMESA(GLfloat x, GLfloat y, GLfloat z) { window_pos4f(x, y, z, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos3iMESA(GLint x, GLint y, GLint z) { window_pos4f((GLfloat) x, (GLfloat) y, (GLfloat) z, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos3sMESA(GLshort x, GLshort y, GLshort z) { window_pos4f(x, y, z, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos4dMESA(GLdouble x, GLdouble y, GLdouble z, GLdouble w) { window_pos4f((GLfloat) x, (GLfloat) y, (GLfloat) z, (GLfloat) w); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos4fMESA(GLfloat x, GLfloat y, GLfloat z, GLfloat w) { window_pos4f(x, y, z, w); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos4iMESA(GLint x, GLint y, GLint z, GLint w) { window_pos4f((GLfloat) x, (GLfloat) y, (GLfloat) z, (GLfloat) w); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos4sMESA(GLshort x, GLshort y, GLshort z, GLshort w) { window_pos4f(x, y, z, w); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos2dvMESA(const GLdouble *v) { window_pos4f((GLfloat) v[0], (GLfloat) v[1], 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos2fvMESA(const GLfloat *v) { window_pos4f(v[0], v[1], 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos2ivMESA(const GLint *v) { window_pos4f((GLfloat) v[0], (GLfloat) v[1], 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos2svMESA(const GLshort *v) { window_pos4f(v[0], v[1], 0.0F, 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos3dvMESA(const GLdouble *v) { window_pos4f((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos3fvMESA(const GLfloat *v) { window_pos4f(v[0], v[1], v[2], 1.0); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos3ivMESA(const GLint *v) { window_pos4f((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos3svMESA(const GLshort *v) { window_pos4f(v[0], v[1], v[2], 1.0F); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos4dvMESA(const GLdouble *v) { window_pos4f((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], (GLfloat) v[3]); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos4fvMESA(const GLfloat *v) { window_pos4f(v[0], v[1], v[2], v[3]); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos4ivMESA(const GLint *v) { window_pos4f((GLfloat) v[0], (GLfloat) v[1], (GLfloat) v[2], (GLfloat) v[3]); } -void GLAPIENTRY +static void GLAPIENTRY _mesa_WindowPos4svMESA(const GLshort *v) { window_pos4f(v[0], v[1], v[2], v[3]); } -#endif #if 0 @@ -477,6 +480,65 @@ void glWindowPos4fMESA( GLfloat x, GLfloat y, GLfloat z, GLfloat w ) #endif +void +_mesa_init_rastpos_dispatch(struct _glapi_table *disp) +{ + SET_RasterPos2f(disp, _mesa_RasterPos2f); + SET_RasterPos2fv(disp, _mesa_RasterPos2fv); + SET_RasterPos2i(disp, _mesa_RasterPos2i); + SET_RasterPos2iv(disp, _mesa_RasterPos2iv); + SET_RasterPos2d(disp, _mesa_RasterPos2d); + SET_RasterPos2dv(disp, _mesa_RasterPos2dv); + SET_RasterPos2s(disp, _mesa_RasterPos2s); + SET_RasterPos2sv(disp, _mesa_RasterPos2sv); + SET_RasterPos3d(disp, _mesa_RasterPos3d); + SET_RasterPos3dv(disp, _mesa_RasterPos3dv); + SET_RasterPos3f(disp, _mesa_RasterPos3f); + SET_RasterPos3fv(disp, _mesa_RasterPos3fv); + SET_RasterPos3i(disp, _mesa_RasterPos3i); + SET_RasterPos3iv(disp, _mesa_RasterPos3iv); + SET_RasterPos3s(disp, _mesa_RasterPos3s); + SET_RasterPos3sv(disp, _mesa_RasterPos3sv); + SET_RasterPos4d(disp, _mesa_RasterPos4d); + SET_RasterPos4dv(disp, _mesa_RasterPos4dv); + SET_RasterPos4f(disp, _mesa_RasterPos4f); + SET_RasterPos4fv(disp, _mesa_RasterPos4fv); + SET_RasterPos4i(disp, _mesa_RasterPos4i); + SET_RasterPos4iv(disp, _mesa_RasterPos4iv); + SET_RasterPos4s(disp, _mesa_RasterPos4s); + SET_RasterPos4sv(disp, _mesa_RasterPos4sv); + + /* 197. GL_MESA_window_pos */ + SET_WindowPos2dMESA(disp, _mesa_WindowPos2dMESA); + SET_WindowPos2dvMESA(disp, _mesa_WindowPos2dvMESA); + SET_WindowPos2fMESA(disp, _mesa_WindowPos2fMESA); + SET_WindowPos2fvMESA(disp, _mesa_WindowPos2fvMESA); + SET_WindowPos2iMESA(disp, _mesa_WindowPos2iMESA); + SET_WindowPos2ivMESA(disp, _mesa_WindowPos2ivMESA); + SET_WindowPos2sMESA(disp, _mesa_WindowPos2sMESA); + SET_WindowPos2svMESA(disp, _mesa_WindowPos2svMESA); + SET_WindowPos3dMESA(disp, _mesa_WindowPos3dMESA); + SET_WindowPos3dvMESA(disp, _mesa_WindowPos3dvMESA); + SET_WindowPos3fMESA(disp, _mesa_WindowPos3fMESA); + SET_WindowPos3fvMESA(disp, _mesa_WindowPos3fvMESA); + SET_WindowPos3iMESA(disp, _mesa_WindowPos3iMESA); + SET_WindowPos3ivMESA(disp, _mesa_WindowPos3ivMESA); + SET_WindowPos3sMESA(disp, _mesa_WindowPos3sMESA); + SET_WindowPos3svMESA(disp, _mesa_WindowPos3svMESA); + SET_WindowPos4dMESA(disp, _mesa_WindowPos4dMESA); + SET_WindowPos4dvMESA(disp, _mesa_WindowPos4dvMESA); + SET_WindowPos4fMESA(disp, _mesa_WindowPos4fMESA); + SET_WindowPos4fvMESA(disp, _mesa_WindowPos4fvMESA); + SET_WindowPos4iMESA(disp, _mesa_WindowPos4iMESA); + SET_WindowPos4ivMESA(disp, _mesa_WindowPos4ivMESA); + SET_WindowPos4sMESA(disp, _mesa_WindowPos4sMESA); + SET_WindowPos4svMESA(disp, _mesa_WindowPos4svMESA); +} + + +#endif /* FEATURE_rastpos */ + + /**********************************************************************/ /** \name Initialization */ /**********************************************************************/ diff --git a/src/mesa/main/rastpos.h b/src/mesa/main/rastpos.h index 363f86ad87b..b2127225b6b 100644 --- a/src/mesa/main/rastpos.h +++ b/src/mesa/main/rastpos.h @@ -32,162 +32,33 @@ #define RASTPOS_H -#include "glheader.h" +#include "main/mtypes.h" -extern void GLAPIENTRY -_mesa_RasterPos2d(GLdouble x, GLdouble y); +#if FEATURE_rastpos -extern void GLAPIENTRY -_mesa_RasterPos2f(GLfloat x, GLfloat y); +#define _MESA_INIT_RASTPOS_FUNCTIONS(driver, impl) \ + do { \ + (driver)->RasterPos = impl ## RasterPos; \ + } while (0) -extern void GLAPIENTRY -_mesa_RasterPos2i(GLint x, GLint y); +extern void +_mesa_init_rastpos_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_RasterPos2s(GLshort x, GLshort y); +#else /* FEATURE_rastpos */ -extern void GLAPIENTRY -_mesa_RasterPos3d(GLdouble x, GLdouble y, GLdouble z); +#define _MESA_INIT_RASTPOS_FUNCTIONS(driver, impl) do { } while (0) -extern void GLAPIENTRY -_mesa_RasterPos3f(GLfloat x, GLfloat y, GLfloat z); +static INLINE void +_mesa_init_rastpos_dispatch(struct _glapi_table *disp) +{ +} -extern void GLAPIENTRY -_mesa_RasterPos3i(GLint x, GLint y, GLint z); - -extern void GLAPIENTRY -_mesa_RasterPos3s(GLshort x, GLshort y, GLshort z); - -extern void GLAPIENTRY -_mesa_RasterPos4d(GLdouble x, GLdouble y, GLdouble z, GLdouble w); - -extern void GLAPIENTRY -_mesa_RasterPos4f(GLfloat x, GLfloat y, GLfloat z, GLfloat w); - -extern void GLAPIENTRY -_mesa_RasterPos4i(GLint x, GLint y, GLint z, GLint w); - -extern void GLAPIENTRY -_mesa_RasterPos4s(GLshort x, GLshort y, GLshort z, GLshort w); - -extern void GLAPIENTRY -_mesa_RasterPos2dv(const GLdouble *v); - -extern void GLAPIENTRY -_mesa_RasterPos2fv(const GLfloat *v); - -extern void GLAPIENTRY -_mesa_RasterPos2iv(const GLint *v); - -extern void GLAPIENTRY -_mesa_RasterPos2sv(const GLshort *v); - -extern void GLAPIENTRY -_mesa_RasterPos3dv(const GLdouble *v); - -extern void GLAPIENTRY -_mesa_RasterPos3fv(const GLfloat *v); - -extern void GLAPIENTRY -_mesa_RasterPos3iv(const GLint *v); - -extern void GLAPIENTRY -_mesa_RasterPos3sv(const GLshort *v); - -extern void GLAPIENTRY -_mesa_RasterPos4dv(const GLdouble *v); - -extern void GLAPIENTRY -_mesa_RasterPos4fv(const GLfloat *v); - -extern void GLAPIENTRY -_mesa_RasterPos4iv(const GLint *v); - -extern void GLAPIENTRY -_mesa_RasterPos4sv(const GLshort *v); - - -/**********************************************************************/ -/** \name GL_MESA_window_pos */ -/**********************************************************************/ -/*@{*/ - -extern void GLAPIENTRY -_mesa_WindowPos2dMESA(GLdouble x, GLdouble y); - -extern void GLAPIENTRY -_mesa_WindowPos2fMESA(GLfloat x, GLfloat y); - -extern void GLAPIENTRY -_mesa_WindowPos2iMESA(GLint x, GLint y); - -extern void GLAPIENTRY -_mesa_WindowPos2sMESA(GLshort x, GLshort y); - -extern void GLAPIENTRY -_mesa_WindowPos3dMESA(GLdouble x, GLdouble y, GLdouble z); - -extern void GLAPIENTRY -_mesa_WindowPos3fMESA(GLfloat x, GLfloat y, GLfloat z); - -extern void GLAPIENTRY -_mesa_WindowPos3iMESA(GLint x, GLint y, GLint z); - -extern void GLAPIENTRY -_mesa_WindowPos3sMESA(GLshort x, GLshort y, GLshort z); - -extern void GLAPIENTRY -_mesa_WindowPos4dMESA(GLdouble x, GLdouble y, GLdouble z, GLdouble w); - -extern void GLAPIENTRY -_mesa_WindowPos4fMESA(GLfloat x, GLfloat y, GLfloat z, GLfloat w); - -extern void GLAPIENTRY -_mesa_WindowPos4iMESA(GLint x, GLint y, GLint z, GLint w); - -extern void GLAPIENTRY -_mesa_WindowPos4sMESA(GLshort x, GLshort y, GLshort z, GLshort w); - -extern void GLAPIENTRY -_mesa_WindowPos2dvMESA(const GLdouble *v); - -extern void GLAPIENTRY -_mesa_WindowPos2fvMESA(const GLfloat *v); - -extern void GLAPIENTRY -_mesa_WindowPos2ivMESA(const GLint *v); - -extern void GLAPIENTRY -_mesa_WindowPos2svMESA(const GLshort *v); - -extern void GLAPIENTRY -_mesa_WindowPos3dvMESA(const GLdouble *v); - -extern void GLAPIENTRY -_mesa_WindowPos3fvMESA(const GLfloat *v); - -extern void GLAPIENTRY -_mesa_WindowPos3ivMESA(const GLint *v); - -extern void GLAPIENTRY -_mesa_WindowPos3svMESA(const GLshort *v); - -extern void GLAPIENTRY -_mesa_WindowPos4dvMESA(const GLdouble *v); - -extern void GLAPIENTRY -_mesa_WindowPos4fvMESA(const GLfloat *v); - -extern void GLAPIENTRY -_mesa_WindowPos4ivMESA(const GLint *v); - -extern void GLAPIENTRY -_mesa_WindowPos4svMESA(const GLshort *v); +#endif /* FEATURE_rastpos */ extern void -_mesa_init_rastpos( GLcontext * ctx ); +_mesa_init_rastpos(GLcontext *ctx); /*@}*/ -#endif +#endif /* RASTPOS_H */ diff --git a/src/mesa/main/shared.c b/src/mesa/main/shared.c index 643ad3354e7..4d01e8abc6e 100644 --- a/src/mesa/main/shared.c +++ b/src/mesa/main/shared.c @@ -37,9 +37,7 @@ #include "shared.h" #include "shader/program.h" #include "shader/shader_api.h" -#if FEATURE_dlist #include "dlist.h" -#endif #if FEATURE_ATI_fragment_shader #include "shader/atifragshader.h" #endif @@ -143,11 +141,9 @@ _mesa_alloc_shared_state(GLcontext *ctx) static void delete_displaylist_cb(GLuint id, void *data, void *userData) { -#if FEATURE_dlist struct gl_display_list *list = (struct gl_display_list *) data; GLcontext *ctx = (GLcontext *) userData; _mesa_delete_list(ctx, list); -#endif } diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c index 140a998df2e..f10e6b04b7c 100644 --- a/src/mesa/main/state.c +++ b/src/mesa/main/state.c @@ -40,9 +40,7 @@ #include "framebuffer.h" #include "light.h" #include "matrix.h" -#if FEATURE_pixel_transfer #include "pixel.h" -#endif #include "shader/program.h" #include "shader/prog_parameter.h" #include "state.h" @@ -585,10 +583,8 @@ _mesa_update_state_locked( GLcontext *ctx ) if (new_state & (_NEW_STENCIL | _NEW_BUFFERS)) _mesa_update_stencil( ctx ); -#if FEATURE_pixel_transfer if (new_state & _MESA_NEW_TRANSFER_STATE) _mesa_update_pixel( ctx, new_state ); -#endif if (new_state & _DD_NEW_SEPARATE_SPECULAR) update_separate_specular( ctx ); diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c index 2f3e47e69ec..d7e77e759e0 100644 --- a/src/mesa/main/texenvprogram.c +++ b/src/mesa/main/texenvprogram.c @@ -276,6 +276,7 @@ need_saturate( GLuint mode ) return GL_TRUE; default: assert(0); + return GL_FALSE; } } diff --git a/src/mesa/main/texgen.c b/src/mesa/main/texgen.c index b3ecfc784e3..733e129fcfe 100644 --- a/src/mesa/main/texgen.c +++ b/src/mesa/main/texgen.c @@ -37,6 +37,10 @@ #include "main/texgen.h" #include "main/texstate.h" #include "math/m_matrix.h" +#include "glapi/dispatch.h" + + +#if FEATURE_texgen /** @@ -162,7 +166,7 @@ _mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_TexGeniv(GLenum coord, GLenum pname, const GLint *params ) { GLfloat p[4]; @@ -179,7 +183,7 @@ _mesa_TexGeniv(GLenum coord, GLenum pname, const GLint *params ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_TexGend(GLenum coord, GLenum pname, GLdouble param ) { GLfloat p = (GLfloat) param; @@ -187,7 +191,7 @@ _mesa_TexGend(GLenum coord, GLenum pname, GLdouble param ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_TexGendv(GLenum coord, GLenum pname, const GLdouble *params ) { GLfloat p[4]; @@ -204,7 +208,7 @@ _mesa_TexGendv(GLenum coord, GLenum pname, const GLdouble *params ) } -void GLAPIENTRY +static void GLAPIENTRY _mesa_TexGenf( GLenum coord, GLenum pname, GLfloat param ) { _mesa_TexGenfv(coord, pname, ¶m); @@ -219,7 +223,7 @@ _mesa_TexGeni( GLenum coord, GLenum pname, GLint param ) -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params ) { struct gl_texture_unit *texUnit; @@ -257,7 +261,7 @@ _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params ) -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params ) { struct gl_texture_unit *texUnit; @@ -295,7 +299,7 @@ _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params ) -void GLAPIENTRY +static void GLAPIENTRY _mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params ) { struct gl_texture_unit *texUnit; @@ -338,3 +342,19 @@ _mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params ) } +void +_mesa_init_texgen_dispatch(struct _glapi_table *disp) +{ + SET_GetTexGendv(disp, _mesa_GetTexGendv); + SET_GetTexGenfv(disp, _mesa_GetTexGenfv); + SET_GetTexGeniv(disp, _mesa_GetTexGeniv); + SET_TexGend(disp, _mesa_TexGend); + SET_TexGendv(disp, _mesa_TexGendv); + SET_TexGenf(disp, _mesa_TexGenf); + SET_TexGenfv(disp, _mesa_TexGenfv); + SET_TexGeni(disp, _mesa_TexGeni); + SET_TexGeniv(disp, _mesa_TexGeniv); +} + + +#endif /* FEATURE_texgen */ diff --git a/src/mesa/main/texgen.h b/src/mesa/main/texgen.h index 073588efcdb..f6924ef7222 100644 --- a/src/mesa/main/texgen.h +++ b/src/mesa/main/texgen.h @@ -27,36 +27,45 @@ #define TEXGEN_H -#include "main/glheader.h" +#include "main/mtypes.h" -extern void GLAPIENTRY -_mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params ); +#if FEATURE_texgen -extern void GLAPIENTRY -_mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params ); +#define _MESA_INIT_TEXGEN_FUNCTIONS(driver, impl) \ + do { \ + (driver)->TexGen = impl ## TexGen; \ + } while (0) extern void GLAPIENTRY -_mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params ); +_mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params ); extern void GLAPIENTRY -_mesa_TexGend( GLenum coord, GLenum pname, GLdouble param ); +_mesa_TexGeni( GLenum coord, GLenum pname, GLint param ); -extern void GLAPIENTRY -_mesa_TexGendv( GLenum coord, GLenum pname, const GLdouble *params ); +extern void +_mesa_init_texgen_dispatch(struct _glapi_table *disp); -extern void GLAPIENTRY -_mesa_TexGenf( GLenum coord, GLenum pname, GLfloat param ); +#else /* FEATURE_texgen */ -extern void GLAPIENTRY -_mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params ); +#define _MESA_INIT_TEXGEN_FUNCTIONS(driver, impl) do { } while (0) -extern void GLAPIENTRY -_mesa_TexGeni( GLenum coord, GLenum pname, GLint param ); +static void +_mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params ) +{ +} -extern void GLAPIENTRY -_mesa_TexGeniv( GLenum coord, GLenum pname, const GLint *params ); +static void INLINE +_mesa_TexGeni( GLenum coord, GLenum pname, GLint param ) +{ +} + +static INLINE void +_mesa_init_texgen_dispatch(struct _glapi_table *disp) +{ +} +#endif /* FEATURE_texgen */ #endif /* TEXGEN_H */ diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c index b0aa04e9aa4..465da6b0469 100644 --- a/src/mesa/main/teximage.c +++ b/src/mesa/main/teximage.c @@ -33,9 +33,8 @@ #include "glheader.h" #include "bufferobj.h" #include "context.h" -#if FEATURE_convolve #include "convolve.h" -#endif +#include "enums.h" #include "fbobject.h" #include "framebuffer.h" #include "hash.h" @@ -2148,6 +2147,13 @@ _mesa_TexImage1D( GLenum target, GLint level, GLint internalFormat, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexImage1D %s %d %s %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), width, border, + _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + internalFormat = override_internal_format(internalFormat, width, 1); #if FEATURE_convolve @@ -2247,6 +2253,13 @@ _mesa_TexImage2D( GLenum target, GLint level, GLint internalFormat, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexImage2D %s %d %s %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), width, height, + border, _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + internalFormat = override_internal_format(internalFormat, width, height); #if FEATURE_convolve @@ -2363,6 +2376,13 @@ _mesa_TexImage3D( GLenum target, GLint level, GLint internalFormat, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexImage3D %s %d %s %d %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), width, height, + depth, border, _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + internalFormat = override_internal_format(internalFormat, width, height); if (target == GL_TEXTURE_3D || @@ -2472,6 +2492,12 @@ _mesa_TexSubImage1D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexSubImage1D %s %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + xoffset, width, _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); @@ -2533,6 +2559,13 @@ _mesa_TexSubImage2D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexSubImage2D %s %d %d %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + xoffset, yoffset, width, height, + _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); @@ -2594,6 +2627,13 @@ _mesa_TexSubImage3D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glTexSubImage3D %s %d %d %d %d %d %d %d %s %s %p\n", + _mesa_lookup_enum_by_nr(target), level, + xoffset, yoffset, zoffset, width, height, depth, + _mesa_lookup_enum_by_nr(format), + _mesa_lookup_enum_by_nr(type), pixels); + if (ctx->NewState & _MESA_NEW_TRANSFER_STATE) _mesa_update_state(ctx); @@ -2652,6 +2692,12 @@ _mesa_CopyTexImage1D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexImage1D %s %d %s %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + x, y, width, border); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2718,6 +2764,12 @@ _mesa_CopyTexImage2D( GLenum target, GLint level, GLenum internalFormat, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexImage2D %s %d %s %d %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + x, y, width, height, border); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2787,6 +2839,11 @@ _mesa_CopyTexSubImage1D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexSubImage1D %s %d %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), + level, xoffset, x, y, width); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2844,6 +2901,11 @@ _mesa_CopyTexSubImage2D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexSubImage2D %s %d %d %d %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), + level, xoffset, yoffset, x, y, width, height); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -2904,6 +2966,11 @@ _mesa_CopyTexSubImage3D( GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCopyTexSubImage3D %s %d %d %d %d %d %d %d %d\n", + _mesa_lookup_enum_by_nr(target), + level, xoffset, yoffset, zoffset, x, y, width, height); + if (ctx->NewState & NEW_COPY_TEX_STATE) _mesa_update_state(ctx); @@ -3155,6 +3222,12 @@ _mesa_CompressedTexImage1DARB(GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCompressedTexImage1DARB %s %d %s %d %d %d %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + width, border, imageSize, data); + if (target == GL_TEXTURE_1D) { /* non-proxy target */ struct gl_texture_unit *texUnit; @@ -3250,6 +3323,12 @@ _mesa_CompressedTexImage2DARB(GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCompressedTexImage2DARB %s %d %s %d %d %d %d %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + width, height, border, imageSize, data); + if (target == GL_TEXTURE_2D || (ctx->Extensions.ARB_texture_cube_map && target >= GL_TEXTURE_CUBE_MAP_POSITIVE_X_ARB && @@ -3350,6 +3429,12 @@ _mesa_CompressedTexImage3DARB(GLenum target, GLint level, GET_CURRENT_CONTEXT(ctx); ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); + if (MESA_VERBOSE & (VERBOSE_API|VERBOSE_TEXTURE)) + _mesa_debug(ctx, "glCompressedTexImage3DARB %s %d %s %d %d %d %d %d %p\n", + _mesa_lookup_enum_by_nr(target), level, + _mesa_lookup_enum_by_nr(internalFormat), + width, height, depth, border, imageSize, data); + if (target == GL_TEXTURE_3D) { /* non-proxy target */ struct gl_texture_unit *texUnit; diff --git a/src/mesa/main/texobj.c b/src/mesa/main/texobj.c index d09c4392506..f8ce811eb29 100644 --- a/src/mesa/main/texobj.c +++ b/src/mesa/main/texobj.c @@ -29,9 +29,7 @@ #include "mfeatures.h" -#if FEATURE_colortable #include "colortab.h" -#endif #include "context.h" #include "enums.h" #include "fbobject.h" @@ -194,9 +192,7 @@ _mesa_delete_texture_object( GLcontext *ctx, struct gl_texture_object *texObj ) */ texObj->Target = 0x99; -#if FEATURE_colortable _mesa_free_colortable_data(&texObj->Palette); -#endif /* free the texture images */ for (face = 0; face < 6; face++) { @@ -495,7 +491,7 @@ _mesa_test_texobj_completeness( const GLcontext *ctx, t->Image[face][baseLevel]->Width2 != w || t->Image[face][baseLevel]->Height2 != h) { t->_Complete = GL_FALSE; - incomplete(t, "Non-quare cubemap image"); + incomplete(t, "Cube face missing or mismatched size"); return; } } diff --git a/src/mesa/main/texrender.c b/src/mesa/main/texrender.c index cc74d58fbd1..53be83b05ce 100644 --- a/src/mesa/main/texrender.c +++ b/src/mesa/main/texrender.c @@ -398,6 +398,14 @@ texture_put_mono_values(GLcontext *ctx, struct gl_renderbuffer *rb, static void +store_nop(struct gl_texture_image *texImage, + GLint col, GLint row, GLint img, + const void *texel) +{ +} + + +static void delete_texture_wrapper(struct gl_renderbuffer *rb) { ASSERT(rb->RefCount == 0); @@ -462,7 +470,10 @@ update_wrapper(GLcontext *ctx, const struct gl_renderbuffer_attachment *att) ASSERT(trb->TexImage); trb->Store = trb->TexImage->TexFormat->StoreTexel; - ASSERT(trb->Store); + if (!trb->Store) { + /* we'll never draw into some textures (compressed formats) */ + trb->Store = store_nop; + } if (att->Texture->Target == GL_TEXTURE_1D_ARRAY_EXT) { trb->Yoffset = att->Zoffset; diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c index 861c5f37c4e..43f26873e0f 100644 --- a/src/mesa/main/texstate.c +++ b/src/mesa/main/texstate.c @@ -31,9 +31,7 @@ #include "glheader.h" #include "mfeatures.h" #include "colormac.h" -#if FEATURE_colortable #include "colortab.h" -#endif #include "context.h" #include "enums.h" #include "macros.h" @@ -99,16 +97,22 @@ _mesa_copy_texture_state( const GLcontext *src, GLcontext *dst ) dst->Texture.Unit[u].BumpTarget = src->Texture.Unit[u].BumpTarget; COPY_4V(dst->Texture.Unit[u].RotMatrix, src->Texture.Unit[u].RotMatrix); + /* + * XXX strictly speaking, we should compare texture names/ids and + * bind textures in the dest context according to id. For now, only + * copy bindings if the contexts share the same pool of textures to + * avoid refcounting bugs. + */ + if (dst->Shared == src->Shared) { + /* copy texture object bindings, not contents of texture objects */ + _mesa_lock_context_textures(dst); - /* copy texture object bindings, not contents of texture objects */ - _mesa_lock_context_textures(dst); - - for (tex = 0; tex < NUM_TEXTURE_TARGETS; tex++) { - _mesa_reference_texobj(&dst->Texture.Unit[u].CurrentTex[tex], - src->Texture.Unit[u].CurrentTex[tex]); + for (tex = 0; tex < NUM_TEXTURE_TARGETS; tex++) { + _mesa_reference_texobj(&dst->Texture.Unit[u].CurrentTex[tex], + src->Texture.Unit[u].CurrentTex[tex]); + } + _mesa_unlock_context_textures(dst); } - - _mesa_unlock_context_textures(dst); } } @@ -753,9 +757,7 @@ _mesa_init_texture(GLcontext *ctx) ctx->Texture.CurrentUnit = 0; /* multitexture */ ctx->Texture._EnabledUnits = 0x0; ctx->Texture.SharedPalette = GL_FALSE; -#if FEATURE_colortable _mesa_init_colortable(&ctx->Texture.Palette); -#endif for (u = 0; u < MAX_TEXTURE_UNITS; u++) init_texture_unit(ctx, u); @@ -796,10 +798,8 @@ _mesa_free_texture_data(GLcontext *ctx) for (tgt = 0; tgt < NUM_TEXTURE_TARGETS; tgt++) ctx->Driver.DeleteTexture(ctx, ctx->Texture.ProxyTex[tgt]); -#if FEATURE_colortable for (u = 0; u < MAX_TEXTURE_IMAGE_UNITS; u++) _mesa_free_colortable_data(&ctx->Texture.Unit[u].ColorTable); -#endif } diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c index ab9973b8101..f553a898f9e 100644 --- a/src/mesa/main/texstore.c +++ b/src/mesa/main/texstore.c @@ -56,9 +56,7 @@ #include "bufferobj.h" #include "colormac.h" #include "context.h" -#if FEATURE_convolve #include "convolve.h" -#endif #include "image.h" #include "macros.h" #include "mipmap.h" diff --git a/src/mesa/main/vtxfmt.c b/src/mesa/main/vtxfmt.c index 91412f138a5..c9eea8ab830 100644 --- a/src/mesa/main/vtxfmt.c +++ b/src/mesa/main/vtxfmt.c @@ -27,12 +27,18 @@ */ #include "glheader.h" +#include "api_arrayelt.h" #include "api_loopback.h" #include "context.h" #include "imports.h" #include "mtypes.h" #include "state.h" #include "vtxfmt.h" +#include "eval.h" +#include "dlist.h" + + +#if FEATURE_beginend /* The neutral vertex format. This wraps all tnl module functions, @@ -82,18 +88,16 @@ static void install_vtxfmt( struct _glapi_table *tab, const GLvertexformat *vfmt ) { - SET_ArrayElement(tab, vfmt->ArrayElement); + _mesa_install_arrayelt_vtxfmt(tab, vfmt); + SET_Color3f(tab, vfmt->Color3f); SET_Color3fv(tab, vfmt->Color3fv); SET_Color4f(tab, vfmt->Color4f); SET_Color4fv(tab, vfmt->Color4fv); SET_EdgeFlag(tab, vfmt->EdgeFlag); - SET_EvalCoord1f(tab, vfmt->EvalCoord1f); - SET_EvalCoord1fv(tab, vfmt->EvalCoord1fv); - SET_EvalCoord2f(tab, vfmt->EvalCoord2f); - SET_EvalCoord2fv(tab, vfmt->EvalCoord2fv); - SET_EvalPoint1(tab, vfmt->EvalPoint1); - SET_EvalPoint2(tab, vfmt->EvalPoint2); + + _mesa_install_eval_vtxfmt(tab, vfmt); + SET_FogCoordfEXT(tab, vfmt->FogCoordfEXT); SET_FogCoordfvEXT(tab, vfmt->FogCoordfvEXT); SET_Indexf(tab, vfmt->Indexf); @@ -125,8 +129,9 @@ install_vtxfmt( struct _glapi_table *tab, const GLvertexformat *vfmt ) SET_Vertex3fv(tab, vfmt->Vertex3fv); SET_Vertex4f(tab, vfmt->Vertex4f); SET_Vertex4fv(tab, vfmt->Vertex4fv); - SET_CallList(tab, vfmt->CallList); - SET_CallLists(tab, vfmt->CallLists); + + _mesa_install_dlist_vtxfmt(tab, vfmt); + SET_Begin(tab, vfmt->Begin); SET_End(tab, vfmt->End); SET_Rectf(tab, vfmt->Rectf); @@ -137,9 +142,6 @@ install_vtxfmt( struct _glapi_table *tab, const GLvertexformat *vfmt ) SET_DrawElementsBaseVertex(tab, vfmt->DrawElementsBaseVertex); SET_DrawRangeElementsBaseVertex(tab, vfmt->DrawRangeElementsBaseVertex); SET_MultiDrawElementsBaseVertex(tab, vfmt->MultiDrawElementsBaseVertex); - SET_EvalMesh1(tab, vfmt->EvalMesh1); - SET_EvalMesh2(tab, vfmt->EvalMesh2); - ASSERT(tab->EvalMesh2); /* GL_NV_vertex_program */ SET_VertexAttrib1fNV(tab, vfmt->VertexAttrib1fNV); @@ -196,3 +198,6 @@ void _mesa_restore_exec_vtxfmt( GLcontext *ctx ) tnl->SwapCount = 0; } + + +#endif /* FEATURE_beginend */ diff --git a/src/mesa/main/vtxfmt.h b/src/mesa/main/vtxfmt.h index 76f108e0232..fb6c23abe98 100644 --- a/src/mesa/main/vtxfmt.h +++ b/src/mesa/main/vtxfmt.h @@ -33,6 +33,8 @@ #ifndef _VTXFMT_H_ #define _VTXFMT_H_ +#if FEATURE_beginend + extern void _mesa_init_exec_vtxfmt( GLcontext *ctx ); extern void _mesa_install_exec_vtxfmt( GLcontext *ctx, const GLvertexformat *vfmt ); @@ -40,4 +42,28 @@ extern void _mesa_install_save_vtxfmt( GLcontext *ctx, const GLvertexformat *vfm extern void _mesa_restore_exec_vtxfmt( GLcontext *ctx ); -#endif +#else /* FEATURE_beginend */ + +static INLINE void +_mesa_init_exec_vtxfmt( GLcontext *ctx ) +{ +} + +static INLINE void +_mesa_install_exec_vtxfmt( GLcontext *ctx, const GLvertexformat *vfmt ) +{ +} + +static INLINE void +_mesa_install_save_vtxfmt( GLcontext *ctx, const GLvertexformat *vfmt ) +{ +} + +static INLINE void +_mesa_restore_exec_vtxfmt( GLcontext *ctx ) +{ +} + +#endif /* FEATURE_beginend */ + +#endif /* _VTXFMT_H_ */ diff --git a/src/mesa/shader/lex.yy.c b/src/mesa/shader/lex.yy.c index 283ba8d26e5..3c86913fdfc 100644 --- a/src/mesa/shader/lex.yy.c +++ b/src/mesa/shader/lex.yy.c @@ -53,7 +53,6 @@ typedef int flex_int32_t; typedef unsigned char flex_uint8_t; typedef unsigned short int flex_uint16_t; typedef unsigned int flex_uint32_t; -#endif /* ! C99 */ /* Limits of integral types. */ #ifndef INT8_MIN @@ -84,6 +83,8 @@ typedef unsigned int flex_uint32_t; #define UINT32_MAX (4294967295U) #endif +#endif /* ! C99 */ + #endif /* ! FLEXINT_H */ #ifdef __cplusplus @@ -157,7 +158,15 @@ typedef void* yyscan_t; /* Size of default input buffer. */ #ifndef YY_BUF_SIZE +#ifdef __ia64__ +/* On IA-64, the buffer size is 16k, not 8k. + * Moreover, YY_BUF_SIZE is 2*YY_READ_BUF_SIZE in the general case. + * Ditto for the __ia64__ case accordingly. + */ +#define YY_BUF_SIZE 32768 +#else #define YY_BUF_SIZE 16384 +#endif /* __ia64__ */ #endif /* The state buf must be large enough to hold one state per character in the main buffer. @@ -348,8 +357,8 @@ static void yy_fatal_error (yyconst char msg[] ,yyscan_t yyscanner ); *yy_cp = '\0'; \ yyg->yy_c_buf_p = yy_cp; -#define YY_NUM_RULES 183 -#define YY_END_OF_BUFFER 184 +#define YY_NUM_RULES 170 +#define YY_END_OF_BUFFER 171 /* This struct is not used in this scanner, but its presence is necessary. */ struct yy_trans_info @@ -357,82 +366,101 @@ struct yy_trans_info flex_int32_t yy_verify; flex_int32_t yy_nxt; }; -static yyconst flex_int16_t yy_accept[675] = +static yyconst flex_int16_t yy_accept[850] = { 0, - 0, 0, 184, 182, 180, 179, 182, 182, 152, 178, - 154, 154, 154, 154, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 152, 180, 0, 0, 181, 152, 0, - 153, 155, 175, 175, 0, 0, 0, 0, 175, 0, - 0, 0, 0, 0, 0, 0, 132, 176, 133, 134, - 166, 166, 166, 166, 0, 154, 0, 140, 141, 142, - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 152, 152, 152, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 174, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 173, 173, 0, 0, 0, + 0, 0, 171, 169, 167, 166, 169, 169, 139, 165, + 141, 141, 141, 141, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 167, 0, 0, 168, 139, + 0, 140, 142, 162, 162, 0, 0, 0, 0, 162, + 0, 0, 0, 0, 0, 0, 0, 119, 163, 120, + 121, 153, 153, 153, 153, 0, 141, 0, 127, 128, + 129, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 161, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 160, 160, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 172, 172, 172, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 163, 163, 163, 164, 164, 165, 156, - 155, 156, 0, 157, 11, 13, 152, 15, 152, 152, - 16, 18, 152, 20, 22, 24, 26, 6, 28, 30, - 31, 33, 35, 38, 36, 40, 41, 43, 45, 47, - - 49, 51, 152, 152, 152, 53, 55, 152, 57, 59, - 61, 152, 63, 65, 67, 69, 152, 71, 73, 75, - 77, 152, 152, 152, 152, 152, 152, 0, 0, 0, - 0, 155, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 93, 94, 96, 0, 171, 0, 0, 0, - 0, 0, 0, 110, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 170, 169, 169, 122, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 0, 160, 160, - 161, 162, 0, 158, 152, 152, 152, 152, 152, 152, - 152, 152, 143, 152, 152, 152, 152, 152, 152, 152, - - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 144, 152, 152, 152, 152, 152, 152, - 152, 152, 10, 152, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 0, 177, 0, 0, 0, 86, 87, - 0, 0, 0, 0, 0, 0, 0, 98, 0, 0, + 159, 159, 159, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 150, 150, 150, 151, 151, 152, 143, + 142, 143, 0, 144, 11, 12, 139, 13, 139, 139, + 14, 15, 139, 16, 17, 18, 19, 20, 21, 6, + + 22, 23, 24, 25, 26, 28, 27, 29, 30, 31, + 32, 33, 34, 35, 139, 139, 139, 139, 139, 40, + 41, 139, 42, 43, 44, 45, 46, 47, 48, 139, + 49, 50, 51, 52, 53, 54, 55, 139, 56, 57, + 58, 59, 139, 139, 64, 65, 139, 139, 139, 139, + 139, 139, 0, 0, 0, 0, 142, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 80, 81, 83, + 0, 158, 0, 0, 0, 0, 0, 0, 97, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 157, + 156, 156, 109, 0, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 147, 147, 148, 149, 0, 145, 11, + 11, 139, 12, 12, 12, 139, 139, 139, 139, 139, + 15, 15, 139, 130, 16, 16, 139, 17, 17, 139, + 18, 18, 139, 19, 19, 139, 20, 20, 139, 21, + 21, 139, 22, 22, 139, 24, 24, 139, 25, 25, + 139, 28, 28, 139, 27, 27, 139, 30, 30, 139, + 31, 31, 139, 32, 32, 139, 33, 33, 139, 34, + 34, 139, 35, 35, 139, 139, 139, 139, 36, 139, + 38, 139, 40, 40, 139, 41, 41, 139, 131, 42, + 42, 139, 43, 43, 139, 139, 45, 45, 139, 46, + + 46, 139, 47, 47, 139, 48, 48, 139, 139, 49, + 49, 139, 50, 50, 139, 51, 51, 139, 52, 52, + 139, 53, 53, 139, 54, 54, 139, 139, 10, 56, + 139, 57, 139, 58, 139, 59, 139, 60, 139, 62, + 139, 64, 64, 139, 139, 139, 139, 139, 139, 139, + 139, 0, 164, 0, 0, 0, 73, 74, 0, 0, + 0, 0, 0, 0, 0, 85, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 168, 0, 0, 0, - 126, 0, 128, 0, 0, 0, 0, 0, 0, 167, - 159, 152, 152, 152, 4, 152, 152, 152, 152, 152, - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - - 152, 152, 152, 152, 152, 152, 9, 152, 152, 152, - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 152, 82, 152, 152, 0, 0, 0, - 0, 0, 88, 89, 0, 0, 0, 0, 97, 0, - 0, 101, 104, 0, 0, 0, 0, 0, 0, 0, - 115, 116, 0, 0, 0, 0, 121, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 0, 152, 152, 152, - 152, 152, 152, 5, 152, 152, 152, 152, 152, 152, - 152, 152, 152, 152, 152, 152, 152, 152, 152, 152, - 152, 7, 8, 152, 152, 152, 152, 152, 152, 152, - - 152, 152, 152, 152, 152, 152, 152, 152, 152, 83, - 152, 79, 0, 0, 0, 0, 137, 0, 0, 0, - 0, 0, 0, 0, 0, 0, 107, 0, 111, 112, - 0, 114, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 130, 131, 0, 0, 138, 12, 3, 14, - 148, 149, 152, 17, 19, 21, 23, 25, 27, 29, - 32, 34, 39, 37, 42, 44, 46, 48, 50, 52, - 54, 56, 58, 60, 62, 152, 152, 152, 64, 66, - 68, 70, 72, 74, 76, 78, 152, 81, 139, 0, - 0, 84, 0, 90, 0, 0, 0, 99, 0, 0, - - 0, 0, 0, 0, 113, 0, 0, 119, 106, 0, - 0, 0, 0, 0, 0, 135, 0, 152, 145, 146, - 152, 80, 0, 0, 0, 0, 92, 95, 100, 0, - 0, 105, 0, 0, 0, 118, 0, 0, 0, 0, - 127, 129, 0, 152, 152, 2, 1, 0, 91, 0, - 103, 0, 109, 117, 0, 0, 124, 125, 136, 152, - 147, 0, 102, 0, 120, 123, 152, 85, 108, 152, - 152, 150, 151, 0 + 0, 0, 0, 0, 155, 0, 0, 0, 113, 0, + 115, 0, 0, 0, 0, 0, 0, 154, 146, 139, + + 139, 139, 4, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 9, 37, 39, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 60, 139, 61, 62, 139, 63, 139, 139, 139, 139, + 139, 69, 139, 139, 0, 0, 0, 0, 0, 75, + 76, 0, 0, 0, 0, 84, 0, 0, 88, 91, + 0, 0, 0, 0, 0, 0, 0, 102, 103, 0, + 0, 0, 0, 108, 0, 0, 0, 0, 0, 0, + + 0, 0, 0, 0, 139, 139, 139, 139, 139, 139, + 5, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 7, 8, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 139, 139, 139, 139, 139, 139, + 139, 139, 139, 139, 61, 139, 139, 63, 139, 139, + 139, 139, 139, 70, 139, 66, 0, 0, 0, 0, + 124, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 94, 0, 98, 99, 0, 101, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 0, 117, 118, 0, 0, + + 125, 11, 3, 12, 135, 136, 139, 14, 15, 16, + 17, 18, 19, 20, 21, 22, 24, 25, 28, 27, + 30, 31, 32, 33, 34, 35, 40, 41, 42, 43, + 44, 45, 46, 47, 48, 139, 139, 139, 49, 50, + 51, 52, 53, 54, 55, 56, 57, 58, 59, 139, + 139, 139, 139, 64, 65, 139, 68, 126, 0, 0, + 71, 0, 77, 0, 0, 0, 86, 0, 0, 0, + 0, 0, 0, 100, 0, 0, 106, 93, 0, 0, + 0, 0, 0, 0, 122, 0, 139, 132, 133, 139, + 60, 139, 62, 139, 67, 0, 0, 0, 0, 79, + + 82, 87, 0, 0, 92, 0, 0, 0, 105, 0, + 0, 0, 0, 114, 116, 0, 139, 139, 61, 63, + 2, 1, 0, 78, 0, 90, 0, 96, 104, 0, + 0, 111, 112, 123, 139, 134, 0, 89, 0, 107, + 110, 139, 72, 95, 139, 139, 137, 138, 0 } ; static yyconst flex_int32_t yy_ec[256] = @@ -478,280 +506,357 @@ static yyconst flex_int32_t yy_meta[68] = 2, 2, 2, 2, 2, 2, 2 } ; -static yyconst flex_int16_t yy_base[678] = +static yyconst flex_int16_t yy_base[853] = { 0, - 0, 0, 954, 955, 66, 955, 948, 949, 0, 69, - 85, 128, 140, 152, 151, 58, 39, 48, 75, 927, - 158, 160, 73, 59, 71, 170, 54, 920, 890, 889, - 901, 885, 899, 898, 142, 927, 939, 955, 0, 206, - 955, 189, 168, 171, 53, 27, 66, 119, 175, 899, - 885, 123, 170, 883, 895, 183, 955, 198, 225, 99, - 212, 219, 223, 227, 285, 297, 308, 955, 955, 955, - 904, 917, 911, 165, 900, 903, 899, 914, 224, 896, - 910, 194, 896, 909, 900, 913, 890, 901, 892, 294, - 893, 884, 893, 884, 883, 884, 878, 884, 895, 881, - - 878, 890, 893, 880, 873, 889, 865, 193, 139, 885, - 861, 846, 841, 858, 834, 839, 865, 167, 854, 259, - 849, 325, 282, 851, 832, 302, 842, 838, 833, 43, - 839, 825, 841, 838, 829, 305, 309, 831, 820, 834, - 837, 819, 834, 821, 818, 825, 275, 833, 254, 299, - 317, 327, 331, 810, 827, 828, 821, 803, 310, 804, - 826, 817, 316, 327, 331, 335, 339, 343, 347, 955, - 405, 416, 422, 428, 825, 240, 849, 0, 848, 831, - 821, 820, 840, 818, 817, 816, 815, 0, 814, 0, - 813, 812, 0, 811, 810, 0, 809, 808, 807, 806, - - 805, 804, 820, 813, 826, 800, 799, 805, 797, 796, - 795, 816, 793, 792, 791, 790, 800, 788, 787, 786, - 785, 777, 776, 761, 761, 760, 759, 802, 774, 762, - 434, 442, 416, 766, 186, 763, 757, 757, 751, 764, - 764, 749, 955, 955, 764, 752, 418, 759, 281, 756, - 762, 308, 757, 955, 748, 755, 754, 757, 743, 742, - 746, 741, 278, 746, 420, 428, 430, 955, 738, 736, - 736, 744, 745, 727, 421, 732, 738, 419, 426, 430, - 434, 438, 496, 502, 752, 764, 750, 749, 742, 756, - 746, 745, 0, 744, 743, 742, 741, 740, 739, 738, - - 737, 736, 735, 734, 733, 732, 731, 730, 733, 726, - 733, 726, 725, 0, 724, 723, 722, 725, 720, 719, - 718, 717, 0, 716, 715, 714, 713, 691, 685, 690, - 696, 679, 694, 315, 955, 693, 683, 687, 955, 955, - 677, 686, 672, 689, 672, 675, 669, 955, 670, 669, - 666, 673, 666, 674, 670, 680, 677, 659, 665, 672, - 656, 655, 673, 655, 667, 666, 955, 665, 655, 659, - 955, 646, 955, 651, 651, 659, 642, 643, 653, 955, - 955, 685, 667, 683, 0, 507, 681, 681, 680, 679, - 678, 677, 676, 675, 674, 673, 672, 671, 670, 669, - - 668, 667, 666, 665, 652, 645, 0, 662, 661, 660, - 659, 658, 636, 656, 655, 654, 653, 652, 651, 650, - 649, 618, 621, 601, 0, 602, 595, 602, 601, 602, - 594, 612, 955, 955, 594, 592, 602, 595, 955, 590, - 607, 330, 955, 598, 582, 583, 592, 583, 582, 582, - 955, 581, 590, 580, 596, 593, 955, 592, 590, 579, - 580, 576, 568, 575, 570, 571, 566, 592, 592, 590, - 604, 603, 598, 0, 586, 585, 584, 583, 582, 581, - 580, 579, 578, 577, 576, 575, 574, 573, 572, 571, - 570, 0, 0, 569, 568, 567, 566, 565, 509, 564, - - 563, 562, 561, 560, 559, 558, 557, 535, 535, 0, - 542, 0, 576, 575, 524, 542, 955, 537, 532, 525, - 521, 533, 523, 521, 517, 533, 524, 523, 955, 955, - 526, 955, 521, 514, 503, 514, 506, 510, 523, 518, - 521, 503, 955, 955, 515, 504, 955, 0, 0, 0, - 0, 0, 543, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 1299, 1300, 66, 1300, 1293, 1294, 0, 69, + 85, 128, 140, 152, 151, 58, 56, 63, 76, 1272, + 158, 160, 39, 163, 173, 189, 52, 1265, 76, 1235, + 1234, 1246, 1230, 1244, 1243, 105, 1272, 1284, 1300, 0, + 225, 1300, 218, 160, 157, 20, 123, 66, 119, 192, + 1244, 1230, 54, 162, 1228, 1240, 194, 1300, 200, 195, + 98, 227, 196, 231, 235, 293, 305, 316, 1300, 1300, + 1300, 1249, 1262, 1256, 223, 1245, 1248, 1244, 1259, 107, + 298, 1241, 1255, 246, 1241, 1254, 1245, 1258, 1235, 1246, + 1237, 182, 1238, 1229, 1238, 1229, 1228, 1229, 144, 1223, + + 1229, 1240, 1231, 1225, 1222, 1223, 1227, 289, 1236, 1223, + 302, 1230, 1217, 1231, 1207, 65, 315, 276, 1227, 1226, + 1202, 1187, 1182, 1199, 1175, 1180, 1206, 279, 1195, 293, + 1190, 342, 299, 1192, 1173, 317, 1183, 1179, 1174, 207, + 1180, 1166, 1182, 1179, 1170, 320, 324, 1172, 1161, 1175, + 1178, 1160, 1175, 1162, 1159, 1166, 284, 1174, 227, 288, + 327, 342, 345, 1151, 1168, 1169, 1162, 1144, 318, 1145, + 1167, 1158, 330, 341, 345, 349, 353, 357, 361, 1300, + 419, 430, 436, 442, 440, 441, 1191, 0, 1190, 1173, + 1163, 443, 1183, 444, 451, 468, 470, 472, 471, 0, + + 496, 0, 497, 498, 0, 499, 500, 0, 524, 525, + 526, 536, 537, 553, 1178, 1171, 1184, 354, 356, 561, + 563, 1165, 564, 565, 1157, 580, 590, 591, 592, 1178, + 593, 617, 618, 619, 629, 630, 1155, 1165, 330, 362, + 419, 483, 445, 364, 646, 1153, 1145, 1144, 1129, 1129, + 1128, 1127, 1170, 1142, 1130, 662, 669, 643, 1134, 487, + 1131, 1125, 1125, 1119, 1132, 1132, 1117, 1300, 1300, 1132, + 1120, 646, 1127, 135, 1124, 1130, 561, 1125, 1300, 1116, + 1123, 1122, 1125, 1111, 1110, 1114, 1109, 448, 1114, 650, + 653, 665, 1300, 1106, 1104, 1104, 1112, 1113, 1095, 670, + + 1100, 1106, 486, 579, 655, 661, 668, 726, 732, 1112, + 682, 1119, 1110, 688, 730, 1117, 1116, 1109, 1123, 1113, + 1104, 712, 1111, 0, 1102, 731, 1109, 1100, 733, 1107, + 1098, 734, 1105, 1096, 736, 1103, 1094, 737, 1101, 1092, + 738, 1099, 1090, 739, 1097, 1088, 740, 1095, 1086, 741, + 1093, 1084, 742, 1091, 1082, 743, 1089, 1080, 744, 1087, + 1078, 745, 1085, 1076, 746, 1083, 1074, 747, 1081, 1072, + 748, 1079, 1070, 749, 1077, 1080, 1073, 1080, 0, 1073, + 0, 1088, 1063, 750, 1070, 1061, 751, 1068, 0, 1059, + 752, 1066, 1057, 755, 1064, 1063, 1054, 758, 1061, 1052, + + 776, 1059, 1050, 777, 1057, 1048, 779, 1055, 1058, 1045, + 780, 1052, 1043, 782, 1050, 1041, 783, 1048, 1039, 784, + 1046, 1037, 785, 1044, 1035, 786, 1042, 1041, 0, 1032, + 1039, 1030, 1037, 1028, 1035, 1026, 1033, 787, 1032, 788, + 1047, 1022, 789, 1029, 1028, 1006, 1000, 1005, 1011, 994, + 1009, 424, 1300, 1008, 998, 1002, 1300, 1300, 992, 1001, + 987, 1004, 987, 990, 984, 1300, 985, 984, 981, 988, + 981, 989, 985, 995, 992, 974, 980, 987, 971, 970, + 988, 970, 982, 981, 1300, 980, 970, 974, 1300, 961, + 1300, 966, 966, 974, 957, 958, 968, 1300, 1300, 1000, + + 982, 998, 0, 798, 996, 996, 995, 994, 993, 992, + 991, 990, 989, 988, 987, 986, 985, 984, 983, 982, + 981, 980, 979, 978, 965, 958, 0, 0, 0, 975, + 974, 973, 972, 971, 970, 969, 968, 967, 945, 965, + 964, 963, 962, 961, 960, 959, 958, 957, 956, 955, + 929, 936, 793, 927, 934, 794, 950, 949, 918, 921, + 901, 0, 902, 895, 902, 901, 902, 894, 912, 1300, + 1300, 894, 892, 902, 895, 1300, 890, 907, 516, 1300, + 898, 882, 883, 892, 883, 882, 882, 1300, 881, 890, + 880, 896, 893, 1300, 892, 890, 879, 880, 876, 868, + + 875, 870, 871, 866, 892, 892, 890, 904, 903, 898, + 0, 886, 885, 884, 883, 882, 881, 880, 879, 878, + 877, 876, 875, 874, 873, 872, 871, 870, 869, 868, + 0, 0, 867, 866, 865, 864, 863, 862, 861, 860, + 859, 804, 858, 857, 856, 855, 854, 853, 852, 851, + 850, 849, 848, 865, 839, 846, 862, 836, 843, 841, + 840, 818, 818, 0, 825, 0, 859, 858, 807, 825, + 1300, 820, 815, 808, 804, 816, 806, 804, 800, 816, + 807, 806, 1300, 1300, 809, 1300, 804, 797, 786, 797, + 789, 793, 806, 801, 804, 786, 1300, 1300, 798, 787, + + 1300, 0, 0, 0, 0, 0, 826, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, 539, 538, 536, 0, 0, - 0, 0, 0, 0, 0, 0, 494, 0, 0, 545, - 544, 955, 491, 955, 495, 495, 504, 955, 488, 502, - - 483, 485, 482, 490, 955, 468, 479, 955, 955, 483, - 479, 472, 470, 470, 483, 955, 467, 507, 0, 0, - 507, 0, 514, 513, 472, 433, 955, 955, 955, 435, - 435, 955, 429, 386, 377, 955, 366, 365, 323, 328, - 955, 955, 339, 348, 337, 955, 955, 307, 955, 305, - 955, 257, 955, 955, 247, 221, 955, 955, 955, 236, - 0, 213, 955, 150, 955, 955, 232, 955, 955, 162, - 138, 0, 0, 955, 541, 108, 544 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, + 0, 0, 0, 0, 0, 814, 813, 802, 0, 0, + 0, 0, 0, 0, 0, 0, 0, 0, 0, 785, + 798, 779, 792, 0, 0, 656, 0, 0, 706, 702, + 1300, 649, 1300, 648, 648, 654, 1300, 637, 645, 610, + 612, 608, 608, 1300, 572, 583, 1300, 1300, 577, 573, + 560, 557, 542, 555, 1300, 539, 573, 0, 0, 572, + 0, 555, 0, 546, 0, 562, 551, 495, 479, 1300, + + 1300, 1300, 481, 481, 1300, 480, 443, 31, 1300, 141, + 166, 171, 186, 1300, 1300, 211, 236, 276, 0, 0, + 1300, 1300, 290, 1300, 325, 1300, 346, 1300, 1300, 343, + 341, 1300, 1300, 1300, 365, 0, 380, 1300, 371, 1300, + 1300, 486, 1300, 1300, 451, 458, 0, 0, 1300, 836, + 503, 839 } ; -static yyconst flex_int16_t yy_def[678] = +static yyconst flex_int16_t yy_def[853] = { 0, - 674, 1, 674, 674, 674, 674, 674, 675, 676, 674, - 674, 674, 674, 674, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 674, 674, 675, 674, 676, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 677, 674, 674, 674, 674, 674, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 676, - 676, 676, 676, 676, 676, 676, 676, 676, 676, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 676, 676, 676, - 676, 676, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 676, 676, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 676, - 676, 674, 674, 674, 674, 674, 676, 674, 674, 676, - 676, 676, 676, 0, 674, 674, 674 + 849, 1, 849, 849, 849, 849, 849, 850, 851, 849, + 849, 849, 849, 849, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 849, 849, 850, 849, 851, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 852, 849, 849, 849, 849, + 849, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + + 849, 849, 849, 849, 849, 849, 849, 849, 849, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 851, + + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + + 849, 849, 849, 849, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + + 849, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 851, 851, 851, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 851, 851, 851, 851, + 851, 851, 851, 851, 851, 849, 849, 849, 849, 849, + + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 851, 851, 851, 851, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 851, 851, 849, 849, 849, 849, + 849, 851, 849, 849, 851, 851, 851, 851, 0, 849, + 849, 849 } ; -static yyconst flex_int16_t yy_nxt[1023] = +static yyconst flex_int16_t yy_nxt[1368] = { 0, 4, 5, 6, 5, 7, 8, 9, 4, 10, 11, 12, 13, 14, 11, 11, 15, 9, 16, 17, 18, 19, 9, 9, 9, 20, 21, 22, 9, 23, 24, - 9, 25, 26, 27, 9, 9, 9, 28, 9, 9, - 9, 9, 9, 9, 9, 9, 29, 9, 9, 9, - 9, 9, 9, 9, 9, 9, 30, 9, 31, 32, - 33, 9, 34, 9, 9, 9, 9, 35, 79, 35, - 40, 80, 129, 108, 96, 81, 130, 41, 42, 42, - 42, 42, 42, 42, 76, 82, 77, 97, 98, 240, - 99, 109, 78, 65, 66, 66, 66, 66, 66, 66, - - 83, 241, 94, 100, 67, 127, 84, 95, 128, 39, - 43, 44, 45, 46, 47, 48, 49, 50, 51, 131, - 132, 52, 53, 54, 55, 56, 57, 58, 59, 60, - 67, 133, 61, 62, 63, 64, 65, 66, 66, 66, - 66, 66, 66, 35, 160, 35, 68, 67, 65, 66, - 66, 66, 66, 66, 66, 219, 673, 161, 69, 67, - 65, 66, 66, 66, 66, 66, 66, 71, 220, 72, - 70, 67, 140, 67, 134, 90, 73, 135, 141, 86, - 672, 87, 74, 91, 75, 67, 88, 101, 92, 89, - 178, 102, 103, 104, 93, 105, 179, 67, 42, 42, - - 42, 42, 42, 42, 106, 189, 107, 40, 122, 123, - 123, 142, 126, 123, 669, 123, 136, 137, 123, 217, - 124, 124, 123, 190, 147, 143, 123, 125, 125, 123, - 218, 337, 144, 123, 122, 148, 184, 185, 149, 151, - 152, 150, 670, 671, 338, 153, 186, 118, 119, 45, - 46, 47, 48, 154, 50, 51, 123, 162, 52, 53, - 54, 55, 56, 57, 120, 59, 60, 668, 155, 121, - 156, 286, 667, 157, 158, 163, 163, 163, 163, 666, - 287, 159, 164, 163, 165, 166, 167, 163, 163, 168, - 169, 163, 163, 163, 171, 171, 171, 171, 171, 171, - - 230, 665, 664, 260, 172, 65, 66, 66, 66, 66, - 66, 66, 198, 261, 154, 173, 67, 174, 174, 174, - 174, 174, 174, 233, 233, 364, 349, 257, 365, 233, - 172, 199, 231, 258, 232, 232, 232, 232, 232, 232, - 233, 350, 67, 233, 233, 236, 233, 233, 262, 233, - 247, 233, 233, 353, 263, 273, 233, 663, 233, 233, - 233, 428, 662, 233, 233, 274, 354, 233, 265, 233, - 661, 264, 266, 267, 233, 233, 660, 429, 233, 278, - 278, 278, 278, 524, 659, 233, 525, 658, 657, 233, - 278, 278, 278, 278, 279, 278, 278, 280, 281, 278, - - 278, 278, 278, 278, 278, 278, 282, 278, 278, 278, - 278, 278, 278, 278, 42, 42, 42, 42, 42, 42, - 656, 655, 654, 283, 122, 284, 284, 284, 284, 284, - 284, 174, 174, 174, 174, 174, 174, 174, 174, 174, - 174, 174, 174, 232, 232, 232, 232, 232, 232, 653, - 122, 232, 232, 232, 232, 232, 232, 335, 335, 335, - 335, 335, 335, 335, 374, 335, 375, 335, 376, 335, - 335, 367, 335, 652, 335, 335, 335, 335, 335, 651, - 650, 377, 380, 380, 380, 380, 335, 649, 335, 380, - 380, 380, 380, 381, 380, 380, 380, 380, 380, 380, - - 380, 380, 380, 380, 380, 284, 284, 284, 284, 284, - 284, 284, 284, 284, 284, 284, 284, 471, 472, 576, - 577, 648, 647, 646, 645, 644, 643, 642, 641, 640, - 639, 638, 637, 636, 635, 634, 633, 632, 631, 473, - 578, 37, 37, 37, 170, 170, 630, 629, 628, 627, - 626, 625, 624, 623, 622, 621, 620, 619, 618, 617, - 616, 615, 614, 613, 612, 611, 610, 609, 608, 607, - 606, 605, 604, 603, 602, 601, 600, 599, 598, 597, - 596, 595, 594, 593, 592, 591, 590, 589, 588, 587, - 586, 585, 584, 583, 582, 581, 580, 579, 575, 574, - - 573, 572, 571, 570, 569, 568, 567, 566, 565, 564, - 563, 562, 561, 560, 559, 558, 557, 556, 555, 554, - 553, 552, 551, 550, 549, 548, 547, 546, 545, 544, - 543, 542, 541, 540, 539, 538, 537, 536, 535, 534, - 533, 532, 531, 530, 529, 528, 527, 526, 523, 522, - 521, 520, 519, 518, 517, 516, 515, 514, 513, 512, - 511, 510, 509, 508, 507, 506, 505, 504, 503, 502, - 501, 500, 499, 498, 497, 496, 495, 494, 493, 492, - 491, 490, 489, 488, 487, 486, 485, 484, 483, 482, - 481, 480, 479, 478, 477, 476, 475, 474, 470, 469, - - 468, 467, 466, 465, 464, 463, 462, 461, 460, 459, - 458, 457, 456, 455, 454, 453, 452, 451, 450, 449, - 448, 447, 446, 445, 444, 443, 442, 441, 440, 439, - 438, 437, 436, 435, 434, 433, 432, 431, 430, 427, - 426, 425, 424, 423, 422, 421, 420, 419, 418, 417, - 416, 415, 414, 413, 412, 411, 410, 409, 408, 407, - 406, 405, 404, 403, 402, 401, 400, 399, 398, 397, - 396, 395, 394, 393, 392, 391, 390, 389, 388, 387, - 386, 385, 384, 383, 382, 379, 378, 373, 372, 371, - 370, 369, 368, 366, 363, 362, 361, 360, 359, 358, - - 357, 356, 355, 352, 351, 348, 347, 346, 345, 344, - 343, 342, 341, 340, 339, 336, 264, 236, 334, 333, - 332, 331, 330, 329, 328, 327, 326, 325, 324, 323, - 322, 321, 320, 319, 318, 317, 316, 315, 314, 313, - 312, 311, 310, 309, 308, 307, 306, 305, 304, 303, - 302, 301, 300, 299, 298, 297, 296, 295, 294, 293, - 292, 291, 290, 289, 288, 285, 277, 276, 275, 272, - 271, 270, 269, 268, 259, 256, 255, 254, 253, 252, - 251, 250, 249, 248, 246, 245, 244, 243, 242, 239, - 238, 237, 235, 234, 162, 229, 228, 227, 226, 225, - - 224, 223, 222, 221, 216, 215, 214, 213, 212, 211, - 210, 209, 208, 207, 206, 205, 204, 203, 202, 201, - 200, 197, 196, 195, 194, 193, 192, 191, 188, 187, - 183, 182, 181, 180, 177, 176, 175, 146, 145, 139, - 138, 38, 117, 116, 115, 114, 113, 112, 111, 110, - 85, 38, 36, 674, 3, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674 + 9, 25, 26, 27, 28, 9, 9, 29, 9, 9, + 9, 9, 9, 9, 9, 9, 30, 9, 9, 9, + 9, 9, 9, 9, 9, 9, 31, 9, 32, 33, + 34, 9, 35, 9, 9, 9, 9, 36, 96, 36, + 41, 116, 137, 97, 80, 138, 829, 42, 43, 43, + 43, 43, 43, 43, 77, 81, 78, 119, 82, 117, + 83, 238, 79, 66, 67, 67, 67, 67, 67, 67, + + 84, 85, 239, 150, 68, 120, 36, 86, 36, 151, + 44, 45, 46, 47, 48, 49, 50, 51, 52, 141, + 142, 53, 54, 55, 56, 57, 58, 59, 60, 61, + 68, 143, 62, 63, 64, 65, 66, 67, 67, 67, + 67, 67, 67, 170, 194, 195, 69, 68, 66, 67, + 67, 67, 67, 67, 67, 218, 171, 219, 70, 68, + 66, 67, 67, 67, 67, 67, 67, 72, 139, 73, + 71, 68, 140, 68, 144, 92, 74, 145, 98, 88, + 467, 89, 75, 93, 76, 68, 90, 99, 94, 91, + 101, 100, 102, 103, 95, 468, 830, 68, 136, 133, + + 210, 133, 133, 152, 133, 104, 105, 133, 106, 107, + 108, 109, 110, 134, 111, 133, 112, 153, 133, 211, + 135, 831, 113, 114, 154, 115, 41, 43, 43, 43, + 43, 43, 43, 146, 147, 157, 832, 132, 165, 133, + 166, 161, 162, 167, 168, 833, 158, 163, 188, 159, + 133, 169, 160, 265, 189, 164, 834, 201, 133, 174, + 173, 175, 176, 132, 835, 266, 128, 129, 46, 47, + 48, 49, 172, 51, 52, 202, 285, 53, 54, 55, + 56, 57, 58, 130, 60, 61, 286, 243, 131, 244, + 173, 173, 173, 173, 177, 173, 173, 178, 179, 173, + + 173, 173, 181, 181, 181, 181, 181, 181, 228, 836, + 196, 197, 182, 66, 67, 67, 67, 67, 67, 67, + 198, 232, 229, 183, 68, 184, 184, 184, 184, 184, + 184, 240, 134, 241, 255, 233, 282, 287, 182, 135, + 258, 258, 283, 288, 242, 837, 258, 430, 164, 256, + 68, 257, 257, 257, 257, 257, 257, 258, 258, 258, + 261, 258, 258, 298, 258, 272, 258, 258, 258, 258, + 431, 258, 381, 299, 258, 258, 379, 838, 258, 432, + 440, 289, 258, 290, 258, 258, 291, 292, 380, 258, + 382, 839, 258, 303, 303, 303, 303, 840, 441, 841, + + 258, 842, 433, 258, 303, 303, 303, 303, 304, 303, + 303, 305, 306, 303, 303, 303, 303, 303, 303, 303, + 307, 303, 303, 303, 303, 303, 303, 303, 43, 43, + 43, 43, 43, 43, 843, 844, 434, 308, 132, 309, + 309, 309, 309, 309, 309, 184, 184, 184, 184, 184, + 184, 184, 184, 184, 184, 184, 184, 310, 313, 435, + 321, 325, 311, 314, 132, 322, 326, 438, 328, 847, + 565, 311, 315, 329, 322, 326, 848, 311, 314, 439, + 312, 316, 329, 323, 327, 331, 566, 334, 340, 337, + 332, 330, 335, 341, 338, 482, 845, 846, 483, 332, + + 436, 335, 341, 338, 40, 332, 828, 335, 333, 338, + 336, 342, 339, 343, 346, 349, 352, 355, 344, 347, + 350, 353, 356, 437, 827, 826, 825, 344, 347, 350, + 353, 356, 455, 824, 347, 350, 345, 348, 351, 354, + 357, 358, 361, 364, 823, 456, 359, 362, 365, 498, + 498, 498, 498, 367, 370, 359, 362, 365, 368, 371, + 822, 359, 362, 365, 360, 363, 366, 368, 371, 678, + 373, 821, 679, 368, 371, 374, 369, 372, 383, 820, + 386, 390, 393, 384, 374, 387, 391, 394, 819, 818, + 374, 817, 384, 375, 387, 391, 394, 397, 816, 815, + + 814, 385, 398, 388, 392, 395, 471, 400, 403, 406, + 410, 398, 401, 404, 407, 411, 813, 398, 812, 472, + 399, 401, 404, 407, 411, 811, 810, 401, 404, 407, + 402, 405, 408, 412, 413, 416, 419, 809, 808, 414, + 417, 420, 498, 498, 498, 498, 422, 425, 414, 417, + 420, 423, 426, 807, 414, 417, 420, 415, 418, 421, + 423, 426, 806, 442, 805, 804, 423, 426, 443, 424, + 427, 257, 257, 257, 257, 257, 257, 443, 257, 257, + 257, 257, 257, 257, 453, 453, 444, 453, 453, 803, + 453, 453, 453, 453, 453, 453, 802, 453, 801, 310, + + 453, 453, 800, 799, 453, 313, 485, 453, 453, 798, + 797, 453, 453, 492, 796, 493, 795, 494, 499, 498, + 498, 498, 312, 453, 498, 498, 498, 498, 316, 321, + 495, 498, 498, 498, 498, 309, 309, 309, 309, 309, + 309, 309, 309, 309, 309, 309, 309, 313, 325, 501, + 328, 331, 323, 334, 337, 340, 343, 346, 349, 352, + 355, 358, 361, 364, 367, 370, 373, 383, 386, 390, + 316, 327, 393, 330, 333, 397, 336, 339, 342, 345, + 348, 351, 354, 357, 360, 363, 366, 369, 372, 375, + 385, 388, 392, 400, 403, 395, 406, 410, 399, 413, + + 416, 419, 422, 425, 551, 554, 442, 794, 608, 609, + 655, 658, 793, 792, 736, 737, 402, 405, 791, 408, + 412, 790, 415, 418, 421, 424, 427, 552, 555, 444, + 610, 789, 788, 656, 659, 738, 38, 38, 38, 180, + 180, 787, 786, 785, 784, 783, 782, 781, 780, 779, + 778, 777, 776, 775, 774, 773, 772, 771, 770, 769, + 768, 767, 766, 765, 764, 763, 762, 761, 760, 759, + 758, 757, 756, 755, 754, 753, 659, 752, 751, 656, + 750, 749, 748, 747, 746, 745, 744, 743, 742, 741, + 740, 739, 735, 734, 733, 732, 731, 730, 729, 728, + + 727, 726, 725, 724, 723, 722, 721, 720, 719, 718, + 717, 716, 715, 714, 713, 712, 711, 710, 709, 708, + 707, 706, 705, 704, 703, 702, 701, 700, 699, 698, + 697, 696, 695, 694, 693, 692, 691, 690, 689, 688, + 687, 686, 685, 684, 683, 682, 681, 680, 677, 676, + 675, 674, 673, 672, 671, 670, 669, 668, 667, 666, + 665, 664, 663, 662, 661, 660, 657, 555, 654, 552, + 653, 652, 651, 650, 649, 648, 647, 646, 645, 644, + 643, 642, 641, 640, 639, 638, 637, 636, 635, 634, + 633, 632, 631, 630, 629, 628, 627, 626, 625, 624, + + 623, 622, 621, 620, 619, 618, 617, 616, 615, 614, + 613, 612, 611, 607, 606, 605, 604, 603, 602, 601, + 600, 599, 598, 597, 596, 595, 594, 593, 592, 591, + 590, 589, 588, 587, 586, 585, 584, 583, 582, 581, + 580, 579, 578, 577, 576, 575, 574, 573, 572, 571, + 570, 569, 568, 567, 564, 563, 562, 561, 560, 559, + 558, 557, 444, 556, 553, 550, 437, 549, 435, 548, + 433, 547, 431, 546, 545, 427, 544, 424, 543, 421, + 542, 418, 541, 415, 540, 412, 539, 538, 408, 537, + 405, 536, 402, 535, 399, 534, 533, 395, 532, 392, + + 531, 388, 530, 385, 529, 528, 527, 526, 525, 524, + 375, 523, 372, 522, 369, 521, 366, 520, 363, 519, + 360, 518, 357, 517, 354, 516, 351, 515, 348, 514, + 345, 513, 342, 512, 339, 511, 336, 510, 333, 509, + 330, 508, 327, 507, 323, 506, 505, 504, 503, 502, + 316, 500, 312, 497, 496, 491, 490, 489, 488, 487, + 486, 484, 481, 480, 479, 478, 477, 476, 475, 474, + 473, 470, 469, 466, 465, 464, 463, 462, 461, 460, + 459, 458, 457, 454, 289, 261, 452, 451, 450, 449, + 448, 447, 446, 445, 429, 428, 409, 396, 389, 378, + + 377, 376, 324, 320, 319, 318, 317, 302, 301, 300, + 297, 296, 295, 294, 293, 284, 281, 280, 279, 278, + 277, 276, 275, 274, 273, 271, 270, 269, 268, 267, + 264, 263, 262, 260, 259, 172, 254, 253, 252, 251, + 250, 249, 248, 247, 246, 245, 237, 236, 235, 234, + 231, 230, 227, 226, 225, 224, 223, 222, 221, 220, + 217, 216, 215, 214, 213, 212, 209, 208, 207, 206, + 205, 204, 203, 200, 199, 193, 192, 191, 190, 187, + 186, 185, 156, 155, 149, 148, 39, 127, 126, 125, + 124, 123, 122, 121, 118, 87, 39, 37, 849, 3, + + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849 } ; -static yyconst flex_int16_t yy_chk[1023] = +static yyconst flex_int16_t yy_chk[1368] = { 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, @@ -759,113 +864,150 @@ static yyconst flex_int16_t yy_chk[1023] = 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, - 1, 1, 1, 1, 1, 1, 1, 5, 17, 5, - 10, 17, 46, 27, 24, 18, 46, 10, 10, 10, - 10, 10, 10, 10, 16, 18, 16, 24, 25, 130, - 25, 27, 16, 11, 11, 11, 11, 11, 11, 11, - - 19, 130, 23, 25, 11, 45, 19, 23, 45, 676, - 10, 10, 10, 10, 10, 10, 10, 10, 10, 47, - 47, 10, 10, 10, 10, 10, 10, 10, 10, 10, - 11, 47, 10, 10, 10, 10, 12, 12, 12, 12, - 12, 12, 12, 35, 60, 35, 12, 12, 13, 13, - 13, 13, 13, 13, 13, 109, 671, 60, 13, 13, - 14, 14, 14, 14, 14, 14, 14, 15, 109, 15, - 14, 14, 52, 12, 48, 22, 15, 48, 52, 21, - 670, 21, 15, 22, 15, 13, 21, 26, 22, 21, - 74, 26, 26, 26, 22, 26, 74, 14, 42, 42, - - 42, 42, 42, 42, 26, 82, 26, 40, 42, 43, - 43, 53, 44, 44, 664, 43, 49, 49, 44, 108, - 118, 43, 49, 82, 56, 53, 43, 118, 43, 44, - 108, 235, 53, 49, 42, 56, 79, 79, 56, 58, - 58, 56, 667, 667, 235, 58, 79, 40, 40, 40, - 40, 40, 40, 58, 40, 40, 58, 61, 40, 40, - 40, 40, 40, 40, 40, 40, 40, 662, 59, 40, - 59, 176, 660, 59, 59, 61, 61, 61, 61, 656, - 176, 59, 62, 62, 62, 62, 63, 63, 63, 63, - 64, 64, 64, 64, 65, 65, 65, 65, 65, 65, - - 120, 655, 652, 149, 65, 66, 66, 66, 66, 66, - 66, 66, 90, 149, 120, 67, 66, 67, 67, 67, - 67, 67, 67, 123, 123, 263, 249, 147, 263, 123, - 65, 90, 122, 147, 122, 122, 122, 122, 122, 122, - 123, 249, 66, 126, 126, 126, 136, 136, 150, 126, - 137, 137, 136, 252, 150, 159, 137, 650, 151, 151, - 126, 334, 648, 136, 151, 159, 252, 137, 152, 152, - 645, 151, 153, 153, 152, 151, 644, 334, 153, 163, - 163, 163, 163, 442, 643, 152, 442, 640, 639, 153, - 164, 164, 164, 164, 165, 165, 165, 165, 166, 166, - - 166, 166, 167, 167, 167, 167, 168, 168, 168, 168, - 169, 169, 169, 169, 171, 171, 171, 171, 171, 171, - 638, 637, 635, 172, 171, 172, 172, 172, 172, 172, - 172, 173, 173, 173, 173, 173, 173, 174, 174, 174, - 174, 174, 174, 231, 231, 231, 231, 231, 231, 634, - 171, 232, 232, 232, 232, 232, 232, 233, 233, 247, - 247, 265, 265, 233, 275, 247, 275, 265, 275, 266, - 266, 267, 267, 633, 233, 266, 247, 267, 265, 631, - 630, 275, 278, 278, 278, 278, 266, 626, 267, 279, - 279, 279, 279, 280, 280, 280, 280, 281, 281, 281, - - 281, 282, 282, 282, 282, 283, 283, 283, 283, 283, - 283, 284, 284, 284, 284, 284, 284, 386, 386, 499, - 499, 625, 624, 623, 621, 618, 617, 615, 614, 613, - 612, 611, 610, 607, 606, 604, 603, 602, 601, 386, - 499, 675, 675, 675, 677, 677, 600, 599, 597, 596, - 595, 593, 591, 590, 587, 578, 577, 576, 553, 546, - 545, 542, 541, 540, 539, 538, 537, 536, 535, 534, - 533, 531, 528, 527, 526, 525, 524, 523, 522, 521, - 520, 519, 518, 516, 515, 514, 513, 511, 509, 508, - 507, 506, 505, 504, 503, 502, 501, 500, 498, 497, - - 496, 495, 494, 491, 490, 489, 488, 487, 486, 485, - 484, 483, 482, 481, 480, 479, 478, 477, 476, 475, - 473, 472, 471, 470, 469, 468, 467, 466, 465, 464, - 463, 462, 461, 460, 459, 458, 456, 455, 454, 453, - 452, 450, 449, 448, 447, 446, 445, 444, 441, 440, - 438, 437, 436, 435, 432, 431, 430, 429, 428, 427, - 426, 424, 423, 422, 421, 420, 419, 418, 417, 416, - 415, 414, 413, 412, 411, 410, 409, 408, 406, 405, - 404, 403, 402, 401, 400, 399, 398, 397, 396, 395, - 394, 393, 392, 391, 390, 389, 388, 387, 384, 383, - - 382, 379, 378, 377, 376, 375, 374, 372, 370, 369, - 368, 366, 365, 364, 363, 362, 361, 360, 359, 358, - 357, 356, 355, 354, 353, 352, 351, 350, 349, 347, - 346, 345, 344, 343, 342, 341, 338, 337, 336, 333, - 332, 331, 330, 329, 328, 327, 326, 325, 324, 322, - 321, 320, 319, 318, 317, 316, 315, 313, 312, 311, - 310, 309, 308, 307, 306, 305, 304, 303, 302, 301, - 300, 299, 298, 297, 296, 295, 294, 292, 291, 290, - 289, 288, 287, 286, 285, 277, 276, 274, 273, 272, - 271, 270, 269, 264, 262, 261, 260, 259, 258, 257, - - 256, 255, 253, 251, 250, 248, 246, 245, 242, 241, - 240, 239, 238, 237, 236, 234, 230, 229, 228, 227, - 226, 225, 224, 223, 222, 221, 220, 219, 218, 217, - 216, 215, 214, 213, 212, 211, 210, 209, 208, 207, - 206, 205, 204, 203, 202, 201, 200, 199, 198, 197, - 195, 194, 192, 191, 189, 187, 186, 185, 184, 183, - 182, 181, 180, 179, 177, 175, 162, 161, 160, 158, - 157, 156, 155, 154, 148, 146, 145, 144, 143, 142, - 141, 140, 139, 138, 135, 134, 133, 132, 131, 129, - 128, 127, 125, 124, 121, 119, 117, 116, 115, 114, - - 113, 112, 111, 110, 107, 106, 105, 104, 103, 102, - 101, 100, 99, 98, 97, 96, 95, 94, 93, 92, - 91, 89, 88, 87, 86, 85, 84, 83, 81, 80, - 78, 77, 76, 75, 73, 72, 71, 55, 54, 51, - 50, 37, 36, 34, 33, 32, 31, 30, 29, 28, - 20, 8, 7, 3, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674, 674, 674, 674, 674, 674, 674, 674, 674, - 674, 674 + 1, 1, 1, 1, 1, 1, 1, 5, 23, 5, + 10, 27, 46, 23, 17, 46, 808, 10, 10, 10, + 10, 10, 10, 10, 16, 17, 16, 29, 17, 27, + 18, 116, 16, 11, 11, 11, 11, 11, 11, 11, + + 18, 19, 116, 53, 11, 29, 36, 19, 36, 53, + 10, 10, 10, 10, 10, 10, 10, 10, 10, 48, + 48, 10, 10, 10, 10, 10, 10, 10, 10, 10, + 11, 48, 10, 10, 10, 10, 12, 12, 12, 12, + 12, 12, 12, 61, 80, 80, 12, 12, 13, 13, + 13, 13, 13, 13, 13, 99, 61, 99, 13, 13, + 14, 14, 14, 14, 14, 14, 14, 15, 47, 15, + 14, 14, 47, 12, 49, 22, 15, 49, 24, 21, + 274, 21, 15, 22, 15, 13, 21, 24, 22, 21, + 25, 24, 25, 25, 22, 274, 810, 14, 45, 45, + + 92, 44, 44, 54, 45, 25, 26, 44, 26, 26, + 26, 26, 26, 44, 26, 45, 26, 54, 44, 92, + 44, 811, 26, 26, 54, 26, 41, 43, 43, 43, + 43, 43, 43, 50, 50, 57, 812, 43, 60, 50, + 60, 59, 59, 60, 60, 813, 57, 59, 75, 57, + 50, 60, 57, 140, 75, 59, 816, 84, 59, 63, + 63, 63, 63, 43, 817, 140, 41, 41, 41, 41, + 41, 41, 62, 41, 41, 84, 159, 41, 41, 41, + 41, 41, 41, 41, 41, 41, 159, 118, 41, 118, + 62, 62, 62, 62, 64, 64, 64, 64, 65, 65, + + 65, 65, 66, 66, 66, 66, 66, 66, 108, 818, + 81, 81, 66, 67, 67, 67, 67, 67, 67, 67, + 81, 111, 108, 68, 67, 68, 68, 68, 68, 68, + 68, 117, 128, 117, 130, 111, 157, 160, 66, 128, + 133, 133, 157, 160, 117, 823, 133, 239, 130, 132, + 67, 132, 132, 132, 132, 132, 132, 133, 136, 136, + 136, 146, 146, 169, 136, 147, 147, 146, 161, 161, + 239, 147, 219, 169, 161, 136, 218, 825, 146, 240, + 244, 161, 147, 162, 162, 161, 163, 163, 218, 162, + 219, 827, 163, 173, 173, 173, 173, 830, 244, 831, + + 162, 835, 240, 163, 174, 174, 174, 174, 175, 175, + 175, 175, 176, 176, 176, 176, 177, 177, 177, 177, + 178, 178, 178, 178, 179, 179, 179, 179, 181, 181, + 181, 181, 181, 181, 837, 839, 241, 182, 181, 182, + 182, 182, 182, 182, 182, 183, 183, 183, 183, 183, + 183, 184, 184, 184, 184, 184, 184, 185, 186, 241, + 192, 194, 185, 186, 181, 192, 194, 243, 195, 845, + 452, 185, 186, 195, 192, 194, 846, 185, 186, 243, + 185, 186, 195, 192, 194, 196, 452, 197, 199, 198, + 196, 195, 197, 199, 198, 288, 842, 842, 288, 196, + + 242, 197, 199, 198, 851, 196, 807, 197, 196, 198, + 197, 199, 198, 201, 203, 204, 206, 207, 201, 203, + 204, 206, 207, 242, 806, 804, 803, 201, 203, 204, + 206, 207, 260, 799, 203, 204, 201, 203, 204, 206, + 207, 209, 210, 211, 798, 260, 209, 210, 211, 303, + 303, 303, 303, 212, 213, 209, 210, 211, 212, 213, + 797, 209, 210, 211, 209, 210, 211, 212, 213, 579, + 214, 796, 579, 212, 213, 214, 212, 213, 220, 794, + 221, 223, 224, 220, 214, 221, 223, 224, 792, 790, + 214, 787, 220, 214, 221, 223, 224, 226, 786, 784, + + 783, 220, 226, 221, 223, 224, 277, 227, 228, 229, + 231, 226, 227, 228, 229, 231, 782, 226, 781, 277, + 226, 227, 228, 229, 231, 780, 779, 227, 228, 229, + 227, 228, 229, 231, 232, 233, 234, 776, 775, 232, + 233, 234, 304, 304, 304, 304, 235, 236, 232, 233, + 234, 235, 236, 773, 232, 233, 234, 232, 233, 234, + 235, 236, 772, 245, 771, 770, 235, 236, 245, 235, + 236, 256, 256, 256, 256, 256, 256, 245, 257, 257, + 257, 257, 257, 257, 258, 258, 245, 272, 272, 769, + 258, 290, 290, 272, 291, 291, 768, 290, 766, 311, + + 291, 258, 765, 764, 272, 314, 292, 292, 290, 762, + 760, 291, 292, 300, 759, 300, 756, 300, 305, 305, + 305, 305, 311, 292, 306, 306, 306, 306, 314, 322, + 300, 307, 307, 307, 307, 308, 308, 308, 308, 308, + 308, 309, 309, 309, 309, 309, 309, 315, 326, 315, + 329, 332, 322, 335, 338, 341, 344, 347, 350, 353, + 356, 359, 362, 365, 368, 371, 374, 384, 387, 391, + 315, 326, 394, 329, 332, 398, 335, 338, 341, 344, + 347, 350, 353, 356, 359, 362, 365, 368, 371, 374, + 384, 387, 391, 401, 404, 394, 407, 411, 398, 414, + + 417, 420, 423, 426, 438, 440, 443, 753, 504, 504, + 553, 556, 752, 751, 642, 642, 401, 404, 750, 407, + 411, 738, 414, 417, 420, 423, 426, 438, 440, 443, + 504, 737, 736, 553, 556, 642, 850, 850, 850, 852, + 852, 707, 700, 699, 696, 695, 694, 693, 692, 691, + 690, 689, 688, 687, 685, 682, 681, 680, 679, 678, + 677, 676, 675, 674, 673, 672, 670, 669, 668, 667, + 665, 663, 662, 661, 660, 659, 658, 657, 656, 655, + 654, 653, 652, 651, 650, 649, 648, 647, 646, 645, + 644, 643, 641, 640, 639, 638, 637, 636, 635, 634, + + 633, 630, 629, 628, 627, 626, 625, 624, 623, 622, + 621, 620, 619, 618, 617, 616, 615, 614, 613, 612, + 610, 609, 608, 607, 606, 605, 604, 603, 602, 601, + 600, 599, 598, 597, 596, 595, 593, 592, 591, 590, + 589, 587, 586, 585, 584, 583, 582, 581, 578, 577, + 575, 574, 573, 572, 569, 568, 567, 566, 565, 564, + 563, 561, 560, 559, 558, 557, 555, 554, 552, 551, + 550, 549, 548, 547, 546, 545, 544, 543, 542, 541, + 540, 539, 538, 537, 536, 535, 534, 533, 532, 531, + 530, 526, 525, 524, 523, 522, 521, 520, 519, 518, + + 517, 516, 515, 514, 513, 512, 511, 510, 509, 508, + 507, 506, 505, 502, 501, 500, 497, 496, 495, 494, + 493, 492, 490, 488, 487, 486, 484, 483, 482, 481, + 480, 479, 478, 477, 476, 475, 474, 473, 472, 471, + 470, 469, 468, 467, 465, 464, 463, 462, 461, 460, + 459, 456, 455, 454, 451, 450, 449, 448, 447, 446, + 445, 444, 442, 441, 439, 437, 436, 435, 434, 433, + 432, 431, 430, 428, 427, 425, 424, 422, 421, 419, + 418, 416, 415, 413, 412, 410, 409, 408, 406, 405, + 403, 402, 400, 399, 397, 396, 395, 393, 392, 390, + + 388, 386, 385, 383, 382, 380, 378, 377, 376, 375, + 373, 372, 370, 369, 367, 366, 364, 363, 361, 360, + 358, 357, 355, 354, 352, 351, 349, 348, 346, 345, + 343, 342, 340, 339, 337, 336, 334, 333, 331, 330, + 328, 327, 325, 323, 321, 320, 319, 318, 317, 316, + 313, 312, 310, 302, 301, 299, 298, 297, 296, 295, + 294, 289, 287, 286, 285, 284, 283, 282, 281, 280, + 278, 276, 275, 273, 271, 270, 267, 266, 265, 264, + 263, 262, 261, 259, 255, 254, 253, 252, 251, 250, + 249, 248, 247, 246, 238, 237, 230, 225, 222, 217, + + 216, 215, 193, 191, 190, 189, 187, 172, 171, 170, + 168, 167, 166, 165, 164, 158, 156, 155, 154, 153, + 152, 151, 150, 149, 148, 145, 144, 143, 142, 141, + 139, 138, 137, 135, 134, 131, 129, 127, 126, 125, + 124, 123, 122, 121, 120, 119, 115, 114, 113, 112, + 110, 109, 107, 106, 105, 104, 103, 102, 101, 100, + 98, 97, 96, 95, 94, 93, 91, 90, 89, 88, + 87, 86, 85, 83, 82, 79, 78, 77, 76, 74, + 73, 72, 56, 55, 52, 51, 38, 37, 35, 34, + 33, 32, 31, 30, 28, 20, 8, 7, 3, 849, + + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849, 849, 849, 849, + 849, 849, 849, 849, 849, 849, 849 } ; /* The intent behind this definition is that it'll catch @@ -903,11 +1045,13 @@ static yyconst flex_int16_t yy_chk[1023] = #include "prog_instruction.h" #include "prog_statevars.h" +#include "symbol_table.h" #include "program_parser.h" #include "program_parse.tab.h" #define require_ARB_vp (yyextra->mode == ARB_vertex) #define require_ARB_fp (yyextra->mode == ARB_fragment) +#define require_NV_fp (yyextra->option.NV_fragment) #define require_shadow (yyextra->option.Shadow) #define require_rect (yyextra->option.TexRect) #define require_texarray (yyextra->option.TexArray) @@ -921,8 +1065,7 @@ static yyconst flex_int16_t yy_chk[1023] = if (condition) { \ return token; \ } else { \ - yylval->string = strdup(yytext); \ - return IDENTIFIER; \ + return handle_ident(yyextra, yytext, yylval); \ } \ } while (0) @@ -937,15 +1080,16 @@ static yyconst flex_int16_t yy_chk[1023] = } while (0) -#define return_opcode(condition, token, opcode, sat) \ +#define return_opcode(condition, token, opcode, len) \ do { \ - if (condition) { \ + if (condition && \ + _mesa_parse_instruction_suffix(yyextra, \ + yytext + len, \ + & yylval->temp_inst)) { \ yylval->temp_inst.Opcode = OPCODE_ ## opcode; \ - yylval->temp_inst.SaturateMode = SATURATE_ ## sat; \ return token; \ } else { \ - yylval->string = strdup(yytext); \ - return IDENTIFIER; \ + return handle_ident(yyextra, yytext, yylval); \ } \ } while (0) @@ -994,6 +1138,15 @@ swiz_from_char(char c) return 0; } +static int +handle_ident(struct asm_parser_state *state, const char *text, YYSTYPE *lval) +{ + lval->string = strdup(text); + + return (_mesa_symbol_table_find_symbol(state->st, 0, text) == NULL) + ? IDENTIFIER : USED_IDENTIFIER; +} + #define YY_USER_ACTION \ do { \ yylloc->first_column = yylloc->last_column; \ @@ -1007,7 +1160,7 @@ swiz_from_char(char c) } while(0); #define YY_EXTRA_TYPE struct asm_parser_state * -#line 1007 "lex.yy.c" +#line 1164 "lex.yy.c" #define INITIAL 0 @@ -1144,7 +1297,12 @@ static int input (yyscan_t yyscanner ); /* Amount of stuff to slurp up with each read. */ #ifndef YY_READ_BUF_SIZE +#ifdef __ia64__ +/* On IA-64, the buffer size is 16k, not 8k */ +#define YY_READ_BUF_SIZE 16384 +#else #define YY_READ_BUF_SIZE 8192 +#endif /* __ia64__ */ #endif /* Copy whatever the last rule matched to the standard output. */ @@ -1152,7 +1310,7 @@ static int input (yyscan_t yyscanner ); /* This used to be an fputs(), but since the string might contain NUL's, * we now use fwrite(). */ -#define ECHO fwrite( yytext, yyleng, 1, yyout ) +#define ECHO do { if (fwrite( yytext, yyleng, 1, yyout )) {} } while (0) #endif /* Gets input and stuffs it into "buf". number of characters read, or YY_NULL, @@ -1163,7 +1321,7 @@ static int input (yyscan_t yyscanner ); if ( YY_CURRENT_BUFFER_LVALUE->yy_is_interactive ) \ { \ int c = '*'; \ - unsigned n; \ + size_t n; \ for ( n = 0; n < max_size && \ (c = getc( yyin )) != EOF && c != '\n'; ++n ) \ buf[n] = (char) c; \ @@ -1248,10 +1406,10 @@ YY_DECL register int yy_act; struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; -#line 136 "program_lexer.l" +#line 156 "program_lexer.l" -#line 1251 "lex.yy.c" +#line 1413 "lex.yy.c" yylval = yylval_param; @@ -1308,13 +1466,13 @@ yy_match: while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state ) { yy_current_state = (int) yy_def[yy_current_state]; - if ( yy_current_state >= 675 ) + if ( yy_current_state >= 850 ) yy_c = yy_meta[(unsigned int) yy_c]; } yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c]; ++yy_cp; } - while ( yy_base[yy_current_state] != 955 ); + while ( yy_base[yy_current_state] != 1300 ); yy_find_action: yy_act = yy_accept[yy_current_state]; @@ -1340,17 +1498,17 @@ do_action: /* This label is used only to access EOF actions. */ case 1: YY_RULE_SETUP -#line 138 "program_lexer.l" +#line 158 "program_lexer.l" { return ARBvp_10; } YY_BREAK case 2: YY_RULE_SETUP -#line 139 "program_lexer.l" +#line 159 "program_lexer.l" { return ARBfp_10; } YY_BREAK case 3: YY_RULE_SETUP -#line 140 "program_lexer.l" +#line 160 "program_lexer.l" { yylval->integer = at_address; return_token_or_IDENTIFIER(require_ARB_vp, ADDRESS); @@ -1358,813 +1516,745 @@ YY_RULE_SETUP YY_BREAK case 4: YY_RULE_SETUP -#line 144 "program_lexer.l" +#line 164 "program_lexer.l" { return ALIAS; } YY_BREAK case 5: YY_RULE_SETUP -#line 145 "program_lexer.l" +#line 165 "program_lexer.l" { return ATTRIB; } YY_BREAK case 6: YY_RULE_SETUP -#line 146 "program_lexer.l" +#line 166 "program_lexer.l" { return END; } YY_BREAK case 7: YY_RULE_SETUP -#line 147 "program_lexer.l" +#line 167 "program_lexer.l" { return OPTION; } YY_BREAK case 8: YY_RULE_SETUP -#line 148 "program_lexer.l" +#line 168 "program_lexer.l" { return OUTPUT; } YY_BREAK case 9: YY_RULE_SETUP -#line 149 "program_lexer.l" +#line 169 "program_lexer.l" { return PARAM; } YY_BREAK case 10: YY_RULE_SETUP -#line 150 "program_lexer.l" +#line 170 "program_lexer.l" { yylval->integer = at_temp; return TEMP; } YY_BREAK case 11: YY_RULE_SETUP -#line 152 "program_lexer.l" -{ return_opcode( 1, VECTOR_OP, ABS, OFF); } +#line 172 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, ABS, 3); } YY_BREAK case 12: YY_RULE_SETUP -#line 153 "program_lexer.l" -{ return_opcode(require_ARB_fp, VECTOR_OP, ABS, ZERO_ONE); } +#line 173 "program_lexer.l" +{ return_opcode( 1, BIN_OP, ADD, 3); } YY_BREAK case 13: YY_RULE_SETUP -#line 154 "program_lexer.l" -{ return_opcode( 1, BIN_OP, ADD, OFF); } +#line 174 "program_lexer.l" +{ return_opcode(require_ARB_vp, ARL, ARL, 3); } YY_BREAK case 14: YY_RULE_SETUP -#line 155 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, ADD, ZERO_ONE); } +#line 176 "program_lexer.l" +{ return_opcode(require_ARB_fp, TRI_OP, CMP, 3); } YY_BREAK case 15: YY_RULE_SETUP -#line 156 "program_lexer.l" -{ return_opcode(require_ARB_vp, ARL, ARL, OFF); } +#line 177 "program_lexer.l" +{ return_opcode(require_ARB_fp, SCALAR_OP, COS, 3); } YY_BREAK case 16: YY_RULE_SETUP -#line 158 "program_lexer.l" -{ return_opcode(require_ARB_fp, TRI_OP, CMP, OFF); } +#line 179 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, DDX, 3); } YY_BREAK case 17: YY_RULE_SETUP -#line 159 "program_lexer.l" -{ return_opcode(require_ARB_fp, TRI_OP, CMP, ZERO_ONE); } +#line 180 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, DDY, 3); } YY_BREAK case 18: YY_RULE_SETUP -#line 160 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, COS, OFF); } +#line 181 "program_lexer.l" +{ return_opcode( 1, BIN_OP, DP3, 3); } YY_BREAK case 19: YY_RULE_SETUP -#line 161 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, COS, ZERO_ONE); } +#line 182 "program_lexer.l" +{ return_opcode( 1, BIN_OP, DP4, 3); } YY_BREAK case 20: YY_RULE_SETUP -#line 163 "program_lexer.l" -{ return_opcode( 1, BIN_OP, DP3, OFF); } +#line 183 "program_lexer.l" +{ return_opcode( 1, BIN_OP, DPH, 3); } YY_BREAK case 21: YY_RULE_SETUP -#line 164 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, DP3, ZERO_ONE); } +#line 184 "program_lexer.l" +{ return_opcode( 1, BIN_OP, DST, 3); } YY_BREAK case 22: YY_RULE_SETUP -#line 165 "program_lexer.l" -{ return_opcode( 1, BIN_OP, DP4, OFF); } +#line 186 "program_lexer.l" +{ return_opcode( 1, SCALAR_OP, EX2, 3); } YY_BREAK case 23: YY_RULE_SETUP -#line 166 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, DP4, ZERO_ONE); } +#line 187 "program_lexer.l" +{ return_opcode(require_ARB_vp, SCALAR_OP, EXP, 3); } YY_BREAK case 24: YY_RULE_SETUP -#line 167 "program_lexer.l" -{ return_opcode( 1, BIN_OP, DPH, OFF); } +#line 189 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, FLR, 3); } YY_BREAK case 25: YY_RULE_SETUP -#line 168 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, DPH, ZERO_ONE); } +#line 190 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, FRC, 3); } YY_BREAK case 26: YY_RULE_SETUP -#line 169 "program_lexer.l" -{ return_opcode( 1, BIN_OP, DST, OFF); } +#line 192 "program_lexer.l" +{ return_opcode(require_ARB_fp, KIL, KIL, 3); } YY_BREAK case 27: YY_RULE_SETUP -#line 170 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, DST, ZERO_ONE); } +#line 194 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, LIT, 3); } YY_BREAK case 28: YY_RULE_SETUP -#line 172 "program_lexer.l" -{ return_opcode( 1, SCALAR_OP, EX2, OFF); } +#line 195 "program_lexer.l" +{ return_opcode( 1, SCALAR_OP, LG2, 3); } YY_BREAK case 29: YY_RULE_SETUP -#line 173 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, EX2, ZERO_ONE); } +#line 196 "program_lexer.l" +{ return_opcode(require_ARB_vp, SCALAR_OP, LOG, 3); } YY_BREAK case 30: YY_RULE_SETUP -#line 174 "program_lexer.l" -{ return_opcode(require_ARB_vp, SCALAR_OP, EXP, OFF); } +#line 197 "program_lexer.l" +{ return_opcode(require_ARB_fp, TRI_OP, LRP, 3); } YY_BREAK case 31: YY_RULE_SETUP -#line 176 "program_lexer.l" -{ return_opcode( 1, VECTOR_OP, FLR, OFF); } +#line 199 "program_lexer.l" +{ return_opcode( 1, TRI_OP, MAD, 3); } YY_BREAK case 32: YY_RULE_SETUP -#line 177 "program_lexer.l" -{ return_opcode(require_ARB_fp, VECTOR_OP, FLR, ZERO_ONE); } +#line 200 "program_lexer.l" +{ return_opcode( 1, BIN_OP, MAX, 3); } YY_BREAK case 33: YY_RULE_SETUP -#line 178 "program_lexer.l" -{ return_opcode( 1, VECTOR_OP, FRC, OFF); } +#line 201 "program_lexer.l" +{ return_opcode( 1, BIN_OP, MIN, 3); } YY_BREAK case 34: YY_RULE_SETUP -#line 179 "program_lexer.l" -{ return_opcode(require_ARB_fp, VECTOR_OP, FRC, ZERO_ONE); } +#line 202 "program_lexer.l" +{ return_opcode( 1, VECTOR_OP, MOV, 3); } YY_BREAK case 35: YY_RULE_SETUP -#line 181 "program_lexer.l" -{ return_opcode(require_ARB_fp, KIL, KIL, OFF); } +#line 203 "program_lexer.l" +{ return_opcode( 1, BIN_OP, MUL, 3); } YY_BREAK case 36: YY_RULE_SETUP -#line 183 "program_lexer.l" -{ return_opcode( 1, VECTOR_OP, LIT, OFF); } +#line 205 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, PK2H, 4); } YY_BREAK case 37: YY_RULE_SETUP -#line 184 "program_lexer.l" -{ return_opcode(require_ARB_fp, VECTOR_OP, LIT, ZERO_ONE); } +#line 206 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, PK2US, 5); } YY_BREAK case 38: YY_RULE_SETUP -#line 185 "program_lexer.l" -{ return_opcode( 1, SCALAR_OP, LG2, OFF); } +#line 207 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, PK4B, 4); } YY_BREAK case 39: YY_RULE_SETUP -#line 186 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, LG2, ZERO_ONE); } +#line 208 "program_lexer.l" +{ return_opcode(require_NV_fp, VECTOR_OP, PK4UB, 5); } YY_BREAK case 40: YY_RULE_SETUP -#line 187 "program_lexer.l" -{ return_opcode(require_ARB_vp, SCALAR_OP, LOG, OFF); } +#line 209 "program_lexer.l" +{ return_opcode( 1, BINSC_OP, POW, 3); } YY_BREAK case 41: YY_RULE_SETUP -#line 188 "program_lexer.l" -{ return_opcode(require_ARB_fp, TRI_OP, LRP, OFF); } +#line 211 "program_lexer.l" +{ return_opcode( 1, SCALAR_OP, RCP, 3); } YY_BREAK case 42: YY_RULE_SETUP -#line 189 "program_lexer.l" -{ return_opcode(require_ARB_fp, TRI_OP, LRP, ZERO_ONE); } +#line 212 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, RFL, 3); } YY_BREAK case 43: YY_RULE_SETUP -#line 191 "program_lexer.l" -{ return_opcode( 1, TRI_OP, MAD, OFF); } +#line 213 "program_lexer.l" +{ return_opcode( 1, SCALAR_OP, RSQ, 3); } YY_BREAK case 44: YY_RULE_SETUP -#line 192 "program_lexer.l" -{ return_opcode(require_ARB_fp, TRI_OP, MAD, ZERO_ONE); } +#line 215 "program_lexer.l" +{ return_opcode(require_ARB_fp, SCALAR_OP, SCS, 3); } YY_BREAK case 45: YY_RULE_SETUP -#line 193 "program_lexer.l" -{ return_opcode( 1, BIN_OP, MAX, OFF); } +#line 216 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SEQ, 3); } YY_BREAK case 46: YY_RULE_SETUP -#line 194 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, MAX, ZERO_ONE); } +#line 217 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SFL, 3); } YY_BREAK case 47: YY_RULE_SETUP -#line 195 "program_lexer.l" -{ return_opcode( 1, BIN_OP, MIN, OFF); } +#line 218 "program_lexer.l" +{ return_opcode( 1, BIN_OP, SGE, 3); } YY_BREAK case 48: YY_RULE_SETUP -#line 196 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, MIN, ZERO_ONE); } +#line 219 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SGT, 3); } YY_BREAK case 49: YY_RULE_SETUP -#line 197 "program_lexer.l" -{ return_opcode( 1, VECTOR_OP, MOV, OFF); } +#line 220 "program_lexer.l" +{ return_opcode(require_ARB_fp, SCALAR_OP, SIN, 3); } YY_BREAK case 50: YY_RULE_SETUP -#line 198 "program_lexer.l" -{ return_opcode(require_ARB_fp, VECTOR_OP, MOV, ZERO_ONE); } +#line 221 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SLE, 3); } YY_BREAK case 51: YY_RULE_SETUP -#line 199 "program_lexer.l" -{ return_opcode( 1, BIN_OP, MUL, OFF); } +#line 222 "program_lexer.l" +{ return_opcode( 1, BIN_OP, SLT, 3); } YY_BREAK case 52: YY_RULE_SETUP -#line 200 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, MUL, ZERO_ONE); } +#line 223 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, SNE, 3); } YY_BREAK case 53: YY_RULE_SETUP -#line 202 "program_lexer.l" -{ return_opcode( 1, BINSC_OP, POW, OFF); } +#line 224 "program_lexer.l" +{ return_opcode(require_NV_fp, BIN_OP, STR, 3); } YY_BREAK case 54: YY_RULE_SETUP -#line 203 "program_lexer.l" -{ return_opcode(require_ARB_fp, BINSC_OP, POW, ZERO_ONE); } +#line 225 "program_lexer.l" +{ return_opcode( 1, BIN_OP, SUB, 3); } YY_BREAK case 55: YY_RULE_SETUP -#line 205 "program_lexer.l" -{ return_opcode( 1, SCALAR_OP, RCP, OFF); } +#line 226 "program_lexer.l" +{ return_opcode( 1, SWZ, SWZ, 3); } YY_BREAK case 56: YY_RULE_SETUP -#line 206 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, RCP, ZERO_ONE); } +#line 228 "program_lexer.l" +{ return_opcode(require_ARB_fp, SAMPLE_OP, TEX, 3); } YY_BREAK case 57: YY_RULE_SETUP -#line 207 "program_lexer.l" -{ return_opcode( 1, SCALAR_OP, RSQ, OFF); } +#line 229 "program_lexer.l" +{ return_opcode(require_ARB_fp, SAMPLE_OP, TXB, 3); } YY_BREAK case 58: YY_RULE_SETUP -#line 208 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, RSQ, ZERO_ONE); } +#line 230 "program_lexer.l" +{ return_opcode(require_NV_fp, TXD_OP, TXD, 3); } YY_BREAK case 59: YY_RULE_SETUP -#line 210 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, SCS, OFF); } +#line 231 "program_lexer.l" +{ return_opcode(require_ARB_fp, SAMPLE_OP, TXP, 3); } YY_BREAK case 60: YY_RULE_SETUP -#line 211 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, SCS, ZERO_ONE); } +#line 233 "program_lexer.l" +{ return_opcode(require_NV_fp, SCALAR_OP, UP2H, 4); } YY_BREAK case 61: YY_RULE_SETUP -#line 212 "program_lexer.l" -{ return_opcode( 1, BIN_OP, SGE, OFF); } +#line 234 "program_lexer.l" +{ return_opcode(require_NV_fp, SCALAR_OP, UP2US, 5); } YY_BREAK case 62: YY_RULE_SETUP -#line 213 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, SGE, ZERO_ONE); } +#line 235 "program_lexer.l" +{ return_opcode(require_NV_fp, SCALAR_OP, UP4B, 4); } YY_BREAK case 63: YY_RULE_SETUP -#line 214 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, SIN, OFF); } +#line 236 "program_lexer.l" +{ return_opcode(require_NV_fp, SCALAR_OP, UP4UB, 5); } YY_BREAK case 64: YY_RULE_SETUP -#line 215 "program_lexer.l" -{ return_opcode(require_ARB_fp, SCALAR_OP, SIN, ZERO_ONE); } +#line 238 "program_lexer.l" +{ return_opcode(require_NV_fp, TRI_OP, X2D, 3); } YY_BREAK case 65: YY_RULE_SETUP -#line 216 "program_lexer.l" -{ return_opcode( 1, BIN_OP, SLT, OFF); } +#line 239 "program_lexer.l" +{ return_opcode( 1, BIN_OP, XPD, 3); } YY_BREAK case 66: YY_RULE_SETUP -#line 217 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, SLT, ZERO_ONE); } +#line 241 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_vp, VERTEX); } YY_BREAK case 67: YY_RULE_SETUP -#line 218 "program_lexer.l" -{ return_opcode( 1, BIN_OP, SUB, OFF); } +#line 242 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, FRAGMENT); } YY_BREAK case 68: YY_RULE_SETUP -#line 219 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, SUB, ZERO_ONE); } +#line 243 "program_lexer.l" +{ return PROGRAM; } YY_BREAK case 69: YY_RULE_SETUP -#line 220 "program_lexer.l" -{ return_opcode( 1, SWZ, SWZ, OFF); } +#line 244 "program_lexer.l" +{ return STATE; } YY_BREAK case 70: YY_RULE_SETUP -#line 221 "program_lexer.l" -{ return_opcode(require_ARB_fp, SWZ, SWZ, ZERO_ONE); } +#line 245 "program_lexer.l" +{ return RESULT; } YY_BREAK case 71: YY_RULE_SETUP -#line 223 "program_lexer.l" -{ return_opcode(require_ARB_fp, SAMPLE_OP, TEX, OFF); } +#line 247 "program_lexer.l" +{ return AMBIENT; } YY_BREAK case 72: YY_RULE_SETUP -#line 224 "program_lexer.l" -{ return_opcode(require_ARB_fp, SAMPLE_OP, TEX, ZERO_ONE); } +#line 248 "program_lexer.l" +{ return ATTENUATION; } YY_BREAK case 73: YY_RULE_SETUP -#line 225 "program_lexer.l" -{ return_opcode(require_ARB_fp, SAMPLE_OP, TXB, OFF); } +#line 249 "program_lexer.l" +{ return BACK; } YY_BREAK case 74: YY_RULE_SETUP -#line 226 "program_lexer.l" -{ return_opcode(require_ARB_fp, SAMPLE_OP, TXB, ZERO_ONE); } +#line 250 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, CLIP); } YY_BREAK case 75: YY_RULE_SETUP -#line 227 "program_lexer.l" -{ return_opcode(require_ARB_fp, SAMPLE_OP, TXP, OFF); } +#line 251 "program_lexer.l" +{ return COLOR; } YY_BREAK case 76: YY_RULE_SETUP -#line 228 "program_lexer.l" -{ return_opcode(require_ARB_fp, SAMPLE_OP, TXP, ZERO_ONE); } +#line 252 "program_lexer.l" +{ return_token_or_DOT(require_ARB_fp, DEPTH); } YY_BREAK case 77: YY_RULE_SETUP -#line 230 "program_lexer.l" -{ return_opcode( 1, BIN_OP, XPD, OFF); } +#line 253 "program_lexer.l" +{ return DIFFUSE; } YY_BREAK case 78: YY_RULE_SETUP -#line 231 "program_lexer.l" -{ return_opcode(require_ARB_fp, BIN_OP, XPD, ZERO_ONE); } +#line 254 "program_lexer.l" +{ return DIRECTION; } YY_BREAK case 79: YY_RULE_SETUP -#line 233 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_vp, VERTEX); } +#line 255 "program_lexer.l" +{ return EMISSION; } YY_BREAK case 80: YY_RULE_SETUP -#line 234 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp, FRAGMENT); } +#line 256 "program_lexer.l" +{ return ENV; } YY_BREAK case 81: YY_RULE_SETUP -#line 235 "program_lexer.l" -{ return PROGRAM; } +#line 257 "program_lexer.l" +{ return EYE; } YY_BREAK case 82: YY_RULE_SETUP -#line 236 "program_lexer.l" -{ return STATE; } +#line 258 "program_lexer.l" +{ return FOGCOORD; } YY_BREAK case 83: YY_RULE_SETUP -#line 237 "program_lexer.l" -{ return RESULT; } +#line 259 "program_lexer.l" +{ return FOG; } YY_BREAK case 84: YY_RULE_SETUP -#line 239 "program_lexer.l" -{ return AMBIENT; } +#line 260 "program_lexer.l" +{ return FRONT; } YY_BREAK case 85: YY_RULE_SETUP -#line 240 "program_lexer.l" -{ return ATTENUATION; } +#line 261 "program_lexer.l" +{ return HALF; } YY_BREAK case 86: YY_RULE_SETUP -#line 241 "program_lexer.l" -{ return BACK; } +#line 262 "program_lexer.l" +{ return INVERSE; } YY_BREAK case 87: YY_RULE_SETUP -#line 242 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, CLIP); } +#line 263 "program_lexer.l" +{ return INVTRANS; } YY_BREAK case 88: YY_RULE_SETUP -#line 243 "program_lexer.l" -{ return COLOR; } +#line 264 "program_lexer.l" +{ return LIGHT; } YY_BREAK case 89: YY_RULE_SETUP -#line 244 "program_lexer.l" -{ return_token_or_DOT(require_ARB_fp, DEPTH); } +#line 265 "program_lexer.l" +{ return LIGHTMODEL; } YY_BREAK case 90: YY_RULE_SETUP -#line 245 "program_lexer.l" -{ return DIFFUSE; } +#line 266 "program_lexer.l" +{ return LIGHTPROD; } YY_BREAK case 91: YY_RULE_SETUP -#line 246 "program_lexer.l" -{ return DIRECTION; } +#line 267 "program_lexer.l" +{ return LOCAL; } YY_BREAK case 92: YY_RULE_SETUP -#line 247 "program_lexer.l" -{ return EMISSION; } +#line 268 "program_lexer.l" +{ return MATERIAL; } YY_BREAK case 93: YY_RULE_SETUP -#line 248 "program_lexer.l" -{ return ENV; } +#line 269 "program_lexer.l" +{ return MAT_PROGRAM; } YY_BREAK case 94: YY_RULE_SETUP -#line 249 "program_lexer.l" -{ return EYE; } +#line 270 "program_lexer.l" +{ return MATRIX; } YY_BREAK case 95: YY_RULE_SETUP -#line 250 "program_lexer.l" -{ return FOGCOORD; } +#line 271 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, MATRIXINDEX); } YY_BREAK case 96: YY_RULE_SETUP -#line 251 "program_lexer.l" -{ return FOG; } +#line 272 "program_lexer.l" +{ return MODELVIEW; } YY_BREAK case 97: YY_RULE_SETUP -#line 252 "program_lexer.l" -{ return FRONT; } +#line 273 "program_lexer.l" +{ return MVP; } YY_BREAK case 98: YY_RULE_SETUP -#line 253 "program_lexer.l" -{ return HALF; } +#line 274 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, NORMAL); } YY_BREAK case 99: YY_RULE_SETUP -#line 254 "program_lexer.l" -{ return INVERSE; } +#line 275 "program_lexer.l" +{ return OBJECT; } YY_BREAK case 100: YY_RULE_SETUP -#line 255 "program_lexer.l" -{ return INVTRANS; } +#line 276 "program_lexer.l" +{ return PALETTE; } YY_BREAK case 101: YY_RULE_SETUP -#line 256 "program_lexer.l" -{ return LIGHT; } +#line 277 "program_lexer.l" +{ return PARAMS; } YY_BREAK case 102: YY_RULE_SETUP -#line 257 "program_lexer.l" -{ return LIGHTMODEL; } +#line 278 "program_lexer.l" +{ return PLANE; } YY_BREAK case 103: YY_RULE_SETUP -#line 258 "program_lexer.l" -{ return LIGHTPROD; } +#line 279 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, POINT_TOK); } YY_BREAK case 104: YY_RULE_SETUP -#line 259 "program_lexer.l" -{ return LOCAL; } +#line 280 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, POINTSIZE); } YY_BREAK case 105: YY_RULE_SETUP -#line 260 "program_lexer.l" -{ return MATERIAL; } +#line 281 "program_lexer.l" +{ return POSITION; } YY_BREAK case 106: YY_RULE_SETUP -#line 261 "program_lexer.l" -{ return MAT_PROGRAM; } +#line 282 "program_lexer.l" +{ return PRIMARY; } YY_BREAK case 107: YY_RULE_SETUP -#line 262 "program_lexer.l" -{ return MATRIX; } +#line 283 "program_lexer.l" +{ return PROJECTION; } YY_BREAK case 108: YY_RULE_SETUP -#line 263 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, MATRIXINDEX); } +#line 284 "program_lexer.l" +{ return_token_or_DOT(require_ARB_fp, RANGE); } YY_BREAK case 109: YY_RULE_SETUP -#line 264 "program_lexer.l" -{ return MODELVIEW; } +#line 285 "program_lexer.l" +{ return ROW; } YY_BREAK case 110: YY_RULE_SETUP -#line 265 "program_lexer.l" -{ return MVP; } +#line 286 "program_lexer.l" +{ return SCENECOLOR; } YY_BREAK case 111: YY_RULE_SETUP -#line 266 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, NORMAL); } +#line 287 "program_lexer.l" +{ return SECONDARY; } YY_BREAK case 112: YY_RULE_SETUP -#line 267 "program_lexer.l" -{ return OBJECT; } +#line 288 "program_lexer.l" +{ return SHININESS; } YY_BREAK case 113: YY_RULE_SETUP -#line 268 "program_lexer.l" -{ return PALETTE; } +#line 289 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, SIZE_TOK); } YY_BREAK case 114: YY_RULE_SETUP -#line 269 "program_lexer.l" -{ return PARAMS; } +#line 290 "program_lexer.l" +{ return SPECULAR; } YY_BREAK case 115: YY_RULE_SETUP -#line 270 "program_lexer.l" -{ return PLANE; } +#line 291 "program_lexer.l" +{ return SPOT; } YY_BREAK case 116: YY_RULE_SETUP -#line 271 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, POINT_TOK); } +#line 292 "program_lexer.l" +{ return TEXCOORD; } YY_BREAK case 117: YY_RULE_SETUP -#line 272 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, POINTSIZE); } +#line 293 "program_lexer.l" +{ return_token_or_DOT(require_ARB_fp, TEXENV); } YY_BREAK case 118: YY_RULE_SETUP -#line 273 "program_lexer.l" -{ return POSITION; } +#line 294 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, TEXGEN); } YY_BREAK case 119: YY_RULE_SETUP -#line 274 "program_lexer.l" -{ return PRIMARY; } +#line 295 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, TEXGEN_Q); } YY_BREAK case 120: YY_RULE_SETUP -#line 275 "program_lexer.l" -{ return PROJECTION; } +#line 296 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, TEXGEN_S); } YY_BREAK case 121: YY_RULE_SETUP -#line 276 "program_lexer.l" -{ return_token_or_DOT(require_ARB_fp, RANGE); } +#line 297 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, TEXGEN_T); } YY_BREAK case 122: YY_RULE_SETUP -#line 277 "program_lexer.l" -{ return ROW; } +#line 298 "program_lexer.l" +{ return TEXTURE; } YY_BREAK case 123: YY_RULE_SETUP -#line 278 "program_lexer.l" -{ return SCENECOLOR; } +#line 299 "program_lexer.l" +{ return TRANSPOSE; } YY_BREAK case 124: YY_RULE_SETUP -#line 279 "program_lexer.l" -{ return SECONDARY; } +#line 300 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, VTXATTRIB); } YY_BREAK case 125: YY_RULE_SETUP -#line 280 "program_lexer.l" -{ return SHININESS; } +#line 301 "program_lexer.l" +{ return_token_or_DOT(require_ARB_vp, WEIGHT); } YY_BREAK case 126: YY_RULE_SETUP -#line 281 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, SIZE_TOK); } +#line 303 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEXTURE_UNIT); } YY_BREAK case 127: YY_RULE_SETUP -#line 282 "program_lexer.l" -{ return SPECULAR; } +#line 304 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_1D); } YY_BREAK case 128: YY_RULE_SETUP -#line 283 "program_lexer.l" -{ return SPOT; } +#line 305 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_2D); } YY_BREAK case 129: YY_RULE_SETUP -#line 284 "program_lexer.l" -{ return TEXCOORD; } +#line 306 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_3D); } YY_BREAK case 130: YY_RULE_SETUP -#line 285 "program_lexer.l" -{ return_token_or_DOT(require_ARB_fp, TEXENV); } +#line 307 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_CUBE); } YY_BREAK case 131: YY_RULE_SETUP -#line 286 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, TEXGEN); } +#line 308 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_rect, TEX_RECT); } YY_BREAK case 132: YY_RULE_SETUP -#line 287 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, TEXGEN_Q); } +#line 309 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW1D); } YY_BREAK case 133: YY_RULE_SETUP -#line 288 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, TEXGEN_S); } +#line 310 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW2D); } YY_BREAK case 134: YY_RULE_SETUP -#line 289 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, TEXGEN_T); } +#line 311 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_rect, TEX_SHADOWRECT); } YY_BREAK case 135: YY_RULE_SETUP -#line 290 "program_lexer.l" -{ return TEXTURE; } +#line 312 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY1D); } YY_BREAK case 136: YY_RULE_SETUP -#line 291 "program_lexer.l" -{ return TRANSPOSE; } +#line 313 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY2D); } YY_BREAK case 137: YY_RULE_SETUP -#line 292 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, VTXATTRIB); } +#line 314 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW1D); } YY_BREAK case 138: YY_RULE_SETUP -#line 293 "program_lexer.l" -{ return_token_or_DOT(require_ARB_vp, WEIGHT); } +#line 315 "program_lexer.l" +{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW2D); } YY_BREAK case 139: YY_RULE_SETUP -#line 295 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp, TEXTURE_UNIT); } +#line 317 "program_lexer.l" +{ return handle_ident(yyextra, yytext, yylval); } YY_BREAK case 140: YY_RULE_SETUP -#line 296 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_1D); } - YY_BREAK -case 141: -YY_RULE_SETUP -#line 297 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_2D); } - YY_BREAK -case 142: -YY_RULE_SETUP -#line 298 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_3D); } - YY_BREAK -case 143: -YY_RULE_SETUP -#line 299 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp, TEX_CUBE); } - YY_BREAK -case 144: -YY_RULE_SETUP -#line 300 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp && require_rect, TEX_RECT); } - YY_BREAK -case 145: -YY_RULE_SETUP -#line 301 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW1D); } - YY_BREAK -case 146: -YY_RULE_SETUP -#line 302 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow, TEX_SHADOW2D); } - YY_BREAK -case 147: -YY_RULE_SETUP -#line 303 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_rect, TEX_SHADOWRECT); } - YY_BREAK -case 148: -YY_RULE_SETUP -#line 304 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY1D); } - YY_BREAK -case 149: -YY_RULE_SETUP -#line 305 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp && require_texarray, TEX_ARRAY2D); } - YY_BREAK -case 150: -YY_RULE_SETUP -#line 306 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW1D); } - YY_BREAK -case 151: -YY_RULE_SETUP -#line 307 "program_lexer.l" -{ return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW2D); } - YY_BREAK -case 152: -YY_RULE_SETUP -#line 309 "program_lexer.l" -{ - yylval->string = strdup(yytext); - return IDENTIFIER; -} - YY_BREAK -case 153: -YY_RULE_SETUP -#line 314 "program_lexer.l" +#line 319 "program_lexer.l" { return DOT_DOT; } YY_BREAK -case 154: +case 141: YY_RULE_SETUP -#line 316 "program_lexer.l" +#line 321 "program_lexer.l" { yylval->integer = strtol(yytext, NULL, 10); return INTEGER; } YY_BREAK -case 155: +case 142: YY_RULE_SETUP -#line 320 "program_lexer.l" +#line 325 "program_lexer.l" { yylval->real = strtod(yytext, NULL); return REAL; } YY_BREAK -case 156: -/* rule 156 can match eol */ +case 143: +/* rule 143 can match eol */ *yy_cp = yyg->yy_hold_char; /* undo effects of setting up yytext */ yyg->yy_c_buf_p = yy_cp -= 1; YY_DO_BEFORE_ACTION; /* set up yytext again */ YY_RULE_SETUP -#line 324 "program_lexer.l" +#line 329 "program_lexer.l" { yylval->real = strtod(yytext, NULL); return REAL; } YY_BREAK -case 157: +case 144: YY_RULE_SETUP -#line 328 "program_lexer.l" +#line 333 "program_lexer.l" { yylval->real = strtod(yytext, NULL); return REAL; } YY_BREAK -case 158: +case 145: YY_RULE_SETUP -#line 332 "program_lexer.l" +#line 337 "program_lexer.l" { yylval->real = strtod(yytext, NULL); return REAL; } YY_BREAK -case 159: +case 146: YY_RULE_SETUP -#line 337 "program_lexer.l" +#line 342 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_NOOP; yylval->swiz_mask.mask = WRITEMASK_XYZW; return MASK4; } YY_BREAK -case 160: +case 147: YY_RULE_SETUP -#line 343 "program_lexer.l" +#line 348 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_XY @@ -2172,27 +2262,27 @@ YY_RULE_SETUP return MASK3; } YY_BREAK -case 161: +case 148: YY_RULE_SETUP -#line 349 "program_lexer.l" +#line 354 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_XZW; return MASK3; } YY_BREAK -case 162: +case 149: YY_RULE_SETUP -#line 354 "program_lexer.l" +#line 359 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_YZW; return MASK3; } YY_BREAK -case 163: +case 150: YY_RULE_SETUP -#line 360 "program_lexer.l" +#line 365 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_X @@ -2200,9 +2290,9 @@ YY_RULE_SETUP return MASK2; } YY_BREAK -case 164: +case 151: YY_RULE_SETUP -#line 366 "program_lexer.l" +#line 371 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_Y @@ -2210,18 +2300,18 @@ YY_RULE_SETUP return MASK2; } YY_BREAK -case 165: +case 152: YY_RULE_SETUP -#line 372 "program_lexer.l" +#line 377 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_ZW; return MASK2; } YY_BREAK -case 166: +case 153: YY_RULE_SETUP -#line 378 "program_lexer.l" +#line 383 "program_lexer.l" { const unsigned s = swiz_from_char(yytext[1]); yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(s, s, s, s); @@ -2229,9 +2319,9 @@ YY_RULE_SETUP return MASK1; } YY_BREAK -case 167: +case 154: YY_RULE_SETUP -#line 385 "program_lexer.l" +#line 390 "program_lexer.l" { yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(swiz_from_char(yytext[1]), swiz_from_char(yytext[2]), @@ -2241,18 +2331,18 @@ YY_RULE_SETUP return SWIZZLE; } YY_BREAK -case 168: +case 155: YY_RULE_SETUP -#line 394 "program_lexer.l" +#line 399 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_NOOP; yylval->swiz_mask.mask = WRITEMASK_XYZW; return_token_or_DOT(require_ARB_fp, MASK4); } YY_BREAK -case 169: +case 156: YY_RULE_SETUP -#line 400 "program_lexer.l" +#line 405 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_XY @@ -2260,27 +2350,27 @@ YY_RULE_SETUP return_token_or_DOT(require_ARB_fp, MASK3); } YY_BREAK -case 170: +case 157: YY_RULE_SETUP -#line 406 "program_lexer.l" +#line 411 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_XZW; return_token_or_DOT(require_ARB_fp, MASK3); } YY_BREAK -case 171: +case 158: YY_RULE_SETUP -#line 411 "program_lexer.l" +#line 416 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_YZW; return_token_or_DOT(require_ARB_fp, MASK3); } YY_BREAK -case 172: +case 159: YY_RULE_SETUP -#line 417 "program_lexer.l" +#line 422 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_X @@ -2288,9 +2378,9 @@ YY_RULE_SETUP return_token_or_DOT(require_ARB_fp, MASK2); } YY_BREAK -case 173: +case 160: YY_RULE_SETUP -#line 423 "program_lexer.l" +#line 428 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_Y @@ -2298,18 +2388,18 @@ YY_RULE_SETUP return_token_or_DOT(require_ARB_fp, MASK2); } YY_BREAK -case 174: +case 161: YY_RULE_SETUP -#line 429 "program_lexer.l" +#line 434 "program_lexer.l" { yylval->swiz_mask.swizzle = SWIZZLE_INVAL; yylval->swiz_mask.mask = WRITEMASK_ZW; return_token_or_DOT(require_ARB_fp, MASK2); } YY_BREAK -case 175: +case 162: YY_RULE_SETUP -#line 435 "program_lexer.l" +#line 440 "program_lexer.l" { const unsigned s = swiz_from_char(yytext[1]); yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(s, s, s, s); @@ -2317,9 +2407,9 @@ YY_RULE_SETUP return_token_or_DOT(require_ARB_fp, MASK1); } YY_BREAK -case 176: +case 163: YY_RULE_SETUP -#line 443 "program_lexer.l" +#line 448 "program_lexer.l" { if (require_ARB_vp) { return TEXGEN_R; @@ -2331,9 +2421,9 @@ YY_RULE_SETUP } } YY_BREAK -case 177: +case 164: YY_RULE_SETUP -#line 454 "program_lexer.l" +#line 459 "program_lexer.l" { yylval->swiz_mask.swizzle = MAKE_SWIZZLE4(swiz_from_char(yytext[1]), swiz_from_char(yytext[2]), @@ -2343,15 +2433,15 @@ YY_RULE_SETUP return_token_or_DOT(require_ARB_fp, SWIZZLE); } YY_BREAK -case 178: +case 165: YY_RULE_SETUP -#line 463 "program_lexer.l" +#line 468 "program_lexer.l" { return DOT; } YY_BREAK -case 179: -/* rule 179 can match eol */ +case 166: +/* rule 166 can match eol */ YY_RULE_SETUP -#line 465 "program_lexer.l" +#line 470 "program_lexer.l" { yylloc->first_line++; yylloc->first_column = 1; @@ -2360,30 +2450,30 @@ YY_RULE_SETUP yylloc->position++; } YY_BREAK -case 180: +case 167: YY_RULE_SETUP -#line 472 "program_lexer.l" +#line 477 "program_lexer.l" /* eat whitespace */ ; YY_BREAK -case 181: +case 168: *yy_cp = yyg->yy_hold_char; /* undo effects of setting up yytext */ yyg->yy_c_buf_p = yy_cp -= 1; YY_DO_BEFORE_ACTION; /* set up yytext again */ YY_RULE_SETUP -#line 473 "program_lexer.l" +#line 478 "program_lexer.l" /* eat comments */ ; YY_BREAK -case 182: +case 169: YY_RULE_SETUP -#line 474 "program_lexer.l" +#line 479 "program_lexer.l" { return yytext[0]; } YY_BREAK -case 183: +case 170: YY_RULE_SETUP -#line 475 "program_lexer.l" +#line 480 "program_lexer.l" ECHO; YY_BREAK -#line 2383 "lex.yy.c" +#line 2477 "lex.yy.c" case YY_STATE_EOF(INITIAL): yyterminate(); @@ -2677,7 +2767,7 @@ static int yy_get_next_buffer (yyscan_t yyscanner) while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state ) { yy_current_state = (int) yy_def[yy_current_state]; - if ( yy_current_state >= 675 ) + if ( yy_current_state >= 850 ) yy_c = yy_meta[(unsigned int) yy_c]; } yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c]; @@ -2706,11 +2796,11 @@ static int yy_get_next_buffer (yyscan_t yyscanner) while ( yy_chk[yy_base[yy_current_state] + yy_c] != yy_current_state ) { yy_current_state = (int) yy_def[yy_current_state]; - if ( yy_current_state >= 675 ) + if ( yy_current_state >= 850 ) yy_c = yy_meta[(unsigned int) yy_c]; } yy_current_state = yy_nxt[yy_base[yy_current_state] + (unsigned int) yy_c]; - yy_is_jam = (yy_current_state == 674); + yy_is_jam = (yy_current_state == 849); return yy_is_jam ? 0 : yy_current_state; } @@ -3151,8 +3241,8 @@ YY_BUFFER_STATE yy_scan_string (yyconst char * yystr , yyscan_t yyscanner) /** Setup the input buffer state to scan the given bytes. The next call to yylex() will * scan from a @e copy of @a bytes. - * @param bytes the byte buffer to scan - * @param len the number of bytes in the buffer pointed to by @a bytes. + * @param yybytes the byte buffer to scan + * @param _yybytes_len the number of bytes in the buffer pointed to by @a bytes. * @param yyscanner The scanner object. * @return the newly allocated buffer state object. */ @@ -3558,7 +3648,7 @@ void yyfree (void * ptr , yyscan_t yyscanner) #define YYTABLES_NAME "yytables" -#line 475 "program_lexer.l" +#line 480 "program_lexer.l" diff --git a/src/mesa/shader/nvprogram.c b/src/mesa/shader/nvprogram.c index d6469b17bea..471a7358a2f 100644 --- a/src/mesa/shader/nvprogram.c +++ b/src/mesa/shader/nvprogram.c @@ -509,7 +509,79 @@ _mesa_GetVertexAttribPointervNV(GLuint index, GLenum pname, GLvoid **pointer) *pointer = (GLvoid *) ctx->Array.ArrayObj->VertexAttrib[index].Ptr; } +void +_mesa_emit_nv_temp_initialization(GLcontext *ctx, + struct gl_program *program) +{ + struct prog_instruction *inst; + int i; + + if (!ctx->Shader.EmitNVTempInitialization) + return; + + /* We'll swizzle up a zero temporary so we can use it for the + * ARL. + */ + if (program->NumTemporaries == 0) + program->NumTemporaries = 1; + + _mesa_insert_instructions(program, 0, program->NumTemporaries + 1); + + for (i = 0; i < program->NumTemporaries; i++) { + struct prog_instruction *inst = &program->Instructions[i]; + + inst->Opcode = OPCODE_SWZ; + inst->DstReg.File = PROGRAM_TEMPORARY; + inst->DstReg.Index = i; + inst->DstReg.WriteMask = WRITEMASK_XYZW; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = 0; + inst->SrcReg[0].Swizzle = MAKE_SWIZZLE4(SWIZZLE_ZERO, + SWIZZLE_ZERO, + SWIZZLE_ZERO, + SWIZZLE_ZERO); + } + + inst = &program->Instructions[i]; + inst->Opcode = OPCODE_ARL; + inst->DstReg.File = PROGRAM_ADDRESS; + inst->DstReg.Index = 0; + inst->DstReg.WriteMask = WRITEMASK_XYZW; + inst->SrcReg[0].File = PROGRAM_TEMPORARY; + inst->SrcReg[0].Index = 0; + inst->SrcReg[0].Swizzle = SWIZZLE_XXXX; + + if (program->NumAddressRegs == 0) + program->NumAddressRegs = 1; +} + +void +_mesa_setup_nv_temporary_count(GLcontext *ctx, struct gl_program *program) +{ + int i; + program->NumTemporaries = 0; + for (i = 0; i < program->NumInstructions; i++) { + struct prog_instruction *inst = &program->Instructions[i]; + + if (inst->DstReg.File == PROGRAM_TEMPORARY) { + program->NumTemporaries = MAX2(program->NumTemporaries, + inst->DstReg.Index + 1); + } + if (inst->SrcReg[0].File == PROGRAM_TEMPORARY) { + program->NumTemporaries = MAX2(program->NumTemporaries, + inst->SrcReg[0].Index + 1); + } + if (inst->SrcReg[1].File == PROGRAM_TEMPORARY) { + program->NumTemporaries = MAX2(program->NumTemporaries, + inst->SrcReg[1].Index + 1); + } + if (inst->SrcReg[2].File == PROGRAM_TEMPORARY) { + program->NumTemporaries = MAX2(program->NumTemporaries, + inst->SrcReg[2].Index + 1); + } + } +} /** * Load/parse/compile a program. diff --git a/src/mesa/shader/nvprogram.h b/src/mesa/shader/nvprogram.h index bfac165b5ed..8ee59661bd0 100644 --- a/src/mesa/shader/nvprogram.h +++ b/src/mesa/shader/nvprogram.h @@ -103,5 +103,11 @@ extern void GLAPIENTRY _mesa_GetProgramNamedParameterdvNV(GLuint id, GLsizei len, const GLubyte *name, GLdouble *params); +extern void +_mesa_setup_nv_temporary_count(GLcontext *ctx, struct gl_program *program); + +extern void +_mesa_emit_nv_temp_initialization(GLcontext *ctx, + struct gl_program *program); #endif diff --git a/src/mesa/shader/nvvertparse.c b/src/mesa/shader/nvvertparse.c index f5e2df26708..8574016050c 100644 --- a/src/mesa/shader/nvvertparse.c +++ b/src/mesa/shader/nvvertparse.c @@ -44,6 +44,7 @@ #include "nvprogram.h" #include "nvvertparse.h" #include "prog_instruction.h" +#include "prog_parameter.h" #include "prog_print.h" #include "program.h" @@ -1345,6 +1346,9 @@ _mesa_parse_nv_vertex_program(GLcontext *ctx, GLenum dstTarget, if (Parse_Program(&parseState, instBuffer)) { + gl_state_index state_tokens[STATE_LENGTH] = {0, 0, 0, 0, 0}; + int i; + /* successful parse! */ if (parseState.isStateProgram) { @@ -1398,6 +1402,29 @@ _mesa_parse_nv_vertex_program(GLcontext *ctx, GLenum dstTarget, _mesa_fprint_program_opt(stdout, &program->Base, PROG_PRINT_NV, 0); _mesa_printf("------------------------------\n"); #endif + + if (program->Base.Parameters) + _mesa_free_parameter_list(program->Base.Parameters); + + program->Base.Parameters = _mesa_new_parameter_list (); + program->Base.NumParameters = 0; + + state_tokens[0] = STATE_VERTEX_PROGRAM; + state_tokens[1] = STATE_ENV; + /* Add refs to all of the potential params, in order. If we want to not + * upload everything, _mesa_layout_parameters is the answer. + */ + for (i = 0; i < MAX_NV_VERTEX_PROGRAM_PARAMS; i++) { + GLint index; + state_tokens[2] = i; + index = _mesa_add_state_reference(program->Base.Parameters, + state_tokens); + assert(index == i); + } + program->Base.NumParameters = program->Base.Parameters->NumParameters; + + _mesa_setup_nv_temporary_count(ctx, &program->Base); + _mesa_emit_nv_temp_initialization(ctx, &program->Base); } else { /* Error! */ diff --git a/src/mesa/shader/prog_execute.c b/src/mesa/shader/prog_execute.c index c8a762f8ff7..192d39aed1e 100644 --- a/src/mesa/shader/prog_execute.c +++ b/src/mesa/shader/prog_execute.c @@ -939,7 +939,7 @@ _mesa_execute_program(GLcontext * ctx, /* The fast LOG2 macro doesn't meet the precision requirements. */ if (a[0] == 0.0F) { - val = 0.0F; + val = -FLT_MAX; } else { val = log(a[0]) * 1.442695F; @@ -1555,17 +1555,12 @@ _mesa_execute_program(GLcontext * ctx, case OPCODE_TXB: /* GL_ARB_fragment_program only */ /* Texel lookup with LOD bias */ { - const GLuint unit = machine->Samplers[inst->TexSrcUnit]; - const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; GLfloat texcoord[4], color[4], lodBias; fetch_vector4(&inst->SrcReg[0], machine, texcoord); /* texcoord[3] is the bias to add to lambda */ - lodBias = texUnit->LodBias + texcoord[3]; - if (texUnit->_Current) { - lodBias += texUnit->_Current->LodBias; - } + lodBias = texcoord[3]; fetch_texel(ctx, machine, inst, texcoord, lodBias, color); diff --git a/src/mesa/shader/prog_parameter.c b/src/mesa/shader/prog_parameter.c index 6b9e73b2cbd..2f029b02e50 100644 --- a/src/mesa/shader/prog_parameter.c +++ b/src/mesa/shader/prog_parameter.c @@ -100,6 +100,7 @@ _mesa_free_parameter_list(struct gl_program_parameter_list *paramList) * \param type type of parameter, such as * \param name the parameter name, will be duplicated/copied! * \param size number of elements in 'values' vector (1..4, or more) + * \param datatype GL_FLOAT, GL_FLOAT_VECx, GL_INT, GL_INT_VECx or GL_NONE. * \param values initial parameter value, up to 4 GLfloats, or NULL * \param state state indexes, or NULL * \return index of new parameter in the list, or -1 if error (out of mem) diff --git a/src/mesa/shader/prog_parameter.h b/src/mesa/shader/prog_parameter.h index d1fcf47e61b..699cb0c7356 100644 --- a/src/mesa/shader/prog_parameter.h +++ b/src/mesa/shader/prog_parameter.h @@ -56,7 +56,13 @@ struct gl_program_parameter const char *Name; /**< Null-terminated string */ gl_register_file Type; /**< PROGRAM_NAMED_PARAM, CONSTANT or STATE_VAR */ GLenum DataType; /**< GL_FLOAT, GL_FLOAT_VEC2, etc */ - GLuint Size; /**< Number of components (1..4) */ + /** + * Number of components (1..4), or more. + * If the number of components is greater than 4, + * this parameter is part of a larger uniform like a GLSL matrix or array. + * The next program parameter's Size will be Size-4 of this parameter. + */ + GLuint Size; GLboolean Used; /**< Helper flag for GLSL uniform tracking */ GLboolean Initialized; /**< Has the ParameterValue[] been set? */ GLbitfield Flags; /**< Bitmask of PROG_PARAM_*_BIT */ diff --git a/src/mesa/shader/program.c b/src/mesa/shader/program.c index 2cd6eb8a389..532adf4d360 100644 --- a/src/mesa/shader/program.c +++ b/src/mesa/shader/program.c @@ -351,13 +351,6 @@ _mesa_delete_program(GLcontext *ctx, struct gl_program *prog) _mesa_free_parameter_list(prog->Attributes); } - /* XXX this is a little ugly */ - if (prog->Target == GL_VERTEX_PROGRAM_ARB) { - struct gl_vertex_program *vprog = (struct gl_vertex_program *) prog; - if (vprog->TnlData) - _mesa_free(vprog->TnlData); - } - _mesa_free(prog); } diff --git a/src/mesa/shader/program_lexer.l b/src/mesa/shader/program_lexer.l index 612f99a42d0..8498c3d8fc0 100644 --- a/src/mesa/shader/program_lexer.l +++ b/src/mesa/shader/program_lexer.l @@ -25,11 +25,13 @@ #include "prog_instruction.h" #include "prog_statevars.h" +#include "symbol_table.h" #include "program_parser.h" #include "program_parse.tab.h" #define require_ARB_vp (yyextra->mode == ARB_vertex) #define require_ARB_fp (yyextra->mode == ARB_fragment) +#define require_NV_fp (yyextra->option.NV_fragment) #define require_shadow (yyextra->option.Shadow) #define require_rect (yyextra->option.TexRect) #define require_texarray (yyextra->option.TexArray) @@ -43,8 +45,7 @@ if (condition) { \ return token; \ } else { \ - yylval->string = strdup(yytext); \ - return IDENTIFIER; \ + return handle_ident(yyextra, yytext, yylval); \ } \ } while (0) @@ -59,15 +60,16 @@ } while (0) -#define return_opcode(condition, token, opcode, sat) \ +#define return_opcode(condition, token, opcode, len) \ do { \ - if (condition) { \ + if (condition && \ + _mesa_parse_instruction_suffix(yyextra, \ + yytext + len, \ + & yylval->temp_inst)) { \ yylval->temp_inst.Opcode = OPCODE_ ## opcode; \ - yylval->temp_inst.SaturateMode = SATURATE_ ## sat; \ return token; \ } else { \ - yylval->string = strdup(yytext); \ - return IDENTIFIER; \ + return handle_ident(yyextra, yytext, yylval); \ } \ } while (0) @@ -116,6 +118,15 @@ swiz_from_char(char c) return 0; } +static int +handle_ident(struct asm_parser_state *state, const char *text, YYSTYPE *lval) +{ + lval->string = strdup(text); + + return (_mesa_symbol_table_find_symbol(state->st, 0, text) == NULL) + ? IDENTIFIER : USED_IDENTIFIER; +} + #define YY_USER_ACTION \ do { \ yylloc->first_column = yylloc->last_column; \ @@ -136,6 +147,11 @@ exp [Ee][-+]?[0-9]+ frac "."[0-9]+ dot "."[ \t]* +sz [HRX]? +szf [HR]? +cc C? +sat (_SAT)? + %option bison-bridge bison-locations reentrant noyywrap %% @@ -153,86 +169,74 @@ OUTPUT { return OUTPUT; } PARAM { return PARAM; } TEMP { yylval->integer = at_temp; return TEMP; } -ABS { return_opcode( 1, VECTOR_OP, ABS, OFF); } -ABS_SAT { return_opcode(require_ARB_fp, VECTOR_OP, ABS, ZERO_ONE); } -ADD { return_opcode( 1, BIN_OP, ADD, OFF); } -ADD_SAT { return_opcode(require_ARB_fp, BIN_OP, ADD, ZERO_ONE); } -ARL { return_opcode(require_ARB_vp, ARL, ARL, OFF); } - -CMP { return_opcode(require_ARB_fp, TRI_OP, CMP, OFF); } -CMP_SAT { return_opcode(require_ARB_fp, TRI_OP, CMP, ZERO_ONE); } -COS { return_opcode(require_ARB_fp, SCALAR_OP, COS, OFF); } -COS_SAT { return_opcode(require_ARB_fp, SCALAR_OP, COS, ZERO_ONE); } - -DP3 { return_opcode( 1, BIN_OP, DP3, OFF); } -DP3_SAT { return_opcode(require_ARB_fp, BIN_OP, DP3, ZERO_ONE); } -DP4 { return_opcode( 1, BIN_OP, DP4, OFF); } -DP4_SAT { return_opcode(require_ARB_fp, BIN_OP, DP4, ZERO_ONE); } -DPH { return_opcode( 1, BIN_OP, DPH, OFF); } -DPH_SAT { return_opcode(require_ARB_fp, BIN_OP, DPH, ZERO_ONE); } -DST { return_opcode( 1, BIN_OP, DST, OFF); } -DST_SAT { return_opcode(require_ARB_fp, BIN_OP, DST, ZERO_ONE); } - -EX2 { return_opcode( 1, SCALAR_OP, EX2, OFF); } -EX2_SAT { return_opcode(require_ARB_fp, SCALAR_OP, EX2, ZERO_ONE); } -EXP { return_opcode(require_ARB_vp, SCALAR_OP, EXP, OFF); } - -FLR { return_opcode( 1, VECTOR_OP, FLR, OFF); } -FLR_SAT { return_opcode(require_ARB_fp, VECTOR_OP, FLR, ZERO_ONE); } -FRC { return_opcode( 1, VECTOR_OP, FRC, OFF); } -FRC_SAT { return_opcode(require_ARB_fp, VECTOR_OP, FRC, ZERO_ONE); } - -KIL { return_opcode(require_ARB_fp, KIL, KIL, OFF); } - -LIT { return_opcode( 1, VECTOR_OP, LIT, OFF); } -LIT_SAT { return_opcode(require_ARB_fp, VECTOR_OP, LIT, ZERO_ONE); } -LG2 { return_opcode( 1, SCALAR_OP, LG2, OFF); } -LG2_SAT { return_opcode(require_ARB_fp, SCALAR_OP, LG2, ZERO_ONE); } -LOG { return_opcode(require_ARB_vp, SCALAR_OP, LOG, OFF); } -LRP { return_opcode(require_ARB_fp, TRI_OP, LRP, OFF); } -LRP_SAT { return_opcode(require_ARB_fp, TRI_OP, LRP, ZERO_ONE); } - -MAD { return_opcode( 1, TRI_OP, MAD, OFF); } -MAD_SAT { return_opcode(require_ARB_fp, TRI_OP, MAD, ZERO_ONE); } -MAX { return_opcode( 1, BIN_OP, MAX, OFF); } -MAX_SAT { return_opcode(require_ARB_fp, BIN_OP, MAX, ZERO_ONE); } -MIN { return_opcode( 1, BIN_OP, MIN, OFF); } -MIN_SAT { return_opcode(require_ARB_fp, BIN_OP, MIN, ZERO_ONE); } -MOV { return_opcode( 1, VECTOR_OP, MOV, OFF); } -MOV_SAT { return_opcode(require_ARB_fp, VECTOR_OP, MOV, ZERO_ONE); } -MUL { return_opcode( 1, BIN_OP, MUL, OFF); } -MUL_SAT { return_opcode(require_ARB_fp, BIN_OP, MUL, ZERO_ONE); } - -POW { return_opcode( 1, BINSC_OP, POW, OFF); } -POW_SAT { return_opcode(require_ARB_fp, BINSC_OP, POW, ZERO_ONE); } - -RCP { return_opcode( 1, SCALAR_OP, RCP, OFF); } -RCP_SAT { return_opcode(require_ARB_fp, SCALAR_OP, RCP, ZERO_ONE); } -RSQ { return_opcode( 1, SCALAR_OP, RSQ, OFF); } -RSQ_SAT { return_opcode(require_ARB_fp, SCALAR_OP, RSQ, ZERO_ONE); } - -SCS { return_opcode(require_ARB_fp, SCALAR_OP, SCS, OFF); } -SCS_SAT { return_opcode(require_ARB_fp, SCALAR_OP, SCS, ZERO_ONE); } -SGE { return_opcode( 1, BIN_OP, SGE, OFF); } -SGE_SAT { return_opcode(require_ARB_fp, BIN_OP, SGE, ZERO_ONE); } -SIN { return_opcode(require_ARB_fp, SCALAR_OP, SIN, OFF); } -SIN_SAT { return_opcode(require_ARB_fp, SCALAR_OP, SIN, ZERO_ONE); } -SLT { return_opcode( 1, BIN_OP, SLT, OFF); } -SLT_SAT { return_opcode(require_ARB_fp, BIN_OP, SLT, ZERO_ONE); } -SUB { return_opcode( 1, BIN_OP, SUB, OFF); } -SUB_SAT { return_opcode(require_ARB_fp, BIN_OP, SUB, ZERO_ONE); } -SWZ { return_opcode( 1, SWZ, SWZ, OFF); } -SWZ_SAT { return_opcode(require_ARB_fp, SWZ, SWZ, ZERO_ONE); } - -TEX { return_opcode(require_ARB_fp, SAMPLE_OP, TEX, OFF); } -TEX_SAT { return_opcode(require_ARB_fp, SAMPLE_OP, TEX, ZERO_ONE); } -TXB { return_opcode(require_ARB_fp, SAMPLE_OP, TXB, OFF); } -TXB_SAT { return_opcode(require_ARB_fp, SAMPLE_OP, TXB, ZERO_ONE); } -TXP { return_opcode(require_ARB_fp, SAMPLE_OP, TXP, OFF); } -TXP_SAT { return_opcode(require_ARB_fp, SAMPLE_OP, TXP, ZERO_ONE); } - -XPD { return_opcode( 1, BIN_OP, XPD, OFF); } -XPD_SAT { return_opcode(require_ARB_fp, BIN_OP, XPD, ZERO_ONE); } +ABS{sz}{cc}{sat} { return_opcode( 1, VECTOR_OP, ABS, 3); } +ADD{sz}{cc}{sat} { return_opcode( 1, BIN_OP, ADD, 3); } +ARL { return_opcode(require_ARB_vp, ARL, ARL, 3); } + +CMP{sat} { return_opcode(require_ARB_fp, TRI_OP, CMP, 3); } +COS{szf}{cc}{sat} { return_opcode(require_ARB_fp, SCALAR_OP, COS, 3); } + +DDX{szf}{cc}{sat} { return_opcode(require_NV_fp, VECTOR_OP, DDX, 3); } +DDY{szf}{cc}{sat} { return_opcode(require_NV_fp, VECTOR_OP, DDY, 3); } +DP3{sz}{cc}{sat} { return_opcode( 1, BIN_OP, DP3, 3); } +DP4{sz}{cc}{sat} { return_opcode( 1, BIN_OP, DP4, 3); } +DPH{sz}{cc}{sat} { return_opcode( 1, BIN_OP, DPH, 3); } +DST{szf}{cc}{sat} { return_opcode( 1, BIN_OP, DST, 3); } + +EX2{szf}{cc}{sat} { return_opcode( 1, SCALAR_OP, EX2, 3); } +EXP { return_opcode(require_ARB_vp, SCALAR_OP, EXP, 3); } + +FLR{sz}{cc}{sat} { return_opcode( 1, VECTOR_OP, FLR, 3); } +FRC{sz}{cc}{sat} { return_opcode( 1, VECTOR_OP, FRC, 3); } + +KIL { return_opcode(require_ARB_fp, KIL, KIL, 3); } + +LIT{szf}{cc}{sat} { return_opcode( 1, VECTOR_OP, LIT, 3); } +LG2{szf}{cc}{sat} { return_opcode( 1, SCALAR_OP, LG2, 3); } +LOG { return_opcode(require_ARB_vp, SCALAR_OP, LOG, 3); } +LRP{sz}{cc}{sat} { return_opcode(require_ARB_fp, TRI_OP, LRP, 3); } + +MAD{sz}{cc}{sat} { return_opcode( 1, TRI_OP, MAD, 3); } +MAX{sz}{cc}{sat} { return_opcode( 1, BIN_OP, MAX, 3); } +MIN{sz}{cc}{sat} { return_opcode( 1, BIN_OP, MIN, 3); } +MOV{sz}{cc}{sat} { return_opcode( 1, VECTOR_OP, MOV, 3); } +MUL{sz}{cc}{sat} { return_opcode( 1, BIN_OP, MUL, 3); } + +PK2H { return_opcode(require_NV_fp, VECTOR_OP, PK2H, 4); } +PK2US { return_opcode(require_NV_fp, VECTOR_OP, PK2US, 5); } +PK4B { return_opcode(require_NV_fp, VECTOR_OP, PK4B, 4); } +PK4UB { return_opcode(require_NV_fp, VECTOR_OP, PK4UB, 5); } +POW{szf}{cc}{sat} { return_opcode( 1, BINSC_OP, POW, 3); } + +RCP{szf}{cc}{sat} { return_opcode( 1, SCALAR_OP, RCP, 3); } +RFL{szf}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, RFL, 3); } +RSQ{szf}{cc}{sat} { return_opcode( 1, SCALAR_OP, RSQ, 3); } + +SCS{sat} { return_opcode(require_ARB_fp, SCALAR_OP, SCS, 3); } +SEQ{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SEQ, 3); } +SFL{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SFL, 3); } +SGE{sz}{cc}{sat} { return_opcode( 1, BIN_OP, SGE, 3); } +SGT{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SGT, 3); } +SIN{szf}{cc}{sat} { return_opcode(require_ARB_fp, SCALAR_OP, SIN, 3); } +SLE{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SLE, 3); } +SLT{sz}{cc}{sat} { return_opcode( 1, BIN_OP, SLT, 3); } +SNE{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, SNE, 3); } +STR{sz}{cc}{sat} { return_opcode(require_NV_fp, BIN_OP, STR, 3); } +SUB{sz}{cc}{sat} { return_opcode( 1, BIN_OP, SUB, 3); } +SWZ{sat} { return_opcode( 1, SWZ, SWZ, 3); } + +TEX{cc}{sat} { return_opcode(require_ARB_fp, SAMPLE_OP, TEX, 3); } +TXB{cc}{sat} { return_opcode(require_ARB_fp, SAMPLE_OP, TXB, 3); } +TXD{cc}{sat} { return_opcode(require_NV_fp, TXD_OP, TXD, 3); } +TXP{cc}{sat} { return_opcode(require_ARB_fp, SAMPLE_OP, TXP, 3); } + +UP2H{cc}{sat} { return_opcode(require_NV_fp, SCALAR_OP, UP2H, 4); } +UP2US{cc}{sat} { return_opcode(require_NV_fp, SCALAR_OP, UP2US, 5); } +UP4B{cc}{sat} { return_opcode(require_NV_fp, SCALAR_OP, UP4B, 4); } +UP4UB{cc}{sat} { return_opcode(require_NV_fp, SCALAR_OP, UP4UB, 5); } + +X2D{szf}{cc}{sat} { return_opcode(require_NV_fp, TRI_OP, X2D, 3); } +XPD{sat} { return_opcode( 1, BIN_OP, XPD, 3); } vertex { return_token_or_IDENTIFIER(require_ARB_vp, VERTEX); } fragment { return_token_or_IDENTIFIER(require_ARB_fp, FRAGMENT); } @@ -310,10 +314,7 @@ ARRAY2D { return_token_or_IDENTIFIER(require_ARB_fp && require ARRAYSHADOW1D { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW1D); } ARRAYSHADOW2D { return_token_or_IDENTIFIER(require_ARB_fp && require_shadow && require_texarray, TEX_ARRAYSHADOW2D); } -[_a-zA-Z$][_a-zA-Z0-9$]* { - yylval->string = strdup(yytext); - return IDENTIFIER; -} +[_a-zA-Z$][_a-zA-Z0-9$]* { return handle_ident(yyextra, yytext, yylval); } ".." { return DOT_DOT; } diff --git a/src/mesa/shader/program_parse.tab.c b/src/mesa/shader/program_parse.tab.c index 2de950b73b6..c0f50147bd9 100644 --- a/src/mesa/shader/program_parse.tab.c +++ b/src/mesa/shader/program_parse.tab.c @@ -139,10 +139,19 @@ static void init_dst_reg(struct prog_dst_register *r); static void init_src_reg(struct asm_src_register *r); +static void asm_instruction_set_operands(struct asm_instruction *inst, + const struct prog_dst_register *dst, const struct asm_src_register *src0, + const struct asm_src_register *src1, const struct asm_src_register *src2); + static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, const struct prog_dst_register *dst, const struct asm_src_register *src0, const struct asm_src_register *src1, const struct asm_src_register *src2); +static struct asm_instruction *asm_instruction_copy_ctor( + const struct prog_instruction *base, const struct prog_dst_register *dst, + const struct asm_src_register *src0, const struct asm_src_register *src1, + const struct asm_src_register *src2); + #ifndef FALSE #define FALSE 0 #define TRUE (!FALSE) @@ -170,7 +179,7 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, /* Line 189 of yacc.c */ -#line 174 "program_parse.tab.c" +#line 183 "program_parse.tab.c" /* Enabling traces. */ #ifndef YYDEBUG @@ -216,90 +225,92 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, ARL = 274, KIL = 275, SWZ = 276, - INTEGER = 277, - REAL = 278, - AMBIENT = 279, - ATTENUATION = 280, - BACK = 281, - CLIP = 282, - COLOR = 283, - DEPTH = 284, - DIFFUSE = 285, - DIRECTION = 286, - EMISSION = 287, - ENV = 288, - EYE = 289, - FOG = 290, - FOGCOORD = 291, - FRAGMENT = 292, - FRONT = 293, - HALF = 294, - INVERSE = 295, - INVTRANS = 296, - LIGHT = 297, - LIGHTMODEL = 298, - LIGHTPROD = 299, - LOCAL = 300, - MATERIAL = 301, - MAT_PROGRAM = 302, - MATRIX = 303, - MATRIXINDEX = 304, - MODELVIEW = 305, - MVP = 306, - NORMAL = 307, - OBJECT = 308, - PALETTE = 309, - PARAMS = 310, - PLANE = 311, - POINT_TOK = 312, - POINTSIZE = 313, - POSITION = 314, - PRIMARY = 315, - PROGRAM = 316, - PROJECTION = 317, - RANGE = 318, - RESULT = 319, - ROW = 320, - SCENECOLOR = 321, - SECONDARY = 322, - SHININESS = 323, - SIZE_TOK = 324, - SPECULAR = 325, - SPOT = 326, - STATE = 327, - TEXCOORD = 328, - TEXENV = 329, - TEXGEN = 330, - TEXGEN_Q = 331, - TEXGEN_R = 332, - TEXGEN_S = 333, - TEXGEN_T = 334, - TEXTURE = 335, - TRANSPOSE = 336, - TEXTURE_UNIT = 337, - TEX_1D = 338, - TEX_2D = 339, - TEX_3D = 340, - TEX_CUBE = 341, - TEX_RECT = 342, - TEX_SHADOW1D = 343, - TEX_SHADOW2D = 344, - TEX_SHADOWRECT = 345, - TEX_ARRAY1D = 346, - TEX_ARRAY2D = 347, - TEX_ARRAYSHADOW1D = 348, - TEX_ARRAYSHADOW2D = 349, - VERTEX = 350, - VTXATTRIB = 351, - WEIGHT = 352, - IDENTIFIER = 353, - MASK4 = 354, - MASK3 = 355, - MASK2 = 356, - MASK1 = 357, - SWIZZLE = 358, - DOT_DOT = 359, - DOT = 360 + TXD_OP = 277, + INTEGER = 278, + REAL = 279, + AMBIENT = 280, + ATTENUATION = 281, + BACK = 282, + CLIP = 283, + COLOR = 284, + DEPTH = 285, + DIFFUSE = 286, + DIRECTION = 287, + EMISSION = 288, + ENV = 289, + EYE = 290, + FOG = 291, + FOGCOORD = 292, + FRAGMENT = 293, + FRONT = 294, + HALF = 295, + INVERSE = 296, + INVTRANS = 297, + LIGHT = 298, + LIGHTMODEL = 299, + LIGHTPROD = 300, + LOCAL = 301, + MATERIAL = 302, + MAT_PROGRAM = 303, + MATRIX = 304, + MATRIXINDEX = 305, + MODELVIEW = 306, + MVP = 307, + NORMAL = 308, + OBJECT = 309, + PALETTE = 310, + PARAMS = 311, + PLANE = 312, + POINT_TOK = 313, + POINTSIZE = 314, + POSITION = 315, + PRIMARY = 316, + PROGRAM = 317, + PROJECTION = 318, + RANGE = 319, + RESULT = 320, + ROW = 321, + SCENECOLOR = 322, + SECONDARY = 323, + SHININESS = 324, + SIZE_TOK = 325, + SPECULAR = 326, + SPOT = 327, + STATE = 328, + TEXCOORD = 329, + TEXENV = 330, + TEXGEN = 331, + TEXGEN_Q = 332, + TEXGEN_R = 333, + TEXGEN_S = 334, + TEXGEN_T = 335, + TEXTURE = 336, + TRANSPOSE = 337, + TEXTURE_UNIT = 338, + TEX_1D = 339, + TEX_2D = 340, + TEX_3D = 341, + TEX_CUBE = 342, + TEX_RECT = 343, + TEX_SHADOW1D = 344, + TEX_SHADOW2D = 345, + TEX_SHADOWRECT = 346, + TEX_ARRAY1D = 347, + TEX_ARRAY2D = 348, + TEX_ARRAYSHADOW1D = 349, + TEX_ARRAYSHADOW2D = 350, + VERTEX = 351, + VTXATTRIB = 352, + WEIGHT = 353, + IDENTIFIER = 354, + USED_IDENTIFIER = 355, + MASK4 = 356, + MASK3 = 357, + MASK2 = 358, + MASK1 = 359, + SWIZZLE = 360, + DOT_DOT = 361, + DOT = 362 }; #endif @@ -310,7 +321,7 @@ typedef union YYSTYPE { /* Line 214 of yacc.c */ -#line 107 "program_parse.y" +#line 116 "program_parse.y" struct asm_instruction *inst; struct asm_symbol *sym; @@ -339,7 +350,7 @@ typedef union YYSTYPE /* Line 214 of yacc.c */ -#line 343 "program_parse.tab.c" +#line 354 "program_parse.tab.c" } YYSTYPE; # define YYSTYPE_IS_TRIVIAL 1 # define yystype YYSTYPE /* obsolescent; will be withdrawn */ @@ -363,14 +374,14 @@ typedef struct YYLTYPE /* Copy the second part of user declarations. */ /* Line 264 of yacc.c */ -#line 249 "program_parse.y" +#line 261 "program_parse.y" extern int yylex(YYSTYPE *yylval_param, YYLTYPE *yylloc_param, void *yyscanner); /* Line 264 of yacc.c */ -#line 374 "program_parse.tab.c" +#line 385 "program_parse.tab.c" #ifdef short # undef short @@ -587,20 +598,20 @@ union yyalloc /* YYFINAL -- State number of the termination state. */ #define YYFINAL 5 /* YYLAST -- Last index in YYTABLE. */ -#define YYLAST 342 +#define YYLAST 396 /* YYNTOKENS -- Number of terminals. */ -#define YYNTOKENS 115 +#define YYNTOKENS 120 /* YYNNTS -- Number of nonterminals. */ -#define YYNNTS 134 +#define YYNNTS 143 /* YYNRULES -- Number of rules. */ -#define YYNRULES 264 +#define YYNRULES 282 /* YYNRULES -- Number of states. */ -#define YYNSTATES 436 +#define YYNSTATES 475 /* YYTRANSLATE(YYLEX) -- Bison symbol number corresponding to YYLEX. */ #define YYUNDEFTOK 2 -#define YYMAXUTOK 360 +#define YYMAXUTOK 362 #define YYTRANSLATE(YYX) \ ((unsigned int) (YYX) <= YYMAXUTOK ? yytranslate[YYX] : YYUNDEFTOK) @@ -612,15 +623,15 @@ static const yytype_uint8 yytranslate[] = 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, - 2, 2, 2, 110, 107, 111, 2, 2, 2, 2, - 2, 2, 2, 2, 2, 2, 2, 2, 2, 106, - 2, 112, 2, 2, 2, 2, 2, 2, 2, 2, + 115, 116, 2, 113, 109, 114, 2, 2, 2, 2, + 2, 2, 2, 2, 2, 2, 2, 2, 2, 108, + 2, 117, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, - 2, 108, 2, 109, 2, 2, 2, 2, 2, 2, + 2, 111, 2, 112, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, - 2, 2, 2, 113, 2, 114, 2, 2, 2, 2, + 2, 2, 2, 118, 110, 119, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, @@ -644,7 +655,7 @@ static const yytype_uint8 yytranslate[] = 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, - 105 + 105, 106, 107 }; #if YYDEBUG @@ -654,142 +665,152 @@ static const yytype_uint16 yyprhs[] = { 0, 0, 3, 8, 10, 12, 15, 16, 20, 23, 24, 27, 30, 32, 34, 36, 38, 40, 42, 44, - 46, 48, 50, 52, 57, 62, 67, 74, 81, 90, - 99, 102, 105, 107, 109, 111, 113, 115, 117, 119, - 121, 123, 125, 127, 129, 136, 140, 144, 147, 150, - 158, 161, 163, 165, 167, 169, 174, 176, 178, 180, - 182, 184, 186, 188, 192, 193, 196, 199, 201, 203, - 205, 207, 209, 211, 213, 215, 217, 218, 220, 222, - 224, 226, 227, 229, 231, 233, 235, 237, 239, 244, - 247, 250, 252, 255, 257, 260, 262, 265, 270, 275, - 277, 278, 282, 284, 286, 289, 291, 294, 296, 298, - 302, 309, 310, 312, 315, 320, 322, 326, 328, 330, - 332, 334, 336, 338, 340, 342, 344, 346, 349, 352, - 355, 358, 361, 364, 367, 370, 373, 376, 379, 382, - 386, 388, 390, 392, 398, 400, 402, 404, 407, 409, - 411, 414, 416, 419, 426, 428, 432, 434, 436, 438, - 440, 442, 447, 449, 451, 453, 455, 457, 459, 462, - 464, 466, 472, 474, 477, 479, 481, 487, 490, 491, - 498, 502, 503, 505, 507, 509, 511, 513, 516, 518, - 520, 523, 528, 533, 534, 538, 540, 542, 544, 547, - 549, 551, 553, 555, 561, 563, 567, 573, 579, 581, - 585, 591, 593, 595, 597, 599, 601, 603, 605, 607, - 609, 613, 619, 627, 637, 640, 643, 645, 647, 648, - 649, 653, 654, 658, 662, 664, 669, 672, 675, 678, - 681, 685, 688, 692, 693, 695, 697, 698, 700, 702, - 703, 705, 707, 708, 710, 712, 713, 717, 718, 722, - 723, 727, 729, 731, 733 + 46, 48, 50, 52, 54, 59, 64, 69, 76, 83, + 92, 101, 104, 107, 120, 123, 125, 127, 129, 131, + 133, 135, 137, 139, 141, 143, 145, 147, 154, 157, + 162, 165, 167, 171, 177, 181, 184, 192, 195, 197, + 199, 201, 203, 208, 210, 212, 214, 216, 218, 220, + 222, 226, 227, 230, 233, 235, 237, 239, 241, 243, + 245, 247, 249, 251, 252, 254, 256, 258, 260, 261, + 265, 269, 270, 273, 276, 278, 280, 282, 284, 286, + 288, 290, 292, 297, 300, 303, 305, 308, 310, 313, + 315, 318, 323, 328, 330, 331, 335, 337, 339, 342, + 344, 347, 349, 351, 355, 362, 363, 365, 368, 373, + 375, 379, 381, 383, 385, 387, 389, 391, 393, 395, + 397, 399, 402, 405, 408, 411, 414, 417, 420, 423, + 426, 429, 432, 435, 439, 441, 443, 445, 451, 453, + 455, 457, 460, 462, 464, 467, 469, 472, 479, 481, + 485, 487, 489, 491, 493, 495, 500, 502, 504, 506, + 508, 510, 512, 515, 517, 519, 525, 527, 530, 532, + 534, 540, 543, 544, 551, 555, 556, 558, 560, 562, + 564, 566, 569, 571, 573, 576, 581, 586, 587, 591, + 593, 595, 597, 600, 602, 604, 606, 608, 614, 616, + 620, 626, 632, 634, 638, 644, 646, 648, 650, 652, + 654, 656, 658, 660, 662, 666, 672, 680, 690, 693, + 696, 698, 700, 701, 702, 707, 709, 710, 711, 715, + 719, 721, 727, 730, 733, 736, 739, 743, 746, 750, + 751, 753, 755, 756, 758, 760, 761, 763, 765, 766, + 768, 770, 771, 775, 776, 780, 781, 785, 787, 789, + 791, 796, 798 }; /* YYRHS -- A `-1'-separated list of the rules' RHS. */ static const yytype_int16 yyrhs[] = { - 116, 0, -1, 117, 118, 120, 12, -1, 3, -1, - 4, -1, 118, 119, -1, -1, 8, 98, 106, -1, - 120, 121, -1, -1, 122, 106, -1, 158, 106, -1, - 123, -1, 124, -1, 125, -1, 126, -1, 127, -1, - 128, -1, 129, -1, 130, -1, 135, -1, 131, -1, - 132, -1, 19, 139, 107, 136, -1, 18, 138, 107, - 137, -1, 16, 138, 107, 136, -1, 14, 138, 107, - 136, 107, 136, -1, 13, 138, 107, 137, 107, 137, - -1, 17, 138, 107, 137, 107, 137, 107, 137, -1, - 15, 138, 107, 137, 107, 133, 107, 134, -1, 20, - 137, -1, 82, 243, -1, 83, -1, 84, -1, 85, - -1, 86, -1, 87, -1, 88, -1, 89, -1, 90, - -1, 91, -1, 92, -1, 93, -1, 94, -1, 21, - 138, 107, 143, 107, 140, -1, 229, 143, 155, -1, - 229, 143, 156, -1, 144, 157, -1, 152, 154, -1, - 141, 107, 141, 107, 141, 107, 141, -1, 229, 142, - -1, 22, -1, 98, -1, 98, -1, 160, -1, 145, - 108, 146, 109, -1, 174, -1, 236, -1, 98, -1, - 98, -1, 147, -1, 148, -1, 22, -1, 152, 153, - 149, -1, -1, 110, 150, -1, 111, 151, -1, 22, - -1, 22, -1, 98, -1, 102, -1, 102, -1, 102, - -1, 102, -1, 99, -1, 103, -1, -1, 99, -1, - 100, -1, 101, -1, 102, -1, -1, 159, -1, 166, - -1, 230, -1, 232, -1, 235, -1, 248, -1, 7, - 98, 112, 160, -1, 95, 161, -1, 37, 165, -1, - 59, -1, 97, 163, -1, 52, -1, 28, 241, -1, - 36, -1, 73, 242, -1, 49, 108, 164, 109, -1, - 96, 108, 162, 109, -1, 22, -1, -1, 108, 164, - 109, -1, 22, -1, 59, -1, 28, 241, -1, 36, - -1, 73, 242, -1, 167, -1, 168, -1, 10, 98, - 170, -1, 10, 98, 108, 169, 109, 171, -1, -1, - 22, -1, 112, 173, -1, 112, 113, 172, 114, -1, - 175, -1, 172, 107, 175, -1, 177, -1, 213, -1, - 223, -1, 177, -1, 213, -1, 224, -1, 176, -1, - 214, -1, 223, -1, 177, -1, 72, 201, -1, 72, - 178, -1, 72, 180, -1, 72, 183, -1, 72, 185, - -1, 72, 191, -1, 72, 187, -1, 72, 194, -1, - 72, 196, -1, 72, 198, -1, 72, 200, -1, 72, - 212, -1, 46, 240, 179, -1, 189, -1, 32, -1, - 68, -1, 42, 108, 190, 109, 181, -1, 189, -1, - 59, -1, 25, -1, 71, 182, -1, 39, -1, 31, - -1, 43, 184, -1, 24, -1, 240, 66, -1, 44, - 108, 190, 109, 240, 186, -1, 189, -1, 74, 244, - 188, -1, 28, -1, 24, -1, 30, -1, 70, -1, - 22, -1, 75, 242, 192, 193, -1, 34, -1, 53, - -1, 78, -1, 79, -1, 77, -1, 76, -1, 35, - 195, -1, 28, -1, 55, -1, 27, 108, 197, 109, - 56, -1, 22, -1, 57, 199, -1, 69, -1, 25, - -1, 203, 65, 108, 206, 109, -1, 203, 202, -1, - -1, 65, 108, 206, 104, 206, 109, -1, 48, 207, - 204, -1, -1, 205, -1, 40, -1, 81, -1, 41, - -1, 22, -1, 50, 208, -1, 62, -1, 51, -1, - 80, 242, -1, 54, 108, 210, 109, -1, 47, 108, - 211, 109, -1, -1, 108, 209, 109, -1, 22, -1, - 22, -1, 22, -1, 29, 63, -1, 217, -1, 220, - -1, 215, -1, 218, -1, 61, 33, 108, 216, 109, - -1, 221, -1, 221, 104, 221, -1, 61, 33, 108, - 221, 109, -1, 61, 45, 108, 219, 109, -1, 222, - -1, 222, 104, 222, -1, 61, 45, 108, 222, 109, - -1, 22, -1, 22, -1, 225, -1, 227, -1, 226, - -1, 227, -1, 228, -1, 23, -1, 22, -1, 113, - 228, 114, -1, 113, 228, 107, 228, 114, -1, 113, - 228, 107, 228, 107, 228, 114, -1, 113, 228, 107, - 228, 107, 228, 107, 228, 114, -1, 229, 23, -1, - 229, 22, -1, 110, -1, 111, -1, -1, -1, 11, - 231, 234, -1, -1, 5, 233, 234, -1, 234, 107, - 98, -1, 98, -1, 9, 98, 112, 236, -1, 64, - 59, -1, 64, 36, -1, 64, 237, -1, 64, 58, - -1, 64, 73, 242, -1, 64, 29, -1, 28, 238, - 239, -1, -1, 38, -1, 26, -1, -1, 60, -1, - 67, -1, -1, 38, -1, 26, -1, -1, 60, -1, - 67, -1, -1, 108, 245, 109, -1, -1, 108, 246, - 109, -1, -1, 108, 247, 109, -1, 22, -1, 22, - -1, 22, -1, 6, 98, 112, 98, -1 + 121, 0, -1, 122, 123, 125, 12, -1, 3, -1, + 4, -1, 123, 124, -1, -1, 8, 262, 108, -1, + 125, 126, -1, -1, 127, 108, -1, 170, 108, -1, + 128, -1, 129, -1, 130, -1, 131, -1, 132, -1, + 133, -1, 134, -1, 135, -1, 141, -1, 136, -1, + 137, -1, 138, -1, 19, 146, 109, 142, -1, 18, + 145, 109, 144, -1, 16, 145, 109, 142, -1, 14, + 145, 109, 142, 109, 142, -1, 13, 145, 109, 144, + 109, 144, -1, 17, 145, 109, 144, 109, 144, 109, + 144, -1, 15, 145, 109, 144, 109, 139, 109, 140, + -1, 20, 144, -1, 20, 166, -1, 22, 145, 109, + 144, 109, 144, 109, 144, 109, 139, 109, 140, -1, + 83, 256, -1, 84, -1, 85, -1, 86, -1, 87, + -1, 88, -1, 89, -1, 90, -1, 91, -1, 92, + -1, 93, -1, 94, -1, 95, -1, 21, 145, 109, + 150, 109, 147, -1, 241, 143, -1, 241, 110, 143, + 110, -1, 150, 162, -1, 238, -1, 241, 150, 163, + -1, 241, 110, 150, 163, 110, -1, 151, 164, 165, + -1, 159, 161, -1, 148, 109, 148, 109, 148, 109, + 148, -1, 241, 149, -1, 23, -1, 262, -1, 100, + -1, 172, -1, 152, 111, 153, 112, -1, 186, -1, + 249, -1, 100, -1, 100, -1, 154, -1, 155, -1, + 23, -1, 159, 160, 156, -1, -1, 113, 157, -1, + 114, 158, -1, 23, -1, 23, -1, 100, -1, 104, + -1, 104, -1, 104, -1, 104, -1, 101, -1, 105, + -1, -1, 101, -1, 102, -1, 103, -1, 104, -1, + -1, 115, 166, 116, -1, 115, 167, 116, -1, -1, + 168, 163, -1, 169, 163, -1, 99, -1, 100, -1, + 171, -1, 178, -1, 242, -1, 245, -1, 248, -1, + 261, -1, 7, 99, 117, 172, -1, 96, 173, -1, + 38, 177, -1, 60, -1, 98, 175, -1, 53, -1, + 29, 254, -1, 37, -1, 74, 255, -1, 50, 111, + 176, 112, -1, 97, 111, 174, 112, -1, 23, -1, + -1, 111, 176, 112, -1, 23, -1, 60, -1, 29, + 254, -1, 37, -1, 74, 255, -1, 179, -1, 180, + -1, 10, 99, 182, -1, 10, 99, 111, 181, 112, + 183, -1, -1, 23, -1, 117, 185, -1, 117, 118, + 184, 119, -1, 187, -1, 184, 109, 187, -1, 189, + -1, 225, -1, 235, -1, 189, -1, 225, -1, 236, + -1, 188, -1, 226, -1, 235, -1, 189, -1, 73, + 213, -1, 73, 190, -1, 73, 192, -1, 73, 195, + -1, 73, 197, -1, 73, 203, -1, 73, 199, -1, + 73, 206, -1, 73, 208, -1, 73, 210, -1, 73, + 212, -1, 73, 224, -1, 47, 253, 191, -1, 201, + -1, 33, -1, 69, -1, 43, 111, 202, 112, 193, + -1, 201, -1, 60, -1, 26, -1, 72, 194, -1, + 40, -1, 32, -1, 44, 196, -1, 25, -1, 253, + 67, -1, 45, 111, 202, 112, 253, 198, -1, 201, + -1, 75, 257, 200, -1, 29, -1, 25, -1, 31, + -1, 71, -1, 23, -1, 76, 255, 204, 205, -1, + 35, -1, 54, -1, 79, -1, 80, -1, 78, -1, + 77, -1, 36, 207, -1, 29, -1, 56, -1, 28, + 111, 209, 112, 57, -1, 23, -1, 58, 211, -1, + 70, -1, 26, -1, 215, 66, 111, 218, 112, -1, + 215, 214, -1, -1, 66, 111, 218, 106, 218, 112, + -1, 49, 219, 216, -1, -1, 217, -1, 41, -1, + 82, -1, 42, -1, 23, -1, 51, 220, -1, 63, + -1, 52, -1, 81, 255, -1, 55, 111, 222, 112, + -1, 48, 111, 223, 112, -1, -1, 111, 221, 112, + -1, 23, -1, 23, -1, 23, -1, 30, 64, -1, + 229, -1, 232, -1, 227, -1, 230, -1, 62, 34, + 111, 228, 112, -1, 233, -1, 233, 106, 233, -1, + 62, 34, 111, 233, 112, -1, 62, 46, 111, 231, + 112, -1, 234, -1, 234, 106, 234, -1, 62, 46, + 111, 234, 112, -1, 23, -1, 23, -1, 237, -1, + 239, -1, 238, -1, 239, -1, 240, -1, 24, -1, + 23, -1, 118, 240, 119, -1, 118, 240, 109, 240, + 119, -1, 118, 240, 109, 240, 109, 240, 119, -1, + 118, 240, 109, 240, 109, 240, 109, 240, 119, -1, + 241, 24, -1, 241, 23, -1, 113, -1, 114, -1, + -1, -1, 244, 11, 243, 247, -1, 262, -1, -1, + -1, 5, 246, 247, -1, 247, 109, 99, -1, 99, + -1, 244, 9, 99, 117, 249, -1, 65, 60, -1, + 65, 37, -1, 65, 250, -1, 65, 59, -1, 65, + 74, 255, -1, 65, 30, -1, 29, 251, 252, -1, + -1, 39, -1, 27, -1, -1, 61, -1, 68, -1, + -1, 39, -1, 27, -1, -1, 61, -1, 68, -1, + -1, 111, 258, 112, -1, -1, 111, 259, 112, -1, + -1, 111, 260, 112, -1, 23, -1, 23, -1, 23, + -1, 6, 99, 117, 100, -1, 99, -1, 100, -1 }; /* YYRLINE[YYN] -- source line where rule number YYN was defined. */ static const yytype_uint16 yyrline[] = { - 0, 256, 256, 259, 267, 279, 280, 283, 305, 306, - 309, 324, 327, 332, 339, 340, 341, 342, 343, 344, - 345, 348, 349, 352, 358, 365, 372, 380, 387, 395, - 440, 447, 453, 454, 455, 456, 457, 458, 459, 460, - 461, 462, 463, 464, 467, 480, 493, 506, 528, 537, - 570, 577, 592, 642, 684, 695, 716, 726, 732, 763, - 780, 780, 782, 789, 801, 802, 803, 806, 820, 834, - 852, 863, 875, 877, 878, 879, 880, 883, 883, 883, - 883, 884, 887, 888, 889, 890, 891, 892, 895, 913, - 917, 923, 927, 931, 935, 944, 953, 957, 962, 968, - 979, 979, 980, 982, 986, 990, 994, 1000, 1000, 1002, - 1018, 1041, 1044, 1055, 1061, 1067, 1068, 1075, 1081, 1087, - 1095, 1101, 1107, 1115, 1121, 1127, 1135, 1136, 1139, 1140, - 1141, 1142, 1143, 1144, 1145, 1146, 1147, 1148, 1149, 1152, - 1161, 1165, 1169, 1175, 1184, 1188, 1192, 1201, 1205, 1211, - 1217, 1224, 1229, 1237, 1247, 1249, 1257, 1263, 1267, 1271, - 1277, 1288, 1297, 1301, 1306, 1310, 1314, 1318, 1324, 1331, - 1335, 1341, 1349, 1360, 1367, 1371, 1377, 1387, 1398, 1402, - 1420, 1429, 1432, 1438, 1442, 1446, 1452, 1463, 1468, 1473, - 1478, 1483, 1488, 1496, 1499, 1504, 1517, 1525, 1536, 1544, - 1544, 1546, 1546, 1548, 1558, 1563, 1570, 1580, 1589, 1594, - 1601, 1611, 1621, 1633, 1633, 1634, 1634, 1636, 1646, 1654, - 1664, 1672, 1680, 1689, 1700, 1704, 1710, 1711, 1712, 1715, - 1715, 1718, 1718, 1721, 1727, 1735, 1748, 1757, 1766, 1770, - 1779, 1788, 1799, 1806, 1811, 1820, 1832, 1835, 1844, 1855, - 1856, 1857, 1860, 1861, 1862, 1865, 1866, 1869, 1870, 1873, - 1874, 1877, 1888, 1899, 1910 + 0, 268, 268, 271, 279, 291, 292, 295, 317, 318, + 321, 336, 339, 344, 351, 352, 353, 354, 355, 356, + 357, 360, 361, 362, 365, 371, 377, 383, 390, 396, + 403, 447, 452, 462, 506, 512, 513, 514, 515, 516, + 517, 518, 519, 520, 521, 522, 523, 526, 538, 546, + 563, 570, 589, 600, 620, 645, 654, 687, 694, 709, + 759, 801, 812, 833, 843, 849, 880, 897, 897, 899, + 906, 918, 919, 920, 923, 937, 951, 969, 980, 992, + 994, 995, 996, 997, 1000, 1000, 1000, 1000, 1001, 1004, + 1008, 1013, 1020, 1027, 1034, 1057, 1080, 1081, 1082, 1083, + 1084, 1085, 1088, 1106, 1110, 1116, 1120, 1124, 1128, 1137, + 1146, 1150, 1155, 1161, 1172, 1172, 1173, 1175, 1179, 1183, + 1187, 1193, 1193, 1195, 1211, 1234, 1237, 1248, 1254, 1260, + 1261, 1268, 1274, 1280, 1288, 1294, 1300, 1308, 1314, 1320, + 1328, 1329, 1332, 1333, 1334, 1335, 1336, 1337, 1338, 1339, + 1340, 1341, 1342, 1345, 1354, 1358, 1362, 1368, 1377, 1381, + 1385, 1394, 1398, 1404, 1410, 1417, 1422, 1430, 1440, 1442, + 1450, 1456, 1460, 1464, 1470, 1481, 1490, 1494, 1499, 1503, + 1507, 1511, 1517, 1524, 1528, 1534, 1542, 1553, 1560, 1564, + 1570, 1580, 1591, 1595, 1613, 1622, 1625, 1631, 1635, 1639, + 1645, 1656, 1661, 1666, 1671, 1676, 1681, 1689, 1692, 1697, + 1710, 1718, 1729, 1737, 1737, 1739, 1739, 1741, 1751, 1756, + 1763, 1773, 1782, 1787, 1794, 1804, 1814, 1826, 1826, 1827, + 1827, 1829, 1839, 1847, 1857, 1865, 1873, 1882, 1893, 1897, + 1903, 1904, 1905, 1908, 1908, 1911, 1946, 1950, 1950, 1953, + 1959, 1967, 1980, 1989, 1998, 2002, 2011, 2020, 2031, 2038, + 2043, 2052, 2064, 2067, 2076, 2087, 2088, 2089, 2092, 2093, + 2094, 2097, 2098, 2101, 2102, 2105, 2106, 2109, 2120, 2131, + 2142, 2163, 2164 }; #endif @@ -801,9 +822,9 @@ static const char *const yytname[] = "$end", "error", "$undefined", "ARBvp_10", "ARBfp_10", "ADDRESS", "ALIAS", "ATTRIB", "OPTION", "OUTPUT", "PARAM", "TEMP", "END", "BIN_OP", "BINSC_OP", "SAMPLE_OP", "SCALAR_OP", "TRI_OP", "VECTOR_OP", "ARL", - "KIL", "SWZ", "INTEGER", "REAL", "AMBIENT", "ATTENUATION", "BACK", - "CLIP", "COLOR", "DEPTH", "DIFFUSE", "DIRECTION", "EMISSION", "ENV", - "EYE", "FOG", "FOGCOORD", "FRAGMENT", "FRONT", "HALF", "INVERSE", + "KIL", "SWZ", "TXD_OP", "INTEGER", "REAL", "AMBIENT", "ATTENUATION", + "BACK", "CLIP", "COLOR", "DEPTH", "DIFFUSE", "DIRECTION", "EMISSION", + "ENV", "EYE", "FOG", "FOGCOORD", "FRAGMENT", "FRONT", "HALF", "INVERSE", "INVTRANS", "LIGHT", "LIGHTMODEL", "LIGHTPROD", "LOCAL", "MATERIAL", "MAT_PROGRAM", "MATRIX", "MATRIXINDEX", "MODELVIEW", "MVP", "NORMAL", "OBJECT", "PALETTE", "PARAMS", "PLANE", "POINT_TOK", "POINTSIZE", @@ -814,49 +835,51 @@ static const char *const yytname[] = "TEX_2D", "TEX_3D", "TEX_CUBE", "TEX_RECT", "TEX_SHADOW1D", "TEX_SHADOW2D", "TEX_SHADOWRECT", "TEX_ARRAY1D", "TEX_ARRAY2D", "TEX_ARRAYSHADOW1D", "TEX_ARRAYSHADOW2D", "VERTEX", "VTXATTRIB", - "WEIGHT", "IDENTIFIER", "MASK4", "MASK3", "MASK2", "MASK1", "SWIZZLE", - "DOT_DOT", "DOT", "';'", "','", "'['", "']'", "'+'", "'-'", "'='", "'{'", - "'}'", "$accept", "program", "language", "optionSequence", "option", - "statementSequence", "statement", "instruction", "ALU_instruction", - "TexInstruction", "ARL_instruction", "VECTORop_instruction", - "SCALARop_instruction", "BINSCop_instruction", "BINop_instruction", - "TRIop_instruction", "SAMPLE_instruction", "KIL_instruction", - "texImageUnit", "texTarget", "SWZ_instruction", "scalarSrcReg", - "swizzleSrcReg", "maskedDstReg", "maskedAddrReg", "extendedSwizzle", - "extSwizComp", "extSwizSel", "srcReg", "dstReg", "progParamArray", - "progParamArrayMem", "progParamArrayAbs", "progParamArrayRel", - "addrRegRelOffset", "addrRegPosOffset", "addrRegNegOffset", "addrReg", - "addrComponent", "addrWriteMask", "scalarSuffix", "swizzleSuffix", - "optionalMask", "namingStatement", "ATTRIB_statement", "attribBinding", - "vtxAttribItem", "vtxAttribNum", "vtxOptWeightNum", "vtxWeightNum", - "fragAttribItem", "PARAM_statement", "PARAM_singleStmt", - "PARAM_multipleStmt", "optArraySize", "paramSingleInit", - "paramMultipleInit", "paramMultInitList", "paramSingleItemDecl", - "paramSingleItemUse", "paramMultipleItem", "stateMultipleItem", - "stateSingleItem", "stateMaterialItem", "stateMatProperty", - "stateLightItem", "stateLightProperty", "stateSpotProperty", - "stateLightModelItem", "stateLModProperty", "stateLightProdItem", - "stateLProdProperty", "stateTexEnvItem", "stateTexEnvProperty", - "ambDiffSpecProperty", "stateLightNumber", "stateTexGenItem", - "stateTexGenType", "stateTexGenCoord", "stateFogItem", - "stateFogProperty", "stateClipPlaneItem", "stateClipPlaneNum", - "statePointItem", "statePointProperty", "stateMatrixRow", - "stateMatrixRows", "optMatrixRows", "stateMatrixItem", - "stateOptMatModifier", "stateMatModifier", "stateMatrixRowNum", - "stateMatrixName", "stateOptModMatNum", "stateModMatNum", - "statePaletteMatNum", "stateProgramMatNum", "stateDepthItem", - "programSingleItem", "programMultipleItem", "progEnvParams", - "progEnvParamNums", "progEnvParam", "progLocalParams", - "progLocalParamNums", "progLocalParam", "progEnvParamNum", - "progLocalParamNum", "paramConstDecl", "paramConstUse", - "paramConstScalarDecl", "paramConstScalarUse", "paramConstVector", - "signedFloatConstant", "optionalSign", "TEMP_statement", "@1", + "WEIGHT", "IDENTIFIER", "USED_IDENTIFIER", "MASK4", "MASK3", "MASK2", + "MASK1", "SWIZZLE", "DOT_DOT", "DOT", "';'", "','", "'|'", "'['", "']'", + "'+'", "'-'", "'('", "')'", "'='", "'{'", "'}'", "$accept", "program", + "language", "optionSequence", "option", "statementSequence", "statement", + "instruction", "ALU_instruction", "TexInstruction", "ARL_instruction", + "VECTORop_instruction", "SCALARop_instruction", "BINSCop_instruction", + "BINop_instruction", "TRIop_instruction", "SAMPLE_instruction", + "KIL_instruction", "TXD_instruction", "texImageUnit", "texTarget", + "SWZ_instruction", "scalarSrcReg", "scalarUse", "swizzleSrcReg", + "maskedDstReg", "maskedAddrReg", "extendedSwizzle", "extSwizComp", + "extSwizSel", "srcReg", "dstReg", "progParamArray", "progParamArrayMem", + "progParamArrayAbs", "progParamArrayRel", "addrRegRelOffset", + "addrRegPosOffset", "addrRegNegOffset", "addrReg", "addrComponent", + "addrWriteMask", "scalarSuffix", "swizzleSuffix", "optionalMask", + "optionalCcMask", "ccTest", "ccTest2", "ccMaskRule", "ccMaskRule2", + "namingStatement", "ATTRIB_statement", "attribBinding", "vtxAttribItem", + "vtxAttribNum", "vtxOptWeightNum", "vtxWeightNum", "fragAttribItem", + "PARAM_statement", "PARAM_singleStmt", "PARAM_multipleStmt", + "optArraySize", "paramSingleInit", "paramMultipleInit", + "paramMultInitList", "paramSingleItemDecl", "paramSingleItemUse", + "paramMultipleItem", "stateMultipleItem", "stateSingleItem", + "stateMaterialItem", "stateMatProperty", "stateLightItem", + "stateLightProperty", "stateSpotProperty", "stateLightModelItem", + "stateLModProperty", "stateLightProdItem", "stateLProdProperty", + "stateTexEnvItem", "stateTexEnvProperty", "ambDiffSpecProperty", + "stateLightNumber", "stateTexGenItem", "stateTexGenType", + "stateTexGenCoord", "stateFogItem", "stateFogProperty", + "stateClipPlaneItem", "stateClipPlaneNum", "statePointItem", + "statePointProperty", "stateMatrixRow", "stateMatrixRows", + "optMatrixRows", "stateMatrixItem", "stateOptMatModifier", + "stateMatModifier", "stateMatrixRowNum", "stateMatrixName", + "stateOptModMatNum", "stateModMatNum", "statePaletteMatNum", + "stateProgramMatNum", "stateDepthItem", "programSingleItem", + "programMultipleItem", "progEnvParams", "progEnvParamNums", + "progEnvParam", "progLocalParams", "progLocalParamNums", + "progLocalParam", "progEnvParamNum", "progLocalParamNum", + "paramConstDecl", "paramConstUse", "paramConstScalarDecl", + "paramConstScalarUse", "paramConstVector", "signedFloatConstant", + "optionalSign", "TEMP_statement", "@1", "optVarSize", "ADDRESS_statement", "@2", "varNameList", "OUTPUT_statement", "resultBinding", "resultColBinding", "optResultFaceType", "optResultColorType", "optFaceType", "optColorType", "optTexCoordUnitNum", "optTexImageUnitNum", "optLegacyTexUnitNum", "texCoordUnitNum", "texImageUnitNum", "legacyTexUnitNum", - "ALIAS_statement", 0 + "ALIAS_statement", "string", 0 }; #endif @@ -875,41 +898,43 @@ static const yytype_uint16 yytoknum[] = 325, 326, 327, 328, 329, 330, 331, 332, 333, 334, 335, 336, 337, 338, 339, 340, 341, 342, 343, 344, 345, 346, 347, 348, 349, 350, 351, 352, 353, 354, - 355, 356, 357, 358, 359, 360, 59, 44, 91, 93, - 43, 45, 61, 123, 125 + 355, 356, 357, 358, 359, 360, 361, 362, 59, 44, + 124, 91, 93, 43, 45, 40, 41, 61, 123, 125 }; # endif /* YYR1[YYN] -- Symbol number of symbol that rule YYN derives. */ -static const yytype_uint8 yyr1[] = +static const yytype_uint16 yyr1[] = { - 0, 115, 116, 117, 117, 118, 118, 119, 120, 120, - 121, 121, 122, 122, 123, 123, 123, 123, 123, 123, - 123, 124, 124, 125, 126, 127, 128, 129, 130, 131, - 132, 133, 134, 134, 134, 134, 134, 134, 134, 134, - 134, 134, 134, 134, 135, 136, 137, 138, 139, 140, - 141, 142, 142, 143, 143, 143, 143, 144, 144, 145, - 146, 146, 147, 148, 149, 149, 149, 150, 151, 152, - 153, 154, 155, 156, 156, 156, 156, 157, 157, 157, - 157, 157, 158, 158, 158, 158, 158, 158, 159, 160, - 160, 161, 161, 161, 161, 161, 161, 161, 161, 162, - 163, 163, 164, 165, 165, 165, 165, 166, 166, 167, - 168, 169, 169, 170, 171, 172, 172, 173, 173, 173, - 174, 174, 174, 175, 175, 175, 176, 176, 177, 177, - 177, 177, 177, 177, 177, 177, 177, 177, 177, 178, - 179, 179, 179, 180, 181, 181, 181, 181, 181, 182, - 183, 184, 184, 185, 186, 187, 188, 189, 189, 189, - 190, 191, 192, 192, 193, 193, 193, 193, 194, 195, - 195, 196, 197, 198, 199, 199, 200, 201, 202, 202, - 203, 204, 204, 205, 205, 205, 206, 207, 207, 207, - 207, 207, 207, 208, 208, 209, 210, 211, 212, 213, - 213, 214, 214, 215, 216, 216, 217, 218, 219, 219, - 220, 221, 222, 223, 223, 224, 224, 225, 226, 226, - 227, 227, 227, 227, 228, 228, 229, 229, 229, 231, - 230, 233, 232, 234, 234, 235, 236, 236, 236, 236, - 236, 236, 237, 238, 238, 238, 239, 239, 239, 240, - 240, 240, 241, 241, 241, 242, 242, 243, 243, 244, - 244, 245, 246, 247, 248 + 0, 120, 121, 122, 122, 123, 123, 124, 125, 125, + 126, 126, 127, 127, 128, 128, 128, 128, 128, 128, + 128, 129, 129, 129, 130, 131, 132, 133, 134, 135, + 136, 137, 137, 138, 139, 140, 140, 140, 140, 140, + 140, 140, 140, 140, 140, 140, 140, 141, 142, 142, + 143, 143, 144, 144, 145, 146, 147, 148, 149, 149, + 150, 150, 150, 150, 151, 151, 152, 153, 153, 154, + 155, 156, 156, 156, 157, 158, 159, 160, 161, 162, + 163, 163, 163, 163, 164, 164, 164, 164, 164, 165, + 165, 165, 166, 167, 168, 169, 170, 170, 170, 170, + 170, 170, 171, 172, 172, 173, 173, 173, 173, 173, + 173, 173, 173, 174, 175, 175, 176, 177, 177, 177, + 177, 178, 178, 179, 180, 181, 181, 182, 183, 184, + 184, 185, 185, 185, 186, 186, 186, 187, 187, 187, + 188, 188, 189, 189, 189, 189, 189, 189, 189, 189, + 189, 189, 189, 190, 191, 191, 191, 192, 193, 193, + 193, 193, 193, 194, 195, 196, 196, 197, 198, 199, + 200, 201, 201, 201, 202, 203, 204, 204, 205, 205, + 205, 205, 206, 207, 207, 208, 209, 210, 211, 211, + 212, 213, 214, 214, 215, 216, 216, 217, 217, 217, + 218, 219, 219, 219, 219, 219, 219, 220, 220, 221, + 222, 223, 224, 225, 225, 226, 226, 227, 228, 228, + 229, 230, 231, 231, 232, 233, 234, 235, 235, 236, + 236, 237, 238, 238, 239, 239, 239, 239, 240, 240, + 241, 241, 241, 243, 242, 244, 244, 246, 245, 247, + 247, 248, 249, 249, 249, 249, 249, 249, 250, 251, + 251, 251, 252, 252, 252, 253, 253, 253, 254, 254, + 254, 255, 255, 256, 256, 257, 257, 258, 259, 260, + 261, 262, 262 }; /* YYR2[YYN] -- Number of symbols composing right hand side of rule YYN. */ @@ -917,31 +942,33 @@ static const yytype_uint8 yyr2[] = { 0, 2, 4, 1, 1, 2, 0, 3, 2, 0, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, - 1, 1, 1, 4, 4, 4, 6, 6, 8, 8, - 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, - 1, 1, 1, 1, 6, 3, 3, 2, 2, 7, - 2, 1, 1, 1, 1, 4, 1, 1, 1, 1, - 1, 1, 1, 3, 0, 2, 2, 1, 1, 1, - 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, - 1, 0, 1, 1, 1, 1, 1, 1, 4, 2, - 2, 1, 2, 1, 2, 1, 2, 4, 4, 1, - 0, 3, 1, 1, 2, 1, 2, 1, 1, 3, - 6, 0, 1, 2, 4, 1, 3, 1, 1, 1, - 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, - 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, - 1, 1, 1, 5, 1, 1, 1, 2, 1, 1, - 2, 1, 2, 6, 1, 3, 1, 1, 1, 1, - 1, 4, 1, 1, 1, 1, 1, 1, 2, 1, - 1, 5, 1, 2, 1, 1, 5, 2, 0, 6, - 3, 0, 1, 1, 1, 1, 1, 2, 1, 1, - 2, 4, 4, 0, 3, 1, 1, 1, 2, 1, - 1, 1, 1, 5, 1, 3, 5, 5, 1, 3, - 5, 1, 1, 1, 1, 1, 1, 1, 1, 1, - 3, 5, 7, 9, 2, 2, 1, 1, 0, 0, - 3, 0, 3, 3, 1, 4, 2, 2, 2, 2, - 3, 2, 3, 0, 1, 1, 0, 1, 1, 0, - 1, 1, 0, 1, 1, 0, 3, 0, 3, 0, - 3, 1, 1, 1, 4 + 1, 1, 1, 1, 4, 4, 4, 6, 6, 8, + 8, 2, 2, 12, 2, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 1, 1, 1, 6, 2, 4, + 2, 1, 3, 5, 3, 2, 7, 2, 1, 1, + 1, 1, 4, 1, 1, 1, 1, 1, 1, 1, + 3, 0, 2, 2, 1, 1, 1, 1, 1, 1, + 1, 1, 1, 0, 1, 1, 1, 1, 0, 3, + 3, 0, 2, 2, 1, 1, 1, 1, 1, 1, + 1, 1, 4, 2, 2, 1, 2, 1, 2, 1, + 2, 4, 4, 1, 0, 3, 1, 1, 2, 1, + 2, 1, 1, 3, 6, 0, 1, 2, 4, 1, + 3, 1, 1, 1, 1, 1, 1, 1, 1, 1, + 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, + 2, 2, 2, 3, 1, 1, 1, 5, 1, 1, + 1, 2, 1, 1, 2, 1, 2, 6, 1, 3, + 1, 1, 1, 1, 1, 4, 1, 1, 1, 1, + 1, 1, 2, 1, 1, 5, 1, 2, 1, 1, + 5, 2, 0, 6, 3, 0, 1, 1, 1, 1, + 1, 2, 1, 1, 2, 4, 4, 0, 3, 1, + 1, 1, 2, 1, 1, 1, 1, 5, 1, 3, + 5, 5, 1, 3, 5, 1, 1, 1, 1, 1, + 1, 1, 1, 1, 3, 5, 7, 9, 2, 2, + 1, 1, 0, 0, 4, 1, 0, 0, 3, 3, + 1, 5, 2, 2, 2, 2, 3, 2, 3, 0, + 1, 1, 0, 1, 1, 0, 1, 1, 0, 1, + 1, 0, 3, 0, 3, 0, 3, 1, 1, 1, + 4, 1, 1 }; /* YYDEFACT[STATE-NAME] -- Default rule to reduce with in state @@ -949,272 +976,296 @@ static const yytype_uint8 yyr2[] = means the default is an error. */ static const yytype_uint16 yydefact[] = { - 0, 3, 4, 0, 6, 1, 9, 0, 5, 0, - 0, 231, 0, 0, 0, 0, 229, 2, 0, 0, - 0, 0, 0, 0, 0, 228, 0, 8, 0, 12, - 13, 14, 15, 16, 17, 18, 19, 21, 22, 20, - 0, 82, 83, 107, 108, 84, 85, 86, 87, 7, - 0, 0, 0, 0, 0, 0, 0, 58, 0, 81, - 57, 0, 0, 0, 0, 0, 69, 0, 0, 226, - 227, 30, 0, 0, 10, 11, 234, 232, 0, 0, - 0, 111, 228, 109, 230, 243, 241, 237, 239, 236, - 255, 238, 228, 77, 78, 79, 80, 47, 228, 228, - 228, 228, 228, 228, 71, 48, 219, 218, 0, 0, - 0, 0, 53, 228, 76, 0, 54, 56, 120, 121, - 199, 200, 122, 215, 216, 0, 0, 264, 88, 235, - 112, 0, 113, 117, 118, 119, 213, 214, 217, 0, - 245, 244, 246, 0, 240, 0, 0, 0, 0, 25, - 0, 24, 23, 252, 105, 103, 255, 90, 0, 0, - 0, 0, 0, 0, 249, 0, 249, 0, 0, 259, - 255, 128, 129, 130, 131, 133, 132, 134, 135, 136, - 137, 0, 138, 252, 95, 0, 93, 91, 255, 0, - 100, 89, 0, 74, 73, 75, 46, 0, 0, 233, - 0, 225, 224, 247, 248, 242, 261, 0, 228, 228, - 0, 0, 228, 253, 254, 104, 106, 0, 0, 0, - 198, 169, 170, 168, 0, 151, 251, 250, 150, 0, - 0, 0, 0, 193, 189, 0, 188, 255, 181, 175, - 174, 173, 0, 0, 0, 0, 94, 0, 96, 0, - 0, 92, 228, 220, 62, 0, 60, 61, 0, 228, - 0, 110, 256, 27, 26, 72, 45, 257, 0, 0, - 211, 0, 212, 0, 172, 0, 160, 0, 152, 0, - 157, 158, 141, 142, 159, 139, 140, 0, 0, 187, - 0, 190, 183, 185, 184, 180, 182, 263, 0, 156, - 155, 162, 163, 0, 0, 102, 0, 99, 0, 0, - 0, 55, 70, 64, 44, 0, 0, 228, 0, 31, - 0, 228, 206, 210, 0, 0, 249, 197, 0, 195, - 0, 196, 0, 260, 167, 166, 164, 165, 161, 186, - 0, 97, 98, 101, 228, 221, 0, 0, 63, 228, - 51, 52, 50, 0, 0, 0, 115, 123, 126, 124, - 201, 202, 125, 262, 0, 32, 33, 34, 35, 36, - 37, 38, 39, 40, 41, 42, 43, 29, 28, 171, - 146, 148, 145, 0, 143, 144, 0, 192, 194, 191, - 176, 0, 67, 65, 68, 66, 0, 0, 0, 127, - 178, 228, 114, 258, 149, 147, 153, 154, 228, 222, - 228, 0, 0, 0, 177, 116, 0, 0, 0, 204, - 0, 208, 0, 223, 228, 203, 0, 207, 0, 0, - 49, 205, 209, 0, 0, 179 + 0, 3, 4, 0, 6, 1, 9, 0, 5, 246, + 281, 282, 0, 247, 0, 0, 0, 2, 0, 0, + 0, 0, 0, 0, 0, 242, 0, 0, 8, 0, + 12, 13, 14, 15, 16, 17, 18, 19, 21, 22, + 23, 20, 0, 96, 97, 121, 122, 98, 0, 99, + 100, 101, 245, 7, 0, 0, 0, 0, 0, 65, + 0, 88, 64, 0, 0, 0, 0, 0, 76, 0, + 0, 94, 240, 241, 31, 32, 83, 0, 0, 0, + 10, 11, 0, 243, 250, 248, 0, 0, 125, 242, + 123, 259, 257, 253, 255, 252, 271, 254, 242, 84, + 85, 86, 87, 91, 242, 242, 242, 242, 242, 242, + 78, 55, 81, 80, 82, 92, 233, 232, 0, 0, + 0, 0, 60, 0, 242, 83, 0, 61, 63, 134, + 135, 213, 214, 136, 229, 230, 0, 242, 0, 0, + 0, 280, 102, 126, 0, 127, 131, 132, 133, 227, + 228, 231, 0, 261, 260, 262, 0, 256, 0, 0, + 54, 0, 0, 0, 26, 0, 25, 24, 268, 119, + 117, 271, 104, 0, 0, 0, 0, 0, 0, 265, + 0, 265, 0, 0, 275, 271, 142, 143, 144, 145, + 147, 146, 148, 149, 150, 151, 0, 152, 268, 109, + 0, 107, 105, 271, 0, 114, 103, 83, 0, 52, + 0, 0, 0, 0, 244, 249, 0, 239, 238, 263, + 264, 258, 277, 0, 242, 95, 0, 0, 83, 242, + 0, 48, 0, 51, 0, 242, 269, 270, 118, 120, + 0, 0, 0, 212, 183, 184, 182, 0, 165, 267, + 266, 164, 0, 0, 0, 0, 207, 203, 0, 202, + 271, 195, 189, 188, 187, 0, 0, 0, 0, 108, + 0, 110, 0, 0, 106, 0, 242, 234, 69, 0, + 67, 68, 0, 242, 242, 251, 0, 124, 272, 28, + 89, 90, 93, 27, 0, 79, 50, 273, 0, 0, + 225, 0, 226, 0, 186, 0, 174, 0, 166, 0, + 171, 172, 155, 156, 173, 153, 154, 0, 0, 201, + 0, 204, 197, 199, 198, 194, 196, 279, 0, 170, + 169, 176, 177, 0, 0, 116, 0, 113, 0, 0, + 53, 0, 62, 77, 71, 47, 0, 0, 0, 242, + 49, 0, 34, 0, 242, 220, 224, 0, 0, 265, + 211, 0, 209, 0, 210, 0, 276, 181, 180, 178, + 179, 175, 200, 0, 111, 112, 115, 242, 235, 0, + 0, 70, 242, 58, 57, 59, 242, 0, 0, 0, + 129, 137, 140, 138, 215, 216, 139, 278, 0, 35, + 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, + 46, 30, 29, 185, 160, 162, 159, 0, 157, 158, + 0, 206, 208, 205, 190, 0, 74, 72, 75, 73, + 0, 0, 0, 0, 141, 192, 242, 128, 274, 163, + 161, 167, 168, 242, 236, 242, 0, 0, 0, 0, + 191, 130, 0, 0, 0, 0, 218, 0, 222, 0, + 237, 242, 0, 217, 0, 221, 0, 0, 56, 33, + 219, 223, 0, 0, 193 }; /* YYDEFGOTO[NTERM-NUM]. */ static const yytype_int16 yydefgoto[] = { - -1, 3, 4, 6, 8, 9, 27, 28, 29, 30, - 31, 32, 33, 34, 35, 36, 37, 38, 268, 377, - 39, 146, 71, 58, 67, 314, 315, 352, 114, 59, - 115, 255, 256, 257, 348, 393, 395, 68, 313, 105, - 266, 196, 97, 40, 41, 116, 191, 308, 251, 306, - 157, 42, 43, 44, 131, 83, 261, 355, 132, 117, - 356, 357, 118, 171, 285, 172, 384, 405, 173, 228, - 174, 406, 175, 300, 286, 277, 176, 303, 338, 177, - 223, 178, 275, 179, 241, 180, 399, 414, 181, 295, - 296, 340, 238, 289, 330, 332, 328, 182, 119, 359, - 360, 418, 120, 361, 420, 121, 271, 273, 362, 122, - 136, 123, 124, 138, 72, 45, 55, 46, 50, 77, - 47, 60, 91, 142, 205, 229, 215, 144, 319, 243, - 207, 364, 298, 48 + -1, 3, 4, 6, 8, 9, 28, 29, 30, 31, + 32, 33, 34, 35, 36, 37, 38, 39, 40, 298, + 411, 41, 161, 231, 74, 60, 69, 345, 346, 384, + 232, 61, 126, 279, 280, 281, 381, 427, 429, 70, + 344, 111, 296, 115, 103, 160, 75, 227, 76, 228, + 42, 43, 127, 206, 338, 274, 336, 172, 44, 45, + 46, 144, 90, 287, 389, 145, 128, 390, 391, 129, + 186, 315, 187, 418, 440, 188, 251, 189, 441, 190, + 330, 316, 307, 191, 333, 371, 192, 246, 193, 305, + 194, 264, 195, 434, 450, 196, 325, 326, 373, 261, + 319, 363, 365, 361, 197, 130, 393, 394, 455, 131, + 395, 457, 132, 301, 303, 396, 133, 149, 134, 135, + 151, 77, 47, 139, 48, 49, 54, 85, 50, 62, + 97, 155, 221, 252, 238, 157, 352, 266, 223, 398, + 328, 51, 12 }; /* YYPACT[STATE-NUM] -- Index in YYTABLE of the portion describing STATE-NUM. */ -#define YYPACT_NINF -369 +#define YYPACT_NINF -401 static const yytype_int16 yypact[] = { - 143, -369, -369, 36, -369, -369, 45, -39, -369, 169, - -33, -369, -19, -6, -4, 12, -369, -369, -34, -34, - -34, -34, -34, -34, 15, 62, -34, -369, 26, -369, - -369, -369, -369, -369, -369, -369, -369, -369, -369, -369, - 60, -369, -369, -369, -369, -369, -369, -369, -369, -369, - 20, 56, 107, 110, 37, 20, -3, -369, 111, 109, - -369, 113, 114, 116, 117, 118, -369, 119, 125, -369, - -369, -369, -15, 121, -369, -369, -369, 122, 132, -18, - 167, 210, -11, -369, 122, 63, -369, -369, -369, -369, - 130, -369, 62, -369, -369, -369, -369, -369, 62, 62, - 62, 62, 62, 62, -369, -369, -369, -369, 9, 72, - 87, -1, 131, 62, 104, 134, -369, -369, -369, -369, - -369, -369, -369, -369, -369, -15, 142, -369, -369, -369, - -369, 135, -369, -369, -369, -369, -369, -369, -369, 182, - -369, -369, 52, 219, -369, 138, 139, -15, 140, -369, - 141, -369, -369, 61, -369, -369, 130, -369, 144, 145, - 146, 180, 11, 147, 85, 148, 99, 89, -2, 149, - 130, -369, -369, -369, -369, -369, -369, -369, -369, -369, - -369, 184, -369, 61, -369, 150, -369, -369, 130, 151, - 152, -369, 27, -369, -369, -369, -369, -10, 154, -369, - 153, -369, -369, -369, -369, -369, -369, 155, 62, 62, - 161, 168, 62, -369, -369, -369, -369, 229, 244, 246, - -369, -369, -369, -369, 247, -369, -369, -369, -369, 204, - 247, 17, 163, 164, -369, 165, -369, 130, 67, -369, - -369, -369, 252, 248, 18, 170, -369, 253, -369, 255, - 253, -369, 62, -369, -369, 171, -369, -369, 177, 62, - 172, -369, -369, -369, -369, -369, -369, 173, 175, 176, - -369, 178, -369, 179, -369, 181, -369, 183, -369, 185, - -369, -369, -369, -369, -369, -369, -369, 262, 264, -369, - 267, -369, -369, -369, -369, -369, -369, -369, 186, -369, - -369, -369, -369, 136, 269, -369, 187, -369, 188, 190, - 43, -369, -369, 106, -369, 193, -5, -7, 271, -369, - 108, 62, -369, -369, 245, 4, 99, -369, 194, -369, - 195, -369, 196, -369, -369, -369, -369, -369, -369, -369, - 197, -369, -369, -369, 62, -369, 280, 285, -369, 62, - -369, -369, -369, 93, 87, 53, -369, -369, -369, -369, - -369, -369, -369, -369, 199, -369, -369, -369, -369, -369, - -369, -369, -369, -369, -369, -369, -369, -369, -369, -369, - -369, -369, -369, 278, -369, -369, 8, -369, -369, -369, - -369, 57, -369, -369, -369, -369, 203, 205, 206, -369, - 250, -7, -369, -369, -369, -369, -369, -369, 62, -369, - 62, 229, 244, 208, -369, -369, 198, 211, 202, 213, - 214, 218, 269, -369, 62, -369, 229, -369, 244, 54, - -369, -369, -369, 269, 215, -369 + 193, -401, -401, 27, -401, -401, 62, 143, -401, 24, + -401, -401, -30, -401, -18, 12, 83, -401, 15, 15, + 15, 15, 15, 15, 67, 61, 15, 15, -401, 127, + -401, -401, -401, -401, -401, -401, -401, -401, -401, -401, + -401, -401, 144, -401, -401, -401, -401, -401, 204, -401, + -401, -401, -401, -401, 155, 136, 138, 34, 140, -401, + 147, 108, -401, 150, 156, 157, 158, 160, -401, 162, + 159, -401, -401, -401, -401, -401, 102, -13, 163, 164, + -401, -401, 165, -401, -401, 166, 170, 10, 235, 0, + -401, 141, -401, -401, -401, -401, 167, -401, 131, -401, + -401, -401, -401, 168, 131, 131, 131, 131, 131, 131, + -401, -401, -401, -401, -401, -401, -401, -401, 104, 97, + 114, 38, 169, 30, 131, 102, 171, -401, -401, -401, + -401, -401, -401, -401, -401, -401, 30, 131, 172, 155, + 175, -401, -401, -401, 173, -401, -401, -401, -401, -401, + -401, -401, 223, -401, -401, 123, 253, -401, 177, 149, + -401, 178, -10, 181, -401, 182, -401, -401, 134, -401, + -401, 167, -401, 183, 184, 185, 213, 99, 186, 154, + 187, 146, 153, 7, 188, 167, -401, -401, -401, -401, + -401, -401, -401, -401, -401, -401, 215, -401, 134, -401, + 190, -401, -401, 167, 191, 192, -401, 102, -48, -401, + 1, 195, 196, 214, 166, -401, 189, -401, -401, -401, + -401, -401, -401, 180, 131, -401, 194, 197, 102, 131, + 30, -401, 203, 205, 201, 131, -401, -401, -401, -401, + 285, 288, 289, -401, -401, -401, -401, 291, -401, -401, + -401, -401, 248, 291, 33, 206, 207, -401, 208, -401, + 167, 14, -401, -401, -401, 293, 292, 92, 209, -401, + 299, -401, 301, 299, -401, 216, 131, -401, -401, 217, + -401, -401, 221, 131, 131, -401, 212, -401, -401, -401, + -401, -401, -401, -401, 218, -401, -401, 220, 224, 225, + -401, 226, -401, 227, -401, 228, -401, 230, -401, 231, + -401, -401, -401, -401, -401, -401, -401, 304, 309, -401, + 312, -401, -401, -401, -401, -401, -401, -401, 232, -401, + -401, -401, -401, 161, 313, -401, 233, -401, 234, 238, + -401, 13, -401, -401, 137, -401, 242, -15, 243, 3, + -401, 314, -401, 133, 131, -401, -401, 296, 94, 146, + -401, 245, -401, 246, -401, 247, -401, -401, -401, -401, + -401, -401, -401, 249, -401, -401, -401, 131, -401, 332, + 337, -401, 131, -401, -401, -401, 131, 142, 114, 28, + -401, -401, -401, -401, -401, -401, -401, -401, 250, -401, + -401, -401, -401, -401, -401, -401, -401, -401, -401, -401, + -401, -401, -401, -401, -401, -401, -401, 331, -401, -401, + 68, -401, -401, -401, -401, 43, -401, -401, -401, -401, + 255, 256, 257, 258, -401, 300, 3, -401, -401, -401, + -401, -401, -401, 131, -401, 131, 201, 285, 288, 259, + -401, -401, 252, 264, 265, 263, 261, 266, 270, 313, + -401, 131, 133, -401, 285, -401, 288, 80, -401, -401, + -401, -401, 313, 267, -401 }; /* YYPGOTO[NTERM-NUM]. */ static const yytype_int16 yypgoto[] = { - -369, -369, -369, -369, -369, -369, -369, -369, -369, -369, - -369, -369, -369, -369, -369, -369, -369, -369, -369, -369, - -369, -94, -88, 133, -369, -369, -334, -369, -85, -369, - -369, -369, -369, -369, -369, -369, -369, 128, -369, -369, - -369, -369, -369, -369, -369, 251, -369, -369, -369, 77, - -369, -369, -369, -369, -369, -369, -369, -369, -369, -369, - -72, -369, -81, -369, -369, -369, -369, -369, -369, -369, - -369, -369, -369, -369, -305, 101, -369, -369, -369, -369, - -369, -369, -369, -369, -369, -369, -369, -369, -22, -369, - -369, -336, -369, -369, -369, -369, -369, -369, 254, -369, - -369, -369, -369, -369, -369, -369, -342, -368, 256, -369, - -369, -369, -80, -110, -82, -369, -369, -369, -369, 279, - -369, 257, -369, -369, -369, -161, 156, -146, -369, -369, - -369, -369, -369, -369 + -401, -401, -401, -401, -401, -401, -401, -401, -401, -401, + -401, -401, -401, -401, -401, -401, -401, -401, -401, -69, + -82, -401, -100, 151, -86, 210, -401, -401, -366, -401, + -54, -401, -401, -401, -401, -401, -401, -401, -401, 174, + -401, -401, -401, -118, -401, -401, 229, -401, -401, -401, + -401, -401, 295, -401, -401, -401, 110, -401, -401, -401, + -401, -401, -401, -401, -401, -401, -401, -51, -401, -88, + -401, -401, -401, -401, -401, -401, -401, -401, -401, -401, + -401, -311, 139, -401, -401, -401, -401, -401, -401, -401, + -401, -401, -401, -401, -401, -2, -401, -401, -400, -401, + -401, -401, -401, -401, -401, 298, -401, -401, -401, -401, + -401, -401, -401, -390, -295, 302, -401, -401, -136, -87, + -120, -89, -401, -401, -401, -401, -401, 251, -401, 176, + -401, -401, -401, -176, 198, -153, -401, -401, -401, -401, + -401, -401, -6 }; /* YYTABLE[YYPACT[STATE-NUM]]. What to do in state STATE-NUM. If positive, shift that token. If negative, reduce the rule which number is the opposite. If zero, do what YYDEFACT says. If YYTABLE_NINF, syntax error. */ -#define YYTABLE_NINF -60 +#define YYTABLE_NINF -230 static const yytype_int16 yytable[] = { - 139, 133, 137, 192, 145, 231, 149, 106, 107, 152, - 216, 148, 254, 150, 151, 396, 147, 350, 147, 108, - 385, 147, 108, 239, 244, 85, 86, 183, 280, 380, - 56, 139, 280, 87, 281, 184, 5, 153, 281, 221, - 198, 280, 248, 381, 421, 154, 109, 281, 185, 282, - 109, 186, 301, 7, 353, 88, 89, 110, 187, 10, - 432, 110, 210, 382, 57, 354, 222, 240, 155, 419, - 90, 302, 188, 49, 284, 383, 417, 111, 284, 51, - 111, 407, 156, 112, 431, 283, 429, 284, 66, 140, - 430, 291, 52, 351, 53, 189, 190, 434, 113, 69, - 70, 141, 113, 69, 70, 158, 113, 292, 293, 225, - 54, 226, 203, 66, 160, 264, 161, 159, 76, 204, - 263, 213, 162, 227, 269, 226, 397, 147, 214, 163, - 164, 165, 74, 166, 252, 167, 232, 227, 398, 233, - 234, 253, 310, 235, 168, 81, 1, 2, 294, 82, - 344, 236, 61, 62, 63, 64, 65, 345, 433, 73, - 401, 169, 170, 390, 408, 386, 75, 402, 78, 237, - 139, 409, 69, 70, 11, 12, 13, 316, 14, 15, - 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, - 26, 365, 366, 367, 368, 369, 370, 371, 372, 373, - 374, 375, 376, 193, 201, 202, 194, 195, 93, 94, - 95, 96, 334, 335, 336, 337, 346, 347, 92, 79, - 98, 99, 80, 100, 101, 102, 103, 104, 125, 126, - 127, 56, 130, 378, 391, 139, 358, 137, 143, -59, - 199, 206, 197, 220, 200, 208, 209, 211, 212, 245, - 267, 270, 217, 218, 219, 224, 230, 242, 247, 249, - 250, 259, 139, 265, 262, 260, 272, 316, 274, 276, - 278, 287, 288, 290, 297, 305, 299, 307, 304, 312, - 311, 318, 320, 321, 327, 317, 329, 322, 323, 331, - 324, 339, 325, 363, 326, 333, 341, 342, 416, 343, - 349, 379, 392, 387, 388, 389, 390, 394, 403, 404, - 410, 425, 423, 411, 412, 413, 422, 426, 424, 139, - 358, 137, 428, 427, 435, 258, 139, 309, 316, 415, - 128, 279, 400, 0, 84, 0, 134, 129, 135, 246, - 0, 0, 316 + 152, 146, 150, 52, 208, 254, 164, 209, 383, 167, + 116, 117, 158, 116, 117, 162, 430, 162, 239, 163, + 162, 165, 166, 125, 278, 118, 233, 5, 118, 13, + 14, 15, 267, 262, 16, 152, 17, 18, 19, 20, + 21, 22, 23, 24, 25, 26, 27, 419, 118, 119, + 271, 212, 119, 116, 117, 322, 323, 456, 310, 467, + 120, 276, 119, 120, 311, 387, 312, 198, 118, 207, + 7, 277, 473, 120, 470, 199, 388, 263, 53, 453, + 58, 55, 211, 121, 10, 11, 121, 122, 200, 275, + 122, 201, 119, 310, 233, 468, 324, 123, 202, 311, + 230, 68, 313, 120, 314, 124, 121, 321, 124, 442, + 292, 56, 203, 72, 73, 59, 72, 73, 124, 310, + 414, 124, 377, 10, 11, 311, 121, 331, 244, 293, + 122, 173, 378, 168, 415, 204, 205, 436, 289, 314, + 162, 169, 175, 174, 176, 88, 332, 437, 124, 299, + 177, 89, 443, 458, 416, 245, 341, 178, 179, 180, + 71, 181, 444, 182, 170, 314, 417, 68, 153, 91, + 92, 471, 183, 249, 72, 73, 432, 93, 171, 248, + 154, 249, 57, 420, 219, 250, 472, 152, 433, 184, + 185, 220, 424, 250, 347, 236, 1, 2, 348, 94, + 95, 255, 237, 112, 256, 257, 113, 114, 258, 99, + 100, 101, 102, 82, 96, 83, 259, 399, 400, 401, + 402, 403, 404, 405, 406, 407, 408, 409, 410, 63, + 64, 65, 66, 67, 260, 80, 78, 79, 367, 368, + 369, 370, 10, 11, 72, 73, 217, 218, 71, 225, + 379, 380, 81, 86, 84, 87, 98, 425, 143, 104, + 152, 392, 150, 110, 138, 105, 106, 107, 412, 108, + 141, 109, 136, 137, 215, 140, 222, 243, 156, 58, + -66, 268, 210, 159, 297, 216, 224, 229, 152, 213, + 234, 235, 288, 347, 240, 241, 242, 247, 253, 265, + 431, 270, 272, 273, 283, 284, 286, 295, 300, -229, + 290, 302, 304, 291, 306, 308, 327, 317, 318, 320, + 334, 329, 335, 452, 337, 343, 340, 360, 350, 342, + 349, 351, 362, 353, 354, 364, 372, 397, 355, 356, + 357, 385, 358, 359, 366, 374, 375, 152, 392, 150, + 376, 382, 386, 413, 152, 426, 347, 421, 422, 423, + 428, 424, 438, 439, 445, 446, 449, 464, 447, 448, + 459, 460, 347, 461, 462, 463, 466, 454, 465, 474, + 469, 294, 142, 339, 282, 451, 435, 147, 226, 285, + 214, 148, 309, 0, 0, 0, 269 }; static const yytype_int16 yycheck[] = { - 82, 82, 82, 113, 92, 166, 100, 22, 23, 103, - 156, 99, 22, 101, 102, 349, 98, 22, 100, 37, - 325, 103, 37, 25, 170, 28, 29, 28, 24, 25, - 64, 113, 24, 36, 30, 36, 0, 28, 30, 28, - 125, 24, 188, 39, 412, 36, 61, 30, 49, 32, - 61, 52, 34, 8, 61, 58, 59, 72, 59, 98, - 428, 72, 147, 59, 98, 72, 55, 69, 59, 411, - 73, 53, 73, 106, 70, 71, 410, 95, 70, 98, - 95, 386, 73, 98, 426, 68, 422, 70, 98, 26, - 424, 237, 98, 98, 98, 96, 97, 433, 113, 110, - 111, 38, 113, 110, 111, 33, 113, 40, 41, 24, - 98, 26, 60, 98, 27, 209, 29, 45, 98, 67, - 208, 60, 35, 38, 212, 26, 33, 209, 67, 42, - 43, 44, 106, 46, 107, 48, 47, 38, 45, 50, - 51, 114, 252, 54, 57, 108, 3, 4, 81, 112, - 107, 62, 19, 20, 21, 22, 23, 114, 104, 26, - 107, 74, 75, 109, 107, 326, 106, 114, 112, 80, - 252, 114, 110, 111, 5, 6, 7, 259, 9, 10, - 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, - 21, 83, 84, 85, 86, 87, 88, 89, 90, 91, - 92, 93, 94, 99, 22, 23, 102, 103, 99, 100, - 101, 102, 76, 77, 78, 79, 110, 111, 107, 112, - 107, 107, 112, 107, 107, 107, 107, 102, 107, 107, - 98, 64, 22, 321, 344, 317, 317, 317, 108, 108, - 98, 22, 108, 63, 109, 107, 107, 107, 107, 65, - 82, 22, 108, 108, 108, 108, 108, 108, 108, 108, - 108, 107, 344, 102, 109, 112, 22, 349, 22, 22, - 66, 108, 108, 108, 22, 22, 28, 22, 108, 102, - 109, 108, 107, 107, 22, 113, 22, 109, 109, 22, - 109, 22, 109, 22, 109, 109, 109, 109, 408, 109, - 107, 56, 22, 109, 109, 109, 109, 22, 109, 31, - 107, 109, 114, 108, 108, 65, 108, 104, 107, 401, - 401, 401, 104, 109, 109, 197, 408, 250, 410, 401, - 79, 230, 354, -1, 55, -1, 82, 80, 82, 183, - -1, -1, 424 + 89, 89, 89, 9, 124, 181, 106, 125, 23, 109, + 23, 24, 98, 23, 24, 104, 382, 106, 171, 105, + 109, 107, 108, 77, 23, 38, 162, 0, 38, 5, + 6, 7, 185, 26, 10, 124, 12, 13, 14, 15, + 16, 17, 18, 19, 20, 21, 22, 358, 38, 62, + 203, 137, 62, 23, 24, 41, 42, 447, 25, 459, + 73, 109, 62, 73, 31, 62, 33, 29, 38, 123, + 8, 119, 472, 73, 464, 37, 73, 70, 108, 445, + 65, 99, 136, 96, 99, 100, 96, 100, 50, 207, + 100, 53, 62, 25, 230, 461, 82, 110, 60, 31, + 110, 100, 69, 73, 71, 118, 96, 260, 118, 420, + 228, 99, 74, 113, 114, 100, 113, 114, 118, 25, + 26, 118, 109, 99, 100, 31, 96, 35, 29, 229, + 100, 34, 119, 29, 40, 97, 98, 109, 224, 71, + 229, 37, 28, 46, 30, 111, 54, 119, 118, 235, + 36, 117, 109, 448, 60, 56, 276, 43, 44, 45, + 99, 47, 119, 49, 60, 71, 72, 100, 27, 29, + 30, 466, 58, 27, 113, 114, 34, 37, 74, 25, + 39, 27, 99, 359, 61, 39, 106, 276, 46, 75, + 76, 68, 112, 39, 283, 61, 3, 4, 284, 59, + 60, 48, 68, 101, 51, 52, 104, 105, 55, 101, + 102, 103, 104, 9, 74, 11, 63, 84, 85, 86, + 87, 88, 89, 90, 91, 92, 93, 94, 95, 19, + 20, 21, 22, 23, 81, 108, 26, 27, 77, 78, + 79, 80, 99, 100, 113, 114, 23, 24, 99, 100, + 113, 114, 108, 117, 99, 117, 109, 377, 23, 109, + 349, 349, 349, 104, 99, 109, 109, 109, 354, 109, + 100, 109, 109, 109, 99, 109, 23, 64, 111, 65, + 111, 66, 111, 115, 83, 112, 109, 109, 377, 117, + 109, 109, 112, 382, 111, 111, 111, 111, 111, 111, + 386, 111, 111, 111, 109, 109, 117, 104, 23, 104, + 116, 23, 23, 116, 23, 67, 23, 111, 111, 111, + 111, 29, 23, 443, 23, 104, 110, 23, 110, 112, + 118, 111, 23, 109, 109, 23, 23, 23, 112, 112, + 112, 347, 112, 112, 112, 112, 112, 436, 436, 436, + 112, 109, 109, 57, 443, 23, 445, 112, 112, 112, + 23, 112, 112, 32, 109, 109, 66, 106, 111, 111, + 111, 119, 461, 109, 109, 112, 106, 446, 112, 112, + 462, 230, 87, 273, 210, 436, 388, 89, 159, 213, + 139, 89, 253, -1, -1, -1, 198 }; /* YYSTOS[STATE-NUM] -- The (internal number of the) accessing symbol of state STATE-NUM. */ -static const yytype_uint8 yystos[] = +static const yytype_uint16 yystos[] = { - 0, 3, 4, 116, 117, 0, 118, 8, 119, 120, - 98, 5, 6, 7, 9, 10, 11, 12, 13, 14, - 15, 16, 17, 18, 19, 20, 21, 121, 122, 123, - 124, 125, 126, 127, 128, 129, 130, 131, 132, 135, - 158, 159, 166, 167, 168, 230, 232, 235, 248, 106, - 233, 98, 98, 98, 98, 231, 64, 98, 138, 144, - 236, 138, 138, 138, 138, 138, 98, 139, 152, 110, - 111, 137, 229, 138, 106, 106, 98, 234, 112, 112, - 112, 108, 112, 170, 234, 28, 29, 36, 58, 59, - 73, 237, 107, 99, 100, 101, 102, 157, 107, 107, - 107, 107, 107, 107, 102, 154, 22, 23, 37, 61, - 72, 95, 98, 113, 143, 145, 160, 174, 177, 213, - 217, 220, 224, 226, 227, 107, 107, 98, 160, 236, - 22, 169, 173, 177, 213, 223, 225, 227, 228, 229, - 26, 38, 238, 108, 242, 137, 136, 229, 137, 136, - 137, 137, 136, 28, 36, 59, 73, 165, 33, 45, - 27, 29, 35, 42, 43, 44, 46, 48, 57, 74, - 75, 178, 180, 183, 185, 187, 191, 194, 196, 198, - 200, 203, 212, 28, 36, 49, 52, 59, 73, 96, - 97, 161, 228, 99, 102, 103, 156, 108, 143, 98, - 109, 22, 23, 60, 67, 239, 22, 245, 107, 107, - 143, 107, 107, 60, 67, 241, 242, 108, 108, 108, - 63, 28, 55, 195, 108, 24, 26, 38, 184, 240, - 108, 240, 47, 50, 51, 54, 62, 80, 207, 25, - 69, 199, 108, 244, 242, 65, 241, 108, 242, 108, - 108, 163, 107, 114, 22, 146, 147, 148, 152, 107, - 112, 171, 109, 137, 136, 102, 155, 82, 133, 137, - 22, 221, 22, 222, 22, 197, 22, 190, 66, 190, - 24, 30, 32, 68, 70, 179, 189, 108, 108, 208, - 108, 242, 40, 41, 81, 204, 205, 22, 247, 28, - 188, 34, 53, 192, 108, 22, 164, 22, 162, 164, - 228, 109, 102, 153, 140, 141, 229, 113, 108, 243, - 107, 107, 109, 109, 109, 109, 109, 22, 211, 22, - 209, 22, 210, 109, 76, 77, 78, 79, 193, 22, - 206, 109, 109, 109, 107, 114, 110, 111, 149, 107, - 22, 98, 142, 61, 72, 172, 175, 176, 177, 214, - 215, 218, 223, 22, 246, 83, 84, 85, 86, 87, - 88, 89, 90, 91, 92, 93, 94, 134, 137, 56, - 25, 39, 59, 71, 181, 189, 240, 109, 109, 109, - 109, 228, 22, 150, 22, 151, 141, 33, 45, 201, - 203, 107, 114, 109, 31, 182, 186, 189, 107, 114, - 107, 108, 108, 65, 202, 175, 228, 141, 216, 221, - 219, 222, 108, 114, 107, 109, 104, 109, 104, 206, - 141, 221, 222, 104, 206, 109 + 0, 3, 4, 121, 122, 0, 123, 8, 124, 125, + 99, 100, 262, 5, 6, 7, 10, 12, 13, 14, + 15, 16, 17, 18, 19, 20, 21, 22, 126, 127, + 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, + 138, 141, 170, 171, 178, 179, 180, 242, 244, 245, + 248, 261, 262, 108, 246, 99, 99, 99, 65, 100, + 145, 151, 249, 145, 145, 145, 145, 145, 100, 146, + 159, 99, 113, 114, 144, 166, 168, 241, 145, 145, + 108, 108, 9, 11, 99, 247, 117, 117, 111, 117, + 182, 29, 30, 37, 59, 60, 74, 250, 109, 101, + 102, 103, 104, 164, 109, 109, 109, 109, 109, 109, + 104, 161, 101, 104, 105, 163, 23, 24, 38, 62, + 73, 96, 100, 110, 118, 150, 152, 172, 186, 189, + 225, 229, 232, 236, 238, 239, 109, 109, 99, 243, + 109, 100, 172, 23, 181, 185, 189, 225, 235, 237, + 239, 240, 241, 27, 39, 251, 111, 255, 144, 115, + 165, 142, 241, 144, 142, 144, 144, 142, 29, 37, + 60, 74, 177, 34, 46, 28, 30, 36, 43, 44, + 45, 47, 49, 58, 75, 76, 190, 192, 195, 197, + 199, 203, 206, 208, 210, 212, 215, 224, 29, 37, + 50, 53, 60, 74, 97, 98, 173, 150, 240, 163, + 111, 150, 144, 117, 247, 99, 112, 23, 24, 61, + 68, 252, 23, 258, 109, 100, 166, 167, 169, 109, + 110, 143, 150, 238, 109, 109, 61, 68, 254, 255, + 111, 111, 111, 64, 29, 56, 207, 111, 25, 27, + 39, 196, 253, 111, 253, 48, 51, 52, 55, 63, + 81, 219, 26, 70, 211, 111, 257, 255, 66, 254, + 111, 255, 111, 111, 175, 163, 109, 119, 23, 153, + 154, 155, 159, 109, 109, 249, 117, 183, 112, 144, + 116, 116, 163, 142, 143, 104, 162, 83, 139, 144, + 23, 233, 23, 234, 23, 209, 23, 202, 67, 202, + 25, 31, 33, 69, 71, 191, 201, 111, 111, 220, + 111, 255, 41, 42, 82, 216, 217, 23, 260, 29, + 200, 35, 54, 204, 111, 23, 176, 23, 174, 176, + 110, 240, 112, 104, 160, 147, 148, 241, 144, 118, + 110, 111, 256, 109, 109, 112, 112, 112, 112, 112, + 23, 223, 23, 221, 23, 222, 112, 77, 78, 79, + 80, 205, 23, 218, 112, 112, 112, 109, 119, 113, + 114, 156, 109, 23, 149, 262, 109, 62, 73, 184, + 187, 188, 189, 226, 227, 230, 235, 23, 259, 84, + 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, + 95, 140, 144, 57, 26, 40, 60, 72, 193, 201, + 253, 112, 112, 112, 112, 240, 23, 157, 23, 158, + 148, 144, 34, 46, 213, 215, 109, 119, 112, 32, + 194, 198, 201, 109, 119, 109, 109, 111, 111, 66, + 214, 187, 240, 148, 139, 228, 233, 231, 234, 111, + 119, 109, 109, 112, 106, 112, 106, 218, 148, 140, + 233, 234, 106, 218, 112 }; #define yyerrok (yyerrstatus = 0) @@ -2068,7 +2119,7 @@ yyreduce: case 3: /* Line 1455 of yacc.c */ -#line 260 "program_parse.y" +#line 272 "program_parse.y" { if (state->prog->Target != GL_VERTEX_PROGRAM_ARB) { yyerror(& (yylsp[(1) - (1)]), state, "invalid fragment program header"); @@ -2081,7 +2132,7 @@ yyreduce: case 4: /* Line 1455 of yacc.c */ -#line 268 "program_parse.y" +#line 280 "program_parse.y" { if (state->prog->Target != GL_FRAGMENT_PROGRAM_ARB) { yyerror(& (yylsp[(1) - (1)]), state, "invalid vertex program header"); @@ -2096,7 +2147,7 @@ yyreduce: case 7: /* Line 1455 of yacc.c */ -#line 284 "program_parse.y" +#line 296 "program_parse.y" { int valid = 0; @@ -2121,7 +2172,7 @@ yyreduce: case 10: /* Line 1455 of yacc.c */ -#line 310 "program_parse.y" +#line 322 "program_parse.y" { if ((yyvsp[(1) - (2)].inst) != NULL) { if (state->inst_tail == NULL) { @@ -2141,7 +2192,7 @@ yyreduce: case 12: /* Line 1455 of yacc.c */ -#line 328 "program_parse.y" +#line 340 "program_parse.y" { (yyval.inst) = (yyvsp[(1) - (1)].inst); state->prog->NumAluInstructions++; @@ -2151,85 +2202,79 @@ yyreduce: case 13: /* Line 1455 of yacc.c */ -#line 333 "program_parse.y" +#line 345 "program_parse.y" { (yyval.inst) = (yyvsp[(1) - (1)].inst); state->prog->NumTexInstructions++; ;} break; - case 23: + case 24: /* Line 1455 of yacc.c */ -#line 353 "program_parse.y" +#line 366 "program_parse.y" { (yyval.inst) = asm_instruction_ctor(OPCODE_ARL, & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL); ;} break; - case 24: + case 25: /* Line 1455 of yacc.c */ -#line 359 "program_parse.y" +#line 372 "program_parse.y" { - (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (4)].temp_inst).Opcode, & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL); - (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (4)].temp_inst).SaturateMode; + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (4)].temp_inst), & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL); ;} break; - case 25: + case 26: /* Line 1455 of yacc.c */ -#line 366 "program_parse.y" +#line 378 "program_parse.y" { - (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (4)].temp_inst).Opcode, & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL); - (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (4)].temp_inst).SaturateMode; + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (4)].temp_inst), & (yyvsp[(2) - (4)].dst_reg), & (yyvsp[(4) - (4)].src_reg), NULL, NULL); ;} break; - case 26: + case 27: /* Line 1455 of yacc.c */ -#line 373 "program_parse.y" +#line 384 "program_parse.y" { - (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (6)].temp_inst).Opcode, & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL); - (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (6)].temp_inst).SaturateMode; + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL); ;} break; - case 27: + case 28: /* Line 1455 of yacc.c */ -#line 381 "program_parse.y" +#line 391 "program_parse.y" { - (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (6)].temp_inst).Opcode, & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL); - (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (6)].temp_inst).SaturateMode; + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), & (yyvsp[(6) - (6)].src_reg), NULL); ;} break; - case 28: + case 29: /* Line 1455 of yacc.c */ -#line 389 "program_parse.y" +#line 398 "program_parse.y" { - (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (8)].temp_inst).Opcode, & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), & (yyvsp[(6) - (8)].src_reg), & (yyvsp[(8) - (8)].src_reg)); - (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (8)].temp_inst).SaturateMode; + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (8)].temp_inst), & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), & (yyvsp[(6) - (8)].src_reg), & (yyvsp[(8) - (8)].src_reg)); ;} break; - case 29: + case 30: /* Line 1455 of yacc.c */ -#line 396 "program_parse.y" +#line 404 "program_parse.y" { - (yyval.inst) = asm_instruction_ctor((yyvsp[(1) - (8)].temp_inst).Opcode, & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), NULL, NULL); + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (8)].temp_inst), & (yyvsp[(2) - (8)].dst_reg), & (yyvsp[(4) - (8)].src_reg), NULL, NULL); if ((yyval.inst) != NULL) { const GLbitfield tex_mask = (1U << (yyvsp[(6) - (8)].integer)); GLbitfield shadow_tex = 0; GLbitfield target_mask = 0; - (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (8)].temp_inst).SaturateMode; (yyval.inst)->Base.TexSrcUnit = (yyvsp[(6) - (8)].integer); if ((yyvsp[(8) - (8)].integer) < 0) { @@ -2265,113 +2310,173 @@ yyreduce: ;} break; - case 30: + case 31: /* Line 1455 of yacc.c */ -#line 441 "program_parse.y" +#line 448 "program_parse.y" { (yyval.inst) = asm_instruction_ctor(OPCODE_KIL, NULL, & (yyvsp[(2) - (2)].src_reg), NULL, NULL); state->fragment.UsesKill = 1; ;} break; - case 31: + case 32: /* Line 1455 of yacc.c */ -#line 448 "program_parse.y" +#line 453 "program_parse.y" + { + (yyval.inst) = asm_instruction_ctor(OPCODE_KIL_NV, NULL, NULL, NULL, NULL); + (yyval.inst)->Base.DstReg.CondMask = (yyvsp[(2) - (2)].dst_reg).CondMask; + (yyval.inst)->Base.DstReg.CondSwizzle = (yyvsp[(2) - (2)].dst_reg).CondSwizzle; + (yyval.inst)->Base.DstReg.CondSrc = (yyvsp[(2) - (2)].dst_reg).CondSrc; + state->fragment.UsesKill = 1; + ;} + break; + + case 33: + +/* Line 1455 of yacc.c */ +#line 463 "program_parse.y" + { + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (12)].temp_inst), & (yyvsp[(2) - (12)].dst_reg), & (yyvsp[(4) - (12)].src_reg), & (yyvsp[(6) - (12)].src_reg), & (yyvsp[(8) - (12)].src_reg)); + if ((yyval.inst) != NULL) { + const GLbitfield tex_mask = (1U << (yyvsp[(10) - (12)].integer)); + GLbitfield shadow_tex = 0; + GLbitfield target_mask = 0; + + + (yyval.inst)->Base.TexSrcUnit = (yyvsp[(10) - (12)].integer); + + if ((yyvsp[(12) - (12)].integer) < 0) { + shadow_tex = tex_mask; + + (yyval.inst)->Base.TexSrcTarget = -(yyvsp[(12) - (12)].integer); + (yyval.inst)->Base.TexShadow = 1; + } else { + (yyval.inst)->Base.TexSrcTarget = (yyvsp[(12) - (12)].integer); + } + + target_mask = (1U << (yyval.inst)->Base.TexSrcTarget); + + /* If this texture unit was previously accessed and that access + * had a different texture target, generate an error. + * + * If this texture unit was previously accessed and that access + * had a different shadow mode, generate an error. + */ + if ((state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] != 0) + && ((state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] != target_mask) + || ((state->prog->ShadowSamplers & tex_mask) + != shadow_tex))) { + yyerror(& (yylsp[(12) - (12)]), state, + "multiple targets used on one texture image unit"); + YYERROR; + } + + + state->prog->TexturesUsed[(yyvsp[(10) - (12)].integer)] |= target_mask; + state->prog->ShadowSamplers |= shadow_tex; + } + ;} + break; + + case 34: + +/* Line 1455 of yacc.c */ +#line 507 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;} break; - case 32: + case 35: /* Line 1455 of yacc.c */ -#line 453 "program_parse.y" +#line 512 "program_parse.y" { (yyval.integer) = TEXTURE_1D_INDEX; ;} break; - case 33: + case 36: /* Line 1455 of yacc.c */ -#line 454 "program_parse.y" +#line 513 "program_parse.y" { (yyval.integer) = TEXTURE_2D_INDEX; ;} break; - case 34: + case 37: /* Line 1455 of yacc.c */ -#line 455 "program_parse.y" +#line 514 "program_parse.y" { (yyval.integer) = TEXTURE_3D_INDEX; ;} break; - case 35: + case 38: /* Line 1455 of yacc.c */ -#line 456 "program_parse.y" +#line 515 "program_parse.y" { (yyval.integer) = TEXTURE_CUBE_INDEX; ;} break; - case 36: + case 39: /* Line 1455 of yacc.c */ -#line 457 "program_parse.y" +#line 516 "program_parse.y" { (yyval.integer) = TEXTURE_RECT_INDEX; ;} break; - case 37: + case 40: /* Line 1455 of yacc.c */ -#line 458 "program_parse.y" +#line 517 "program_parse.y" { (yyval.integer) = -TEXTURE_1D_INDEX; ;} break; - case 38: + case 41: /* Line 1455 of yacc.c */ -#line 459 "program_parse.y" +#line 518 "program_parse.y" { (yyval.integer) = -TEXTURE_2D_INDEX; ;} break; - case 39: + case 42: /* Line 1455 of yacc.c */ -#line 460 "program_parse.y" +#line 519 "program_parse.y" { (yyval.integer) = -TEXTURE_RECT_INDEX; ;} break; - case 40: + case 43: /* Line 1455 of yacc.c */ -#line 461 "program_parse.y" +#line 520 "program_parse.y" { (yyval.integer) = TEXTURE_1D_ARRAY_INDEX; ;} break; - case 41: + case 44: /* Line 1455 of yacc.c */ -#line 462 "program_parse.y" +#line 521 "program_parse.y" { (yyval.integer) = TEXTURE_2D_ARRAY_INDEX; ;} break; - case 42: + case 45: /* Line 1455 of yacc.c */ -#line 463 "program_parse.y" +#line 522 "program_parse.y" { (yyval.integer) = -TEXTURE_1D_ARRAY_INDEX; ;} break; - case 43: + case 46: /* Line 1455 of yacc.c */ -#line 464 "program_parse.y" +#line 523 "program_parse.y" { (yyval.integer) = -TEXTURE_2D_ARRAY_INDEX; ;} break; - case 44: + case 47: /* Line 1455 of yacc.c */ -#line 468 "program_parse.y" +#line 527 "program_parse.y" { /* FIXME: Is this correct? Should the extenedSwizzle be applied * FIXME: to the existing swizzle? @@ -2379,31 +2484,81 @@ yyreduce: (yyvsp[(4) - (6)].src_reg).Base.Swizzle = (yyvsp[(6) - (6)].swiz_mask).swizzle; (yyvsp[(4) - (6)].src_reg).Base.Negate = (yyvsp[(6) - (6)].swiz_mask).mask; - (yyval.inst) = asm_instruction_ctor(OPCODE_SWZ, & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), NULL, NULL); - (yyval.inst)->Base.SaturateMode = (yyvsp[(1) - (6)].temp_inst).SaturateMode; + (yyval.inst) = asm_instruction_copy_ctor(& (yyvsp[(1) - (6)].temp_inst), & (yyvsp[(2) - (6)].dst_reg), & (yyvsp[(4) - (6)].src_reg), NULL, NULL); ;} break; - case 45: + case 48: /* Line 1455 of yacc.c */ -#line 481 "program_parse.y" +#line 539 "program_parse.y" { - (yyval.src_reg) = (yyvsp[(2) - (3)].src_reg); + (yyval.src_reg) = (yyvsp[(2) - (2)].src_reg); - if ((yyvsp[(1) - (3)].negate)) { + if ((yyvsp[(1) - (2)].negate)) { + (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate; + } + ;} + break; + + case 49: + +/* Line 1455 of yacc.c */ +#line 547 "program_parse.y" + { + (yyval.src_reg) = (yyvsp[(3) - (4)].src_reg); + + if (!state->option.NV_fragment) { + yyerror(& (yylsp[(2) - (4)]), state, "unexpected character '|'"); + YYERROR; + } + + if ((yyvsp[(1) - (4)].negate)) { (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate; } + (yyval.src_reg).Base.Abs = 1; + ;} + break; + + case 50: + +/* Line 1455 of yacc.c */ +#line 564 "program_parse.y" + { + (yyval.src_reg) = (yyvsp[(1) - (2)].src_reg); + (yyval.src_reg).Base.Swizzle = _mesa_combine_swizzles((yyval.src_reg).Base.Swizzle, - (yyvsp[(3) - (3)].swiz_mask).swizzle); + (yyvsp[(2) - (2)].swiz_mask).swizzle); ;} break; - case 46: + case 51: + +/* Line 1455 of yacc.c */ +#line 571 "program_parse.y" + { + struct asm_symbol temp_sym; + + if (!state->option.NV_fragment) { + yyerror(& (yylsp[(1) - (1)]), state, "expected scalar suffix"); + YYERROR; + } + + memset(& temp_sym, 0, sizeof(temp_sym)); + temp_sym.param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & temp_sym, & (yyvsp[(1) - (1)].vector)); + + init_src_reg(& (yyval.src_reg)); + (yyval.src_reg).Base.File = PROGRAM_CONSTANT; + (yyval.src_reg).Base.Index = temp_sym.param_binding_begin; + ;} + break; + + case 52: /* Line 1455 of yacc.c */ -#line 494 "program_parse.y" +#line 590 "program_parse.y" { (yyval.src_reg) = (yyvsp[(2) - (3)].src_reg); @@ -2416,13 +2571,38 @@ yyreduce: ;} break; - case 47: + case 53: /* Line 1455 of yacc.c */ -#line 507 "program_parse.y" +#line 601 "program_parse.y" { - (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg); - (yyval.dst_reg).WriteMask = (yyvsp[(2) - (2)].swiz_mask).mask; + (yyval.src_reg) = (yyvsp[(3) - (5)].src_reg); + + if (!state->option.NV_fragment) { + yyerror(& (yylsp[(2) - (5)]), state, "unexpected character '|'"); + YYERROR; + } + + if ((yyvsp[(1) - (5)].negate)) { + (yyval.src_reg).Base.Negate = ~(yyval.src_reg).Base.Negate; + } + + (yyval.src_reg).Base.Abs = 1; + (yyval.src_reg).Base.Swizzle = _mesa_combine_swizzles((yyval.src_reg).Base.Swizzle, + (yyvsp[(4) - (5)].swiz_mask).swizzle); + ;} + break; + + case 54: + +/* Line 1455 of yacc.c */ +#line 621 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(1) - (3)].dst_reg); + (yyval.dst_reg).WriteMask = (yyvsp[(2) - (3)].swiz_mask).mask; + (yyval.dst_reg).CondMask = (yyvsp[(3) - (3)].dst_reg).CondMask; + (yyval.dst_reg).CondSwizzle = (yyvsp[(3) - (3)].dst_reg).CondSwizzle; + (yyval.dst_reg).CondSrc = (yyvsp[(3) - (3)].dst_reg).CondSrc; if ((yyval.dst_reg).File == PROGRAM_OUTPUT) { /* Technically speaking, this should check that it is in @@ -2431,7 +2611,7 @@ yyreduce: */ if (state->option.PositionInvariant && ((yyval.dst_reg).Index == VERT_RESULT_HPOS)) { - yyerror(& (yylsp[(1) - (2)]), state, "position-invariant programs cannot " + yyerror(& (yylsp[(1) - (3)]), state, "position-invariant programs cannot " "write position"); YYERROR; } @@ -2441,10 +2621,10 @@ yyreduce: ;} break; - case 48: + case 55: /* Line 1455 of yacc.c */ -#line 529 "program_parse.y" +#line 646 "program_parse.y" { init_dst_reg(& (yyval.dst_reg)); (yyval.dst_reg).File = PROGRAM_ADDRESS; @@ -2453,10 +2633,10 @@ yyreduce: ;} break; - case 49: + case 56: /* Line 1455 of yacc.c */ -#line 538 "program_parse.y" +#line 655 "program_parse.y" { const unsigned xyzw_valid = ((yyvsp[(1) - (7)].ext_swizzle).xyzw_valid << 0) @@ -2489,20 +2669,20 @@ yyreduce: ;} break; - case 50: + case 57: /* Line 1455 of yacc.c */ -#line 571 "program_parse.y" +#line 688 "program_parse.y" { (yyval.ext_swizzle) = (yyvsp[(2) - (2)].ext_swizzle); (yyval.ext_swizzle).negate = ((yyvsp[(1) - (2)].negate)) ? 1 : 0; ;} break; - case 51: + case 58: /* Line 1455 of yacc.c */ -#line 578 "program_parse.y" +#line 695 "program_parse.y" { if (((yyvsp[(1) - (1)].integer) != 0) && ((yyvsp[(1) - (1)].integer) != 1)) { yyerror(& (yylsp[(1) - (1)]), state, "invalid extended swizzle selector"); @@ -2519,10 +2699,10 @@ yyreduce: ;} break; - case 52: + case 59: /* Line 1455 of yacc.c */ -#line 593 "program_parse.y" +#line 710 "program_parse.y" { if (strlen((yyvsp[(1) - (1)].string)) > 1) { yyerror(& (yylsp[(1) - (1)]), state, "invalid extended swizzle selector"); @@ -2572,10 +2752,10 @@ yyreduce: ;} break; - case 53: + case 60: /* Line 1455 of yacc.c */ -#line 643 "program_parse.y" +#line 760 "program_parse.y" { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); @@ -2619,10 +2799,10 @@ yyreduce: ;} break; - case 54: + case 61: /* Line 1455 of yacc.c */ -#line 685 "program_parse.y" +#line 802 "program_parse.y" { init_src_reg(& (yyval.src_reg)); (yyval.src_reg).Base.File = PROGRAM_INPUT; @@ -2635,10 +2815,10 @@ yyreduce: ;} break; - case 55: + case 62: /* Line 1455 of yacc.c */ -#line 696 "program_parse.y" +#line 813 "program_parse.y" { if (! (yyvsp[(3) - (4)].src_reg).Base.RelAddr && ((unsigned) (yyvsp[(3) - (4)].src_reg).Base.Index >= (yyvsp[(1) - (4)].sym)->param_binding_length)) { @@ -2661,10 +2841,10 @@ yyreduce: ;} break; - case 56: + case 63: /* Line 1455 of yacc.c */ -#line 717 "program_parse.y" +#line 834 "program_parse.y" { init_src_reg(& (yyval.src_reg)); (yyval.src_reg).Base.File = ((yyvsp[(1) - (1)].temp_sym).name != NULL) @@ -2674,10 +2854,10 @@ yyreduce: ;} break; - case 57: + case 64: /* Line 1455 of yacc.c */ -#line 727 "program_parse.y" +#line 844 "program_parse.y" { init_dst_reg(& (yyval.dst_reg)); (yyval.dst_reg).File = PROGRAM_OUTPUT; @@ -2685,10 +2865,10 @@ yyreduce: ;} break; - case 58: + case 65: /* Line 1455 of yacc.c */ -#line 733 "program_parse.y" +#line 850 "program_parse.y" { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); @@ -2719,10 +2899,10 @@ yyreduce: ;} break; - case 59: + case 66: /* Line 1455 of yacc.c */ -#line 764 "program_parse.y" +#line 881 "program_parse.y" { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); @@ -2739,20 +2919,20 @@ yyreduce: ;} break; - case 62: + case 69: /* Line 1455 of yacc.c */ -#line 783 "program_parse.y" +#line 900 "program_parse.y" { init_src_reg(& (yyval.src_reg)); (yyval.src_reg).Base.Index = (yyvsp[(1) - (1)].integer); ;} break; - case 63: + case 70: /* Line 1455 of yacc.c */ -#line 790 "program_parse.y" +#line 907 "program_parse.y" { /* FINISHME: Add support for multiple address registers. */ @@ -2764,31 +2944,31 @@ yyreduce: ;} break; - case 64: + case 71: /* Line 1455 of yacc.c */ -#line 801 "program_parse.y" +#line 918 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 65: + case 72: /* Line 1455 of yacc.c */ -#line 802 "program_parse.y" +#line 919 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;} break; - case 66: + case 73: /* Line 1455 of yacc.c */ -#line 803 "program_parse.y" +#line 920 "program_parse.y" { (yyval.integer) = -(yyvsp[(2) - (2)].integer); ;} break; - case 67: + case 74: /* Line 1455 of yacc.c */ -#line 807 "program_parse.y" +#line 924 "program_parse.y" { if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 63)) { char s[100]; @@ -2802,10 +2982,10 @@ yyreduce: ;} break; - case 68: + case 75: /* Line 1455 of yacc.c */ -#line 821 "program_parse.y" +#line 938 "program_parse.y" { if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 64)) { char s[100]; @@ -2819,10 +2999,10 @@ yyreduce: ;} break; - case 69: + case 76: /* Line 1455 of yacc.c */ -#line 835 "program_parse.y" +#line 952 "program_parse.y" { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string)); @@ -2840,10 +3020,10 @@ yyreduce: ;} break; - case 70: + case 77: /* Line 1455 of yacc.c */ -#line 853 "program_parse.y" +#line 970 "program_parse.y" { if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) { yyerror(& (yylsp[(1) - (1)]), state, "invalid address component selector"); @@ -2854,10 +3034,10 @@ yyreduce: ;} break; - case 71: + case 78: /* Line 1455 of yacc.c */ -#line 864 "program_parse.y" +#line 981 "program_parse.y" { if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) { yyerror(& (yylsp[(1) - (1)]), state, @@ -2869,24 +3049,125 @@ yyreduce: ;} break; - case 76: + case 83: /* Line 1455 of yacc.c */ -#line 880 "program_parse.y" +#line 997 "program_parse.y" { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;} break; - case 81: + case 88: /* Line 1455 of yacc.c */ -#line 884 "program_parse.y" +#line 1001 "program_parse.y" { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;} break; - case 88: + case 89: /* Line 1455 of yacc.c */ -#line 896 "program_parse.y" +#line 1005 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg); + ;} + break; + + case 90: + +/* Line 1455 of yacc.c */ +#line 1009 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg); + ;} + break; + + case 91: + +/* Line 1455 of yacc.c */ +#line 1013 "program_parse.y" + { + (yyval.dst_reg).CondMask = COND_TR; + (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP; + (yyval.dst_reg).CondSrc = 0; + ;} + break; + + case 92: + +/* Line 1455 of yacc.c */ +#line 1021 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg); + (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle; + ;} + break; + + case 93: + +/* Line 1455 of yacc.c */ +#line 1028 "program_parse.y" + { + (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg); + (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle; + ;} + break; + + case 94: + +/* Line 1455 of yacc.c */ +#line 1035 "program_parse.y" + { + const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string)); + if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) { + char *const err_str = + make_error_string("invalid condition code \"%s\"", (yyvsp[(1) - (1)].string)); + + yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL) + ? err_str : "invalid condition code"); + + if (err_str != NULL) { + _mesa_free(err_str); + } + + YYERROR; + } + + (yyval.dst_reg).CondMask = cond; + (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP; + (yyval.dst_reg).CondSrc = 0; + ;} + break; + + case 95: + +/* Line 1455 of yacc.c */ +#line 1058 "program_parse.y" + { + const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string)); + if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) { + char *const err_str = + make_error_string("invalid condition code \"%s\"", (yyvsp[(1) - (1)].string)); + + yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL) + ? err_str : "invalid condition code"); + + if (err_str != NULL) { + _mesa_free(err_str); + } + + YYERROR; + } + + (yyval.dst_reg).CondMask = cond; + (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP; + (yyval.dst_reg).CondSrc = 0; + ;} + break; + + case 102: + +/* Line 1455 of yacc.c */ +#line 1089 "program_parse.y" { struct asm_symbol *const s = declare_variable(state, (yyvsp[(2) - (4)].string), at_attrib, & (yylsp[(2) - (4)])); @@ -2904,55 +3185,55 @@ yyreduce: ;} break; - case 89: + case 103: /* Line 1455 of yacc.c */ -#line 914 "program_parse.y" +#line 1107 "program_parse.y" { (yyval.attrib) = (yyvsp[(2) - (2)].attrib); ;} break; - case 90: + case 104: /* Line 1455 of yacc.c */ -#line 918 "program_parse.y" +#line 1111 "program_parse.y" { (yyval.attrib) = (yyvsp[(2) - (2)].attrib); ;} break; - case 91: + case 105: /* Line 1455 of yacc.c */ -#line 924 "program_parse.y" +#line 1117 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_POS; ;} break; - case 92: + case 106: /* Line 1455 of yacc.c */ -#line 928 "program_parse.y" +#line 1121 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_WEIGHT; ;} break; - case 93: + case 107: /* Line 1455 of yacc.c */ -#line 932 "program_parse.y" +#line 1125 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_NORMAL; ;} break; - case 94: + case 108: /* Line 1455 of yacc.c */ -#line 936 "program_parse.y" +#line 1129 "program_parse.y" { if (!state->ctx->Extensions.EXT_secondary_color) { yyerror(& (yylsp[(2) - (2)]), state, "GL_EXT_secondary_color not supported"); @@ -2963,10 +3244,10 @@ yyreduce: ;} break; - case 95: + case 109: /* Line 1455 of yacc.c */ -#line 945 "program_parse.y" +#line 1138 "program_parse.y" { if (!state->ctx->Extensions.EXT_fog_coord) { yyerror(& (yylsp[(1) - (1)]), state, "GL_EXT_fog_coord not supported"); @@ -2977,38 +3258,38 @@ yyreduce: ;} break; - case 96: + case 110: /* Line 1455 of yacc.c */ -#line 954 "program_parse.y" +#line 1147 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer); ;} break; - case 97: + case 111: /* Line 1455 of yacc.c */ -#line 958 "program_parse.y" +#line 1151 "program_parse.y" { yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported"); YYERROR; ;} break; - case 98: + case 112: /* Line 1455 of yacc.c */ -#line 963 "program_parse.y" +#line 1156 "program_parse.y" { (yyval.attrib) = VERT_ATTRIB_GENERIC0 + (yyvsp[(3) - (4)].integer); ;} break; - case 99: + case 113: /* Line 1455 of yacc.c */ -#line 969 "program_parse.y" +#line 1162 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxAttribs) { yyerror(& (yylsp[(1) - (1)]), state, "invalid vertex attribute reference"); @@ -3019,46 +3300,46 @@ yyreduce: ;} break; - case 103: + case 117: /* Line 1455 of yacc.c */ -#line 983 "program_parse.y" +#line 1176 "program_parse.y" { (yyval.attrib) = FRAG_ATTRIB_WPOS; ;} break; - case 104: + case 118: /* Line 1455 of yacc.c */ -#line 987 "program_parse.y" +#line 1180 "program_parse.y" { (yyval.attrib) = FRAG_ATTRIB_COL0 + (yyvsp[(2) - (2)].integer); ;} break; - case 105: + case 119: /* Line 1455 of yacc.c */ -#line 991 "program_parse.y" +#line 1184 "program_parse.y" { (yyval.attrib) = FRAG_ATTRIB_FOGC; ;} break; - case 106: + case 120: /* Line 1455 of yacc.c */ -#line 995 "program_parse.y" +#line 1188 "program_parse.y" { (yyval.attrib) = FRAG_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer); ;} break; - case 109: + case 123: /* Line 1455 of yacc.c */ -#line 1003 "program_parse.y" +#line 1196 "program_parse.y" { struct asm_symbol *const s = declare_variable(state, (yyvsp[(2) - (3)].string), at_param, & (yylsp[(2) - (3)])); @@ -3074,10 +3355,10 @@ yyreduce: ;} break; - case 110: + case 124: /* Line 1455 of yacc.c */ -#line 1019 "program_parse.y" +#line 1212 "program_parse.y" { if (((yyvsp[(4) - (6)].integer) != 0) && ((unsigned) (yyvsp[(4) - (6)].integer) != (yyvsp[(6) - (6)].temp_sym).param_binding_length)) { yyerror(& (yylsp[(4) - (6)]), state, @@ -3099,19 +3380,19 @@ yyreduce: ;} break; - case 111: + case 125: /* Line 1455 of yacc.c */ -#line 1041 "program_parse.y" +#line 1234 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 112: + case 126: /* Line 1455 of yacc.c */ -#line 1045 "program_parse.y" +#line 1238 "program_parse.y" { if (((yyvsp[(1) - (1)].integer) < 1) || ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxParameters)) { yyerror(& (yylsp[(1) - (1)]), state, "invalid parameter array size"); @@ -3122,38 +3403,38 @@ yyreduce: ;} break; - case 113: + case 127: /* Line 1455 of yacc.c */ -#line 1056 "program_parse.y" +#line 1249 "program_parse.y" { (yyval.temp_sym) = (yyvsp[(2) - (2)].temp_sym); ;} break; - case 114: + case 128: /* Line 1455 of yacc.c */ -#line 1062 "program_parse.y" +#line 1255 "program_parse.y" { (yyval.temp_sym) = (yyvsp[(3) - (4)].temp_sym); ;} break; - case 116: + case 130: /* Line 1455 of yacc.c */ -#line 1069 "program_parse.y" +#line 1262 "program_parse.y" { (yyvsp[(1) - (3)].temp_sym).param_binding_length += (yyvsp[(3) - (3)].temp_sym).param_binding_length; (yyval.temp_sym) = (yyvsp[(1) - (3)].temp_sym); ;} break; - case 117: + case 131: /* Line 1455 of yacc.c */ -#line 1076 "program_parse.y" +#line 1269 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3161,10 +3442,10 @@ yyreduce: ;} break; - case 118: + case 132: /* Line 1455 of yacc.c */ -#line 1082 "program_parse.y" +#line 1275 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3172,10 +3453,10 @@ yyreduce: ;} break; - case 119: + case 133: /* Line 1455 of yacc.c */ -#line 1088 "program_parse.y" +#line 1281 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3183,10 +3464,10 @@ yyreduce: ;} break; - case 120: + case 134: /* Line 1455 of yacc.c */ -#line 1096 "program_parse.y" +#line 1289 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3194,10 +3475,10 @@ yyreduce: ;} break; - case 121: + case 135: /* Line 1455 of yacc.c */ -#line 1102 "program_parse.y" +#line 1295 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3205,10 +3486,10 @@ yyreduce: ;} break; - case 122: + case 136: /* Line 1455 of yacc.c */ -#line 1108 "program_parse.y" +#line 1301 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3216,10 +3497,10 @@ yyreduce: ;} break; - case 123: + case 137: /* Line 1455 of yacc.c */ -#line 1116 "program_parse.y" +#line 1309 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3227,10 +3508,10 @@ yyreduce: ;} break; - case 124: + case 138: /* Line 1455 of yacc.c */ -#line 1122 "program_parse.y" +#line 1315 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3238,10 +3519,10 @@ yyreduce: ;} break; - case 125: + case 139: /* Line 1455 of yacc.c */ -#line 1128 "program_parse.y" +#line 1321 "program_parse.y" { memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym))); (yyval.temp_sym).param_binding_begin = ~0; @@ -3249,101 +3530,101 @@ yyreduce: ;} break; - case 126: + case 140: /* Line 1455 of yacc.c */ -#line 1135 "program_parse.y" +#line 1328 "program_parse.y" { memcpy((yyval.state), (yyvsp[(1) - (1)].state), sizeof((yyval.state))); ;} break; - case 127: + case 141: /* Line 1455 of yacc.c */ -#line 1136 "program_parse.y" +#line 1329 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 128: + case 142: /* Line 1455 of yacc.c */ -#line 1139 "program_parse.y" +#line 1332 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 129: + case 143: /* Line 1455 of yacc.c */ -#line 1140 "program_parse.y" +#line 1333 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 130: + case 144: /* Line 1455 of yacc.c */ -#line 1141 "program_parse.y" +#line 1334 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 131: + case 145: /* Line 1455 of yacc.c */ -#line 1142 "program_parse.y" +#line 1335 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 132: + case 146: /* Line 1455 of yacc.c */ -#line 1143 "program_parse.y" +#line 1336 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 133: + case 147: /* Line 1455 of yacc.c */ -#line 1144 "program_parse.y" +#line 1337 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 134: + case 148: /* Line 1455 of yacc.c */ -#line 1145 "program_parse.y" +#line 1338 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 135: + case 149: /* Line 1455 of yacc.c */ -#line 1146 "program_parse.y" +#line 1339 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 136: + case 150: /* Line 1455 of yacc.c */ -#line 1147 "program_parse.y" +#line 1340 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 137: + case 151: /* Line 1455 of yacc.c */ -#line 1148 "program_parse.y" +#line 1341 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 138: + case 152: /* Line 1455 of yacc.c */ -#line 1149 "program_parse.y" +#line 1342 "program_parse.y" { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;} break; - case 139: + case 153: /* Line 1455 of yacc.c */ -#line 1153 "program_parse.y" +#line 1346 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_MATERIAL; @@ -3352,37 +3633,37 @@ yyreduce: ;} break; - case 140: + case 154: /* Line 1455 of yacc.c */ -#line 1162 "program_parse.y" +#line 1355 "program_parse.y" { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} break; - case 141: + case 155: /* Line 1455 of yacc.c */ -#line 1166 "program_parse.y" +#line 1359 "program_parse.y" { (yyval.integer) = STATE_EMISSION; ;} break; - case 142: + case 156: /* Line 1455 of yacc.c */ -#line 1170 "program_parse.y" +#line 1363 "program_parse.y" { (yyval.integer) = STATE_SHININESS; ;} break; - case 143: + case 157: /* Line 1455 of yacc.c */ -#line 1176 "program_parse.y" +#line 1369 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_LIGHT; @@ -3391,28 +3672,28 @@ yyreduce: ;} break; - case 144: + case 158: /* Line 1455 of yacc.c */ -#line 1185 "program_parse.y" +#line 1378 "program_parse.y" { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} break; - case 145: + case 159: /* Line 1455 of yacc.c */ -#line 1189 "program_parse.y" +#line 1382 "program_parse.y" { (yyval.integer) = STATE_POSITION; ;} break; - case 146: + case 160: /* Line 1455 of yacc.c */ -#line 1193 "program_parse.y" +#line 1386 "program_parse.y" { if (!state->ctx->Extensions.EXT_point_parameters) { yyerror(& (yylsp[(1) - (1)]), state, "GL_ARB_point_parameters not supported"); @@ -3423,57 +3704,57 @@ yyreduce: ;} break; - case 147: + case 161: /* Line 1455 of yacc.c */ -#line 1202 "program_parse.y" +#line 1395 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;} break; - case 148: + case 162: /* Line 1455 of yacc.c */ -#line 1206 "program_parse.y" +#line 1399 "program_parse.y" { (yyval.integer) = STATE_HALF_VECTOR; ;} break; - case 149: + case 163: /* Line 1455 of yacc.c */ -#line 1212 "program_parse.y" +#line 1405 "program_parse.y" { (yyval.integer) = STATE_SPOT_DIRECTION; ;} break; - case 150: + case 164: /* Line 1455 of yacc.c */ -#line 1218 "program_parse.y" +#line 1411 "program_parse.y" { (yyval.state)[0] = (yyvsp[(2) - (2)].state)[0]; (yyval.state)[1] = (yyvsp[(2) - (2)].state)[1]; ;} break; - case 151: + case 165: /* Line 1455 of yacc.c */ -#line 1225 "program_parse.y" +#line 1418 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_LIGHTMODEL_AMBIENT; ;} break; - case 152: + case 166: /* Line 1455 of yacc.c */ -#line 1230 "program_parse.y" +#line 1423 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_LIGHTMODEL_SCENECOLOR; @@ -3481,10 +3762,10 @@ yyreduce: ;} break; - case 153: + case 167: /* Line 1455 of yacc.c */ -#line 1238 "program_parse.y" +#line 1431 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_LIGHTPROD; @@ -3494,10 +3775,10 @@ yyreduce: ;} break; - case 155: + case 169: /* Line 1455 of yacc.c */ -#line 1250 "program_parse.y" +#line 1443 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = (yyvsp[(3) - (3)].integer); @@ -3505,46 +3786,46 @@ yyreduce: ;} break; - case 156: + case 170: /* Line 1455 of yacc.c */ -#line 1258 "program_parse.y" +#line 1451 "program_parse.y" { (yyval.integer) = STATE_TEXENV_COLOR; ;} break; - case 157: + case 171: /* Line 1455 of yacc.c */ -#line 1264 "program_parse.y" +#line 1457 "program_parse.y" { (yyval.integer) = STATE_AMBIENT; ;} break; - case 158: + case 172: /* Line 1455 of yacc.c */ -#line 1268 "program_parse.y" +#line 1461 "program_parse.y" { (yyval.integer) = STATE_DIFFUSE; ;} break; - case 159: + case 173: /* Line 1455 of yacc.c */ -#line 1272 "program_parse.y" +#line 1465 "program_parse.y" { (yyval.integer) = STATE_SPECULAR; ;} break; - case 160: + case 174: /* Line 1455 of yacc.c */ -#line 1278 "program_parse.y" +#line 1471 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxLights) { yyerror(& (yylsp[(1) - (1)]), state, "invalid light selector"); @@ -3555,10 +3836,10 @@ yyreduce: ;} break; - case 161: + case 175: /* Line 1455 of yacc.c */ -#line 1289 "program_parse.y" +#line 1482 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_TEXGEN; @@ -3567,92 +3848,92 @@ yyreduce: ;} break; - case 162: + case 176: /* Line 1455 of yacc.c */ -#line 1298 "program_parse.y" +#line 1491 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_S; ;} break; - case 163: + case 177: /* Line 1455 of yacc.c */ -#line 1302 "program_parse.y" +#line 1495 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_OBJECT_S; ;} break; - case 164: + case 178: /* Line 1455 of yacc.c */ -#line 1307 "program_parse.y" +#line 1500 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_S - STATE_TEXGEN_EYE_S; ;} break; - case 165: + case 179: /* Line 1455 of yacc.c */ -#line 1311 "program_parse.y" +#line 1504 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_T - STATE_TEXGEN_EYE_S; ;} break; - case 166: + case 180: /* Line 1455 of yacc.c */ -#line 1315 "program_parse.y" +#line 1508 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_R - STATE_TEXGEN_EYE_S; ;} break; - case 167: + case 181: /* Line 1455 of yacc.c */ -#line 1319 "program_parse.y" +#line 1512 "program_parse.y" { (yyval.integer) = STATE_TEXGEN_EYE_Q - STATE_TEXGEN_EYE_S; ;} break; - case 168: + case 182: /* Line 1455 of yacc.c */ -#line 1325 "program_parse.y" +#line 1518 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = (yyvsp[(2) - (2)].integer); ;} break; - case 169: + case 183: /* Line 1455 of yacc.c */ -#line 1332 "program_parse.y" +#line 1525 "program_parse.y" { (yyval.integer) = STATE_FOG_COLOR; ;} break; - case 170: + case 184: /* Line 1455 of yacc.c */ -#line 1336 "program_parse.y" +#line 1529 "program_parse.y" { (yyval.integer) = STATE_FOG_PARAMS; ;} break; - case 171: + case 185: /* Line 1455 of yacc.c */ -#line 1342 "program_parse.y" +#line 1535 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_CLIPPLANE; @@ -3660,10 +3941,10 @@ yyreduce: ;} break; - case 172: + case 186: /* Line 1455 of yacc.c */ -#line 1350 "program_parse.y" +#line 1543 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxClipPlanes) { yyerror(& (yylsp[(1) - (1)]), state, "invalid clip plane selector"); @@ -3674,38 +3955,38 @@ yyreduce: ;} break; - case 173: + case 187: /* Line 1455 of yacc.c */ -#line 1361 "program_parse.y" +#line 1554 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = (yyvsp[(2) - (2)].integer); ;} break; - case 174: + case 188: /* Line 1455 of yacc.c */ -#line 1368 "program_parse.y" +#line 1561 "program_parse.y" { (yyval.integer) = STATE_POINT_SIZE; ;} break; - case 175: + case 189: /* Line 1455 of yacc.c */ -#line 1372 "program_parse.y" +#line 1565 "program_parse.y" { (yyval.integer) = STATE_POINT_ATTENUATION; ;} break; - case 176: + case 190: /* Line 1455 of yacc.c */ -#line 1378 "program_parse.y" +#line 1571 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (5)].state)[0]; (yyval.state)[1] = (yyvsp[(1) - (5)].state)[1]; @@ -3715,10 +3996,10 @@ yyreduce: ;} break; - case 177: + case 191: /* Line 1455 of yacc.c */ -#line 1388 "program_parse.y" +#line 1581 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (2)].state)[0]; (yyval.state)[1] = (yyvsp[(1) - (2)].state)[1]; @@ -3728,20 +4009,20 @@ yyreduce: ;} break; - case 178: + case 192: /* Line 1455 of yacc.c */ -#line 1398 "program_parse.y" +#line 1591 "program_parse.y" { (yyval.state)[2] = 0; (yyval.state)[3] = 3; ;} break; - case 179: + case 193: /* Line 1455 of yacc.c */ -#line 1403 "program_parse.y" +#line 1596 "program_parse.y" { /* It seems logical that the matrix row range specifier would have * to specify a range or more than one row (i.e., $5 > $3). @@ -3759,10 +4040,10 @@ yyreduce: ;} break; - case 180: + case 194: /* Line 1455 of yacc.c */ -#line 1421 "program_parse.y" +#line 1614 "program_parse.y" { (yyval.state)[0] = (yyvsp[(2) - (3)].state)[0]; (yyval.state)[1] = (yyvsp[(2) - (3)].state)[1]; @@ -3770,55 +4051,55 @@ yyreduce: ;} break; - case 181: + case 195: /* Line 1455 of yacc.c */ -#line 1429 "program_parse.y" +#line 1622 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 182: + case 196: /* Line 1455 of yacc.c */ -#line 1433 "program_parse.y" +#line 1626 "program_parse.y" { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} break; - case 183: + case 197: /* Line 1455 of yacc.c */ -#line 1439 "program_parse.y" +#line 1632 "program_parse.y" { (yyval.integer) = STATE_MATRIX_INVERSE; ;} break; - case 184: + case 198: /* Line 1455 of yacc.c */ -#line 1443 "program_parse.y" +#line 1636 "program_parse.y" { (yyval.integer) = STATE_MATRIX_TRANSPOSE; ;} break; - case 185: + case 199: /* Line 1455 of yacc.c */ -#line 1447 "program_parse.y" +#line 1640 "program_parse.y" { (yyval.integer) = STATE_MATRIX_INVTRANS; ;} break; - case 186: + case 200: /* Line 1455 of yacc.c */ -#line 1453 "program_parse.y" +#line 1646 "program_parse.y" { if ((yyvsp[(1) - (1)].integer) > 3) { yyerror(& (yylsp[(1) - (1)]), state, "invalid matrix row reference"); @@ -3829,88 +4110,88 @@ yyreduce: ;} break; - case 187: + case 201: /* Line 1455 of yacc.c */ -#line 1464 "program_parse.y" +#line 1657 "program_parse.y" { (yyval.state)[0] = STATE_MODELVIEW_MATRIX; (yyval.state)[1] = (yyvsp[(2) - (2)].integer); ;} break; - case 188: + case 202: /* Line 1455 of yacc.c */ -#line 1469 "program_parse.y" +#line 1662 "program_parse.y" { (yyval.state)[0] = STATE_PROJECTION_MATRIX; (yyval.state)[1] = 0; ;} break; - case 189: + case 203: /* Line 1455 of yacc.c */ -#line 1474 "program_parse.y" +#line 1667 "program_parse.y" { (yyval.state)[0] = STATE_MVP_MATRIX; (yyval.state)[1] = 0; ;} break; - case 190: + case 204: /* Line 1455 of yacc.c */ -#line 1479 "program_parse.y" +#line 1672 "program_parse.y" { (yyval.state)[0] = STATE_TEXTURE_MATRIX; (yyval.state)[1] = (yyvsp[(2) - (2)].integer); ;} break; - case 191: + case 205: /* Line 1455 of yacc.c */ -#line 1484 "program_parse.y" +#line 1677 "program_parse.y" { yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported"); YYERROR; ;} break; - case 192: + case 206: /* Line 1455 of yacc.c */ -#line 1489 "program_parse.y" +#line 1682 "program_parse.y" { (yyval.state)[0] = STATE_PROGRAM_MATRIX; (yyval.state)[1] = (yyvsp[(3) - (4)].integer); ;} break; - case 193: + case 207: /* Line 1455 of yacc.c */ -#line 1496 "program_parse.y" +#line 1689 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 194: + case 208: /* Line 1455 of yacc.c */ -#line 1500 "program_parse.y" +#line 1693 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} break; - case 195: + case 209: /* Line 1455 of yacc.c */ -#line 1505 "program_parse.y" +#line 1698 "program_parse.y" { /* Since GL_ARB_vertex_blend isn't supported, only modelview matrix * zero is valid. @@ -3924,10 +4205,10 @@ yyreduce: ;} break; - case 196: + case 210: /* Line 1455 of yacc.c */ -#line 1518 "program_parse.y" +#line 1711 "program_parse.y" { /* Since GL_ARB_matrix_palette isn't supported, just let any value * through here. The error will be generated later. @@ -3936,10 +4217,10 @@ yyreduce: ;} break; - case 197: + case 211: /* Line 1455 of yacc.c */ -#line 1526 "program_parse.y" +#line 1719 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxProgramMatrices) { yyerror(& (yylsp[(1) - (1)]), state, "invalid program matrix selector"); @@ -3950,20 +4231,20 @@ yyreduce: ;} break; - case 198: + case 212: /* Line 1455 of yacc.c */ -#line 1537 "program_parse.y" +#line 1730 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = STATE_DEPTH_RANGE; ;} break; - case 203: + case 217: /* Line 1455 of yacc.c */ -#line 1549 "program_parse.y" +#line 1742 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = state->state_param_enum; @@ -3973,30 +4254,30 @@ yyreduce: ;} break; - case 204: + case 218: /* Line 1455 of yacc.c */ -#line 1559 "program_parse.y" +#line 1752 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (1)].integer); (yyval.state)[1] = (yyvsp[(1) - (1)].integer); ;} break; - case 205: + case 219: /* Line 1455 of yacc.c */ -#line 1564 "program_parse.y" +#line 1757 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (3)].integer); (yyval.state)[1] = (yyvsp[(3) - (3)].integer); ;} break; - case 206: + case 220: /* Line 1455 of yacc.c */ -#line 1571 "program_parse.y" +#line 1764 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = state->state_param_enum; @@ -4006,10 +4287,10 @@ yyreduce: ;} break; - case 207: + case 221: /* Line 1455 of yacc.c */ -#line 1581 "program_parse.y" +#line 1774 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = state->state_param_enum; @@ -4019,30 +4300,30 @@ yyreduce: ;} break; - case 208: + case 222: /* Line 1455 of yacc.c */ -#line 1590 "program_parse.y" +#line 1783 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (1)].integer); (yyval.state)[1] = (yyvsp[(1) - (1)].integer); ;} break; - case 209: + case 223: /* Line 1455 of yacc.c */ -#line 1595 "program_parse.y" +#line 1788 "program_parse.y" { (yyval.state)[0] = (yyvsp[(1) - (3)].integer); (yyval.state)[1] = (yyvsp[(3) - (3)].integer); ;} break; - case 210: + case 224: /* Line 1455 of yacc.c */ -#line 1602 "program_parse.y" +#line 1795 "program_parse.y" { memset((yyval.state), 0, sizeof((yyval.state))); (yyval.state)[0] = state->state_param_enum; @@ -4052,10 +4333,10 @@ yyreduce: ;} break; - case 211: + case 225: /* Line 1455 of yacc.c */ -#line 1612 "program_parse.y" +#line 1805 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxEnvParams) { yyerror(& (yylsp[(1) - (1)]), state, "invalid environment parameter reference"); @@ -4065,10 +4346,10 @@ yyreduce: ;} break; - case 212: + case 226: /* Line 1455 of yacc.c */ -#line 1622 "program_parse.y" +#line 1815 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxLocalParams) { yyerror(& (yylsp[(1) - (1)]), state, "invalid local parameter reference"); @@ -4078,10 +4359,10 @@ yyreduce: ;} break; - case 217: + case 231: /* Line 1455 of yacc.c */ -#line 1637 "program_parse.y" +#line 1830 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(1) - (1)].real); @@ -4091,10 +4372,10 @@ yyreduce: ;} break; - case 218: + case 232: /* Line 1455 of yacc.c */ -#line 1647 "program_parse.y" +#line 1840 "program_parse.y" { (yyval.vector).count = 1; (yyval.vector).data[0] = (yyvsp[(1) - (1)].real); @@ -4104,10 +4385,10 @@ yyreduce: ;} break; - case 219: + case 233: /* Line 1455 of yacc.c */ -#line 1655 "program_parse.y" +#line 1848 "program_parse.y" { (yyval.vector).count = 1; (yyval.vector).data[0] = (float) (yyvsp[(1) - (1)].integer); @@ -4117,10 +4398,10 @@ yyreduce: ;} break; - case 220: + case 234: /* Line 1455 of yacc.c */ -#line 1665 "program_parse.y" +#line 1858 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(2) - (3)].real); @@ -4130,10 +4411,10 @@ yyreduce: ;} break; - case 221: + case 235: /* Line 1455 of yacc.c */ -#line 1673 "program_parse.y" +#line 1866 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(2) - (5)].real); @@ -4143,10 +4424,10 @@ yyreduce: ;} break; - case 222: + case 236: /* Line 1455 of yacc.c */ -#line 1682 "program_parse.y" +#line 1875 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(2) - (7)].real); @@ -4156,10 +4437,10 @@ yyreduce: ;} break; - case 223: + case 237: /* Line 1455 of yacc.c */ -#line 1691 "program_parse.y" +#line 1884 "program_parse.y" { (yyval.vector).count = 4; (yyval.vector).data[0] = (yyvsp[(2) - (9)].real); @@ -4169,63 +4450,110 @@ yyreduce: ;} break; - case 224: + case 238: /* Line 1455 of yacc.c */ -#line 1701 "program_parse.y" +#line 1894 "program_parse.y" { (yyval.real) = ((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].real) : (yyvsp[(2) - (2)].real); ;} break; - case 225: + case 239: /* Line 1455 of yacc.c */ -#line 1705 "program_parse.y" +#line 1898 "program_parse.y" { (yyval.real) = (float)(((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].integer) : (yyvsp[(2) - (2)].integer)); ;} break; - case 226: + case 240: /* Line 1455 of yacc.c */ -#line 1710 "program_parse.y" +#line 1903 "program_parse.y" { (yyval.negate) = FALSE; ;} break; - case 227: + case 241: /* Line 1455 of yacc.c */ -#line 1711 "program_parse.y" +#line 1904 "program_parse.y" { (yyval.negate) = TRUE; ;} break; - case 228: + case 242: /* Line 1455 of yacc.c */ -#line 1712 "program_parse.y" +#line 1905 "program_parse.y" { (yyval.negate) = FALSE; ;} break; - case 229: + case 243: /* Line 1455 of yacc.c */ -#line 1715 "program_parse.y" - { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} +#line 1908 "program_parse.y" + { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;} break; - case 231: + case 245: + +/* Line 1455 of yacc.c */ +#line 1912 "program_parse.y" + { + /* NV_fragment_program_option defines the size qualifiers in a + * fairly broken way. "SHORT" or "LONG" can optionally be used + * before TEMP or OUTPUT. However, neither is a reserved word! + * This means that we have to parse it as an identifier, then check + * to make sure it's one of the valid values. *sigh* + * + * In addition, the grammar in the extension spec does *not* allow + * the size specifier to be optional, but all known implementations + * do. + */ + if (!state->option.NV_fragment) { + yyerror(& (yylsp[(1) - (1)]), state, "unexpected IDENTIFIER"); + YYERROR; + } + + if (strcmp("SHORT", (yyvsp[(1) - (1)].string)) == 0) { + } else if (strcmp("LONG", (yyvsp[(1) - (1)].string)) == 0) { + } else { + char *const err_str = + make_error_string("invalid storage size specifier \"%s\"", + (yyvsp[(1) - (1)].string)); + + yyerror(& (yylsp[(1) - (1)]), state, (err_str != NULL) + ? err_str : "invalid storage size specifier"); + + if (err_str != NULL) { + _mesa_free(err_str); + } + + YYERROR; + } + ;} + break; + + case 246: + +/* Line 1455 of yacc.c */ +#line 1946 "program_parse.y" + { + ;} + break; + + case 247: /* Line 1455 of yacc.c */ -#line 1718 "program_parse.y" +#line 1950 "program_parse.y" { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;} break; - case 233: + case 249: /* Line 1455 of yacc.c */ -#line 1722 "program_parse.y" +#line 1954 "program_parse.y" { if (!declare_variable(state, (yyvsp[(3) - (3)].string), (yyvsp[(0) - (3)].integer), & (yylsp[(3) - (3)]))) { YYERROR; @@ -4233,10 +4561,10 @@ yyreduce: ;} break; - case 234: + case 250: /* Line 1455 of yacc.c */ -#line 1728 "program_parse.y" +#line 1960 "program_parse.y" { if (!declare_variable(state, (yyvsp[(1) - (1)].string), (yyvsp[(0) - (1)].integer), & (yylsp[(1) - (1)]))) { YYERROR; @@ -4244,26 +4572,26 @@ yyreduce: ;} break; - case 235: + case 251: /* Line 1455 of yacc.c */ -#line 1736 "program_parse.y" +#line 1968 "program_parse.y" { struct asm_symbol *const s = - declare_variable(state, (yyvsp[(2) - (4)].string), at_output, & (yylsp[(2) - (4)])); + declare_variable(state, (yyvsp[(3) - (5)].string), at_output, & (yylsp[(3) - (5)])); if (s == NULL) { YYERROR; } else { - s->output_binding = (yyvsp[(4) - (4)].result); + s->output_binding = (yyvsp[(5) - (5)].result); } ;} break; - case 236: + case 252: /* Line 1455 of yacc.c */ -#line 1749 "program_parse.y" +#line 1981 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.result) = VERT_RESULT_HPOS; @@ -4274,10 +4602,10 @@ yyreduce: ;} break; - case 237: + case 253: /* Line 1455 of yacc.c */ -#line 1758 "program_parse.y" +#line 1990 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.result) = VERT_RESULT_FOGC; @@ -4288,19 +4616,19 @@ yyreduce: ;} break; - case 238: + case 254: /* Line 1455 of yacc.c */ -#line 1767 "program_parse.y" +#line 1999 "program_parse.y" { (yyval.result) = (yyvsp[(2) - (2)].result); ;} break; - case 239: + case 255: /* Line 1455 of yacc.c */ -#line 1771 "program_parse.y" +#line 2003 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.result) = VERT_RESULT_PSIZ; @@ -4311,10 +4639,10 @@ yyreduce: ;} break; - case 240: + case 256: /* Line 1455 of yacc.c */ -#line 1780 "program_parse.y" +#line 2012 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.result) = VERT_RESULT_TEX0 + (yyvsp[(3) - (3)].integer); @@ -4325,10 +4653,10 @@ yyreduce: ;} break; - case 241: + case 257: /* Line 1455 of yacc.c */ -#line 1789 "program_parse.y" +#line 2021 "program_parse.y" { if (state->mode == ARB_fragment) { (yyval.result) = FRAG_RESULT_DEPTH; @@ -4339,19 +4667,19 @@ yyreduce: ;} break; - case 242: + case 258: /* Line 1455 of yacc.c */ -#line 1800 "program_parse.y" +#line 2032 "program_parse.y" { (yyval.result) = (yyvsp[(2) - (3)].integer) + (yyvsp[(3) - (3)].integer); ;} break; - case 243: + case 259: /* Line 1455 of yacc.c */ -#line 1806 "program_parse.y" +#line 2038 "program_parse.y" { (yyval.integer) = (state->mode == ARB_vertex) ? VERT_RESULT_COL0 @@ -4359,10 +4687,10 @@ yyreduce: ;} break; - case 244: + case 260: /* Line 1455 of yacc.c */ -#line 1812 "program_parse.y" +#line 2044 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.integer) = VERT_RESULT_COL0; @@ -4373,10 +4701,10 @@ yyreduce: ;} break; - case 245: + case 261: /* Line 1455 of yacc.c */ -#line 1821 "program_parse.y" +#line 2053 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.integer) = VERT_RESULT_BFC0; @@ -4387,19 +4715,19 @@ yyreduce: ;} break; - case 246: + case 262: /* Line 1455 of yacc.c */ -#line 1832 "program_parse.y" +#line 2064 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 247: + case 263: /* Line 1455 of yacc.c */ -#line 1836 "program_parse.y" +#line 2068 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.integer) = 0; @@ -4410,10 +4738,10 @@ yyreduce: ;} break; - case 248: + case 264: /* Line 1455 of yacc.c */ -#line 1845 "program_parse.y" +#line 2077 "program_parse.y" { if (state->mode == ARB_vertex) { (yyval.integer) = 1; @@ -4424,94 +4752,94 @@ yyreduce: ;} break; - case 249: + case 265: /* Line 1455 of yacc.c */ -#line 1855 "program_parse.y" +#line 2087 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 250: + case 266: /* Line 1455 of yacc.c */ -#line 1856 "program_parse.y" +#line 2088 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 251: + case 267: /* Line 1455 of yacc.c */ -#line 1857 "program_parse.y" +#line 2089 "program_parse.y" { (yyval.integer) = 1; ;} break; - case 252: + case 268: /* Line 1455 of yacc.c */ -#line 1860 "program_parse.y" +#line 2092 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 253: + case 269: /* Line 1455 of yacc.c */ -#line 1861 "program_parse.y" +#line 2093 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 254: + case 270: /* Line 1455 of yacc.c */ -#line 1862 "program_parse.y" +#line 2094 "program_parse.y" { (yyval.integer) = 1; ;} break; - case 255: + case 271: /* Line 1455 of yacc.c */ -#line 1865 "program_parse.y" +#line 2097 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 256: + case 272: /* Line 1455 of yacc.c */ -#line 1866 "program_parse.y" +#line 2098 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} break; - case 257: + case 273: /* Line 1455 of yacc.c */ -#line 1869 "program_parse.y" +#line 2101 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 258: + case 274: /* Line 1455 of yacc.c */ -#line 1870 "program_parse.y" +#line 2102 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} break; - case 259: + case 275: /* Line 1455 of yacc.c */ -#line 1873 "program_parse.y" +#line 2105 "program_parse.y" { (yyval.integer) = 0; ;} break; - case 260: + case 276: /* Line 1455 of yacc.c */ -#line 1874 "program_parse.y" +#line 2106 "program_parse.y" { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;} break; - case 261: + case 277: /* Line 1455 of yacc.c */ -#line 1878 "program_parse.y" +#line 2110 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureCoordUnits) { yyerror(& (yylsp[(1) - (1)]), state, "invalid texture coordinate unit selector"); @@ -4522,10 +4850,10 @@ yyreduce: ;} break; - case 262: + case 278: /* Line 1455 of yacc.c */ -#line 1889 "program_parse.y" +#line 2121 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureImageUnits) { yyerror(& (yylsp[(1) - (1)]), state, "invalid texture image unit selector"); @@ -4536,10 +4864,10 @@ yyreduce: ;} break; - case 263: + case 279: /* Line 1455 of yacc.c */ -#line 1900 "program_parse.y" +#line 2132 "program_parse.y" { if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureUnits) { yyerror(& (yylsp[(1) - (1)]), state, "invalid texture unit selector"); @@ -4550,10 +4878,10 @@ yyreduce: ;} break; - case 264: + case 280: /* Line 1455 of yacc.c */ -#line 1911 "program_parse.y" +#line 2143 "program_parse.y" { struct asm_symbol *exist = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(2) - (4)].string)); @@ -4577,7 +4905,7 @@ yyreduce: /* Line 1455 of yacc.c */ -#line 4581 "program_parse.tab.c" +#line 4909 "program_parse.tab.c" default: break; } YY_SYMBOL_PRINT ("-> $$ =", yyr1[yyn], &yyval, &yyloc); @@ -4796,7 +5124,49 @@ yyreturn: /* Line 1675 of yacc.c */ -#line 1931 "program_parse.y" +#line 2167 "program_parse.y" + + +void +asm_instruction_set_operands(struct asm_instruction *inst, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + /* In the core ARB extensions only the KIL instruction doesn't have a + * destination register. + */ + if (dst == NULL) { + init_dst_reg(& inst->Base.DstReg); + } else { + inst->Base.DstReg = *dst; + } + + /* The only instruction that doesn't have any source registers is the + * condition-code based KIL instruction added by NV_fragment_program_option. + */ + if (src0 != NULL) { + inst->Base.SrcReg[0] = src0->Base; + inst->SrcReg[0] = *src0; + } else { + init_src_reg(& inst->SrcReg[0]); + } + + if (src1 != NULL) { + inst->Base.SrcReg[1] = src1->Base; + inst->SrcReg[1] = *src1; + } else { + init_src_reg(& inst->SrcReg[1]); + } + + if (src2 != NULL) { + inst->Base.SrcReg[2] = src2->Base; + inst->SrcReg[2] = *src2; + } else { + init_src_reg(& inst->SrcReg[2]); + } +} struct asm_instruction * @@ -4806,37 +5176,37 @@ asm_instruction_ctor(gl_inst_opcode op, const struct asm_src_register *src1, const struct asm_src_register *src2) { - struct asm_instruction *inst = calloc(1, sizeof(struct asm_instruction)); + struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction); if (inst) { _mesa_init_instructions(& inst->Base, 1); inst->Base.Opcode = op; - /* In the core ARB extensions only the KIL instruction doesn't have a - * destination register. - */ - if (dst == NULL) { - init_dst_reg(& inst->Base.DstReg); - } else { - inst->Base.DstReg = *dst; - } + asm_instruction_set_operands(inst, dst, src0, src1, src2); + } - inst->Base.SrcReg[0] = src0->Base; - inst->SrcReg[0] = *src0; + return inst; +} - if (src1 != NULL) { - inst->Base.SrcReg[1] = src1->Base; - inst->SrcReg[1] = *src1; - } else { - init_src_reg(& inst->SrcReg[1]); - } - if (src2 != NULL) { - inst->Base.SrcReg[2] = src2->Base; - inst->SrcReg[2] = *src2; - } else { - init_src_reg(& inst->SrcReg[2]); - } +struct asm_instruction * +asm_instruction_copy_ctor(const struct prog_instruction *base, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction); + + if (inst) { + _mesa_init_instructions(& inst->Base, 1); + inst->Base.Opcode = base->Opcode; + inst->Base.CondUpdate = base->CondUpdate; + inst->Base.CondDst = base->CondDst; + inst->Base.SaturateMode = base->SaturateMode; + inst->Base.Precision = base->Precision; + + asm_instruction_set_operands(inst, dst, src0, src1, src2); } return inst; diff --git a/src/mesa/shader/program_parse.tab.h b/src/mesa/shader/program_parse.tab.h index dabb3bf1a36..860b6f886e9 100644 --- a/src/mesa/shader/program_parse.tab.h +++ b/src/mesa/shader/program_parse.tab.h @@ -58,90 +58,92 @@ ARL = 274, KIL = 275, SWZ = 276, - INTEGER = 277, - REAL = 278, - AMBIENT = 279, - ATTENUATION = 280, - BACK = 281, - CLIP = 282, - COLOR = 283, - DEPTH = 284, - DIFFUSE = 285, - DIRECTION = 286, - EMISSION = 287, - ENV = 288, - EYE = 289, - FOG = 290, - FOGCOORD = 291, - FRAGMENT = 292, - FRONT = 293, - HALF = 294, - INVERSE = 295, - INVTRANS = 296, - LIGHT = 297, - LIGHTMODEL = 298, - LIGHTPROD = 299, - LOCAL = 300, - MATERIAL = 301, - MAT_PROGRAM = 302, - MATRIX = 303, - MATRIXINDEX = 304, - MODELVIEW = 305, - MVP = 306, - NORMAL = 307, - OBJECT = 308, - PALETTE = 309, - PARAMS = 310, - PLANE = 311, - POINT_TOK = 312, - POINTSIZE = 313, - POSITION = 314, - PRIMARY = 315, - PROGRAM = 316, - PROJECTION = 317, - RANGE = 318, - RESULT = 319, - ROW = 320, - SCENECOLOR = 321, - SECONDARY = 322, - SHININESS = 323, - SIZE_TOK = 324, - SPECULAR = 325, - SPOT = 326, - STATE = 327, - TEXCOORD = 328, - TEXENV = 329, - TEXGEN = 330, - TEXGEN_Q = 331, - TEXGEN_R = 332, - TEXGEN_S = 333, - TEXGEN_T = 334, - TEXTURE = 335, - TRANSPOSE = 336, - TEXTURE_UNIT = 337, - TEX_1D = 338, - TEX_2D = 339, - TEX_3D = 340, - TEX_CUBE = 341, - TEX_RECT = 342, - TEX_SHADOW1D = 343, - TEX_SHADOW2D = 344, - TEX_SHADOWRECT = 345, - TEX_ARRAY1D = 346, - TEX_ARRAY2D = 347, - TEX_ARRAYSHADOW1D = 348, - TEX_ARRAYSHADOW2D = 349, - VERTEX = 350, - VTXATTRIB = 351, - WEIGHT = 352, - IDENTIFIER = 353, - MASK4 = 354, - MASK3 = 355, - MASK2 = 356, - MASK1 = 357, - SWIZZLE = 358, - DOT_DOT = 359, - DOT = 360 + TXD_OP = 277, + INTEGER = 278, + REAL = 279, + AMBIENT = 280, + ATTENUATION = 281, + BACK = 282, + CLIP = 283, + COLOR = 284, + DEPTH = 285, + DIFFUSE = 286, + DIRECTION = 287, + EMISSION = 288, + ENV = 289, + EYE = 290, + FOG = 291, + FOGCOORD = 292, + FRAGMENT = 293, + FRONT = 294, + HALF = 295, + INVERSE = 296, + INVTRANS = 297, + LIGHT = 298, + LIGHTMODEL = 299, + LIGHTPROD = 300, + LOCAL = 301, + MATERIAL = 302, + MAT_PROGRAM = 303, + MATRIX = 304, + MATRIXINDEX = 305, + MODELVIEW = 306, + MVP = 307, + NORMAL = 308, + OBJECT = 309, + PALETTE = 310, + PARAMS = 311, + PLANE = 312, + POINT_TOK = 313, + POINTSIZE = 314, + POSITION = 315, + PRIMARY = 316, + PROGRAM = 317, + PROJECTION = 318, + RANGE = 319, + RESULT = 320, + ROW = 321, + SCENECOLOR = 322, + SECONDARY = 323, + SHININESS = 324, + SIZE_TOK = 325, + SPECULAR = 326, + SPOT = 327, + STATE = 328, + TEXCOORD = 329, + TEXENV = 330, + TEXGEN = 331, + TEXGEN_Q = 332, + TEXGEN_R = 333, + TEXGEN_S = 334, + TEXGEN_T = 335, + TEXTURE = 336, + TRANSPOSE = 337, + TEXTURE_UNIT = 338, + TEX_1D = 339, + TEX_2D = 340, + TEX_3D = 341, + TEX_CUBE = 342, + TEX_RECT = 343, + TEX_SHADOW1D = 344, + TEX_SHADOW2D = 345, + TEX_SHADOWRECT = 346, + TEX_ARRAY1D = 347, + TEX_ARRAY2D = 348, + TEX_ARRAYSHADOW1D = 349, + TEX_ARRAYSHADOW2D = 350, + VERTEX = 351, + VTXATTRIB = 352, + WEIGHT = 353, + IDENTIFIER = 354, + USED_IDENTIFIER = 355, + MASK4 = 356, + MASK3 = 357, + MASK2 = 358, + MASK1 = 359, + SWIZZLE = 360, + DOT_DOT = 361, + DOT = 362 }; #endif @@ -152,7 +154,7 @@ typedef union YYSTYPE { /* Line 1676 of yacc.c */ -#line 107 "program_parse.y" +#line 116 "program_parse.y" struct asm_instruction *inst; struct asm_symbol *sym; @@ -181,7 +183,7 @@ typedef union YYSTYPE /* Line 1676 of yacc.c */ -#line 185 "program_parse.tab.h" +#line 187 "program_parse.tab.h" } YYSTYPE; # define YYSTYPE_IS_TRIVIAL 1 # define yystype YYSTYPE /* obsolescent; will be withdrawn */ diff --git a/src/mesa/shader/program_parse.y b/src/mesa/shader/program_parse.y index 89da91064b2..9703e8e670f 100644 --- a/src/mesa/shader/program_parse.y +++ b/src/mesa/shader/program_parse.y @@ -68,10 +68,19 @@ static void init_dst_reg(struct prog_dst_register *r); static void init_src_reg(struct asm_src_register *r); +static void asm_instruction_set_operands(struct asm_instruction *inst, + const struct prog_dst_register *dst, const struct asm_src_register *src0, + const struct asm_src_register *src1, const struct asm_src_register *src2); + static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, const struct prog_dst_register *dst, const struct asm_src_register *src0, const struct asm_src_register *src1, const struct asm_src_register *src2); +static struct asm_instruction *asm_instruction_copy_ctor( + const struct prog_instruction *base, const struct prog_dst_register *dst, + const struct asm_src_register *src0, const struct asm_src_register *src1, + const struct asm_src_register *src2); + #ifndef FALSE #define FALSE 0 #define TRUE (!FALSE) @@ -142,7 +151,7 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, /* Tokens for instructions */ %token <temp_inst> BIN_OP BINSC_OP SAMPLE_OP SCALAR_OP TRI_OP VECTOR_OP -%token <temp_inst> ARL KIL SWZ +%token <temp_inst> ARL KIL SWZ TXD_OP %token <integer> INTEGER %token <real> REAL @@ -169,7 +178,8 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, %token VERTEX VTXATTRIB %token WEIGHT -%token <string> IDENTIFIER +%token <string> IDENTIFIER USED_IDENTIFIER +%type <string> string %token <swiz_mask> MASK4 MASK3 MASK2 MASK1 SWIZZLE %token DOT_DOT %token DOT @@ -177,11 +187,11 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, %type <inst> instruction ALU_instruction TexInstruction %type <inst> ARL_instruction VECTORop_instruction %type <inst> SCALARop_instruction BINSCop_instruction BINop_instruction -%type <inst> TRIop_instruction SWZ_instruction SAMPLE_instruction +%type <inst> TRIop_instruction TXD_instruction SWZ_instruction SAMPLE_instruction %type <inst> KIL_instruction %type <dst_reg> dstReg maskedDstReg maskedAddrReg -%type <src_reg> srcReg scalarSrcReg swizzleSrcReg +%type <src_reg> srcReg scalarUse scalarSrcReg swizzleSrcReg %type <swiz_mask> scalarSuffix swizzleSuffix extendedSwizzle %type <ext_swizzle> extSwizComp extSwizSel %type <swiz_mask> optionalMask @@ -192,6 +202,8 @@ static struct asm_instruction *asm_instruction_ctor(gl_inst_opcode op, %type <sym> addrReg %type <swiz_mask> addrComponent addrWriteMask +%type <dst_reg> ccMaskRule ccTest ccMaskRule2 ccTest2 optionalCcMask + %type <result> resultBinding resultColBinding %type <integer> optFaceType optColorType %type <integer> optResultFaceType optResultColorType @@ -280,7 +292,7 @@ optionSequence: optionSequence option | ; -option: OPTION IDENTIFIER ';' +option: OPTION string ';' { int valid = 0; @@ -347,6 +359,7 @@ ALU_instruction: ARL_instruction TexInstruction: SAMPLE_instruction | KIL_instruction + | TXD_instruction ; ARL_instruction: ARL maskedAddrReg ',' scalarSrcReg @@ -357,51 +370,45 @@ ARL_instruction: ARL maskedAddrReg ',' scalarSrcReg VECTORop_instruction: VECTOR_OP maskedDstReg ',' swizzleSrcReg { - $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, NULL, NULL); - $$->Base.SaturateMode = $1.SaturateMode; + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL); } ; SCALARop_instruction: SCALAR_OP maskedDstReg ',' scalarSrcReg { - $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, NULL, NULL); - $$->Base.SaturateMode = $1.SaturateMode; + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL); } ; BINSCop_instruction: BINSC_OP maskedDstReg ',' scalarSrcReg ',' scalarSrcReg { - $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, & $6, NULL); - $$->Base.SaturateMode = $1.SaturateMode; + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, NULL); } ; BINop_instruction: BIN_OP maskedDstReg ',' swizzleSrcReg ',' swizzleSrcReg { - $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, & $6, NULL); - $$->Base.SaturateMode = $1.SaturateMode; + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, NULL); } ; TRIop_instruction: TRI_OP maskedDstReg ',' swizzleSrcReg ',' swizzleSrcReg ',' swizzleSrcReg { - $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, & $6, & $8); - $$->Base.SaturateMode = $1.SaturateMode; + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, & $8); } ; SAMPLE_instruction: SAMPLE_OP maskedDstReg ',' swizzleSrcReg ',' texImageUnit ',' texTarget { - $$ = asm_instruction_ctor($1.Opcode, & $2, & $4, NULL, NULL); + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL); if ($$ != NULL) { const GLbitfield tex_mask = (1U << $6); GLbitfield shadow_tex = 0; GLbitfield target_mask = 0; - $$->Base.SaturateMode = $1.SaturateMode; $$->Base.TexSrcUnit = $6; if ($8 < 0) { @@ -442,6 +449,58 @@ KIL_instruction: KIL swizzleSrcReg $$ = asm_instruction_ctor(OPCODE_KIL, NULL, & $2, NULL, NULL); state->fragment.UsesKill = 1; } + | KIL ccTest + { + $$ = asm_instruction_ctor(OPCODE_KIL_NV, NULL, NULL, NULL, NULL); + $$->Base.DstReg.CondMask = $2.CondMask; + $$->Base.DstReg.CondSwizzle = $2.CondSwizzle; + $$->Base.DstReg.CondSrc = $2.CondSrc; + state->fragment.UsesKill = 1; + } + ; + +TXD_instruction: TXD_OP maskedDstReg ',' swizzleSrcReg ',' swizzleSrcReg ',' swizzleSrcReg ',' texImageUnit ',' texTarget + { + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, & $6, & $8); + if ($$ != NULL) { + const GLbitfield tex_mask = (1U << $10); + GLbitfield shadow_tex = 0; + GLbitfield target_mask = 0; + + + $$->Base.TexSrcUnit = $10; + + if ($12 < 0) { + shadow_tex = tex_mask; + + $$->Base.TexSrcTarget = -$12; + $$->Base.TexShadow = 1; + } else { + $$->Base.TexSrcTarget = $12; + } + + target_mask = (1U << $$->Base.TexSrcTarget); + + /* If this texture unit was previously accessed and that access + * had a different texture target, generate an error. + * + * If this texture unit was previously accessed and that access + * had a different shadow mode, generate an error. + */ + if ((state->prog->TexturesUsed[$10] != 0) + && ((state->prog->TexturesUsed[$10] != target_mask) + || ((state->prog->ShadowSamplers & tex_mask) + != shadow_tex))) { + yyerror(& @12, state, + "multiple targets used on one texture image unit"); + YYERROR; + } + + + state->prog->TexturesUsed[$10] |= target_mask; + state->prog->ShadowSamplers |= shadow_tex; + } + } ; texImageUnit: TEXTURE_UNIT optTexImageUnitNum @@ -472,21 +531,58 @@ SWZ_instruction: SWZ maskedDstReg ',' srcReg ',' extendedSwizzle $4.Base.Swizzle = $6.swizzle; $4.Base.Negate = $6.mask; - $$ = asm_instruction_ctor(OPCODE_SWZ, & $2, & $4, NULL, NULL); - $$->Base.SaturateMode = $1.SaturateMode; + $$ = asm_instruction_copy_ctor(& $1, & $2, & $4, NULL, NULL); } ; -scalarSrcReg: optionalSign srcReg scalarSuffix +scalarSrcReg: optionalSign scalarUse { $$ = $2; if ($1) { $$.Base.Negate = ~$$.Base.Negate; } + } + | optionalSign '|' scalarUse '|' + { + $$ = $3; + + if (!state->option.NV_fragment) { + yyerror(& @2, state, "unexpected character '|'"); + YYERROR; + } + + if ($1) { + $$.Base.Negate = ~$$.Base.Negate; + } + + $$.Base.Abs = 1; + } + ; + +scalarUse: srcReg scalarSuffix + { + $$ = $1; $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle, - $3.swizzle); + $2.swizzle); + } + | paramConstScalarUse + { + struct asm_symbol temp_sym; + + if (!state->option.NV_fragment) { + yyerror(& @1, state, "expected scalar suffix"); + YYERROR; + } + + memset(& temp_sym, 0, sizeof(temp_sym)); + temp_sym.param_binding_begin = ~0; + initialize_symbol_from_const(state->prog, & temp_sym, & $1); + + init_src_reg(& $$); + $$.Base.File = PROGRAM_CONSTANT; + $$.Base.Index = temp_sym.param_binding_begin; } ; @@ -501,12 +597,33 @@ swizzleSrcReg: optionalSign srcReg swizzleSuffix $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle, $3.swizzle); } + | optionalSign '|' srcReg swizzleSuffix '|' + { + $$ = $3; + + if (!state->option.NV_fragment) { + yyerror(& @2, state, "unexpected character '|'"); + YYERROR; + } + + if ($1) { + $$.Base.Negate = ~$$.Base.Negate; + } + + $$.Base.Abs = 1; + $$.Base.Swizzle = _mesa_combine_swizzles($$.Base.Swizzle, + $4.swizzle); + } + ; -maskedDstReg: dstReg optionalMask +maskedDstReg: dstReg optionalMask optionalCcMask { $$ = $1; $$.WriteMask = $2.mask; + $$.CondMask = $3.CondMask; + $$.CondSwizzle = $3.CondSwizzle; + $$.CondSrc = $3.CondSrc; if ($$.File == PROGRAM_OUTPUT) { /* Technically speaking, this should check that it is in @@ -589,7 +706,7 @@ extSwizSel: INTEGER $$.xyzw_valid = 1; $$.rgba_valid = 1; } - | IDENTIFIER + | string { if (strlen($1) > 1) { yyerror(& @1, state, "invalid extended swizzle selector"); @@ -639,7 +756,7 @@ extSwizSel: INTEGER } ; -srcReg: IDENTIFIER /* temporaryReg | progParamSingle */ +srcReg: USED_IDENTIFIER /* temporaryReg | progParamSingle */ { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, $1); @@ -729,7 +846,7 @@ dstReg: resultBinding $$.File = PROGRAM_OUTPUT; $$.Index = $1; } - | IDENTIFIER /* temporaryReg | vertexResultReg */ + | USED_IDENTIFIER /* temporaryReg | vertexResultReg */ { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, $1); @@ -760,7 +877,7 @@ dstReg: resultBinding } ; -progParamArray: IDENTIFIER +progParamArray: USED_IDENTIFIER { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, $1); @@ -831,7 +948,7 @@ addrRegNegOffset: INTEGER } ; -addrReg: IDENTIFIER +addrReg: USED_IDENTIFIER { struct asm_symbol *const s = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, $1); @@ -884,6 +1001,82 @@ optionalMask: MASK4 | MASK3 | MASK2 | MASK1 | { $$.swizzle = SWIZZLE_NOOP; $$.mask = WRITEMASK_XYZW; } ; +optionalCcMask: '(' ccTest ')' + { + $$ = $2; + } + | '(' ccTest2 ')' + { + $$ = $2; + } + | + { + $$.CondMask = COND_TR; + $$.CondSwizzle = SWIZZLE_NOOP; + $$.CondSrc = 0; + } + ; + +ccTest: ccMaskRule swizzleSuffix + { + $$ = $1; + $$.CondSwizzle = $2.swizzle; + } + ; + +ccTest2: ccMaskRule2 swizzleSuffix + { + $$ = $1; + $$.CondSwizzle = $2.swizzle; + } + ; + +ccMaskRule: IDENTIFIER + { + const int cond = _mesa_parse_cc($1); + if ((cond == 0) || ($1[2] != '\0')) { + char *const err_str = + make_error_string("invalid condition code \"%s\"", $1); + + yyerror(& @1, state, (err_str != NULL) + ? err_str : "invalid condition code"); + + if (err_str != NULL) { + _mesa_free(err_str); + } + + YYERROR; + } + + $$.CondMask = cond; + $$.CondSwizzle = SWIZZLE_NOOP; + $$.CondSrc = 0; + } + ; + +ccMaskRule2: USED_IDENTIFIER + { + const int cond = _mesa_parse_cc($1); + if ((cond == 0) || ($1[2] != '\0')) { + char *const err_str = + make_error_string("invalid condition code \"%s\"", $1); + + yyerror(& @1, state, (err_str != NULL) + ? err_str : "invalid condition code"); + + if (err_str != NULL) { + _mesa_free(err_str); + } + + YYERROR; + } + + $$.CondMask = cond; + $$.CondSwizzle = SWIZZLE_NOOP; + $$.CondSrc = 0; + } + ; + namingStatement: ATTRIB_statement | PARAM_statement | TEMP_statement @@ -1712,7 +1905,46 @@ optionalSign: '+' { $$ = FALSE; } | { $$ = FALSE; } ; -TEMP_statement: TEMP { $<integer>$ = $1; } varNameList +TEMP_statement: optVarSize TEMP { $<integer>$ = $2; } varNameList + ; + +optVarSize: string + { + /* NV_fragment_program_option defines the size qualifiers in a + * fairly broken way. "SHORT" or "LONG" can optionally be used + * before TEMP or OUTPUT. However, neither is a reserved word! + * This means that we have to parse it as an identifier, then check + * to make sure it's one of the valid values. *sigh* + * + * In addition, the grammar in the extension spec does *not* allow + * the size specifier to be optional, but all known implementations + * do. + */ + if (!state->option.NV_fragment) { + yyerror(& @1, state, "unexpected IDENTIFIER"); + YYERROR; + } + + if (strcmp("SHORT", $1) == 0) { + } else if (strcmp("LONG", $1) == 0) { + } else { + char *const err_str = + make_error_string("invalid storage size specifier \"%s\"", + $1); + + yyerror(& @1, state, (err_str != NULL) + ? err_str : "invalid storage size specifier"); + + if (err_str != NULL) { + _mesa_free(err_str); + } + + YYERROR; + } + } + | + { + } ; ADDRESS_statement: ADDRESS { $<integer>$ = $1; } varNameList @@ -1732,15 +1964,15 @@ varNameList: varNameList ',' IDENTIFIER } ; -OUTPUT_statement: OUTPUT IDENTIFIER '=' resultBinding +OUTPUT_statement: optVarSize OUTPUT IDENTIFIER '=' resultBinding { struct asm_symbol *const s = - declare_variable(state, $2, at_output, & @2); + declare_variable(state, $3, at_output, & @3); if (s == NULL) { YYERROR; } else { - s->output_binding = $4; + s->output_binding = $5; } } ; @@ -1907,7 +2139,7 @@ legacyTexUnitNum: INTEGER } ; -ALIAS_statement: ALIAS IDENTIFIER '=' IDENTIFIER +ALIAS_statement: ALIAS IDENTIFIER '=' USED_IDENTIFIER { struct asm_symbol *exist = (struct asm_symbol *) _mesa_symbol_table_find_symbol(state->st, 0, $2); @@ -1928,8 +2160,54 @@ ALIAS_statement: ALIAS IDENTIFIER '=' IDENTIFIER } ; +string: IDENTIFIER + | USED_IDENTIFIER + ; + %% +void +asm_instruction_set_operands(struct asm_instruction *inst, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + /* In the core ARB extensions only the KIL instruction doesn't have a + * destination register. + */ + if (dst == NULL) { + init_dst_reg(& inst->Base.DstReg); + } else { + inst->Base.DstReg = *dst; + } + + /* The only instruction that doesn't have any source registers is the + * condition-code based KIL instruction added by NV_fragment_program_option. + */ + if (src0 != NULL) { + inst->Base.SrcReg[0] = src0->Base; + inst->SrcReg[0] = *src0; + } else { + init_src_reg(& inst->SrcReg[0]); + } + + if (src1 != NULL) { + inst->Base.SrcReg[1] = src1->Base; + inst->SrcReg[1] = *src1; + } else { + init_src_reg(& inst->SrcReg[1]); + } + + if (src2 != NULL) { + inst->Base.SrcReg[2] = src2->Base; + inst->SrcReg[2] = *src2; + } else { + init_src_reg(& inst->SrcReg[2]); + } +} + + struct asm_instruction * asm_instruction_ctor(gl_inst_opcode op, const struct prog_dst_register *dst, @@ -1937,37 +2215,37 @@ asm_instruction_ctor(gl_inst_opcode op, const struct asm_src_register *src1, const struct asm_src_register *src2) { - struct asm_instruction *inst = calloc(1, sizeof(struct asm_instruction)); + struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction); if (inst) { _mesa_init_instructions(& inst->Base, 1); inst->Base.Opcode = op; - /* In the core ARB extensions only the KIL instruction doesn't have a - * destination register. - */ - if (dst == NULL) { - init_dst_reg(& inst->Base.DstReg); - } else { - inst->Base.DstReg = *dst; - } + asm_instruction_set_operands(inst, dst, src0, src1, src2); + } - inst->Base.SrcReg[0] = src0->Base; - inst->SrcReg[0] = *src0; + return inst; +} - if (src1 != NULL) { - inst->Base.SrcReg[1] = src1->Base; - inst->SrcReg[1] = *src1; - } else { - init_src_reg(& inst->SrcReg[1]); - } - if (src2 != NULL) { - inst->Base.SrcReg[2] = src2->Base; - inst->SrcReg[2] = *src2; - } else { - init_src_reg(& inst->SrcReg[2]); - } +struct asm_instruction * +asm_instruction_copy_ctor(const struct prog_instruction *base, + const struct prog_dst_register *dst, + const struct asm_src_register *src0, + const struct asm_src_register *src1, + const struct asm_src_register *src2) +{ + struct asm_instruction *inst = CALLOC_STRUCT(asm_instruction); + + if (inst) { + _mesa_init_instructions(& inst->Base, 1); + inst->Base.Opcode = base->Opcode; + inst->Base.CondUpdate = base->CondUpdate; + inst->Base.CondDst = base->CondDst; + inst->Base.SaturateMode = base->SaturateMode; + inst->Base.Precision = base->Precision; + + asm_instruction_set_operands(inst, dst, src0, src1, src2); } return inst; diff --git a/src/mesa/shader/program_parse_extra.c b/src/mesa/shader/program_parse_extra.c index 8e4be606f16..0656c5eaa8e 100644 --- a/src/mesa/shader/program_parse_extra.c +++ b/src/mesa/shader/program_parse_extra.c @@ -34,6 +34,121 @@ */ int +_mesa_parse_instruction_suffix(const struct asm_parser_state *state, + const char *suffix, + struct prog_instruction *inst) +{ + inst->CondUpdate = 0; + inst->CondDst = 0; + inst->SaturateMode = SATURATE_OFF; + inst->Precision = FLOAT32; + + + /* The first possible suffix element is the precision specifier from + * NV_fragment_program_option. + */ + if (state->option.NV_fragment) { + switch (suffix[0]) { + case 'H': + inst->Precision = FLOAT16; + suffix++; + break; + case 'R': + inst->Precision = FLOAT32; + suffix++; + break; + case 'X': + inst->Precision = FIXED12; + suffix++; + break; + default: + break; + } + } + + /* The next possible suffix element is the condition code modifier selection + * from NV_fragment_program_option. + */ + if (state->option.NV_fragment) { + if (suffix[0] == 'C') { + inst->CondUpdate = 1; + suffix++; + } + } + + + /* The final possible suffix element is the saturation selector from + * ARB_fragment_program. + */ + if (state->mode == ARB_fragment) { + if (strcmp(suffix, "_SAT") == 0) { + inst->SaturateMode = SATURATE_ZERO_ONE; + suffix += 4; + } + } + + + /* It is an error for all of the suffix string not to be consumed. + */ + return suffix[0] == '\0'; +} + + +int +_mesa_parse_cc(const char *s) +{ + int cond = 0; + + switch (s[0]) { + case 'E': + if (s[1] == 'Q') { + cond = COND_EQ; + } + break; + + case 'F': + if (s[1] == 'L') { + cond = COND_FL; + } + break; + + case 'G': + if (s[1] == 'E') { + cond = COND_GE; + } else if (s[1] == 'T') { + cond = COND_GT; + } + break; + + case 'L': + if (s[1] == 'E') { + cond = COND_LE; + } else if (s[1] == 'T') { + cond = COND_LT; + } + break; + + case 'N': + if (s[1] == 'E') { + cond = COND_NE; + } + break; + + case 'T': + if (s[1] == 'R') { + cond = COND_TR; + } + break; + + default: + break; + } + + return ((cond == 0) || (s[2] != '\0')) ? 0 : cond; +} + + +int _mesa_ARBvp_parse_option(struct asm_parser_state *state, const char *option) { if (strcmp(option, "ARB_position_invariant") == 0) { @@ -102,6 +217,17 @@ _mesa_ARBfp_parse_option(struct asm_parser_state *state, const char *option) return 1; } } + } else if (strncmp(option, "NV_fragment_program", 19) == 0) { + option += 19; + + /* Other NV_fragment_program strings may be supported later. + */ + if (option[0] == '\0') { + if (state->ctx->Extensions.NV_fragment_program_option) { + state->option.NV_fragment = 1; + return 1; + } + } } else if (strncmp(option, "MESA_", 5) == 0) { option += 5; diff --git a/src/mesa/shader/program_parser.h b/src/mesa/shader/program_parser.h index fa47d84565a..bce6041381f 100644 --- a/src/mesa/shader/program_parser.h +++ b/src/mesa/shader/program_parser.h @@ -202,6 +202,7 @@ struct asm_parser_state { unsigned Shadow:1; unsigned TexRect:1; unsigned TexArray:1; + unsigned NV_fragment:1; } option; struct { @@ -263,4 +264,31 @@ extern int _mesa_ARBvp_parse_option(struct asm_parser_state *state, extern int _mesa_ARBfp_parse_option(struct asm_parser_state *state, const char *option); +/** + * Parses and processes instruction suffixes + * + * Instruction suffixes, such as \c _SAT, are processed. The relevant bits + * are set in \c inst. If suffixes are encountered that are either not known + * or not supported by the modes and options set in \c state, zero will be + * returned. + * + * \return + * Non-zero on success, zero on failure. + */ +extern int _mesa_parse_instruction_suffix(const struct asm_parser_state *state, + const char *suffix, struct prog_instruction *inst); + +/** + * Parses a condition code name + * + * The condition code names (e.g., \c LT, \c GT, \c NE) were added to assembly + * shaders with the \c GL_NV_fragment_program_option extension. This function + * converts a string representation into one of the \c COND_ macros. + * + * \return + * One of the \c COND_ macros defined in prog_instruction.h on success or zero + * on failure. + */ +extern int _mesa_parse_cc(const char *s); + /*@}*/ diff --git a/src/mesa/shader/programopt.c b/src/mesa/shader/programopt.c index f70c75cec8e..3b8529592dd 100644 --- a/src/mesa/shader/programopt.c +++ b/src/mesa/shader/programopt.c @@ -573,3 +573,94 @@ _mesa_remove_output_reads(struct gl_program *prog, gl_register_file type) } } } + + +/** + * Make the given fragment program into a "no-op" shader. + * Actually, just copy the incoming fragment color (or texcoord) + * to the output color. + * This is for debug/test purposes. + */ +void +_mesa_nop_fragment_program(GLcontext *ctx, struct gl_fragment_program *prog) +{ + struct prog_instruction *inst; + GLuint inputAttr; + + inst = _mesa_alloc_instructions(2); + if (!inst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "_mesa_nop_fragment_program"); + return; + } + + _mesa_init_instructions(inst, 2); + + inst[0].Opcode = OPCODE_MOV; + inst[0].DstReg.File = PROGRAM_OUTPUT; + inst[0].DstReg.Index = FRAG_RESULT_COLOR; + inst[0].SrcReg[0].File = PROGRAM_INPUT; + if (prog->Base.InputsRead & FRAG_BIT_COL0) + inputAttr = FRAG_ATTRIB_COL0; + else + inputAttr = FRAG_ATTRIB_TEX0; + inst[0].SrcReg[0].Index = inputAttr; + + inst[1].Opcode = OPCODE_END; + + _mesa_free_instructions(prog->Base.Instructions, + prog->Base.NumInstructions); + + prog->Base.Instructions = inst; + prog->Base.NumInstructions = 2; + prog->Base.InputsRead = 1 << inputAttr; + prog->Base.OutputsWritten = 1 << FRAG_RESULT_COLOR; +} + + +/** + * \sa _mesa_nop_fragment_program + * Replace the given vertex program with a "no-op" program that just + * transforms vertex position and emits color. + */ +void +_mesa_nop_vertex_program(GLcontext *ctx, struct gl_vertex_program *prog) +{ + struct prog_instruction *inst; + GLuint inputAttr; + + /* + * Start with a simple vertex program that emits color. + */ + inst = _mesa_alloc_instructions(2); + if (!inst) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "_mesa_nop_vertex_program"); + return; + } + + _mesa_init_instructions(inst, 2); + + inst[0].Opcode = OPCODE_MOV; + inst[0].DstReg.File = PROGRAM_OUTPUT; + inst[0].DstReg.Index = VERT_RESULT_COL0; + inst[0].SrcReg[0].File = PROGRAM_INPUT; + if (prog->Base.InputsRead & VERT_BIT_COLOR0) + inputAttr = VERT_ATTRIB_COLOR0; + else + inputAttr = VERT_ATTRIB_TEX0; + inst[0].SrcReg[0].Index = inputAttr; + + inst[1].Opcode = OPCODE_END; + + _mesa_free_instructions(prog->Base.Instructions, + prog->Base.NumInstructions); + + prog->Base.Instructions = inst; + prog->Base.NumInstructions = 2; + prog->Base.InputsRead = 1 << inputAttr; + prog->Base.OutputsWritten = 1 << VERT_RESULT_COL0; + + /* + * Now insert code to do standard modelview/projection transformation. + */ + _mesa_insert_mvp_code(ctx, prog); +} diff --git a/src/mesa/shader/programopt.h b/src/mesa/shader/programopt.h index 96acaf9566c..21fac07849a 100644 --- a/src/mesa/shader/programopt.h +++ b/src/mesa/shader/programopt.h @@ -42,4 +42,11 @@ _mesa_count_texture_instructions(struct gl_program *prog); extern void _mesa_remove_output_reads(struct gl_program *prog, gl_register_file type); +extern void +_mesa_nop_fragment_program(GLcontext *ctx, struct gl_fragment_program *prog); + +extern void +_mesa_nop_vertex_program(GLcontext *ctx, struct gl_vertex_program *prog); + + #endif /* PROGRAMOPT_H */ diff --git a/src/mesa/shader/shader_api.c b/src/mesa/shader/shader_api.c index 178b7d0dbaf..6b19b4c46b3 100644 --- a/src/mesa/shader/shader_api.c +++ b/src/mesa/shader/shader_api.c @@ -380,6 +380,10 @@ get_shader_flags(void) flags |= GLSL_DUMP; if (_mesa_strstr(env, "log")) flags |= GLSL_LOG; + if (_mesa_strstr(env, "nopvert")) + flags |= GLSL_NOP_VERT; + if (_mesa_strstr(env, "nopfrag")) + flags |= GLSL_NOP_FRAG; if (_mesa_strstr(env, "nopt")) flags |= GLSL_NO_OPT; else if (_mesa_strstr(env, "opt")) diff --git a/src/mesa/shader/slang/library/slang_common_builtin.gc b/src/mesa/shader/slang/library/slang_common_builtin.gc index 9764fc25b09..56de47ee8d7 100644 --- a/src/mesa/shader/slang/library/slang_common_builtin.gc +++ b/src/mesa/shader/slang/library/slang_common_builtin.gc @@ -602,42 +602,50 @@ vec4 exp2(const vec4 a) float sqrt(const float x) { + const float nx = -x; float r; __asm float_rsq r, x; - __asm float_rcp __retVal, r; -} - -vec2 sqrt(const vec2 v) -{ - float r; - __asm float_rsq r, v.x; - __asm float_rcp __retVal.x, r; - __asm float_rsq r, v.y; - __asm float_rcp __retVal.y, r; -} - -vec3 sqrt(const vec3 v) -{ - float r; - __asm float_rsq r, v.x; - __asm float_rcp __retVal.x, r; - __asm float_rsq r, v.y; - __asm float_rcp __retVal.y, r; - __asm float_rsq r, v.z; - __asm float_rcp __retVal.z, r; -} - -vec4 sqrt(const vec4 v) -{ - float r; - __asm float_rsq r, v.x; - __asm float_rcp __retVal.x, r; - __asm float_rsq r, v.y; - __asm float_rcp __retVal.y, r; - __asm float_rsq r, v.z; - __asm float_rcp __retVal.z, r; - __asm float_rsq r, v.w; - __asm float_rcp __retVal.w, r; + __asm float_rcp r, r; + __asm vec4_cmp __retVal, nx, r, 0.0; +} + +vec2 sqrt(const vec2 x) +{ + const vec2 nx = -x, zero = vec2(0.0); + vec2 r; + __asm float_rsq r.x, x.x; + __asm float_rsq r.y, x.y; + __asm float_rcp r.x, r.x; + __asm float_rcp r.y, r.y; + __asm vec4_cmp __retVal, nx, r, zero; +} + +vec3 sqrt(const vec3 x) +{ + const vec3 nx = -x, zero = vec3(0.0); + vec3 r; + __asm float_rsq r.x, x.x; + __asm float_rsq r.y, x.y; + __asm float_rsq r.z, x.z; + __asm float_rcp r.x, r.x; + __asm float_rcp r.y, r.y; + __asm float_rcp r.z, r.z; + __asm vec4_cmp __retVal, nx, r, zero; +} + +vec4 sqrt(const vec4 x) +{ + const vec4 nx = -x, zero = vec4(0.0); + vec4 r; + __asm float_rsq r.x, x.x; + __asm float_rsq r.y, x.y; + __asm float_rsq r.z, x.z; + __asm float_rsq r.w, x.w; + __asm float_rcp r.x, r.x; + __asm float_rcp r.y, r.y; + __asm float_rcp r.z, r.z; + __asm float_rcp r.w, r.w; + __asm vec4_cmp __retVal, nx, r, zero; } diff --git a/src/mesa/shader/slang/library/slang_common_builtin_gc.h b/src/mesa/shader/slang/library/slang_common_builtin_gc.h index 78a7b83ec18..3c3666e4ea2 100644 --- a/src/mesa/shader/slang/library/slang_common_builtin_gc.h +++ b/src/mesa/shader/slang/library/slang_common_builtin_gc.h @@ -307,55 +307,63 @@ 0,0,18,97,0,59,121,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86,97,108,0, 59,122,0,0,18,97,0,59,122,0,0,0,4,102,108,111,97,116,95,101,120,112,50,0,18,95,95,114,101,116,86, 97,108,0,59,119,0,0,18,97,0,59,119,0,0,0,0,1,90,95,0,0,9,0,0,115,113,114,116,0,1,1,0,0,9,0,120,0,0, -0,1,3,2,90,95,0,0,9,0,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,120,0,0,0,4, -102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,0,18,114,0,0,0,0,1,90,95,0,0, -10,0,0,115,113,114,116,0,1,1,0,0,10,0,118,0,0,0,1,3,2,90,95,0,0,9,0,1,114,0,0,0,4,102,108,111,97, -116,95,114,115,113,0,18,114,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95, -95,114,101,116,86,97,108,0,59,120,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0, -0,18,118,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59, -121,0,0,18,114,0,0,0,0,1,90,95,0,0,11,0,0,115,113,114,116,0,1,1,0,0,11,0,118,0,0,0,1,3,2,90,95,0,0, -9,0,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,120,0,0,0,4,102,108, -111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,0,0,0,4,102,108,111, -97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18, -95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114, -0,0,18,118,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59, -122,0,0,18,114,0,0,0,0,1,90,95,0,0,12,0,0,115,113,114,116,0,1,1,0,0,12,0,118,0,0,0,1,3,2,90,95,0,0, -9,0,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,120,0,0,0,4,102,108, -111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,114,0,0,0,4,102,108,111, -97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,121,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18, -95,95,114,101,116,86,97,108,0,59,121,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114, -0,0,18,118,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59, -122,0,0,18,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,0,18,118,0,59,119,0,0,0,4,102, -108,111,97,116,95,114,99,112,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18,114,0,0,0,0,1,90,95, -0,0,9,0,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,9,0,120,0,0,0,1,4,102,108,111,97, -116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,120,0,0,0,0,1,90,95,0,0,10,0,0, -105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,10,0,118,0,0,0,1,4,102,108,111,97,116,95,114, -115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,116, -95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,0,1,90,95,0,0, -11,0,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,11,0,118,0,0,0,1,4,102,108,111,97,116, -95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111, -97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,4,102, -108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0,59,122,0,0,0, -0,1,90,95,0,0,12,0,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,12,0,118,0,0,0,1,4,102, -108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0, -4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121, -0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0, -59,122,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18, -118,0,59,119,0,0,0,0,1,90,95,0,0,9,0,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,9,0,120,0,0,0, -1,9,18,95,95,114,101,116,86,97,108,0,17,49,0,48,0,0,20,0,0,1,90,95,0,0,10,0,0,110,111,114,109,97, -108,105,122,101,0,1,1,0,0,10,0,118,0,0,0,1,3,2,90,95,1,0,9,0,1,115,0,2,58,105,110,118,101,114,115, -101,115,113,114,116,0,0,58,100,111,116,0,0,18,118,0,0,18,118,0,0,0,0,0,0,0,4,118,101,99,52,95,109, -117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,18,115,0, -0,0,0,1,90,95,0,0,11,0,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,11,0,118,0,0,0,1,3,2,90,95,0, -0,9,0,1,116,109,112,0,0,0,4,118,101,99,51,95,100,111,116,0,18,116,109,112,0,0,18,118,0,0,18,118,0, -0,0,4,102,108,111,97,116,95,114,115,113,0,18,116,109,112,0,0,18,116,109,112,0,0,0,4,118,101,99,52, -95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0, -0,18,116,109,112,0,0,0,0,1,90,95,0,0,12,0,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,12,0,118, -0,0,0,1,3,2,90,95,0,0,9,0,1,116,109,112,0,0,0,4,118,101,99,52,95,100,111,116,0,18,116,109,112,0,0, -18,118,0,0,18,118,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,116,109,112,0,0,18,116,109,112,0, -0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120, -121,122,0,0,18,118,0,0,18,116,109,112,0,0,0,0,1,90,95,0,0,9,0,0,97,98,115,0,1,1,0,0,9,0,97,0,0,0,1, -4,118,101,99,52,95,97,98,115,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,90,95,0,0,10,0,0, +0,1,3,2,90,95,1,0,9,0,1,110,120,0,2,18,120,0,54,0,0,3,2,90,95,0,0,9,0,1,114,0,0,0,4,102,108,111,97, +116,95,114,115,113,0,18,114,0,0,18,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,114,0,0,18, +114,0,0,0,4,118,101,99,52,95,99,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,110,120,0,0,18,114, +0,0,17,48,0,48,0,0,0,0,0,1,90,95,0,0,10,0,0,115,113,114,116,0,1,1,0,0,10,0,120,0,0,0,1,3,2,90,95,1, +0,10,0,1,110,120,0,2,18,120,0,54,0,1,1,122,101,114,111,0,2,58,118,101,99,50,0,0,17,48,0,48,0,0,0,0, +0,0,3,2,90,95,0,0,10,0,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,59,120,0,0,18, +120,0,59,120,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,59,121,0,0,18,120,0,59,121,0,0,0, +4,102,108,111,97,116,95,114,99,112,0,18,114,0,59,120,0,0,18,114,0,59,120,0,0,0,4,102,108,111,97, +116,95,114,99,112,0,18,114,0,59,121,0,0,18,114,0,59,121,0,0,0,4,118,101,99,52,95,99,109,112,0,18, +95,95,114,101,116,86,97,108,0,0,18,110,120,0,0,18,114,0,0,18,122,101,114,111,0,0,0,0,1,90,95,0,0, +11,0,0,115,113,114,116,0,1,1,0,0,11,0,120,0,0,0,1,3,2,90,95,1,0,11,0,1,110,120,0,2,18,120,0,54,0,1, +1,122,101,114,111,0,2,58,118,101,99,51,0,0,17,48,0,48,0,0,0,0,0,0,3,2,90,95,0,0,11,0,1,114,0,0,0,4, +102,108,111,97,116,95,114,115,113,0,18,114,0,59,120,0,0,18,120,0,59,120,0,0,0,4,102,108,111,97,116, +95,114,115,113,0,18,114,0,59,121,0,0,18,120,0,59,121,0,0,0,4,102,108,111,97,116,95,114,115,113,0, +18,114,0,59,122,0,0,18,120,0,59,122,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,114,0,59,120,0,0, +18,114,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,114,0,59,121,0,0,18,114,0,59,121,0,0, +0,4,102,108,111,97,116,95,114,99,112,0,18,114,0,59,122,0,0,18,114,0,59,122,0,0,0,4,118,101,99,52, +95,99,109,112,0,18,95,95,114,101,116,86,97,108,0,0,18,110,120,0,0,18,114,0,0,18,122,101,114,111,0, +0,0,0,1,90,95,0,0,12,0,0,115,113,114,116,0,1,1,0,0,12,0,120,0,0,0,1,3,2,90,95,1,0,12,0,1,110,120,0, +2,18,120,0,54,0,1,1,122,101,114,111,0,2,58,118,101,99,52,0,0,17,48,0,48,0,0,0,0,0,0,3,2,90,95,0,0, +12,0,1,114,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,114,0,59,120,0,0,18,120,0,59,120,0,0,0,4, +102,108,111,97,116,95,114,115,113,0,18,114,0,59,121,0,0,18,120,0,59,121,0,0,0,4,102,108,111,97,116, +95,114,115,113,0,18,114,0,59,122,0,0,18,120,0,59,122,0,0,0,4,102,108,111,97,116,95,114,115,113,0, +18,114,0,59,119,0,0,18,120,0,59,119,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,114,0,59,120,0,0, +18,114,0,59,120,0,0,0,4,102,108,111,97,116,95,114,99,112,0,18,114,0,59,121,0,0,18,114,0,59,121,0,0, +0,4,102,108,111,97,116,95,114,99,112,0,18,114,0,59,122,0,0,18,114,0,59,122,0,0,0,4,102,108,111,97, +116,95,114,99,112,0,18,114,0,59,119,0,0,18,114,0,59,119,0,0,0,4,118,101,99,52,95,99,109,112,0,18, +95,95,114,101,116,86,97,108,0,0,18,110,120,0,0,18,114,0,0,18,122,101,114,111,0,0,0,0,1,90,95,0,0,9, +0,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,9,0,120,0,0,0,1,4,102,108,111,97,116,95, +114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,120,0,0,0,0,1,90,95,0,0,10,0,0,105, +110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,10,0,118,0,0,0,1,4,102,108,111,97,116,95,114,115, +113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97,116,95, +114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,0,1,90,95,0,0,11,0, +0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,11,0,118,0,0,0,1,4,102,108,111,97,116,95, +114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4,102,108,111,97, +116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0,0,0,4,102,108, +111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0,59,122,0,0,0,0,1, +90,95,0,0,12,0,0,105,110,118,101,114,115,101,115,113,114,116,0,1,1,0,0,12,0,118,0,0,0,1,4,102,108, +111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,120,0,0,18,118,0,59,120,0,0,0,4, +102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,121,0,0,18,118,0,59,121,0, +0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,122,0,0,18,118,0,59, +122,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,95,95,114,101,116,86,97,108,0,59,119,0,0,18,118, +0,59,119,0,0,0,0,1,90,95,0,0,9,0,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,9,0,120,0,0,0,1,9, +18,95,95,114,101,116,86,97,108,0,17,49,0,48,0,0,20,0,0,1,90,95,0,0,10,0,0,110,111,114,109,97,108, +105,122,101,0,1,1,0,0,10,0,118,0,0,0,1,3,2,90,95,1,0,9,0,1,115,0,2,58,105,110,118,101,114,115,101, +115,113,114,116,0,0,58,100,111,116,0,0,18,118,0,0,18,118,0,0,0,0,0,0,0,4,118,101,99,52,95,109,117, +108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,0,0,18,118,0,0,18,115,0,0,0, +0,1,90,95,0,0,11,0,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,11,0,118,0,0,0,1,3,2,90,95,0,0,9, +0,1,116,109,112,0,0,0,4,118,101,99,51,95,100,111,116,0,18,116,109,112,0,0,18,118,0,0,18,118,0,0,0, +4,102,108,111,97,116,95,114,115,113,0,18,116,109,112,0,0,18,116,109,112,0,0,0,4,118,101,99,52,95, +109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,118,0,0, +18,116,109,112,0,0,0,0,1,90,95,0,0,12,0,0,110,111,114,109,97,108,105,122,101,0,1,1,0,0,12,0,118,0, +0,0,1,3,2,90,95,0,0,9,0,1,116,109,112,0,0,0,4,118,101,99,52,95,100,111,116,0,18,116,109,112,0,0,18, +118,0,0,18,118,0,0,0,4,102,108,111,97,116,95,114,115,113,0,18,116,109,112,0,0,18,116,109,112,0,0,0, +4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,120,121, +122,0,0,18,118,0,0,18,116,109,112,0,0,0,0,1,90,95,0,0,9,0,0,97,98,115,0,1,1,0,0,9,0,97,0,0,0,1,4, +118,101,99,52,95,97,98,115,0,18,95,95,114,101,116,86,97,108,0,0,18,97,0,0,0,0,1,90,95,0,0,10,0,0, 97,98,115,0,1,1,0,0,10,0,97,0,0,0,1,4,118,101,99,52,95,97,98,115,0,18,95,95,114,101,116,86,97,108, 0,59,120,121,0,0,18,97,0,0,0,0,1,90,95,0,0,11,0,0,97,98,115,0,1,1,0,0,11,0,97,0,0,0,1,4,118,101,99, 52,95,97,98,115,0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,97,0,0,0,0,1,90,95,0,0,12, diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c index bef0f856534..e5809509c93 100644 --- a/src/mesa/shader/slang/slang_builtin.c +++ b/src/mesa/shader/slang/slang_builtin.c @@ -741,7 +741,7 @@ static const struct input_info vertInputs[] = { { "gl_MultiTexCoord5", VERT_ATTRIB_TEX5, GL_FLOAT_VEC4, SWIZZLE_NOOP }, { "gl_MultiTexCoord6", VERT_ATTRIB_TEX6, GL_FLOAT_VEC4, SWIZZLE_NOOP }, { "gl_MultiTexCoord7", VERT_ATTRIB_TEX7, GL_FLOAT_VEC4, SWIZZLE_NOOP }, - { NULL, 0, SWIZZLE_NOOP } + { NULL, 0, GL_NONE, SWIZZLE_NOOP } }; /** Predefined fragment shader inputs */ @@ -754,7 +754,7 @@ static const struct input_info fragInputs[] = { { "gl_FogFragCoord", FRAG_ATTRIB_FOGC, GL_FLOAT, SWIZZLE_XXXX }, { "gl_FrontFacing", FRAG_ATTRIB_FACE, GL_FLOAT, SWIZZLE_XXXX }, { "gl_PointCoord", FRAG_ATTRIB_PNTC, GL_FLOAT_VEC2, SWIZZLE_XYZW }, - { NULL, 0, SWIZZLE_NOOP } + { NULL, 0, GL_NONE, SWIZZLE_NOOP } }; diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c index 349f432deca..344dfdc6804 100644 --- a/src/mesa/shader/slang/slang_codegen.c +++ b/src/mesa/shader/slang/slang_codegen.c @@ -422,6 +422,7 @@ static slang_asm_info AsmInfo[] = { { "vec4_lrp", IR_LRP, 1, 3 }, { "vec4_min", IR_MIN, 1, 2 }, { "vec4_max", IR_MAX, 1, 2 }, + { "vec4_cmp", IR_CMP, 1, 3 }, { "vec4_clamp", IR_CLAMP, 1, 3 }, { "vec4_seq", IR_SEQUAL, 1, 2 }, { "vec4_sne", IR_SNEQUAL, 1, 2 }, @@ -2775,7 +2776,7 @@ _slang_gen_while(slang_assemble_ctx * A, slang_operation *oper) * body code (child[1]) */ slang_ir_node *loop, *breakIf, *body; - GLboolean isConst, constTrue; + GLboolean isConst, constTrue = GL_FALSE; if (!A->EmitContReturn) { /* We don't want to emit CONT instructions. If this while-loop has diff --git a/src/mesa/shader/slang/slang_compile.c b/src/mesa/shader/slang/slang_compile.c index c1b97c7cb70..a2708884437 100644 --- a/src/mesa/shader/slang/slang_compile.c +++ b/src/mesa/shader/slang/slang_compile.c @@ -2814,6 +2814,16 @@ _slang_compile(GLcontext *ctx, struct gl_shader *shader) (ctx->Shader.Flags & GLSL_NO_OPT) == 0) { _mesa_optimize_program(ctx, shader->Program); } + if ((ctx->Shader.Flags & GLSL_NOP_VERT) && + shader->Program->Target == GL_VERTEX_PROGRAM_ARB) { + _mesa_nop_vertex_program(ctx, + (struct gl_vertex_program *) shader->Program); + } + if ((ctx->Shader.Flags & GLSL_NOP_FRAG) && + shader->Program->Target == GL_FRAGMENT_PROGRAM_ARB) { + _mesa_nop_fragment_program(ctx, + (struct gl_fragment_program *) shader->Program); + } } if (ctx->Shader.Flags & GLSL_LOG) { diff --git a/src/mesa/shader/slang/slang_emit.c b/src/mesa/shader/slang/slang_emit.c index 3f455e0640b..3af301eacdf 100644 --- a/src/mesa/shader/slang/slang_emit.c +++ b/src/mesa/shader/slang/slang_emit.c @@ -2287,6 +2287,7 @@ emit(slang_emit_info *emitInfo, slang_ir_node *n) case IR_POW: /* trinary operators */ case IR_LRP: + case IR_CMP: return emit_arith(emitInfo, n); case IR_EQUAL: diff --git a/src/mesa/shader/slang/slang_ir.c b/src/mesa/shader/slang/slang_ir.c index 1c7f7474e71..62603503dd9 100644 --- a/src/mesa/shader/slang/slang_ir.c +++ b/src/mesa/shader/slang/slang_ir.c @@ -80,6 +80,7 @@ static const slang_ir_info IrInfo[] = { { IR_NOISE4, "IR_NOISE4", OPCODE_NOISE4, 1, 1 }, /* other */ + { IR_CMP, "IR_CMP", OPCODE_CMP, 4, 3 }, /* compare/select */ { IR_SEQ, "IR_SEQ", OPCODE_NOP, 0, 0 }, { IR_SCOPE, "IR_SCOPE", OPCODE_NOP, 0, 0 }, { IR_LABEL, "IR_LABEL", OPCODE_NOP, 0, 0 }, diff --git a/src/mesa/shader/slang/slang_ir.h b/src/mesa/shader/slang/slang_ir.h index e796693ed58..166b4e80436 100644 --- a/src/mesa/shader/slang/slang_ir.h +++ b/src/mesa/shader/slang/slang_ir.h @@ -91,6 +91,7 @@ typedef enum IR_CLAMP, IR_MIN, IR_MAX, + IR_CMP, /* = (op0 < 0) ? op1 : op2 */ IR_SEQUAL, /* Set if args are equal (vector) */ IR_SNEQUAL, /* Set if args are not equal (vector) */ IR_SGE, /* Set if greater or equal (vector) */ diff --git a/src/mesa/state_tracker/st_atom_depth.c b/src/mesa/state_tracker/st_atom_depth.c index 0aa128f9479..88b80a07fc9 100644 --- a/src/mesa/state_tracker/st_atom_depth.c +++ b/src/mesa/state_tracker/st_atom_depth.c @@ -104,10 +104,6 @@ update_depth_stencil_alpha(struct st_context *st) dsa->depth.func = st_compare_func_to_pipe(ctx->Depth.Func); } - if (ctx->Query.CurrentOcclusionObject && - ctx->Query.CurrentOcclusionObject->Active) - dsa->depth.occlusion_count = 1; - if (ctx->Stencil.Enabled && ctx->DrawBuffer->Visual.stencilBits > 0) { dsa->stencil[0].enabled = 1; dsa->stencil[0].func = st_compare_func_to_pipe(ctx->Stencil.Function[0]); diff --git a/src/mesa/state_tracker/st_cb_accum.c b/src/mesa/state_tracker/st_cb_accum.c index 95181578f63..3d1d0f71d53 100644 --- a/src/mesa/state_tracker/st_cb_accum.c +++ b/src/mesa/state_tracker/st_cb_accum.c @@ -241,7 +241,7 @@ accum_return(GLcontext *ctx, GLfloat value, xpos, ypos, width, height); - if (usage != PIPE_TRANSFER_WRITE) + if (usage & PIPE_TRANSFER_READ) pipe_get_tile_rgba(color_trans, 0, 0, width, height, buf); switch (acc_strb->format) { diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c index 902fb38d1a7..a22fa68299b 100644 --- a/src/mesa/state_tracker/st_cb_bitmap.c +++ b/src/mesa/state_tracker/st_cb_bitmap.c @@ -330,7 +330,18 @@ setup_bitmap_vertex_data(struct st_context *st, const GLfloat clip_y0 = (GLfloat)(y0 / fb_height * 2.0 - 1.0); const GLfloat clip_x1 = (GLfloat)(x1 / fb_width * 2.0 - 1.0); const GLfloat clip_y1 = (GLfloat)(y1 / fb_height * 2.0 - 1.0); - const GLuint max_slots = 4096 / sizeof(st->bitmap.vertices); + + /* XXX: Need to improve buffer_write to allow NO_WAIT (as well as + * no_flush) updates to buffers where we know there is no conflict + * with previous data. Currently using max_slots > 1 will cause + * synchronous rendering if the driver flushes its command buffers + * between one bitmap and the next. Our flush hook below isn't + * sufficient to catch this as the driver doesn't tell us when it + * flushes its own command buffers. Until this gets fixed, pay the + * price of allocating a new buffer for each bitmap cache-flush to + * avoid synchronous rendering. + */ + const GLuint max_slots = 1; /* 4096 / sizeof(st->bitmap.vertices); */ GLuint i; if (st->bitmap.vbuf_slot >= max_slots) { diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c index 8a8c99f7e17..36510720a4c 100644 --- a/src/mesa/state_tracker/st_cb_clear.c +++ b/src/mesa/state_tracker/st_cb_clear.c @@ -116,7 +116,18 @@ draw_quad(GLcontext *ctx, { struct st_context *st = ctx->st; struct pipe_context *pipe = st->pipe; - const GLuint max_slots = 1024 / sizeof(st->clear.vertices); + + /* XXX: Need to improve buffer_write to allow NO_WAIT (as well as + * no_flush) updates to buffers where we know there is no conflict + * with previous data. Currently using max_slots > 1 will cause + * synchronous rendering if the driver flushes its command buffers + * between one bitmap and the next. Our flush hook below isn't + * sufficient to catch this as the driver doesn't tell us when it + * flushes its own command buffers. Until this gets fixed, pay the + * price of allocating a new buffer for each bitmap cache-flush to + * avoid synchronous rendering. + */ + const GLuint max_slots = 1; /* 1024 / sizeof(st->clear.vertices); */ GLuint i; if (st->clear.vbuf_slot >= max_slots) { diff --git a/src/mesa/state_tracker/st_cb_program.c b/src/mesa/state_tracker/st_cb_program.c index 4398ab28392..b2d5c39a3a0 100644 --- a/src/mesa/state_tracker/st_cb_program.c +++ b/src/mesa/state_tracker/st_cb_program.c @@ -45,6 +45,7 @@ #include "st_context.h" #include "st_program.h" #include "st_atom_shader.h" +#include "st_mesa_to_tgsi.h" #include "st_cb_program.h" @@ -152,7 +153,7 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog) } if (stvp->state.tokens) { - _mesa_free((void *) stvp->state.tokens); + st_free_tokens(stvp->state.tokens); stvp->state.tokens = NULL; } } @@ -167,7 +168,7 @@ st_delete_program(GLcontext *ctx, struct gl_program *prog) } if (stfp->state.tokens) { - _mesa_free((void *) stfp->state.tokens); + st_free_tokens(stfp->state.tokens); stfp->state.tokens = NULL; } @@ -214,7 +215,7 @@ static void st_program_string_notify( GLcontext *ctx, } if (stfp->state.tokens) { - _mesa_free((void *) stfp->state.tokens); + st_free_tokens(stfp->state.tokens); stfp->state.tokens = NULL; } @@ -242,7 +243,7 @@ static void st_program_string_notify( GLcontext *ctx, } if (stvp->state.tokens) { - _mesa_free((void *) stvp->state.tokens); + st_free_tokens(stvp->state.tokens); stvp->state.tokens = NULL; } diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 3520c986b47..9a634bb9302 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -328,10 +328,13 @@ guess_and_alloc_texture(struct st_context *st, stObj->base.MinFilter == GL_LINEAR || stImage->base._BaseFormat == GL_DEPTH_COMPONENT || stImage->base._BaseFormat == GL_DEPTH_STENCIL_EXT) && + !stObj->base.GenerateMipmap && stImage->level == firstLevel) { + /* only alloc space for a single mipmap level */ lastLevel = firstLevel; } else { + /* alloc space for a full mipmap */ GLuint l2width = util_logbase2(width); GLuint l2height = util_logbase2(height); GLuint l2depth = util_logbase2(depth); @@ -1693,53 +1696,6 @@ st_CopyTexSubImage3D(GLcontext * ctx, GLenum target, GLint level, } -/** - * Compute which mipmap levels that really need to be sent to the hardware. - * This depends on the base image size, GL_TEXTURE_MIN_LOD, - * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL. - */ -static void -calculate_first_last_level(struct st_texture_object *stObj) -{ - struct gl_texture_object *tObj = &stObj->base; - - /* These must be signed values. MinLod and MaxLod can be negative numbers, - * and having firstLevel and lastLevel as signed prevents the need for - * extra sign checks. - */ - GLint firstLevel; - GLint lastLevel; - - /* Yes, this looks overly complicated, but it's all needed. - */ - switch (tObj->Target) { - case GL_TEXTURE_1D: - case GL_TEXTURE_2D: - case GL_TEXTURE_3D: - case GL_TEXTURE_CUBE_MAP: - if (tObj->MinFilter == GL_NEAREST || tObj->MinFilter == GL_LINEAR) { - /* GL_NEAREST and GL_LINEAR only care about GL_TEXTURE_BASE_LEVEL. - */ - firstLevel = lastLevel = tObj->BaseLevel; - } - else { - firstLevel = 0; - lastLevel = MIN2(tObj->MaxLevel, - (int) tObj->Image[0][tObj->BaseLevel]->WidthLog2); - } - break; - case GL_TEXTURE_RECTANGLE_NV: - case GL_TEXTURE_4D_SGIS: - firstLevel = lastLevel = 0; - break; - default: - return; - } - - stObj->lastLevel = lastLevel; -} - - static void copy_image_data_to_texture(struct st_context *st, struct st_texture_object *stObj, @@ -1803,13 +1759,20 @@ st_finalize_texture(GLcontext *ctx, *needFlush = GL_FALSE; - /* We know/require this is true by now: - */ - assert(stObj->base._Complete); + if (stObj->base._Complete) { + /* The texture is complete and we know exactly how many mipmap levels + * are present/needed. This is conditional because we may be called + * from the st_generate_mipmap() function when the texture object is + * incomplete. In that case, we'll have set stObj->lastLevel before + * we get here. + */ + if (stObj->base.MinFilter == GL_LINEAR || + stObj->base.MinFilter == GL_NEAREST) + stObj->lastLevel = stObj->base.BaseLevel; + else + stObj->lastLevel = stObj->base._MaxLevel - stObj->base.BaseLevel; + } - /* What levels must the texture include at a minimum? - */ - calculate_first_last_level(stObj); firstImage = st_texture_image(stObj->base.Image[0][stObj->base.BaseLevel]); /* If both firstImage and stObj point to a texture which can contain diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c index 8514b6b3756..96969c736c0 100644 --- a/src/mesa/state_tracker/st_context.c +++ b/src/mesa/state_tracker/st_context.c @@ -332,7 +332,7 @@ void st_init_driver_functions(struct dd_function_table *functions) st_init_feedback_functions(functions); #endif st_init_program_functions(functions); -#if FEATURE_ARB_occlusion_query +#if FEATURE_queryobj st_init_query_functions(functions); #endif st_init_readpixels_functions(functions); diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index 503a5f34a3f..c76bff91819 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -328,23 +328,29 @@ get_arrays_bounds(const struct st_vertex_program *vp, const GLubyte **low, const GLubyte **high) { const GLubyte *low_addr = NULL; + const GLubyte *high_addr = NULL; GLuint attr; - GLint stride; for (attr = 0; attr < vp->num_inputs; attr++) { const GLuint mesaAttr = vp->index_to_input[attr]; + const GLint stride = arrays[mesaAttr]->StrideB; const GLubyte *start = arrays[mesaAttr]->Ptr; - stride = arrays[mesaAttr]->StrideB; + const unsigned sz = (arrays[mesaAttr]->Size * + _mesa_sizeof_type(arrays[mesaAttr]->Type)); + const GLubyte *end = start + (max_index * stride) + sz; + if (attr == 0) { low_addr = start; + high_addr = end; } else { low_addr = MIN2(low_addr, start); + high_addr = MAX2(high_addr, end); } } *low = low_addr; - *high = low_addr + (max_index + 1) * stride; + *high = high_addr; } diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index 58f69336523..f75b2348b82 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -27,6 +27,7 @@ #include "main/imports.h" +#include "main/macros.h" #include "main/mipmap.h" #include "main/teximage.h" #include "main/texformat.h" @@ -161,6 +162,43 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, } +/** + * Compute the expected number of mipmap levels in the texture given + * the width/height/depth of the base image and the GL_TEXTURE_BASE_LEVEL/ + * GL_TEXTURE_MAX_LEVEL settings. This will tell us how many mipmap + * level should be generated. + */ +static GLuint +compute_num_levels(GLcontext *ctx, + struct gl_texture_object *texObj, + GLenum target) +{ + if (target == GL_TEXTURE_RECTANGLE_ARB) { + return 1; + } + else { + const GLuint maxLevels = texObj->MaxLevel - texObj->BaseLevel + 1; + const struct gl_texture_image *baseImage = + _mesa_get_tex_image(ctx, texObj, target, texObj->BaseLevel); + GLuint size, numLevels; + + size = MAX2(baseImage->Width2, baseImage->Height2); + size = MAX2(size, baseImage->Depth2); + + numLevels = 0; + + while (size > 0) { + numLevels++; + size >>= 1; + } + + numLevels = MIN2(numLevels, maxLevels); + + return numLevels; + } +} + + void st_generate_mipmap(GLcontext *ctx, GLenum target, struct gl_texture_object *texObj) @@ -174,9 +212,49 @@ st_generate_mipmap(GLcontext *ctx, GLenum target, if (!pt) return; - lastLevel = pt->last_level; + /* find expected last mipmap level */ + lastLevel = compute_num_levels(ctx, texObj, target) - 1; + + if (pt->last_level < lastLevel) { + /* The current gallium texture doesn't have space for all the + * mipmap levels we need to generate. So allocate a new texture. + */ + struct st_texture_object *stObj = st_texture_object(texObj); + struct pipe_texture *oldTex = stObj->pt; + GLboolean needFlush; + + /* create new texture with space for more levels */ + stObj->pt = st_texture_create(st, + oldTex->target, + oldTex->format, + lastLevel, + oldTex->width[0], + oldTex->height[0], + oldTex->depth[0], + oldTex->tex_usage); + + /* The texture isn't in a "complete" state yet so set the expected + * lastLevel here, since it won't get done in st_finalize_texture(). + */ + stObj->lastLevel = lastLevel; + + /* This will copy the old texture's base image into the new texture + * which we just allocated. + */ + st_finalize_texture(ctx, st->pipe, texObj, &needFlush); + + /* release the old tex (will likely be freed too) */ + pipe_texture_reference(&oldTex, NULL); + + pt = stObj->pt; + } + + assert(lastLevel <= pt->last_level); - if (!st_render_mipmap(st, target, pt, baseLevel, lastLevel)) { + /* Recall that the Mesa BaseLevel image is stored in the gallium + * texture's level[0] position. So pass baseLevel=0 here. + */ + if (!st_render_mipmap(st, target, pt, 0, lastLevel)) { fallback_generate_mipmap(ctx, target, texObj); } diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c index 04be57f8ffa..b0a1b529f1e 100644 --- a/src/mesa/state_tracker/st_mesa_to_tgsi.c +++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c @@ -869,3 +869,14 @@ out: return tokens; } + + +/** + * Tokens cannot be free with _mesa_free otherwise the builtin gallium + * malloc debugging will get confused. + */ +void +st_free_tokens(const struct tgsi_token *tokens) +{ + FREE((void *)tokens); +} diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.h b/src/mesa/state_tracker/st_mesa_to_tgsi.h index 679d0ddd41b..c0d1ff59e1f 100644 --- a/src/mesa/state_tracker/st_mesa_to_tgsi.h +++ b/src/mesa/state_tracker/st_mesa_to_tgsi.h @@ -56,6 +56,9 @@ st_translate_mesa_program( const ubyte outputSemanticIndex[], const GLbitfield outputFlags[] ); +void +st_free_tokens(const struct tgsi_token *tokens); + #if defined __cplusplus } /* extern "C" */ diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index 5f9d2a6dad9..927f60cc7e9 100644 --- a/src/mesa/state_tracker/st_program.c +++ b/src/mesa/state_tracker/st_program.c @@ -317,7 +317,7 @@ st_translate_vertex_program(struct st_context *st, /* free old shader state, if any */ if (stvp->state.tokens) { - _mesa_free((void *) stvp->state.tokens); + st_free_tokens(stvp->state.tokens); stvp->state.tokens = NULL; } if (stvp->driver_shader) { diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index bbc2830e694..ba8d1e8cc16 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -577,7 +577,6 @@ st_teximage_flush_before_map(struct st_context *st, pipe->is_texture_referenced(pipe, pt, face, level); if (referenced && ((referenced & PIPE_REFERENCED_FOR_WRITE) || - usage == PIPE_TRANSFER_WRITE || - usage == PIPE_TRANSFER_READ_WRITE)) + (usage & PIPE_TRANSFER_WRITE))) st_flush(st, PIPE_FLUSH_RENDER_CACHE, NULL); } diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c index 613a91b0ecd..77a77f0bcbb 100644 --- a/src/mesa/swrast/s_fragprog.c +++ b/src/mesa/swrast/s_fragprog.c @@ -89,6 +89,8 @@ fetch_texel_lod( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda, * Fetch a texel with the given partial derivatives to compute a level * of detail in the mipmap. * Called via machine->FetchTexelDeriv() + * \param lodBias the lod bias which may be specified by a TXB instruction, + * otherwise zero. */ static void fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], @@ -96,7 +98,8 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lodBias, GLuint unit, GLfloat color[4] ) { SWcontext *swrast = SWRAST_CONTEXT(ctx); - const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current; + const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; + const struct gl_texture_object *texObj = texUnit->_Current; if (texObj) { const struct gl_texture_image *texImg = @@ -108,10 +111,12 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */ texdx[1], texdy[1], /* dt/dx, dt/dy */ - texdx[3], texdy[2], /* dq/dx, dq/dy */ + texdx[3], texdy[3], /* dq/dx, dq/dy */ texW, texH, texcoord[0], texcoord[1], texcoord[3], - 1.0F / texcoord[3]) + lodBias; + 1.0F / texcoord[3]); + + lambda += lodBias + texUnit->LodBias + texObj->LodBias; lambda = CLAMP(lambda, texObj->MinLod, texObj->MaxLod); diff --git a/src/mesa/swrast/s_readpix.c b/src/mesa/swrast/s_readpix.c index 48b9408d242..a1aeb2e01fb 100644 --- a/src/mesa/swrast/s_readpix.c +++ b/src/mesa/swrast/s_readpix.c @@ -555,15 +555,15 @@ _swrast_ReadPixels( GLcontext *ctx, SWcontext *swrast = SWRAST_CONTEXT(ctx); struct gl_pixelstore_attrib clippedPacking = *packing; + if (ctx->NewState) + _mesa_update_state(ctx); + /* Need to do swrast_render_start() before clipping or anything else * since this is where a driver may grab the hw lock and get an updated * window size. */ swrast_render_start(ctx); - if (ctx->NewState) - _mesa_update_state(ctx); - if (swrast->NewState) _swrast_validate_derived( ctx ); diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c index a45eac438e4..704230d1d7b 100644 --- a/src/mesa/swrast/s_span.c +++ b/src/mesa/swrast/s_span.c @@ -904,7 +904,6 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span) } } -#if FEATURE_ARB_occlusion_query if (ctx->Query.CurrentOcclusionObject) { /* update count of 'passed' fragments */ struct gl_query_object *q = ctx->Query.CurrentOcclusionObject; @@ -912,7 +911,6 @@ _swrast_write_index_span( GLcontext *ctx, SWspan *span) for (i = 0; i < span->end; i++) q->Result += span->array->mask[i]; } -#endif /* we have to wait until after occlusion to do this test */ if (ctx->Color.IndexMask == 0) { @@ -1376,7 +1374,6 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span) } } -#if FEATURE_ARB_occlusion_query if (ctx->Query.CurrentOcclusionObject) { /* update count of 'passed' fragments */ struct gl_query_object *q = ctx->Query.CurrentOcclusionObject; @@ -1384,7 +1381,6 @@ _swrast_write_rgba_span( GLcontext *ctx, SWspan *span) for (i = 0; i < span->end; i++) q->Result += span->array->mask[i]; } -#endif /* We had to wait until now to check for glColorMask(0,0,0,0) because of * the occlusion test. diff --git a/src/mesa/swrast/s_texfilter.c b/src/mesa/swrast/s_texfilter.c index 6b1f9346475..004d4e05ae5 100644 --- a/src/mesa/swrast/s_texfilter.c +++ b/src/mesa/swrast/s_texfilter.c @@ -1862,7 +1862,7 @@ choose_cube_face(const struct gl_texture_object *texObj, GLuint face; GLfloat sc, tc, ma; - if (arx > ary && arx > arz) { + if (arx >= ary && arx >= arz) { if (rx >= 0.0F) { face = FACE_POS_X; sc = -rz; @@ -1876,7 +1876,7 @@ choose_cube_face(const struct gl_texture_object *texObj, ma = arx; } } - else if (ary > arx && ary > arz) { + else if (ary >= arx && ary >= arz) { if (ry >= 0.0F) { face = FACE_POS_Y; sc = rx; @@ -1905,8 +1905,12 @@ choose_cube_face(const struct gl_texture_object *texObj, } } - newCoord[0] = ( sc / ma + 1.0F ) * 0.5F; - newCoord[1] = ( tc / ma + 1.0F ) * 0.5F; + { + const float ima = 1.0F / ma; + newCoord[0] = ( sc * ima + 1.0F ) * 0.5F; + newCoord[1] = ( tc * ima + 1.0F ) * 0.5F; + } + return (const struct gl_texture_image **) texObj->Image[face]; } diff --git a/src/mesa/vbo/vbo_exec.h b/src/mesa/vbo/vbo_exec.h index e0f44892cff..516be0995b2 100644 --- a/src/mesa/vbo/vbo_exec.h +++ b/src/mesa/vbo/vbo_exec.h @@ -161,8 +161,26 @@ void vbo_exec_array_destroy( struct vbo_exec_context *exec ); void vbo_exec_vtx_init( struct vbo_exec_context *exec ); void vbo_exec_vtx_destroy( struct vbo_exec_context *exec ); + +#if FEATURE_beginend + void vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ); void vbo_exec_vtx_map( struct vbo_exec_context *exec ); + +#else /* FEATURE_beginend */ + +static INLINE void +vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) +{ +} + +static INLINE void +vbo_exec_vtx_map( struct vbo_exec_context *exec ) +{ +} + +#endif /* FEATURE_beginend */ + void vbo_exec_vtx_wrap( struct vbo_exec_context *exec ); void vbo_exec_eval_update( struct vbo_exec_context *exec ); diff --git a/src/mesa/vbo/vbo_exec_api.c b/src/mesa/vbo/vbo_exec_api.c index 387d4ee3d4a..bfa6d76886a 100644 --- a/src/mesa/vbo/vbo_exec_api.c +++ b/src/mesa/vbo/vbo_exec_api.c @@ -35,9 +35,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. #include "main/context.h" #include "main/macros.h" #include "main/vtxfmt.h" -#if FEATURE_dlist #include "main/dlist.h" -#endif +#include "main/eval.h" #include "main/state.h" #include "main/light.h" #include "main/api_arrayelt.h" @@ -349,7 +348,7 @@ static void vbo_exec_fixup_vertex( GLcontext *ctx, } - +#if FEATURE_beginend /* */ @@ -392,6 +391,7 @@ do { \ +#if FEATURE_evaluators /* Eval */ static void GLAPIENTRY vbo_exec_EvalCoord1f( GLfloat u ) @@ -485,6 +485,12 @@ static void GLAPIENTRY vbo_exec_EvalPoint2( GLint i, GLint j ) vbo_exec_EvalCoord2f( u, v ); } +/* use noop eval mesh */ +#define vbo_exec_EvalMesh1 _mesa_noop_EvalMesh1 +#define vbo_exec_EvalMesh2 _mesa_noop_EvalMesh2 + +#endif /* FEATURE_evaluators */ + /* Build a list of primitives on the fly. Keep * ctx->Driver.CurrentExecPrimitive uptodate as well. @@ -557,24 +563,15 @@ static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec ) { GLvertexformat *vfmt = &exec->vtxfmt; - vfmt->ArrayElement = _ae_loopback_array_elt; /* generic helper */ + _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_); + vfmt->Begin = vbo_exec_Begin; -#if FEATURE_dlist - vfmt->CallList = _mesa_CallList; - vfmt->CallLists = _mesa_CallLists; -#endif vfmt->End = vbo_exec_End; - vfmt->EvalCoord1f = vbo_exec_EvalCoord1f; - vfmt->EvalCoord1fv = vbo_exec_EvalCoord1fv; - vfmt->EvalCoord2f = vbo_exec_EvalCoord2f; - vfmt->EvalCoord2fv = vbo_exec_EvalCoord2fv; - vfmt->EvalPoint1 = vbo_exec_EvalPoint1; - vfmt->EvalPoint2 = vbo_exec_EvalPoint2; - vfmt->Rectf = _mesa_noop_Rectf; - vfmt->EvalMesh1 = _mesa_noop_EvalMesh1; - vfmt->EvalMesh2 = _mesa_noop_EvalMesh2; + _MESA_INIT_DLIST_VTXFMT(vfmt, _mesa_); + _MESA_INIT_EVAL_VTXFMT(vfmt, vbo_exec_); + vfmt->Rectf = _mesa_noop_Rectf; /* from attrib_tmp.h: */ @@ -638,6 +635,98 @@ static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec ) } +#else /* FEATURE_beginend */ + + +#define ATTR( A, N, V0, V1, V2, V3 ) \ +do { \ + struct vbo_exec_context *exec = &vbo_context(ctx)->exec; \ + \ + /* FLUSH_UPDATE_CURRENT needs to be set manually */ \ + exec->ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT; \ + \ + if (exec->vtx.active_sz[A] != N) \ + vbo_exec_fixup_vertex(ctx, A, N); \ + \ + { \ + GLfloat *dest = exec->vtx.attrptr[A]; \ + if (N>0) dest[0] = V0; \ + if (N>1) dest[1] = V1; \ + if (N>2) dest[2] = V2; \ + if (N>3) dest[3] = V3; \ + } \ +} while (0) + +#define ERROR() _mesa_error( ctx, GL_INVALID_ENUM, __FUNCTION__ ) +#define TAG(x) vbo_##x + +#include "vbo_attrib_tmp.h" + +static void vbo_exec_vtxfmt_init( struct vbo_exec_context *exec ) +{ + /* silence warnings */ + (void) vbo_Color3f; + (void) vbo_Color3fv; + (void) vbo_Color4f; + (void) vbo_Color4fv; + (void) vbo_FogCoordfEXT; + (void) vbo_FogCoordfvEXT; + (void) vbo_MultiTexCoord1f; + (void) vbo_MultiTexCoord1fv; + (void) vbo_MultiTexCoord2f; + (void) vbo_MultiTexCoord2fv; + (void) vbo_MultiTexCoord3f; + (void) vbo_MultiTexCoord3fv; + (void) vbo_MultiTexCoord4f; + (void) vbo_MultiTexCoord4fv; + (void) vbo_Normal3f; + (void) vbo_Normal3fv; + (void) vbo_SecondaryColor3fEXT; + (void) vbo_SecondaryColor3fvEXT; + (void) vbo_TexCoord1f; + (void) vbo_TexCoord1fv; + (void) vbo_TexCoord2f; + (void) vbo_TexCoord2fv; + (void) vbo_TexCoord3f; + (void) vbo_TexCoord3fv; + (void) vbo_TexCoord4f; + (void) vbo_TexCoord4fv; + (void) vbo_Vertex2f; + (void) vbo_Vertex2fv; + (void) vbo_Vertex3f; + (void) vbo_Vertex3fv; + (void) vbo_Vertex4f; + (void) vbo_Vertex4fv; + + (void) vbo_VertexAttrib1fARB; + (void) vbo_VertexAttrib1fvARB; + (void) vbo_VertexAttrib2fARB; + (void) vbo_VertexAttrib2fvARB; + (void) vbo_VertexAttrib3fARB; + (void) vbo_VertexAttrib3fvARB; + (void) vbo_VertexAttrib4fARB; + (void) vbo_VertexAttrib4fvARB; + + (void) vbo_VertexAttrib1fNV; + (void) vbo_VertexAttrib1fvNV; + (void) vbo_VertexAttrib2fNV; + (void) vbo_VertexAttrib2fvNV; + (void) vbo_VertexAttrib3fNV; + (void) vbo_VertexAttrib3fvNV; + (void) vbo_VertexAttrib4fNV; + (void) vbo_VertexAttrib4fvNV; + + (void) vbo_Materialfv; + + (void) vbo_EdgeFlag; + (void) vbo_Indexf; + (void) vbo_Indexfv; +} + + +#endif /* FEATURE_beginend */ + + /** * Tell the VBO module to use a real OpenGL vertex buffer object to * store accumulated immediate-mode vertex data. diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c index b9550d6106c..774cffc451b 100644 --- a/src/mesa/vbo/vbo_exec_array.c +++ b/src/mesa/vbo/vbo_exec_array.c @@ -672,29 +672,69 @@ vbo_exec_DrawRangeElementsBaseVertex(GLenum mode, const GLvoid *indices, GLint basevertex) { + static GLuint warnCount = 0; GET_CURRENT_CONTEXT(ctx); if (!_mesa_validate_DrawRangeElements( ctx, mode, start, end, count, type, indices, basevertex )) return; + /* NOTE: It's important that 'end' is a reasonable value. + * in _tnl_draw_prims(), we use end to determine how many vertices + * to transform. If it's too large, we can unnecessarily split prims + * or we can read/write out of memory in several different places! + */ + if (end >= ctx->Array.ArrayObj->_MaxElement) { /* the max element is out of bounds of one or more enabled arrays */ - _mesa_warning(ctx, "glDraw[Range]Elements{,BaseVertex}(start %u, end %u, " - "count %d, type 0x%x, indices=%p, base=%d)\n" - "\tindex=%u is out of bounds (max=%u) " - "Element Buffer %u (size %d)", - start, end, count, type, indices, end, basevertex, - ctx->Array.ArrayObj->_MaxElement - 1, - ctx->Array.ElementArrayBufferObj->Name, - ctx->Array.ElementArrayBufferObj->Size); + warnCount++; + + if (warnCount < 10) { + _mesa_warning(ctx, "glDraw[Range]Elements(start %u, end %u, count %d, " + "type 0x%x, indices=%p)\n" + "\tend is out of bounds (max=%u) " + "Element Buffer %u (size %d)\n" + "\tThis should probably be fixed in the application.", + start, end, count, type, indices, + ctx->Array.ArrayObj->_MaxElement - 1, + ctx->Array.ElementArrayBufferObj->Name, + ctx->Array.ElementArrayBufferObj->Size); + } if (0) dump_element_buffer(ctx, type); if (0) _mesa_print_arrays(ctx); - return; + +#ifdef DEBUG + /* 'end' was out of bounds, but now let's check the actual array + * indexes to see if any of them are out of bounds. If so, warn + * and skip the draw to avoid potential segfault, etc. + */ + { + GLuint max = _mesa_max_buffer_index(ctx, count, type, indices, + ctx->Array.ElementArrayBufferObj); + if (max >= ctx->Array.ArrayObj->_MaxElement) { + if (warnCount < 10) { + _mesa_warning(ctx, "glDraw[Range]Elements(start %u, end %u, " + "count %d, type 0x%x, indices=%p)\n" + "\tindex=%u is out of bounds (max=%u) " + "Element Buffer %u (size %d)\n" + "\tSkipping the glDrawRangeElements() call", + start, end, count, type, indices, max, + ctx->Array.ArrayObj->_MaxElement - 1, + ctx->Array.ElementArrayBufferObj->Name, + ctx->Array.ElementArrayBufferObj->Size); + } + return; + } + /* XXX we could also find the min index and compare to 'start' + * to see if start is correct. But it's more likely to get the + * upper bound wrong. + */ + } +#endif } else if (0) { _mesa_printf("glDraw[Range]Elements{,BaseVertex}" diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c index 0c258c535e0..ee148df4a12 100644 --- a/src/mesa/vbo/vbo_exec_draw.c +++ b/src/mesa/vbo/vbo_exec_draw.c @@ -35,6 +35,9 @@ #include "vbo_context.h" +#if FEATURE_beginend + + static void vbo_exec_debug_verts( struct vbo_exec_context *exec ) { @@ -411,3 +414,6 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap ) exec->vtx.prim_count = 0; exec->vtx.vert_count = 0; } + + +#endif /* FEATURE_beginend */ diff --git a/src/mesa/vbo/vbo_rebase.c b/src/mesa/vbo/vbo_rebase.c index 799a25fc1cb..55a82ee369e 100644 --- a/src/mesa/vbo/vbo_rebase.c +++ b/src/mesa/vbo/vbo_rebase.c @@ -127,7 +127,10 @@ void vbo_rebase_prims( GLcontext *ctx, _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index); - if (ib && ctx->Extensions.ARB_draw_elements_base_vertex) { + /* XXX this path is disabled for now. + * There's rendering corruption in some apps when it's enabled. + */ + if (0 && ib && ctx->Extensions.ARB_draw_elements_base_vertex) { /* If we can just tell the hardware or the TNL to interpret our * indices with a different base, do so. */ diff --git a/src/mesa/vbo/vbo_save_api.c b/src/mesa/vbo/vbo_save_api.c index 41cd21d04b7..4da248e2b88 100644 --- a/src/mesa/vbo/vbo_save_api.c +++ b/src/mesa/vbo/vbo_save_api.c @@ -72,6 +72,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. #include "main/context.h" #include "main/dlist.h" #include "main/enums.h" +#include "main/eval.h" #include "main/macros.h" #include "main/api_noop.h" #include "main/api_validate.h" @@ -988,7 +989,8 @@ static void _save_vtxfmt_init( GLcontext *ctx ) struct vbo_save_context *save = &vbo_context(ctx)->save; GLvertexformat *vfmt = &save->vtxfmt; - vfmt->ArrayElement = _ae_loopback_array_elt; /* generic helper */ + _MESA_INIT_ARRAYELT_VTXFMT(vfmt, _ae_); + vfmt->Begin = _save_Begin; vfmt->Color3f = _save_Color3f; vfmt->Color3fv = _save_Color3fv; @@ -1047,20 +1049,13 @@ static void _save_vtxfmt_init( GLcontext *ctx ) /* This will all require us to fallback to saving the list as opcodes: */ - vfmt->CallList = _save_CallList; /* inside begin/end */ - vfmt->CallLists = _save_CallLists; /* inside begin/end */ - vfmt->EvalCoord1f = _save_EvalCoord1f; - vfmt->EvalCoord1fv = _save_EvalCoord1fv; - vfmt->EvalCoord2f = _save_EvalCoord2f; - vfmt->EvalCoord2fv = _save_EvalCoord2fv; - vfmt->EvalPoint1 = _save_EvalPoint1; - vfmt->EvalPoint2 = _save_EvalPoint2; + _MESA_INIT_DLIST_VTXFMT(vfmt, _save_); /* inside begin/end */ + + _MESA_INIT_EVAL_VTXFMT(vfmt, _save_); /* These are all errors as we at least know we are in some sort of * begin/end pair: */ - vfmt->EvalMesh1 = _save_EvalMesh1; - vfmt->EvalMesh2 = _save_EvalMesh2; vfmt->Begin = _save_Begin; vfmt->Rectf = _save_Rectf; vfmt->DrawArrays = _save_DrawArrays; diff --git a/src/xvmc/Makefile b/src/xvmc/Makefile deleted file mode 100644 index 7badcfd2645..00000000000 --- a/src/xvmc/Makefile +++ /dev/null @@ -1,73 +0,0 @@ -TARGET = libXvMCg3dvl.so -SONAME = libXvMCg3dvl.so.1 -GALLIUMDIR = ../gallium - -OBJECTS = block.o surface.o context.o subpicture.o attributes.o - -ifeq (${DRIVER}, softpipe) -OBJECTS += ${GALLIUMDIR}/winsys/g3dvl/xsp_winsys.o -endif - -CFLAGS += -g -fPIC -Wall -Werror=implicit-function-declaration \ - -I${GALLIUMDIR}/state_trackers/g3dvl \ - -I${GALLIUMDIR}/winsys/g3dvl \ - -I${GALLIUMDIR}/include \ - -I${GALLIUMDIR}/auxiliary \ - -I${GALLIUMDIR}/drivers - -ifeq (${DRIVER}, softpipe) -LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ - -L${GALLIUMDIR}/drivers/softpipe \ - -L${GALLIUMDIR}/auxiliary/tgsi \ - -L${GALLIUMDIR}/auxiliary/draw \ - -L${GALLIUMDIR}/auxiliary/translate \ - -L${GALLIUMDIR}/auxiliary/cso_cache \ - -L${GALLIUMDIR}/auxiliary/util \ - -L${GALLIUMDIR}/auxiliary/rtasm -else -LDFLAGS += -L${GALLIUMDIR}/state_trackers/g3dvl \ - -L${GALLIUMDIR}/winsys/g3dvl/nouveau \ - -L${GALLIUMDIR}/auxiliary/util -endif - -ifeq (${DRIVER}, softpipe) -LIBS += -lg3dvl -lsoftpipe -ldraw -ltgsi -ltranslate -lrtasm -lcso_cache -lutil -lm -else -LIBS += -lg3dvl -lnouveau_dri -lutil -endif - -############################################# - -ifeq (${DRIVER}, softpipe) -.PHONY = all clean g3dvl -else -.PHONY = all clean g3dvl nouveau_winsys -endif - -all: ${TARGET} - -ifeq (${DRIVER}, softpipe) -${TARGET}: ${OBJECTS} g3dvl - $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS} - -g3dvl: - cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} - -clean: - cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean - rm -rf ${OBJECTS} ${TARGET} -else -${TARGET}: ${OBJECTS} g3dvl nouveau_winsys - $(CC) ${LDFLAGS} -shared -Wl,-soname,${SONAME} -o $@ ${OBJECTS} ${LIBS} - -g3dvl: - cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} - -nouveau_winsys: - cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE} - -clean: - cd ${GALLIUMDIR}/state_trackers/g3dvl; ${MAKE} clean - cd ${GALLIUMDIR}/winsys/g3dvl/nouveau; ${MAKE} clean - rm -rf ${OBJECTS} ${TARGET} -endif diff --git a/src/xvmc/attributes.c b/src/xvmc/attributes.c deleted file mode 100644 index 674524b8b82..00000000000 --- a/src/xvmc/attributes.c +++ /dev/null @@ -1,20 +0,0 @@ -#include <assert.h> -#include <X11/Xlib.h> -#include <X11/extensions/Xvlib.h> -#include <X11/extensions/XvMC.h> - -XvAttribute* XvMCQueryAttributes(Display *display, XvMCContext *context, int *number) -{ - return NULL; -} - -Status XvMCSetAttribute(Display *display, XvMCContext *context, Atom attribute, int value) -{ - return BadImplementation; -} - -Status XvMCGetAttribute(Display *display, XvMCContext *context, Atom attribute, int *value) -{ - return BadImplementation; -} - diff --git a/src/xvmc/block.c b/src/xvmc/block.c deleted file mode 100644 index b38a89be092..00000000000 --- a/src/xvmc/block.c +++ /dev/null @@ -1,79 +0,0 @@ -#include <assert.h> -#include <X11/Xlib.h> -#include <X11/extensions/XvMC.h> -#include <util/u_memory.h> -#include <vl_display.h> -#include <vl_screen.h> -#include <vl_context.h> - -#define BLOCK_SIZE (64 * 2) - -Status XvMCCreateBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCBlockArray *blocks) -{ - struct vlContext *vl_ctx; - - assert(display); - - if (!context) - return XvMCBadContext; - if (num_blocks == 0) - return BadValue; - - assert(blocks); - - vl_ctx = context->privData; - assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); - - blocks->context_id = context->context_id; - blocks->num_blocks = num_blocks; - blocks->blocks = MALLOC(BLOCK_SIZE * num_blocks); - /* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */ - blocks->privData = display; - - return Success; -} - -Status XvMCDestroyBlocks(Display *display, XvMCBlockArray *blocks) -{ - assert(display); - assert(blocks); - assert(display == blocks->privData); - FREE(blocks->blocks); - - return Success; -} - -Status XvMCCreateMacroBlocks(Display *display, XvMCContext *context, unsigned int num_blocks, XvMCMacroBlockArray *blocks) -{ - struct vlContext *vl_ctx; - - assert(display); - - if (!context) - return XvMCBadContext; - if (num_blocks == 0) - return BadValue; - - assert(blocks); - - vl_ctx = context->privData; - assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); - - blocks->context_id = context->context_id; - blocks->num_blocks = num_blocks; - blocks->macro_blocks = MALLOC(sizeof(XvMCMacroBlock) * num_blocks); - /* Since we don't have a VL type for blocks, set privData to the display so we can catch mismatches */ - blocks->privData = display; - - return Success; -} - -Status XvMCDestroyMacroBlocks(Display *display, XvMCMacroBlockArray *blocks) -{ - assert(display); - assert(blocks); - assert(display == blocks->privData); - FREE(blocks->macro_blocks); - - return Success; -} diff --git a/src/xvmc/context.c b/src/xvmc/context.c deleted file mode 100644 index 9c2b6648bb5..00000000000 --- a/src/xvmc/context.c +++ /dev/null @@ -1,210 +0,0 @@ -#include <assert.h> -#include <X11/Xlib.h> -#include <X11/extensions/XvMClib.h> -#include <X11/Xlibint.h> -#include <pipe/p_context.h> -#include <vl_display.h> -#include <vl_screen.h> -#include <vl_context.h> -#include <vl_winsys.h> - -static Status Validate -( - Display *display, - XvPortID port, - int surface_type_id, - unsigned int width, - unsigned int height, - int flags, - int *found_port, - int *chroma_format, - int *mc_type -) -{ - unsigned int found_surface = 0; - XvAdaptorInfo *adaptor_info; - unsigned int num_adaptors; - int num_types; - unsigned int max_width, max_height; - Status ret; - unsigned int i, j, k; - - assert(display && chroma_format); - - *found_port = 0; - - ret = XvQueryAdaptors(display, XDefaultRootWindow(display), &num_adaptors, &adaptor_info); - if (ret != Success) - return ret; - - /* Scan through all adaptors looking for this port and surface */ - for (i = 0; i < num_adaptors && !*found_port; ++i) - { - /* Scan through all ports of this adaptor looking for our port */ - for (j = 0; j < adaptor_info[i].num_ports && !*found_port; ++j) - { - /* If this is our port, scan through all its surfaces looking for our surface */ - if (adaptor_info[i].base_id + j == port) - { - XvMCSurfaceInfo *surface_info; - - *found_port = 1; - surface_info = XvMCListSurfaceTypes(display, adaptor_info[i].base_id, &num_types); - - if (surface_info) - { - for (k = 0; k < num_types && !found_surface; ++k) - { - if (surface_info[k].surface_type_id == surface_type_id) - { - found_surface = 1; - max_width = surface_info[k].max_width; - max_height = surface_info[k].max_height; - *chroma_format = surface_info[k].chroma_format; - *mc_type = surface_info[k].mc_type; - } - } - - XFree(surface_info); - } - else - { - XvFreeAdaptorInfo(adaptor_info); - return BadAlloc; - } - } - } - } - - XvFreeAdaptorInfo(adaptor_info); - - if (!*found_port) - return XvBadPort; - if (!found_surface) - return BadMatch; - if (width > max_width || height > max_height) - return BadValue; - if (flags != XVMC_DIRECT && flags != 0) - return BadValue; - - return Success; -} - -static enum vlProfile ProfileToVL(int xvmc_profile) -{ - if (xvmc_profile & XVMC_MPEG_1) - assert(0); - else if (xvmc_profile & XVMC_MPEG_2) - return vlProfileMpeg2Main; - else if (xvmc_profile & XVMC_H263) - assert(0); - else if (xvmc_profile & XVMC_MPEG_4) - assert(0); - else - assert(0); - - return -1; -} - -static enum vlEntryPoint EntryToVL(int xvmc_entry) -{ - return xvmc_entry & XVMC_IDCT ? vlEntryPointIDCT : vlEntryPointMC; -} - -static enum vlFormat FormatToVL(int xvmc_format) -{ - switch (xvmc_format) - { - case XVMC_CHROMA_FORMAT_420: - return vlFormatYCbCr420; - case XVMC_CHROMA_FORMAT_422: - return vlFormatYCbCr422; - case XVMC_CHROMA_FORMAT_444: - return vlFormatYCbCr444; - default: - assert(0); - } - - return -1; -} - -Status XvMCCreateContext(Display *display, XvPortID port, int surface_type_id, int width, int height, int flags, XvMCContext *context) -{ - int found_port; - int chroma_format; - int mc_type; - Status ret; - struct vlDisplay *vl_dpy; - struct vlScreen *vl_scrn; - struct vlContext *vl_ctx; - struct pipe_context *pipe; - Display *dpy = display; - - assert(display); - - if (!context) - return XvMCBadContext; - - ret = Validate(display, port, surface_type_id, width, height, flags, &found_port, &chroma_format, &mc_type); - - /* XXX: Success and XvBadPort have the same value */ - if (ret != Success || !found_port) - return ret; - - /* XXX: Assumes default screen, should check which screen port is on */ - pipe = create_pipe_context(display, XDefaultScreen(display)); - - assert(pipe); - - vlCreateDisplay(display, &vl_dpy); - vlCreateScreen(vl_dpy, XDefaultScreen(display), pipe->screen, &vl_scrn); - vlCreateContext - ( - vl_scrn, - pipe, - width, - height, - FormatToVL(chroma_format), - ProfileToVL(mc_type), - EntryToVL(mc_type), - &vl_ctx - ); - - context->context_id = XAllocID(display); - context->surface_type_id = surface_type_id; - context->width = width; - context->height = height; - context->flags = flags; - context->port = port; - context->privData = vl_ctx; - - SyncHandle(); - return Success; -} - -Status XvMCDestroyContext(Display *display, XvMCContext *context) -{ - struct vlContext *vl_ctx; - struct vlScreen *vl_screen; - struct vlDisplay *vl_dpy; - struct pipe_context *pipe; - - assert(display); - - if (!context) - return XvMCBadContext; - - vl_ctx = context->privData; - - assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); - - pipe = vlGetPipeContext(vl_ctx); - vl_screen = vlContextGetScreen(vl_ctx); - vl_dpy = vlGetDisplay(vl_screen); - vlDestroyContext(vl_ctx); - vlDestroyScreen(vl_screen); - vlDestroyDisplay(vl_dpy); - destroy_pipe_context(pipe); - - return Success; -} diff --git a/src/xvmc/subpicture.c b/src/xvmc/subpicture.c deleted file mode 100644 index 09e9c98e6af..00000000000 --- a/src/xvmc/subpicture.c +++ /dev/null @@ -1,218 +0,0 @@ -#include <assert.h> -#include <X11/Xlib.h> -#include <X11/extensions/Xvlib.h> -#include <X11/extensions/XvMC.h> -#include <X11/Xlibint.h> - -Status XvMCCreateSubpicture -( - Display *display, - XvMCContext *context, - XvMCSubpicture *subpicture, - unsigned short width, - unsigned short height, - int xvimage_id -) -{ - Display *dpy = display; - assert(display); - - if (!context) - return XvMCBadContext; - - assert(subpicture); - - if (width > 2048 || height > 2048) - return BadValue; - - if (xvimage_id != 123) - return BadMatch; - - subpicture->subpicture_id = XAllocID(display); - subpicture->context_id = context->context_id; - subpicture->xvimage_id = xvimage_id; - subpicture->width = width; - subpicture->height = height; - subpicture->num_palette_entries = 0; - subpicture->entry_bytes = 0; - subpicture->component_order[0] = 0; - subpicture->component_order[1] = 0; - subpicture->component_order[2] = 0; - subpicture->component_order[3] = 0; - /* TODO: subpicture->privData = ;*/ - - SyncHandle(); - return Success; -} - -Status XvMCClearSubpicture -( - Display *display, - XvMCSubpicture *subpicture, - short x, - short y, - unsigned short width, - unsigned short height, - unsigned int color -) -{ - assert(display); - - if (!subpicture) - return XvMCBadSubpicture; - - /* TODO: Assert clear rect is within bounds? Or clip? */ - - return Success; -} - -Status XvMCCompositeSubpicture -( - Display *display, - XvMCSubpicture *subpicture, - XvImage *image, - short srcx, - short srcy, - unsigned short width, - unsigned short height, - short dstx, - short dsty -) -{ - assert(display); - - if (!subpicture) - return XvMCBadSubpicture; - - assert(image); - - if (subpicture->xvimage_id != image->id) - return BadMatch; - - /* TODO: Assert rects are within bounds? Or clip? */ - - return Success; -} - -Status XvMCDestroySubpicture(Display *display, XvMCSubpicture *subpicture) -{ - assert(display); - - if (!subpicture) - return XvMCBadSubpicture; - - return BadImplementation; -} - -Status XvMCSetSubpicturePalette(Display *display, XvMCSubpicture *subpicture, unsigned char *palette) -{ - assert(display); - - if (!subpicture) - return XvMCBadSubpicture; - - assert(palette); - - /* We don't support paletted subpictures */ - return BadMatch; -} - -Status XvMCBlendSubpicture -( - Display *display, - XvMCSurface *target_surface, - XvMCSubpicture *subpicture, - short subx, - short suby, - unsigned short subw, - unsigned short subh, - short surfx, - short surfy, - unsigned short surfw, - unsigned short surfh -) -{ - assert(display); - - if (!target_surface) - return XvMCBadSurface; - - if (!subpicture) - return XvMCBadSubpicture; - - if (target_surface->context_id != subpicture->context_id) - return BadMatch; - - /* TODO: Assert rects are within bounds? Or clip? */ - return Success; -} - -Status XvMCBlendSubpicture2 -( - Display *display, - XvMCSurface *source_surface, - XvMCSurface *target_surface, - XvMCSubpicture *subpicture, - short subx, - short suby, - unsigned short subw, - unsigned short subh, - short surfx, - short surfy, - unsigned short surfw, - unsigned short surfh -) -{ - assert(display); - - if (!source_surface || !target_surface) - return XvMCBadSurface; - - if (!subpicture) - return XvMCBadSubpicture; - - if (source_surface->context_id != subpicture->context_id) - return BadMatch; - - if (source_surface->context_id != subpicture->context_id) - return BadMatch; - - /* TODO: Assert rects are within bounds? Or clip? */ - return Success; -} - -Status XvMCSyncSubpicture(Display *display, XvMCSubpicture *subpicture) -{ - assert(display); - - if (!subpicture) - return XvMCBadSubpicture; - - return Success; -} - -Status XvMCFlushSubpicture(Display *display, XvMCSubpicture *subpicture) -{ - assert(display); - - if (!subpicture) - return XvMCBadSubpicture; - - return Success; -} - -Status XvMCGetSubpictureStatus(Display *display, XvMCSubpicture *subpicture, int *status) -{ - assert(display); - - if (!subpicture) - return XvMCBadSubpicture; - - assert(status); - - /* TODO */ - *status = 0; - - return Success; -} - diff --git a/src/xvmc/surface.c b/src/xvmc/surface.c deleted file mode 100644 index fea351b84f0..00000000000 --- a/src/xvmc/surface.c +++ /dev/null @@ -1,358 +0,0 @@ -#include <assert.h> -#include <X11/Xlib.h> -#include <X11/extensions/XvMC.h> -#include <X11/Xlibint.h> -#include <vl_display.h> -#include <vl_screen.h> -#include <vl_context.h> -#include <vl_surface.h> -#include <vl_types.h> - -static enum vlMacroBlockType TypeToVL(int xvmc_mb_type) -{ - if (xvmc_mb_type & XVMC_MB_TYPE_INTRA) - return vlMacroBlockTypeIntra; - if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_FORWARD) - return vlMacroBlockTypeFwdPredicted; - if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == XVMC_MB_TYPE_MOTION_BACKWARD) - return vlMacroBlockTypeBkwdPredicted; - if ((xvmc_mb_type & (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) == (XVMC_MB_TYPE_MOTION_FORWARD | XVMC_MB_TYPE_MOTION_BACKWARD)) - return vlMacroBlockTypeBiPredicted; - - assert(0); - - return -1; -} - -static enum vlPictureType PictureToVL(int xvmc_pic) -{ - switch (xvmc_pic) - { - case XVMC_TOP_FIELD: - return vlPictureTypeTopField; - case XVMC_BOTTOM_FIELD: - return vlPictureTypeBottomField; - case XVMC_FRAME_PICTURE: - return vlPictureTypeFrame; - default: - assert(0); - } - - return -1; -} - -static enum vlMotionType MotionToVL(int xvmc_motion_type, int xvmc_dct_type) -{ - switch (xvmc_motion_type) - { - case XVMC_PREDICTION_FRAME: - return xvmc_dct_type == XVMC_DCT_TYPE_FIELD ? vlMotionType16x8 : vlMotionTypeFrame; - case XVMC_PREDICTION_FIELD: - return vlMotionTypeField; - case XVMC_PREDICTION_DUAL_PRIME: - return vlMotionTypeDualPrime; - default: - assert(0); - } - - return -1; -} - -Status XvMCCreateSurface(Display *display, XvMCContext *context, XvMCSurface *surface) -{ - struct vlContext *vl_ctx; - struct vlSurface *vl_sfc; - Display *dpy = display; - - assert(display); - - if (!context) - return XvMCBadContext; - if (!surface) - return XvMCBadSurface; - - vl_ctx = context->privData; - - assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); - - if (vlCreateSurface(vlContextGetScreen(vl_ctx), - context->width, context->height, - vlGetPictureFormat(vl_ctx), - &vl_sfc)) - { - return BadAlloc; - } - - vlBindToContext(vl_sfc, vl_ctx); - - surface->surface_id = XAllocID(display); - surface->context_id = context->context_id; - surface->surface_type_id = context->surface_type_id; - surface->width = context->width; - surface->height = context->height; - surface->privData = vl_sfc; - - SyncHandle(); - return Success; -} - -Status XvMCRenderSurface -( - Display *display, - XvMCContext *context, - unsigned int picture_structure, - XvMCSurface *target_surface, - XvMCSurface *past_surface, - XvMCSurface *future_surface, - unsigned int flags, - unsigned int num_macroblocks, - unsigned int first_macroblock, - XvMCMacroBlockArray *macroblocks, - XvMCBlockArray *blocks -) -{ - struct vlContext *vl_ctx; - struct vlSurface *target_vl_surface; - struct vlSurface *past_vl_surface; - struct vlSurface *future_vl_surface; - struct vlMpeg2MacroBlockBatch batch; - struct vlMpeg2MacroBlock vl_macroblocks[num_macroblocks]; - unsigned int i; - - assert(display); - - if (!context) - return XvMCBadContext; - if (!target_surface) - return XvMCBadSurface; - - if - ( - picture_structure != XVMC_TOP_FIELD && - picture_structure != XVMC_BOTTOM_FIELD && - picture_structure != XVMC_FRAME_PICTURE - ) - return BadValue; - if (future_surface && !past_surface) - return BadMatch; - - vl_ctx = context->privData; - - assert(display == vlGetNativeDisplay(vlGetDisplay(vlContextGetScreen(vl_ctx)))); - - target_vl_surface = target_surface->privData; - past_vl_surface = past_surface ? past_surface->privData : NULL; - future_vl_surface = future_surface ? future_surface->privData : NULL; - - assert(context->context_id == target_surface->context_id); - assert(!past_surface || context->context_id == past_surface->context_id); - assert(!future_surface || context->context_id == future_surface->context_id); - - assert(macroblocks); - assert(blocks); - - assert(macroblocks->context_id == context->context_id); - assert(blocks->context_id == context->context_id); - - assert(flags == 0 || flags == XVMC_SECOND_FIELD); - - batch.past_surface = past_vl_surface; - batch.future_surface = future_vl_surface; - batch.picture_type = PictureToVL(picture_structure); - batch.field_order = flags & XVMC_SECOND_FIELD ? vlFieldOrderSecond : vlFieldOrderFirst; - batch.num_macroblocks = num_macroblocks; - batch.macroblocks = vl_macroblocks; - - for (i = 0; i < num_macroblocks; ++i) - { - unsigned int j = first_macroblock + i; - - unsigned int k, l, m; - - batch.macroblocks[i].mbx = macroblocks->macro_blocks[j].x; - batch.macroblocks[i].mby = macroblocks->macro_blocks[j].y; - batch.macroblocks[i].mb_type = TypeToVL(macroblocks->macro_blocks[j].macroblock_type); - if (batch.macroblocks[i].mb_type != vlMacroBlockTypeIntra) - batch.macroblocks[i].mo_type = MotionToVL(macroblocks->macro_blocks[j].motion_type, macroblocks->macro_blocks[j].dct_type); - batch.macroblocks[i].dct_type = macroblocks->macro_blocks[j].dct_type == XVMC_DCT_TYPE_FIELD ? vlDCTTypeFieldCoded : vlDCTTypeFrameCoded; - - for (k = 0; k < 2; ++k) - for (l = 0; l < 2; ++l) - for (m = 0; m < 2; ++m) - batch.macroblocks[i].PMV[k][l][m] = macroblocks->macro_blocks[j].PMV[k][l][m]; - - batch.macroblocks[i].cbp = macroblocks->macro_blocks[j].coded_block_pattern; - batch.macroblocks[i].blocks = blocks->blocks + (macroblocks->macro_blocks[j].index * 64); - } - - vlRenderMacroBlocksMpeg2(&batch, target_vl_surface); - - return Success; -} - -Status XvMCFlushSurface(Display *display, XvMCSurface *surface) -{ - struct vlSurface *vl_sfc; - - assert(display); - - if (!surface) - return XvMCBadSurface; - - vl_sfc = surface->privData; - - assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); - - vlSurfaceFlush(vl_sfc); - - return Success; -} - -Status XvMCSyncSurface(Display *display, XvMCSurface *surface) -{ - struct vlSurface *vl_sfc; - - assert(display); - - if (!surface) - return XvMCBadSurface; - - vl_sfc = surface->privData; - - assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); - - vlSurfaceSync(vl_sfc); - - return Success; -} - -Status XvMCPutSurface -( - Display *display, - XvMCSurface *surface, - Drawable drawable, - short srcx, - short srcy, - unsigned short srcw, - unsigned short srch, - short destx, - short desty, - unsigned short destw, - unsigned short desth, - int flags -) -{ - Window root; - int x, y; - unsigned int width, height; - unsigned int border_width; - unsigned int depth; - struct vlSurface *vl_sfc; - - assert(display); - - if (!surface) - return XvMCBadSurface; - - if (XGetGeometry(display, drawable, &root, &x, &y, &width, &height, &border_width, &depth) == BadDrawable) - return BadDrawable; - - assert(flags == XVMC_TOP_FIELD || flags == XVMC_BOTTOM_FIELD || flags == XVMC_FRAME_PICTURE); - - /* TODO: Correct for negative srcx,srcy & destx,desty by clipping */ - - assert(srcx + srcw - 1 < surface->width); - assert(srcy + srch - 1 < surface->height); - /* XXX: Some apps (mplayer) hit these asserts because they call - * this function after the window has been resized by the WM - * but before they've handled the corresponding XEvent and - * know about the new dimensions. The output will be clipped - * for a few frames until the app updates destw and desth. - */ - /*assert(destx + destw - 1 < width); - assert(desty + desth - 1 < height);*/ - - vl_sfc = surface->privData; - - vlPutPicture(vl_sfc, drawable, srcx, srcy, srcw, srch, destx, desty, destw, desth, width, height, PictureToVL(flags)); - - return Success; -} - -Status XvMCGetSurfaceStatus(Display *display, XvMCSurface *surface, int *status) -{ - struct vlSurface *vl_sfc; - enum vlResourceStatus res_status; - - assert(display); - - if (!surface) - return XvMCBadSurface; - - assert(status); - - vl_sfc = surface->privData; - - assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); - - vlSurfaceGetStatus(vl_sfc, &res_status); - - switch (res_status) - { - case vlResourceStatusFree: - { - *status = 0; - break; - } - case vlResourceStatusRendering: - { - *status = XVMC_RENDERING; - break; - } - case vlResourceStatusDisplaying: - { - *status = XVMC_DISPLAYING; - break; - } - default: - assert(0); - } - - return Success; -} - -Status XvMCDestroySurface(Display *display, XvMCSurface *surface) -{ - struct vlSurface *vl_sfc; - - assert(display); - - if (!surface) - return XvMCBadSurface; - - vl_sfc = surface->privData; - - assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); - - vlDestroySurface(vl_sfc); - - return Success; -} - -Status XvMCHideSurface(Display *display, XvMCSurface *surface) -{ - struct vlSurface *vl_sfc; - - assert(display); - - if (!surface) - return XvMCBadSurface; - - vl_sfc = surface->privData; - - assert(display == vlGetNativeDisplay(vlGetDisplay(vlSurfaceGetScreen(vl_sfc)))); - - /* No op, only for overlaid rendering */ - - return Success; -} diff --git a/src/xvmc/tests/Makefile b/src/xvmc/tests/Makefile deleted file mode 100644 index de095161d29..00000000000 --- a/src/xvmc/tests/Makefile +++ /dev/null @@ -1,27 +0,0 @@ -CFLAGS += -g -Wall -LDFLAGS += -LIBS += -lXvMCW -lXvMC -lXv - -############################################# - -.PHONY = all clean - -all: test_context test_surface test_blocks test_rendering xvmc_bench - -test_context: test_context.o testlib.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -test_surface: test_surface.o testlib.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -test_blocks: test_blocks.o testlib.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -test_rendering: test_rendering.o testlib.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -xvmc_bench: xvmc_bench.o testlib.o - $(CC) ${LDFLAGS} -o $@ $^ ${LIBS} - -clean: - rm -rf *.o test_context test_surface test_blocks test_rendering xvmc_bench diff --git a/src/xvmc/tests/testlib.h b/src/xvmc/tests/testlib.h deleted file mode 100644 index af71ad74e17..00000000000 --- a/src/xvmc/tests/testlib.h +++ /dev/null @@ -1,42 +0,0 @@ -#ifndef testlib_h -#define testlib_h - -/* -#define TEST(pred, doc) test(pred, #pred, doc, __FILE__, __LINE__) - -void test(int pred, const char *pred_string, const char *doc_string, const char *file, unsigned int line); -*/ - -#include <sys/time.h> -#include <X11/Xlib.h> -#include <X11/extensions/XvMClib.h> - -/* - * display: IN A valid X display - * width, height: IN Surface size that the port must display - * chroma_format: IN Chroma format that the port must display - * mc_types, num_mc_types: IN List of MC types that the port must support, first port that matches the first mc_type will be returned - * port_id: OUT Your port's ID - * surface_type_id: OUT Your port's surface ID - * is_overlay: OUT If 1, port uses overlay surfaces, you need to set a colorkey - * intra_unsigned: OUT If 1, port uses unsigned values for intra-coded blocks - */ -int GetPort -( - Display *display, - unsigned int width, - unsigned int height, - unsigned int chroma_format, - const unsigned int *mc_types, - unsigned int num_mc_types, - XvPortID *port_id, - int *surface_type_id, - unsigned int *is_overlay, - unsigned int *intra_unsigned -); - -unsigned int align(unsigned int value, unsigned int alignment); - -int timeval_subtract(struct timeval *result, struct timeval *x, struct timeval *y); - -#endif |