diff options
author | Christian König <[email protected]> | 2010-12-22 21:12:24 +0100 |
---|---|---|
committer | Christian König <[email protected]> | 2010-12-22 21:12:24 +0100 |
commit | e179a8bf34022c200deee92d328cb1bae3c3aa4a (patch) | |
tree | 4c67616380e7c0efa2d7ff0fe67fd18af2fa11ad /src | |
parent | 1b03996b1269fb8b2f89e0a013d8427da660ff83 (diff) | |
parent | 38c8b034e2775640c588711290c244632dcc2475 (diff) |
Merge remote branch 'origin/master' into pipe-video
Diffstat (limited to 'src')
189 files changed, 3112 insertions, 2554 deletions
diff --git a/src/egl/main/egldriver.c b/src/egl/main/egldriver.c index ff0011c4b15..e5d8a39a4c8 100644 --- a/src/egl/main/egldriver.c +++ b/src/egl/main/egldriver.c @@ -186,11 +186,22 @@ _eglLoadModule(_EGLModule *mod) static void _eglUnloadModule(_EGLModule *mod) { +#if defined(_EGL_OS_UNIX) /* destroy the driver */ if (mod->Driver && mod->Driver->Unload) mod->Driver->Unload(mod->Driver); + + /* + * XXX At this point (atexit), the module might be the last reference to + * libEGL. Closing the module might unmap libEGL and give problems. + */ +#if 0 if (mod->Handle) close_library(mod->Handle); +#endif +#elif defined(_EGL_OS_WINDOWS) + /* XXX Windows unloads DLLs before atexit */ +#endif mod->Driver = NULL; mod->Handle = NULL; @@ -670,12 +681,7 @@ _eglUnloadDrivers(void) { /* this is called at atexit time */ if (_eglModules) { -#if defined(_EGL_OS_UNIX) _eglDestroyArray(_eglModules, _eglFreeModule); -#elif defined(_EGL_OS_WINDOWS) - /* XXX Windows unloads DLLs before atexit */ - _eglDestroyArray(_eglModules, NULL); -#endif _eglModules = NULL; } } diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c index 7c198c6026d..c98fb3d5205 100644 --- a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c +++ b/src/gallium/auxiliary/draw/draw_pt_fetch_shade_emit.c @@ -58,8 +58,8 @@ struct fetch_shade_emit { const ubyte *src[PIPE_MAX_ATTRIBS]; unsigned prim; - struct draw_vs_varient_key key; - struct draw_vs_varient *active; + struct draw_vs_variant_key key; + struct draw_vs_variant *active; const struct vertex_info *vinfo; @@ -150,7 +150,7 @@ static void fse_prepare( struct draw_pt_middle_end *middle, } - fse->active = draw_vs_lookup_varient( draw->vs.vertex_shader, + fse->active = draw_vs_lookup_variant( draw->vs.vertex_shader, &fse->key ); if (!fse->active) { diff --git a/src/gallium/auxiliary/draw/draw_vs.c b/src/gallium/auxiliary/draw/draw_vs.c index fb665b08fff..7caad6f0053 100644 --- a/src/gallium/auxiliary/draw/draw_vs.c +++ b/src/gallium/auxiliary/draw/draw_vs.c @@ -165,10 +165,10 @@ draw_delete_vertex_shader(struct draw_context *draw, { unsigned i; - for (i = 0; i < dvs->nr_varients; i++) - dvs->varient[i]->destroy( dvs->varient[i] ); + for (i = 0; i < dvs->nr_variants; i++) + dvs->variant[i]->destroy( dvs->variant[i] ); - dvs->nr_varients = 0; + dvs->nr_variants = 0; dvs->delete( dvs ); } @@ -225,40 +225,40 @@ draw_vs_destroy( struct draw_context *draw ) } -struct draw_vs_varient * -draw_vs_lookup_varient( struct draw_vertex_shader *vs, - const struct draw_vs_varient_key *key ) +struct draw_vs_variant * +draw_vs_lookup_variant( struct draw_vertex_shader *vs, + const struct draw_vs_variant_key *key ) { - struct draw_vs_varient *varient; + struct draw_vs_variant *variant; unsigned i; - /* Lookup existing varient: + /* Lookup existing variant: */ - for (i = 0; i < vs->nr_varients; i++) - if (draw_vs_varient_key_compare(key, &vs->varient[i]->key) == 0) - return vs->varient[i]; + for (i = 0; i < vs->nr_variants; i++) + if (draw_vs_variant_key_compare(key, &vs->variant[i]->key) == 0) + return vs->variant[i]; /* Else have to create a new one: */ - varient = vs->create_varient( vs, key ); - if (varient == NULL) + variant = vs->create_variant( vs, key ); + if (variant == NULL) return NULL; /* Add it to our list, could be smarter: */ - if (vs->nr_varients < Elements(vs->varient)) { - vs->varient[vs->nr_varients++] = varient; + if (vs->nr_variants < Elements(vs->variant)) { + vs->variant[vs->nr_variants++] = variant; } else { - vs->last_varient++; - vs->last_varient %= Elements(vs->varient); - vs->varient[vs->last_varient]->destroy(vs->varient[vs->last_varient]); - vs->varient[vs->last_varient] = varient; + vs->last_variant++; + vs->last_variant %= Elements(vs->variant); + vs->variant[vs->last_variant]->destroy(vs->variant[vs->last_variant]); + vs->variant[vs->last_variant] = variant; } /* Done */ - return varient; + return variant; } diff --git a/src/gallium/auxiliary/draw/draw_vs.h b/src/gallium/auxiliary/draw/draw_vs.h index f9a038788fb..bfb72d50efa 100644 --- a/src/gallium/auxiliary/draw/draw_vs.h +++ b/src/gallium/auxiliary/draw/draw_vs.h @@ -38,7 +38,7 @@ struct draw_context; struct pipe_shader_state; -struct draw_varient_input +struct draw_variant_input { enum pipe_format format; unsigned buffer; @@ -46,19 +46,19 @@ struct draw_varient_input unsigned instance_divisor; }; -struct draw_varient_output +struct draw_variant_output { enum pipe_format format; /* output format */ unsigned vs_output:8; /* which vertex shader output is this? */ unsigned offset:24; /* offset into output vertex */ }; -struct draw_varient_element { - struct draw_varient_input in; - struct draw_varient_output out; +struct draw_variant_element { + struct draw_variant_input in; + struct draw_variant_output out; }; -struct draw_vs_varient_key { +struct draw_vs_variant_key { unsigned output_stride; unsigned nr_elements:8; /* max2(nr_inputs, nr_outputs) */ unsigned nr_inputs:8; @@ -66,34 +66,34 @@ struct draw_vs_varient_key { unsigned viewport:1; unsigned clip:1; unsigned const_vbuffers:5; - struct draw_varient_element element[PIPE_MAX_ATTRIBS]; + struct draw_variant_element element[PIPE_MAX_ATTRIBS]; }; -struct draw_vs_varient; +struct draw_vs_variant; -struct draw_vs_varient { - struct draw_vs_varient_key key; +struct draw_vs_variant { + struct draw_vs_variant_key key; struct draw_vertex_shader *vs; - void (*set_buffer)( struct draw_vs_varient *, + void (*set_buffer)( struct draw_vs_variant *, unsigned i, const void *ptr, unsigned stride, unsigned max_stride ); - void (PIPE_CDECL *run_linear)( struct draw_vs_varient *shader, + void (PIPE_CDECL *run_linear)( struct draw_vs_variant *shader, unsigned start, unsigned count, void *output_buffer ); - void (PIPE_CDECL *run_elts)( struct draw_vs_varient *shader, + void (PIPE_CDECL *run_elts)( struct draw_vs_variant *shader, const unsigned *elts, unsigned count, void *output_buffer ); - void (*destroy)( struct draw_vs_varient * ); + void (*destroy)( struct draw_vs_variant * ); }; @@ -117,11 +117,11 @@ struct draw_vertex_shader { /* */ - struct draw_vs_varient *varient[16]; - unsigned nr_varients; - unsigned last_varient; - struct draw_vs_varient *(*create_varient)( struct draw_vertex_shader *shader, - const struct draw_vs_varient_key *key ); + struct draw_vs_variant *variant[16]; + unsigned nr_variants; + unsigned last_variant; + struct draw_vs_variant *(*create_variant)( struct draw_vertex_shader *shader, + const struct draw_vs_variant_key *key ); void (*prepare)( struct draw_vertex_shader *shader, @@ -144,9 +144,9 @@ struct draw_vertex_shader { }; -struct draw_vs_varient * -draw_vs_lookup_varient( struct draw_vertex_shader *base, - const struct draw_vs_varient_key *key ); +struct draw_vs_variant * +draw_vs_lookup_variant( struct draw_vertex_shader *base, + const struct draw_vs_variant_key *key ); /******************************************************************************** @@ -166,12 +166,12 @@ draw_create_vs_ppc(struct draw_context *draw, const struct pipe_shader_state *templ); -struct draw_vs_varient_key; +struct draw_vs_variant_key; struct draw_vertex_shader; -struct draw_vs_varient * -draw_vs_create_varient_aos_sse( struct draw_vertex_shader *vs, - const struct draw_vs_varient_key *key ); +struct draw_vs_variant * +draw_vs_create_variant_aos_sse( struct draw_vertex_shader *vs, + const struct draw_vs_variant_key *key ); #if HAVE_LLVM struct draw_vertex_shader * @@ -181,7 +181,7 @@ draw_create_vs_llvm(struct draw_context *draw, /******************************************************************************** - * Helpers for vs implementations that don't do their own fetch/emit varients. + * Helpers for vs implementations that don't do their own fetch/emit variants. * Means these can be shared between shaders. */ struct translate; @@ -194,21 +194,21 @@ struct translate *draw_vs_get_fetch( struct draw_context *draw, struct translate *draw_vs_get_emit( struct draw_context *draw, struct translate_key *key ); -struct draw_vs_varient * -draw_vs_create_varient_generic( struct draw_vertex_shader *vs, - const struct draw_vs_varient_key *key ); +struct draw_vs_variant * +draw_vs_create_variant_generic( struct draw_vertex_shader *vs, + const struct draw_vs_variant_key *key ); -static INLINE int draw_vs_varient_keysize( const struct draw_vs_varient_key *key ) +static INLINE int draw_vs_variant_keysize( const struct draw_vs_variant_key *key ) { - return 2 * sizeof(int) + key->nr_elements * sizeof(struct draw_varient_element); + return 2 * sizeof(int) + key->nr_elements * sizeof(struct draw_variant_element); } -static INLINE int draw_vs_varient_key_compare( const struct draw_vs_varient_key *a, - const struct draw_vs_varient_key *b ) +static INLINE int draw_vs_variant_key_compare( const struct draw_vs_variant_key *a, + const struct draw_vs_variant_key *b ) { - int keysize = draw_vs_varient_keysize(a); + int keysize = draw_vs_variant_keysize(a); return memcmp(a, b, keysize); } diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.c b/src/gallium/auxiliary/draw/draw_vs_aos.c index 19f49e34c8b..7b90dba0cd5 100644 --- a/src/gallium/auxiliary/draw/draw_vs_aos.c +++ b/src/gallium/auxiliary/draw/draw_vs_aos.c @@ -1918,7 +1918,7 @@ static void find_last_write_outputs( struct aos_compilation *cp ) #define ARG_OUTBUF 4 -static boolean build_vertex_program( struct draw_vs_varient_aos_sse *varient, +static boolean build_vertex_program( struct draw_vs_variant_aos_sse *variant, boolean linear ) { struct tgsi_parse_context parse; @@ -1927,14 +1927,14 @@ static boolean build_vertex_program( struct draw_vs_varient_aos_sse *varient, util_init_math(); - tgsi_parse_init( &parse, varient->base.vs->state.tokens ); + tgsi_parse_init( &parse, variant->base.vs->state.tokens ); memset(&cp, 0, sizeof(cp)); cp.insn_counter = 1; - cp.vaos = varient; + cp.vaos = variant; cp.have_sse2 = 1; - cp.func = &varient->func[ linear ? 0 : 1 ]; + cp.func = &variant->func[ linear ? 0 : 1 ]; cp.tmp_EAX = x86_make_reg(file_REG32, reg_AX); cp.idx_EBX = x86_make_reg(file_REG32, reg_BX); @@ -2090,20 +2090,20 @@ static boolean build_vertex_program( struct draw_vs_varient_aos_sse *varient, /** cast wrapper */ -static INLINE struct draw_vs_varient_aos_sse * -draw_vs_varient_aos_sse(struct draw_vs_varient *varient) +static INLINE struct draw_vs_variant_aos_sse * +draw_vs_variant_aos_sse(struct draw_vs_variant *variant) { - return (struct draw_vs_varient_aos_sse *) varient; + return (struct draw_vs_variant_aos_sse *) variant; } -static void vaos_set_buffer( struct draw_vs_varient *varient, +static void vaos_set_buffer( struct draw_vs_variant *variant, unsigned buf, const void *ptr, unsigned stride, unsigned max_stride) { - struct draw_vs_varient_aos_sse *vaos = draw_vs_varient_aos_sse(varient); + struct draw_vs_variant_aos_sse *vaos = draw_vs_variant_aos_sse(variant); if (buf < vaos->nr_vb) { vaos->buffer[buf].base_ptr = (char *)ptr; @@ -2115,12 +2115,12 @@ static void vaos_set_buffer( struct draw_vs_varient *varient, -static void PIPE_CDECL vaos_run_elts( struct draw_vs_varient *varient, +static void PIPE_CDECL vaos_run_elts( struct draw_vs_variant *variant, const unsigned *elts, unsigned count, void *output_buffer ) { - struct draw_vs_varient_aos_sse *vaos = draw_vs_varient_aos_sse(varient); + struct draw_vs_variant_aos_sse *vaos = draw_vs_variant_aos_sse(variant); struct aos_machine *machine = vaos->draw->vs.aos_machine; unsigned i; @@ -2139,12 +2139,12 @@ static void PIPE_CDECL vaos_run_elts( struct draw_vs_varient *varient, output_buffer ); } -static void PIPE_CDECL vaos_run_linear( struct draw_vs_varient *varient, +static void PIPE_CDECL vaos_run_linear( struct draw_vs_variant *variant, unsigned start, unsigned count, void *output_buffer ) { - struct draw_vs_varient_aos_sse *vaos = draw_vs_varient_aos_sse(varient); + struct draw_vs_variant_aos_sse *vaos = draw_vs_variant_aos_sse(variant); struct aos_machine *machine = vaos->draw->vs.aos_machine; unsigned i; @@ -2171,9 +2171,9 @@ static void PIPE_CDECL vaos_run_linear( struct draw_vs_varient *varient, -static void vaos_destroy( struct draw_vs_varient *varient ) +static void vaos_destroy( struct draw_vs_variant *variant ) { - struct draw_vs_varient_aos_sse *vaos = draw_vs_varient_aos_sse(varient); + struct draw_vs_variant_aos_sse *vaos = draw_vs_variant_aos_sse(variant); FREE( vaos->buffer ); @@ -2185,11 +2185,11 @@ static void vaos_destroy( struct draw_vs_varient *varient ) -static struct draw_vs_varient *varient_aos_sse( struct draw_vertex_shader *vs, - const struct draw_vs_varient_key *key ) +static struct draw_vs_variant *variant_aos_sse( struct draw_vertex_shader *vs, + const struct draw_vs_variant_key *key ) { unsigned i; - struct draw_vs_varient_aos_sse *vaos = CALLOC_STRUCT(draw_vs_varient_aos_sse); + struct draw_vs_variant_aos_sse *vaos = CALLOC_STRUCT(draw_vs_variant_aos_sse); if (!vaos) goto fail; @@ -2249,17 +2249,17 @@ static struct draw_vs_varient *varient_aos_sse( struct draw_vertex_shader *vs, } -struct draw_vs_varient * -draw_vs_create_varient_aos_sse( struct draw_vertex_shader *vs, - const struct draw_vs_varient_key *key ) +struct draw_vs_variant * +draw_vs_create_variant_aos_sse( struct draw_vertex_shader *vs, + const struct draw_vs_variant_key *key ) { - struct draw_vs_varient *varient = varient_aos_sse( vs, key ); + struct draw_vs_variant *variant = variant_aos_sse( vs, key ); - if (varient == NULL) { - varient = draw_vs_create_varient_generic( vs, key ); + if (variant == NULL) { + variant = draw_vs_create_variant_generic( vs, key ); } - return varient; + return variant; } diff --git a/src/gallium/auxiliary/draw/draw_vs_aos.h b/src/gallium/auxiliary/draw/draw_vs_aos.h index 68e8295b5e1..55e63d8b9fa 100644 --- a/src/gallium/auxiliary/draw/draw_vs_aos.h +++ b/src/gallium/auxiliary/draw/draw_vs_aos.h @@ -98,9 +98,9 @@ struct aos_buffer { -/* This is the temporary storage used by all the aos_sse vs varients. +/* This is the temporary storage used by all the aos_sse vs variants. * Create one per context and reuse by passing a pointer in at - * vs_varient creation?? + * vs_variant creation?? */ struct aos_machine { float input [MAX_INPUTS ][4]; @@ -134,7 +134,7 @@ struct aos_machine { struct aos_compilation { struct x86_function *func; - struct draw_vs_varient_aos_sse *vaos; + struct draw_vs_variant_aos_sse *vaos; unsigned insn_counter; unsigned num_immediates; @@ -234,8 +234,8 @@ typedef void (PIPE_CDECL *vaos_run_linear_func)( struct aos_machine *, void *output_buffer); -struct draw_vs_varient_aos_sse { - struct draw_vs_varient base; +struct draw_vs_variant_aos_sse { + struct draw_vs_variant base; struct draw_context *draw; struct aos_buffer *buffer; diff --git a/src/gallium/auxiliary/draw/draw_vs_exec.c b/src/gallium/auxiliary/draw/draw_vs_exec.c index dab3eb1ca8e..667eb507855 100644 --- a/src/gallium/auxiliary/draw/draw_vs_exec.c +++ b/src/gallium/auxiliary/draw/draw_vs_exec.c @@ -203,7 +203,7 @@ draw_create_vs_exec(struct draw_context *draw, vs->base.prepare = vs_exec_prepare; vs->base.run_linear = vs_exec_run_linear; vs->base.delete = vs_exec_delete; - vs->base.create_varient = draw_vs_create_varient_generic; + vs->base.create_variant = draw_vs_create_variant_generic; vs->machine = draw->vs.machine; return &vs->base; diff --git a/src/gallium/auxiliary/draw/draw_vs_llvm.c b/src/gallium/auxiliary/draw/draw_vs_llvm.c index de074be0926..54a5574f73f 100644 --- a/src/gallium/auxiliary/draw/draw_vs_llvm.c +++ b/src/gallium/auxiliary/draw/draw_vs_llvm.c @@ -107,7 +107,7 @@ draw_create_vs_llvm(struct draw_context *draw, vs->base.prepare = vs_llvm_prepare; vs->base.run_linear = vs_llvm_run_linear; vs->base.delete = vs_llvm_delete; - vs->base.create_varient = draw_vs_create_varient_generic; + vs->base.create_variant = draw_vs_create_variant_generic; make_empty_list(&vs->variants); diff --git a/src/gallium/auxiliary/draw/draw_vs_ppc.c b/src/gallium/auxiliary/draw/draw_vs_ppc.c index 5df84916c51..cf894bbe8af 100644 --- a/src/gallium/auxiliary/draw/draw_vs_ppc.c +++ b/src/gallium/auxiliary/draw/draw_vs_ppc.c @@ -187,10 +187,10 @@ draw_create_vs_ppc(struct draw_context *draw, vs->base.draw = draw; #if 0 if (1) - vs->base.create_varient = draw_vs_varient_aos_ppc; + vs->base.create_variant = draw_vs_variant_aos_ppc; else #endif - vs->base.create_varient = draw_vs_create_varient_generic; + vs->base.create_variant = draw_vs_create_variant_generic; vs->base.prepare = vs_ppc_prepare; vs->base.run_linear = vs_ppc_run_linear; vs->base.delete = vs_ppc_delete; diff --git a/src/gallium/auxiliary/draw/draw_vs_sse.c b/src/gallium/auxiliary/draw/draw_vs_sse.c index 0b0c6077c6f..dee7c0da9b6 100644 --- a/src/gallium/auxiliary/draw/draw_vs_sse.c +++ b/src/gallium/auxiliary/draw/draw_vs_sse.c @@ -166,9 +166,9 @@ draw_create_vs_sse(struct draw_context *draw, vs->base.draw = draw; if (1) - vs->base.create_varient = draw_vs_create_varient_aos_sse; + vs->base.create_variant = draw_vs_create_variant_aos_sse; else - vs->base.create_varient = draw_vs_create_varient_generic; + vs->base.create_variant = draw_vs_create_variant_generic; vs->base.prepare = vs_sse_prepare; vs->base.run_linear = vs_sse_run_linear; vs->base.delete = vs_sse_delete; diff --git a/src/gallium/auxiliary/draw/draw_vs_varient.c b/src/gallium/auxiliary/draw/draw_vs_varient.c index eacd1601877..d8f030f61eb 100644 --- a/src/gallium/auxiliary/draw/draw_vs_varient.c +++ b/src/gallium/auxiliary/draw/draw_vs_varient.c @@ -41,8 +41,8 @@ /* A first pass at incorporating vertex fetch/emit functionality into */ -struct draw_vs_varient_generic { - struct draw_vs_varient base; +struct draw_vs_variant_generic { + struct draw_vs_variant base; struct draw_vertex_shader *shader; struct draw_context *draw; @@ -63,13 +63,13 @@ struct draw_vs_varient_generic { -static void vsvg_set_buffer( struct draw_vs_varient *varient, +static void vsvg_set_buffer( struct draw_vs_variant *variant, unsigned buffer, const void *ptr, unsigned stride, unsigned max_index ) { - struct draw_vs_varient_generic *vsvg = (struct draw_vs_varient_generic *)varient; + struct draw_vs_variant_generic *vsvg = (struct draw_vs_variant_generic *)variant; vsvg->fetch->set_buffer(vsvg->fetch, buffer, @@ -81,7 +81,7 @@ static void vsvg_set_buffer( struct draw_vs_varient *varient, /* Mainly for debug at this stage: */ -static void do_rhw_viewport( struct draw_vs_varient_generic *vsvg, +static void do_rhw_viewport( struct draw_vs_variant_generic *vsvg, unsigned count, void *output_buffer ) { @@ -104,7 +104,7 @@ static void do_rhw_viewport( struct draw_vs_varient_generic *vsvg, } } -static void do_viewport( struct draw_vs_varient_generic *vsvg, +static void do_viewport( struct draw_vs_variant_generic *vsvg, unsigned count, void *output_buffer ) { @@ -126,12 +126,12 @@ static void do_viewport( struct draw_vs_varient_generic *vsvg, } -static void PIPE_CDECL vsvg_run_elts( struct draw_vs_varient *varient, +static void PIPE_CDECL vsvg_run_elts( struct draw_vs_variant *variant, const unsigned *elts, unsigned count, void *output_buffer) { - struct draw_vs_varient_generic *vsvg = (struct draw_vs_varient_generic *)varient; + struct draw_vs_variant_generic *vsvg = (struct draw_vs_variant_generic *)variant; unsigned temp_vertex_stride = vsvg->temp_vertex_stride; void *temp_buffer = MALLOC( align(count,4) * temp_vertex_stride ); @@ -193,12 +193,12 @@ static void PIPE_CDECL vsvg_run_elts( struct draw_vs_varient *varient, } -static void PIPE_CDECL vsvg_run_linear( struct draw_vs_varient *varient, +static void PIPE_CDECL vsvg_run_linear( struct draw_vs_variant *variant, unsigned start, unsigned count, void *output_buffer ) { - struct draw_vs_varient_generic *vsvg = (struct draw_vs_varient_generic *)varient; + struct draw_vs_variant_generic *vsvg = (struct draw_vs_variant_generic *)variant; unsigned temp_vertex_stride = vsvg->temp_vertex_stride; void *temp_buffer = MALLOC( align(count,4) * temp_vertex_stride ); @@ -259,20 +259,20 @@ static void PIPE_CDECL vsvg_run_linear( struct draw_vs_varient *varient, -static void vsvg_destroy( struct draw_vs_varient *varient ) +static void vsvg_destroy( struct draw_vs_variant *variant ) { - FREE(varient); + FREE(variant); } -struct draw_vs_varient * -draw_vs_create_varient_generic( struct draw_vertex_shader *vs, - const struct draw_vs_varient_key *key ) +struct draw_vs_variant * +draw_vs_create_variant_generic( struct draw_vertex_shader *vs, + const struct draw_vs_variant_key *key ) { unsigned i; struct translate_key fetch, emit; - struct draw_vs_varient_generic *vsvg = CALLOC_STRUCT( draw_vs_varient_generic ); + struct draw_vs_variant_generic *vsvg = CALLOC_STRUCT( draw_vs_variant_generic ); if (vsvg == NULL) return NULL; diff --git a/src/gallium/auxiliary/gallivm/lp_bld_const.h b/src/gallium/auxiliary/gallivm/lp_bld_const.h index c749a7a3150..680211fbbd7 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_const.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_const.h @@ -125,4 +125,22 @@ lp_build_const_float(struct gallivm_state *gallivm, float x) } +/** Return constant-valued pointer to int */ +static INLINE LLVMValueRef +lp_build_const_int_pointer(struct gallivm_state *gallivm, const void *ptr) +{ + LLVMTypeRef int_type; + LLVMValueRef v; + + /* int type large enough to hold a pointer */ + int_type = LLVMIntTypeInContext(gallivm->context, 8 * sizeof(void *)); + v = LLVMConstInt(int_type, (unsigned long long) ptr, 0); + v = LLVMBuildIntToPtr(gallivm->builder, v, + LLVMPointerType(int_type, 0), + "cast int to ptr"); + return v; +} + + + #endif /* !LP_BLD_CONST_H */ diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c b/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c index 75d2e666f09..82ab19eda14 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_format_aos.c @@ -36,6 +36,7 @@ #include "util/u_format.h" #include "util/u_memory.h" #include "util/u_math.h" +#include "util/u_pointer.h" #include "util/u_string.h" #include "lp_bld_arit.h" @@ -511,8 +512,6 @@ lp_build_fetch_rgba_aos(struct gallivm_state *gallivm, * or incentive to optimize. */ - LLVMModuleRef module = LLVMGetGlobalParent(LLVMGetBasicBlockParent(LLVMGetInsertBlock(gallivm->builder))); - char name[256]; LLVMTypeRef i8t = LLVMInt8TypeInContext(gallivm->context); LLVMTypeRef pi8t = LLVMPointerType(i8t, 0); LLVMTypeRef i32t = LLVMInt32TypeInContext(gallivm->context); @@ -522,19 +521,20 @@ lp_build_fetch_rgba_aos(struct gallivm_state *gallivm, LLVMValueRef res; unsigned k; - util_snprintf(name, sizeof name, "util_format_%s_fetch_rgba_8unorm", - format_desc->short_name); - if (gallivm_debug & GALLIVM_DEBUG_PERF) { - debug_printf("%s: falling back to %s\n", __FUNCTION__, name); + debug_printf("%s: falling back to util_format_%s_fetch_rgba_8unorm\n", + __FUNCTION__, format_desc->short_name); } /* * Declare and bind format_desc->fetch_rgba_8unorm(). */ - function = LLVMGetNamedFunction(module, name); - if (!function) { + { + /* + * Function to call looks like: + * fetch(uint8_t *dst, const uint8_t *src, unsigned i, unsigned j) + */ LLVMTypeRef ret_type; LLVMTypeRef arg_types[4]; LLVMTypeRef function_type; @@ -542,17 +542,19 @@ lp_build_fetch_rgba_aos(struct gallivm_state *gallivm, ret_type = LLVMVoidTypeInContext(gallivm->context); arg_types[0] = pi8t; arg_types[1] = pi8t; - arg_types[3] = arg_types[2] = LLVMIntTypeInContext(gallivm->context, sizeof(unsigned) * 8); - function_type = LLVMFunctionType(ret_type, arg_types, Elements(arg_types), 0); - function = LLVMAddFunction(module, name, function_type); - - LLVMSetFunctionCallConv(function, LLVMCCallConv); - LLVMSetLinkage(function, LLVMExternalLinkage); - - assert(LLVMIsDeclaration(function)); - - LLVMAddGlobalMapping(gallivm->engine, function, - func_to_pointer((func_pointer)format_desc->fetch_rgba_8unorm)); + arg_types[2] = i32t; + arg_types[3] = i32t; + function_type = LLVMFunctionType(ret_type, arg_types, + Elements(arg_types), 0); + + /* make const pointer for the C fetch_rgba_8unorm function */ + function = lp_build_const_int_pointer(gallivm, + func_to_pointer((func_pointer) format_desc->fetch_rgba_8unorm)); + + /* cast the callee pointer to the function's type */ + function = LLVMBuildBitCast(builder, function, + LLVMPointerType(function_type, 0), + "cast callee"); } tmp_ptr = lp_build_alloca(gallivm, i32t, ""); @@ -614,48 +616,55 @@ lp_build_fetch_rgba_aos(struct gallivm_state *gallivm, * or incentive to optimize. */ - LLVMModuleRef module = LLVMGetGlobalParent(LLVMGetBasicBlockParent(LLVMGetInsertBlock(builder))); - char name[256]; LLVMTypeRef f32t = LLVMFloatTypeInContext(gallivm->context); LLVMTypeRef f32x4t = LLVMVectorType(f32t, 4); LLVMTypeRef pf32t = LLVMPointerType(f32t, 0); + LLVMTypeRef pi8t = LLVMPointerType(LLVMInt8TypeInContext(gallivm->context), 0); + LLVMTypeRef i32t = LLVMInt32TypeInContext(gallivm->context); LLVMValueRef function; LLVMValueRef tmp_ptr; LLVMValueRef tmps[LP_MAX_VECTOR_LENGTH/4]; LLVMValueRef res; unsigned k; - util_snprintf(name, sizeof name, "util_format_%s_fetch_rgba_float", - format_desc->short_name); - if (gallivm_debug & GALLIVM_DEBUG_PERF) { - debug_printf("%s: falling back to %s\n", __FUNCTION__, name); + debug_printf("%s: falling back to util_format_%s_fetch_rgba_float\n", + __FUNCTION__, format_desc->short_name); } /* * Declare and bind format_desc->fetch_rgba_float(). */ - function = LLVMGetNamedFunction(module, name); - if (!function) { + { + /* + * Function to call looks like: + * fetch(float *dst, const uint8_t *src, unsigned i, unsigned j) + */ LLVMTypeRef ret_type; LLVMTypeRef arg_types[4]; LLVMTypeRef function_type; ret_type = LLVMVoidTypeInContext(gallivm->context); arg_types[0] = pf32t; - arg_types[1] = LLVMPointerType(LLVMInt8TypeInContext(gallivm->context), 0); - arg_types[3] = arg_types[2] = LLVMIntTypeInContext(gallivm->context, sizeof(unsigned) * 8); - function_type = LLVMFunctionType(ret_type, arg_types, Elements(arg_types), 0); - function = LLVMAddFunction(module, name, function_type); + arg_types[1] = pi8t; + arg_types[2] = i32t; + arg_types[3] = i32t; + function_type = LLVMFunctionType(ret_type, arg_types, + Elements(arg_types), 0); - LLVMSetFunctionCallConv(function, LLVMCCallConv); - LLVMSetLinkage(function, LLVMExternalLinkage); + /* Note: we're using this casting here instead of LLVMAddGlobalMapping() + * to work around a bug in LLVM 2.6, and for efficiency/simplicity. + */ - assert(LLVMIsDeclaration(function)); + /* make const pointer for the C fetch_rgba_float function */ + function = lp_build_const_int_pointer(gallivm, + func_to_pointer((func_pointer) format_desc->fetch_rgba_float)); - LLVMAddGlobalMapping(gallivm->engine, function, - func_to_pointer((func_pointer)format_desc->fetch_rgba_float)); + /* cast the callee pointer to the function's type */ + function = LLVMBuildBitCast(builder, function, + LLVMPointerType(function_type, 0), + "cast callee"); } tmp_ptr = lp_build_alloca(gallivm, f32x4t, ""); diff --git a/src/gallium/auxiliary/gallivm/lp_bld_init.c b/src/gallium/auxiliary/gallivm/lp_bld_init.c index efe8d38b8f0..7504cb5cf2f 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_init.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_init.c @@ -30,6 +30,7 @@ #include "util/u_cpu_detect.h" #include "util/u_debug.h" #include "util/u_memory.h" +#include "util/u_simple_list.h" #include "lp_bld_debug.h" #include "lp_bld_init.h" @@ -291,12 +292,12 @@ struct callback { garbage_collect_callback_func func; void *cb_data; + struct callback *prev, *next; }; -#define MAX_CALLBACKS 32 -static struct callback Callbacks[MAX_CALLBACKS]; -static unsigned NumCallbacks = 0; +/** list of all garbage collector callbacks */ +static struct callback callback_list = {NULL, NULL, NULL, NULL}; /** @@ -307,20 +308,24 @@ void gallivm_register_garbage_collector_callback(garbage_collect_callback_func func, void *cb_data) { - unsigned i; + struct callback *cb; - for (i = 0; i < NumCallbacks; i++) { - if (Callbacks[i].func == func && Callbacks[i].cb_data == cb_data) { - /* already in list: no-op */ + if (!callback_list.prev) { + make_empty_list(&callback_list); + } + + /* see if already in list */ + foreach(cb, &callback_list) { + if (cb->func == func && cb->cb_data == cb_data) return; - } } - assert(NumCallbacks < MAX_CALLBACKS); - if (NumCallbacks < MAX_CALLBACKS) { - Callbacks[NumCallbacks].func = func; - Callbacks[NumCallbacks].cb_data = cb_data; - NumCallbacks++; + /* add to list */ + cb = CALLOC_STRUCT(callback); + if (cb) { + cb->func = func; + cb->cb_data = cb_data; + insert_at_head(&callback_list, cb); } } @@ -332,15 +337,13 @@ void gallivm_remove_garbage_collector_callback(garbage_collect_callback_func func, void *cb_data) { - unsigned i; - - for (i = 0; i < NumCallbacks; i++) { - if (Callbacks[i].func == func && Callbacks[i].cb_data == cb_data) { - /* found, now remove it */ - NumCallbacks--; - for ( ; i < NumCallbacks; i++) { - Callbacks[i] = Callbacks[i + 1]; - } + struct callback *cb; + + /* search list */ + foreach(cb, &callback_list) { + if (cb->func == func && cb->cb_data == cb_data) { + /* found, remove it */ + remove_from_list(cb); return; } } @@ -354,10 +357,9 @@ gallivm_remove_garbage_collector_callback(garbage_collect_callback_func func, static void call_garbage_collector_callbacks(void) { - unsigned i; - - for (i = 0; i < NumCallbacks; i++) { - Callbacks[i].func(Callbacks[i].cb_data); + struct callback *cb; + foreach(cb, &callback_list) { + cb->func(cb->cb_data); } } diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c b/src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c index 991f6fa5ef7..e61cf9541ea 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c @@ -1098,6 +1098,4 @@ lp_build_sample_aos(struct lp_build_sample_context *bld, texel_out[2] = unswizzled[2]; texel_out[3] = unswizzled[3]; } - - apply_sampler_swizzle(bld, texel_out); } diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c index cf46e2be832..e685f4b73f0 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c @@ -187,8 +187,6 @@ lp_build_sample_texel_soa(struct lp_build_sample_context *bld, border_chan_vec, texel_out[chan]); } } - - apply_sampler_swizzle(bld, texel_out); } @@ -1268,4 +1266,6 @@ lp_build_sample_soa(struct gallivm_state *gallivm, } lp_build_sample_compare(&bld, r, texel_out); + + apply_sampler_swizzle(&bld, texel_out); } diff --git a/src/gallium/auxiliary/tgsi/tgsi_parse.h b/src/gallium/auxiliary/tgsi/tgsi_parse.h index d4df5851764..2aafa2a6e83 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_parse.h +++ b/src/gallium/auxiliary/tgsi/tgsi_parse.h @@ -136,7 +136,8 @@ tgsi_parse_token( static INLINE unsigned tgsi_num_tokens(const struct tgsi_token *tokens) { - struct tgsi_header header = *(const struct tgsi_header *) tokens; + struct tgsi_header header; + memcpy(&header, tokens, sizeof(header)); return header.HeaderSize + header.BodySize; } diff --git a/src/gallium/auxiliary/util/u_blitter.c b/src/gallium/auxiliary/util/u_blitter.c index 4c986e3565c..545021d2642 100644 --- a/src/gallium/auxiliary/util/u_blitter.c +++ b/src/gallium/auxiliary/util/u_blitter.c @@ -738,8 +738,8 @@ void util_blitter_copy_region(struct blitter_context *blitter, assert(!is_overlap(srcbox->x, srcbox->x + width, srcbox->y, srcbox->y + height, dstx, dstx + width, dsty, dsty + height)); } else { - assert(util_is_format_compatible(util_format_description(dst->format), - util_format_description(src->format))); + assert(util_is_format_compatible(util_format_description(src->format), + util_format_description(dst->format))); } assert(src->target < PIPE_MAX_TEXTURE_TYPES); /* XXX should handle 3d regions */ diff --git a/src/gallium/auxiliary/util/u_index_modify.c b/src/gallium/auxiliary/util/u_index_modify.c index 65b079ed537..3822f60e71d 100644 --- a/src/gallium/auxiliary/util/u_index_modify.c +++ b/src/gallium/auxiliary/util/u_index_modify.c @@ -52,8 +52,8 @@ void util_shorten_ubyte_elts(struct pipe_context *context, out_map++; } - pipe_buffer_unmap(context, *elts, src_transfer); - pipe_buffer_unmap(context, new_elts, dst_transfer); + pipe_buffer_unmap(context, src_transfer); + pipe_buffer_unmap(context, dst_transfer); *elts = new_elts; } @@ -86,8 +86,8 @@ void util_rebuild_ushort_elts(struct pipe_context *context, out_map++; } - pipe_buffer_unmap(context, *elts, in_transfer); - pipe_buffer_unmap(context, new_elts, out_transfer); + pipe_buffer_unmap(context, in_transfer); + pipe_buffer_unmap(context, out_transfer); *elts = new_elts; } @@ -120,8 +120,8 @@ void util_rebuild_uint_elts(struct pipe_context *context, out_map++; } - pipe_buffer_unmap(context, *elts, in_transfer); - pipe_buffer_unmap(context, new_elts, out_transfer); + pipe_buffer_unmap(context, in_transfer); + pipe_buffer_unmap(context, out_transfer); *elts = new_elts; } diff --git a/src/gallium/auxiliary/util/u_inlines.h b/src/gallium/auxiliary/util/u_inlines.h index e55aafe90f0..9184b6aa4db 100644 --- a/src/gallium/auxiliary/util/u_inlines.h +++ b/src/gallium/auxiliary/util/u_inlines.h @@ -242,7 +242,6 @@ pipe_buffer_map(struct pipe_context *pipe, static INLINE void pipe_buffer_unmap(struct pipe_context *pipe, - struct pipe_resource *buf, struct pipe_transfer *transfer) { if (transfer) { @@ -341,7 +340,7 @@ pipe_buffer_read(struct pipe_context *pipe, if (map) memcpy(data, map + offset, size); - pipe_buffer_unmap(pipe, buf, src_transfer); + pipe_buffer_unmap(pipe, src_transfer); } static INLINE struct pipe_transfer * diff --git a/src/gallium/auxiliary/util/u_upload_mgr.c b/src/gallium/auxiliary/util/u_upload_mgr.c index af229e61a00..4daa55d6638 100644 --- a/src/gallium/auxiliary/util/u_upload_mgr.c +++ b/src/gallium/auxiliary/util/u_upload_mgr.c @@ -108,7 +108,7 @@ my_buffer_write(struct pipe_context *pipe, memcpy(map + offset, data, size); pipe_buffer_flush_mapped_range(pipe, transfer, offset, dirty_size); - pipe_buffer_unmap(pipe, buf, transfer); + pipe_buffer_unmap(pipe, transfer); return PIPE_OK; } @@ -243,7 +243,7 @@ enum pipe_error u_upload_buffer( struct u_upload_mgr *upload, done: if (map) - pipe_buffer_unmap( upload->pipe, inbuf, transfer ); + pipe_buffer_unmap( upload->pipe, transfer ); return ret; } diff --git a/src/gallium/docs/source/tgsi.rst b/src/gallium/docs/source/tgsi.rst index d99ed7c6d6b..7eb6bd05d7b 100644 --- a/src/gallium/docs/source/tgsi.rst +++ b/src/gallium/docs/source/tgsi.rst @@ -577,17 +577,45 @@ This instruction replicates its result. .. opcode:: TEX - Texture Lookup - TBD +.. math:: + + coord = src0 + + bias = 0.0 + + dst = texture_sample(unit, coord, bias) .. opcode:: TXD - Texture Lookup with Derivatives - TBD +.. math:: + + coord = src0 + + ddx = src1 + + ddy = src2 + + bias = 0.0 + + dst = texture_sample_deriv(unit, coord, bias, ddx, ddy) .. opcode:: TXP - Projective Texture Lookup - TBD +.. math:: + + coord.x = src0.x / src.w + + coord.y = src0.y / src.w + + coord.z = src0.z / src.w + + coord.w = src0.w + + bias = 0.0 + + dst = texture_sample(unit, coord, bias) .. opcode:: UP2H - Unpack Two 16-Bit Floats @@ -729,7 +757,19 @@ This instruction replicates its result. .. opcode:: TXB - Texture Lookup With Bias - TBD +.. math:: + + coord.x = src.x + + coord.y = src.y + + coord.z = src.z + + coord.w = 1.0 + + bias = src.z + + dst = texture_sample(unit, coord, bias) .. opcode:: NRM - 3-component Vector Normalise @@ -767,9 +807,21 @@ This instruction replicates its result. dst = src0.x \times src1.x + src0.y \times src1.y -.. opcode:: TXL - Texture Lookup With LOD +.. opcode:: TXL - Texture Lookup With explicit LOD - TBD +.. math:: + + coord.x = src0.x + + coord.y = src0.y + + coord.z = src0.z + + coord.w = 1.0 + + lod = src0.w + + dst = texture_sample(unit, coord, lod) .. opcode:: BRK - Break diff --git a/src/gallium/drivers/i915/i915_context.h b/src/gallium/drivers/i915/i915_context.h index 7103a1b8c16..d15e1723d83 100644 --- a/src/gallium/drivers/i915/i915_context.h +++ b/src/gallium/drivers/i915/i915_context.h @@ -272,7 +272,7 @@ struct i915_context { #define I915_HW_PROGRAM (1<<I915_CACHE_PROGRAM) #define I915_HW_CONSTANTS (1<<I915_CACHE_CONSTANTS) #define I915_HW_IMMEDIATE (1<<(I915_MAX_CACHE+0)) -#define I915_HW_INVARIENT (1<<(I915_MAX_CACHE+1)) +#define I915_HW_INVARIANT (1<<(I915_MAX_CACHE+1)) /*********************************************************************** diff --git a/src/gallium/drivers/i915/i915_debug.c b/src/gallium/drivers/i915/i915_debug.c index d7150c99c4e..87c435a2f36 100644 --- a/src/gallium/drivers/i915/i915_debug.c +++ b/src/gallium/drivers/i915/i915_debug.c @@ -976,7 +976,7 @@ i915_dump_hardware_dirty(struct i915_context *i915, const char *func) {I915_HW_PROGRAM, "program"}, {I915_HW_CONSTANTS, "constants"}, {I915_HW_IMMEDIATE, "immediate"}, - {I915_HW_INVARIENT, "invarient"}, + {I915_HW_INVARIANT, "invariant"}, {0, NULL}, }; int i; diff --git a/src/gallium/drivers/i915/i915_screen.c b/src/gallium/drivers/i915/i915_screen.c index f66478e729c..bdbc08e8086 100644 --- a/src/gallium/drivers/i915/i915_screen.c +++ b/src/gallium/drivers/i915/i915_screen.c @@ -35,7 +35,6 @@ #include "i915_debug.h" #include "i915_context.h" #include "i915_screen.h" -#include "i915_surface.h" #include "i915_resource.h" #include "i915_winsys.h" #include "i915_public.h" diff --git a/src/gallium/drivers/i915/i915_state_emit.c b/src/gallium/drivers/i915/i915_state_emit.c index c48d53ffbb2..86c02976495 100644 --- a/src/gallium/drivers/i915/i915_state_emit.c +++ b/src/gallium/drivers/i915/i915_state_emit.c @@ -142,7 +142,7 @@ i915_emit_hardware_state(struct i915_context *i915 ) save_relocs = i915->batch->relocs; /* 14 dwords, 0 relocs */ - if (i915->hardware_dirty & I915_HW_INVARIENT) + if (i915->hardware_dirty & I915_HW_INVARIANT) { OUT_BATCH(_3DSTATE_AA_CMD | AA_LINE_ECAAR_WIDTH_ENABLE | diff --git a/src/gallium/drivers/i965/brw_misc_state.c b/src/gallium/drivers/i965/brw_misc_state.c index 6d89b5d2baf..c635d696617 100644 --- a/src/gallium/drivers/i965/brw_misc_state.c +++ b/src/gallium/drivers/i965/brw_misc_state.c @@ -363,10 +363,10 @@ const struct brw_tracked_state brw_line_stipple = { /*********************************************************************** - * Misc invarient state packets + * Misc invariant state packets */ -static int upload_invarient_state( struct brw_context *brw ) +static int upload_invariant_state( struct brw_context *brw ) { { /* 0x61040000 Pipeline Select */ @@ -439,7 +439,7 @@ static int upload_invarient_state( struct brw_context *brw ) { struct brw_polygon_stipple_offset bpso; - /* This is invarient state in gallium: + /* This is invariant state in gallium: */ memset(&bpso, 0, sizeof(bpso)); bpso.header.opcode = CMD_POLY_STIPPLE_OFFSET; @@ -453,13 +453,13 @@ static int upload_invarient_state( struct brw_context *brw ) return 0; } -const struct brw_tracked_state brw_invarient_state = { +const struct brw_tracked_state brw_invariant_state = { .dirty = { .mesa = 0, .brw = BRW_NEW_CONTEXT, .cache = 0 }, - .emit = upload_invarient_state + .emit = upload_invariant_state }; diff --git a/src/gallium/drivers/i965/brw_state.h b/src/gallium/drivers/i965/brw_state.h index d2bbd0123d1..380d511f9bb 100644 --- a/src/gallium/drivers/i965/brw_state.h +++ b/src/gallium/drivers/i965/brw_state.h @@ -56,7 +56,7 @@ const struct brw_tracked_state brw_clip_prog; const struct brw_tracked_state brw_clip_unit; const struct brw_tracked_state brw_curbe_buffer; const struct brw_tracked_state brw_curbe_offsets; -const struct brw_tracked_state brw_invarient_state; +const struct brw_tracked_state brw_invariant_state; const struct brw_tracked_state brw_gs_prog; const struct brw_tracked_state brw_gs_unit; const struct brw_tracked_state brw_line_stipple; diff --git a/src/gallium/drivers/i965/brw_state_upload.c b/src/gallium/drivers/i965/brw_state_upload.c index f8b91eff816..cdbf270e06a 100644 --- a/src/gallium/drivers/i965/brw_state_upload.c +++ b/src/gallium/drivers/i965/brw_state_upload.c @@ -69,7 +69,7 @@ const struct brw_tracked_state *atoms[] = /* Command packets: */ - &brw_invarient_state, + &brw_invariant_state, &brw_state_base_address, &brw_binding_table_pointers, diff --git a/src/gallium/drivers/i965/brw_wm_fp.c b/src/gallium/drivers/i965/brw_wm_fp.c index f7ee55cc1c8..a65e16edec0 100644 --- a/src/gallium/drivers/i965/brw_wm_fp.c +++ b/src/gallium/drivers/i965/brw_wm_fp.c @@ -812,7 +812,7 @@ static void precalc_tex( struct brw_wm_compile *c, } /* XXX: add GL_EXT_texture_swizzle support to gallium -- by - * generating shader varients in mesa state tracker. + * generating shader variants in mesa state tracker. */ /* Release this temp if we ended up allocating it: diff --git a/src/gallium/drivers/llvmpipe/lp_jit.c b/src/gallium/drivers/llvmpipe/lp_jit.c index a775990f92a..482a902dd23 100644 --- a/src/gallium/drivers/llvmpipe/lp_jit.c +++ b/src/gallium/drivers/llvmpipe/lp_jit.c @@ -36,9 +36,7 @@ #include "util/u_memory.h" #include "gallivm/lp_bld_init.h" #include "gallivm/lp_bld_debug.h" -#include "gallivm/lp_bld_intr.h" #include "lp_context.h" -#include "lp_screen.h" #include "lp_jit.h" diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c index 2c4943a69f6..ae207617cc1 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_fs.c +++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c @@ -90,7 +90,6 @@ #include "lp_context.h" #include "lp_debug.h" #include "lp_perf.h" -#include "lp_screen.h" #include "lp_setup.h" #include "lp_state.h" #include "lp_tex_sample.h" diff --git a/src/gallium/drivers/nouveau/nouveau_winsys.h b/src/gallium/drivers/nouveau/nouveau_winsys.h index ab480cabd09..747b084bc45 100644 --- a/src/gallium/drivers/nouveau/nouveau_winsys.h +++ b/src/gallium/drivers/nouveau/nouveau_winsys.h @@ -10,7 +10,7 @@ #include "nouveau/nouveau_grobj.h" #include "nouveau/nouveau_notifier.h" #include "nouveau/nouveau_resource.h" -#include "nouveau/nouveau_pushbuf.h" +#include "nouveau/nv04_pushbuf.h" #ifndef NV04_PFIFO_MAX_PACKET_LEN #define NV04_PFIFO_MAX_PACKET_LEN 2047 diff --git a/src/gallium/drivers/nv50/nv50_shader_state.c b/src/gallium/drivers/nv50/nv50_shader_state.c index 306aa81d985..1c1b66deb3c 100644 --- a/src/gallium/drivers/nv50/nv50_shader_state.c +++ b/src/gallium/drivers/nv50/nv50_shader_state.c @@ -71,7 +71,7 @@ nv50_transfer_constbuf(struct nv50_context *nv50, map += nr; } - pipe_buffer_unmap(pipe, buf, transfer); + pipe_buffer_unmap(pipe, transfer); } static void diff --git a/src/gallium/drivers/nv50/nv50_surface.c b/src/gallium/drivers/nv50/nv50_surface.c index ce48022db4e..a99df76cee3 100644 --- a/src/gallium/drivers/nv50/nv50_surface.c +++ b/src/gallium/drivers/nv50/nv50_surface.c @@ -22,7 +22,7 @@ #define __NOUVEAU_PUSH_H__ #include <stdint.h> -#include "nouveau/nouveau_pushbuf.h" +#include "nouveau/nv04_pushbuf.h" #include "nv50_context.h" #include "nv50_resource.h" #include "pipe/p_defines.h" diff --git a/src/gallium/drivers/nv50/nv50_vbo.c b/src/gallium/drivers/nv50/nv50_vbo.c index d41a59d05db..53f319acf46 100644 --- a/src/gallium/drivers/nv50/nv50_vbo.c +++ b/src/gallium/drivers/nv50/nv50_vbo.c @@ -284,7 +284,7 @@ nv50_draw_elements_inline(struct pipe_context *pipe, nzi = TRUE; } - pipe_buffer_unmap(pipe, indexBuffer, transfer); + pipe_buffer_unmap(pipe, transfer); } static void diff --git a/src/gallium/drivers/nvfx/nv04_2d.c b/src/gallium/drivers/nvfx/nv04_2d.c index e0e65e7a87f..e2fadd33e1c 100644 --- a/src/gallium/drivers/nvfx/nv04_2d.c +++ b/src/gallium/drivers/nvfx/nv04_2d.c @@ -34,11 +34,11 @@ #include <stdio.h> #include <stdint.h> #include <nouveau/nouveau_device.h> -#include <nouveau/nouveau_pushbuf.h> #include <nouveau/nouveau_channel.h> #include <nouveau/nouveau_bo.h> #include <nouveau/nouveau_notifier.h> #include <nouveau/nouveau_grobj.h> +#include <nouveau/nv04_pushbuf.h> #include "nv04_2d.h" #include "nouveau/nv_object.xml.h" diff --git a/src/gallium/drivers/nvfx/nvfx_fragprog.c b/src/gallium/drivers/nvfx/nvfx_fragprog.c index 13e8beed479..1740d72a8ae 100644 --- a/src/gallium/drivers/nvfx/nvfx_fragprog.c +++ b/src/gallium/drivers/nvfx/nvfx_fragprog.c @@ -1189,12 +1189,12 @@ out_err: static inline void nvfx_fp_memcpy(void* dst, const void* src, size_t len) { -#ifndef WORDS_BIGENDIAN +#ifndef PIPE_ARCH_BIG_ENDIAN memcpy(dst, src, len); #else size_t i; for(i = 0; i < len; i += 4) { - uint32_t v = (uint32_t*)((char*)src + i); + uint32_t v = *(uint32_t*)((char*)src + i); *(uint32_t*)((char*)dst + i) = (v >> 16) | (v << 16); } #endif diff --git a/src/gallium/drivers/nvfx/nvfx_vbo.c b/src/gallium/drivers/nvfx/nvfx_vbo.c index 597664e7716..339b31786d6 100644 --- a/src/gallium/drivers/nvfx/nvfx_vbo.c +++ b/src/gallium/drivers/nvfx/nvfx_vbo.c @@ -9,8 +9,7 @@ #include "nvfx_resource.h" #include "nouveau/nouveau_channel.h" - -#include "nouveau/nouveau_pushbuf.h" +#include "nouveau/nv04_pushbuf.h" static inline unsigned util_guess_unique_indices_count(unsigned mode, unsigned indices) diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c index 67b011a145c..bf1b8c33c00 100644 --- a/src/gallium/drivers/r300/r300_context.c +++ b/src/gallium/drivers/r300/r300_context.c @@ -35,8 +35,6 @@ #include "r300_screen_buffer.h" #include "r300_winsys.h" -#include <inttypes.h> - static void r300_update_num_contexts(struct r300_screen *r300screen, int diff) { diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c index 04a5bd92d12..9e0df30e527 100644 --- a/src/gallium/drivers/r300/r300_emit.c +++ b/src/gallium/drivers/r300/r300_emit.c @@ -26,7 +26,6 @@ #include "util/u_format.h" #include "util/u_math.h" #include "util/u_mm.h" -#include "util/u_simple_list.h" #include "r300_context.h" #include "r300_cb.h" diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c index b4197e03520..f58d511e11b 100644 --- a/src/gallium/drivers/r300/r300_render.c +++ b/src/gallium/drivers/r300/r300_render.c @@ -425,7 +425,7 @@ static void r300_emit_draw_arrays_immediate(struct r300_context *r300, if (transfer[vbi]) { vbuf = &r300->vertex_buffer[vbi]; - pipe_buffer_unmap(&r300->context, vbuf->buffer, transfer[vbi]); + pipe_buffer_unmap(&r300->context, transfer[vbi]); transfer[vbi] = NULL; } } @@ -551,7 +551,27 @@ static void r300_draw_range_elements(struct pipe_context* pipe, &start, count); r300_update_derived_state(r300); - r300_upload_index_buffer(r300, &indexBuffer, indexSize, start, count, &new_offset); + + /* Fallback for misaligned ushort indices. */ + if (indexSize == 2 && start % 2 == 1) { + struct pipe_transfer *transfer; + struct pipe_resource *userbuf; + uint16_t *ptr = pipe_buffer_map(pipe, indexBuffer, + PIPE_TRANSFER_READ, &transfer); + + /* Copy the mapped index buffer directly to the upload buffer. + * The start index will be aligned simply from the fact that + * every sub-buffer in u_upload_mgr is aligned. */ + userbuf = pipe->screen->user_buffer_create(pipe->screen, + ptr + start, count * 2, + PIPE_BIND_INDEX_BUFFER); + indexBuffer = userbuf; + r300_upload_index_buffer(r300, &indexBuffer, indexSize, 0, count, &new_offset); + pipe_resource_reference(&userbuf, NULL); + pipe_buffer_unmap(pipe, transfer); + } else { + r300_upload_index_buffer(r300, &indexBuffer, indexSize, start, count, &new_offset); + } start = new_offset; @@ -750,14 +770,13 @@ static void r300_swtcl_draw_vbo(struct pipe_context* pipe, for (i = 0; i < r300->vertex_buffer_count; i++) { if (r300->vertex_buffer[i].buffer) { - pipe_buffer_unmap(pipe, r300->vertex_buffer[i].buffer, - vb_transfer[i]); + pipe_buffer_unmap(pipe, vb_transfer[i]); draw_set_mapped_vertex_buffer(r300->draw, i, NULL); } } if (indexed) { - pipe_buffer_unmap(pipe, r300->index_buffer.buffer, ib_transfer); + pipe_buffer_unmap(pipe, ib_transfer); draw_set_mapped_index_buffer(r300->draw, NULL); } } @@ -857,7 +876,7 @@ static void r300_render_unmap_vertices(struct vbuf_render* render, r300render->vbo_max_used = MAX2(r300render->vbo_max_used, r300render->vertex_size * (max + 1)); - pipe_buffer_unmap(context, r300->vbo, r300render->vbo_transfer); + pipe_buffer_unmap(context, r300render->vbo_transfer); r300render->vbo_transfer = NULL; } diff --git a/src/gallium/drivers/r300/r300_render_translate.c b/src/gallium/drivers/r300/r300_render_translate.c index 41a43b04de7..26e00a2cad9 100644 --- a/src/gallium/drivers/r300/r300_render_translate.c +++ b/src/gallium/drivers/r300/r300_render_translate.c @@ -128,12 +128,11 @@ void r300_begin_vertex_translate(struct r300_context *r300) /* Unmap all buffers. */ for (i = 0; i < r300->vertex_buffer_count; i++) { if (vb_translated[i]) { - pipe_buffer_unmap(pipe, r300->vertex_buffer[i].buffer, - vb_transfer[i]); + pipe_buffer_unmap(pipe, vb_transfer[i]); } } - pipe_buffer_unmap(pipe, out_buffer, out_transfer); + pipe_buffer_unmap(pipe, out_transfer); /* Setup the new vertex buffer in the first free slot. */ for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { @@ -204,7 +203,7 @@ void r300_translate_index_buffer(struct r300_context *r300, break; case 2: - if (*start % 2 != 0 || index_offset) { + if (index_offset) { util_rebuild_ushort_elts(&r300->context, index_buffer, index_offset, *start, count); *start = 0; r300->validate_buffers = TRUE; diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c index 921d6f1e676..36060ab4d08 100644 --- a/src/gallium/drivers/r300/r300_screen.c +++ b/src/gallium/drivers/r300/r300_screen.c @@ -309,7 +309,9 @@ static boolean r300_is_format_supported(struct pipe_screen* screen, unsigned usage, unsigned geom_flags) { + struct r300_winsys_screen *rws = r300_screen(screen)->rws; uint32_t retval = 0; + boolean drm_2_8_0 = rws->get_value(rws, R300_VID_DRM_2_8_0); boolean is_r500 = r300_screen(screen)->caps.is_r500; boolean is_r400 = r300_screen(screen)->caps.is_r400; boolean is_color2101010 = format == PIPE_FORMAT_R10G10B10A2_UNORM || @@ -363,7 +365,7 @@ static boolean r300_is_format_supported(struct pipe_screen* screen, PIPE_BIND_SCANOUT | PIPE_BIND_SHARED)) && /* 2101010 cannot be rendered to on non-r5xx. */ - (is_r500 || !is_color2101010) && + (!is_color2101010 || (is_r500 && drm_2_8_0)) && r300_is_colorbuffer_format_supported(format)) { retval |= usage & (PIPE_BIND_RENDER_TARGET | diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c index 75292532405..f902db54cc1 100644 --- a/src/gallium/drivers/r300/r300_state.c +++ b/src/gallium/drivers/r300/r300_state.c @@ -1298,29 +1298,27 @@ static void r300_set_fragment_sampler_views(struct pipe_context* pipe, } for (i = 0; i < count; i++) { - if (&state->sampler_views[i]->base != views[i]) { - pipe_sampler_view_reference( - (struct pipe_sampler_view**)&state->sampler_views[i], - views[i]); + pipe_sampler_view_reference( + (struct pipe_sampler_view**)&state->sampler_views[i], + views[i]); - if (!views[i]) { - continue; - } + if (!views[i]) { + continue; + } - /* A new sampler view (= texture)... */ - dirty_tex = TRUE; + /* A new sampler view (= texture)... */ + dirty_tex = TRUE; - /* Set the texrect factor in the fragment shader. + /* Set the texrect factor in the fragment shader. * Needed for RECT and NPOT fallback. */ - texture = r300_texture(views[i]->texture); - if (texture->desc.is_npot) { - r300_mark_atom_dirty(r300, &r300->fs_rc_constant_state); - } + texture = r300_texture(views[i]->texture); + if (texture->desc.is_npot) { + r300_mark_atom_dirty(r300, &r300->fs_rc_constant_state); + } - state->sampler_views[i]->texcache_region = + state->sampler_views[i]->texcache_region = r300_assign_texture_cache_region(view_index, real_num_views); - view_index++; - } + view_index++; } for (i = count; i < tex_units; i++) { @@ -1496,14 +1494,14 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe, any_user_buffer = TRUE; } + /* The stride of zero means we will be fetching only the first + * vertex, so don't care about max_index. */ + if (!vbo->stride) + continue; + if (vbo->max_index == ~0) { - /* if no VBO stride then only one vertex value so max index is 1 */ - /* should think about converting to VS constants like svga does */ - if (!vbo->stride) - vbo->max_index = 1; - else - vbo->max_index = - (vbo->buffer->width0 - vbo->buffer_offset) / vbo->stride; + vbo->max_index = + (vbo->buffer->width0 - vbo->buffer_offset) / vbo->stride; } max_index = MIN2(vbo->max_index, max_index); diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c index 70fc5d96d83..24f1d68f4a7 100644 --- a/src/gallium/drivers/r300/r300_texture.c +++ b/src/gallium/drivers/r300/r300_texture.c @@ -481,6 +481,8 @@ static uint32_t r300_translate_out_fmt(enum pipe_format format) } else { if (desc->channel[i].size == 16) { modifier |= R300_US_OUT_FMT_C4_16; + } else if (desc->channel[i].size == 10) { + modifier |= R300_US_OUT_FMT_C4_10; } else { /* C4_8 seems to be used for the formats whose pixel size * is <= 32 bits. */ @@ -899,7 +901,7 @@ struct pipe_surface* r300_create_surface(struct pipe_context * ctx, tex->desc.b.b.nr_samples, tex->desc.microtile, tex->desc.macrotile[level], - DIM_HEIGHT); + DIM_HEIGHT, 0); surface->cbzb_height = align((surface->base.height + 1) / 2, tile_height); diff --git a/src/gallium/drivers/r300/r300_texture_desc.c b/src/gallium/drivers/r300/r300_texture_desc.c index aa82c47151a..7b1739142d4 100644 --- a/src/gallium/drivers/r300/r300_texture_desc.c +++ b/src/gallium/drivers/r300/r300_texture_desc.c @@ -34,7 +34,7 @@ unsigned r300_get_pixel_alignment(enum pipe_format format, unsigned num_samples, enum r300_buffer_tiling microtile, enum r300_buffer_tiling macrotile, - enum r300_dim dim) + enum r300_dim dim, boolean is_rs690) { static const unsigned table[2][5][3][2] = { @@ -57,6 +57,7 @@ unsigned r300_get_pixel_alignment(enum pipe_format format, {{ 16, 8}, { 0, 0}, { 0, 0}} /* 128 bits per pixel */ } }; + static const unsigned aa_block[2] = {4, 8}; unsigned tile = 0; unsigned pixsize = util_format_get_blocksize(format); @@ -74,6 +75,14 @@ unsigned r300_get_pixel_alignment(enum pipe_format format, } else { /* Standard alignment. */ tile = table[macrotile][util_logbase2(pixsize)][microtile][dim]; + if (macrotile == 0 && is_rs690 && dim == DIM_WIDTH) { + int align; + int h_tile; + h_tile = table[macrotile][util_logbase2(pixsize)][microtile][DIM_HEIGHT]; + align = 64 / (pixsize * h_tile); + if (tile < align) + tile = align; + } } assert(tile); @@ -89,7 +98,7 @@ static boolean r300_texture_macro_switch(struct r300_texture_desc *desc, unsigned tile, texdim; tile = r300_get_pixel_alignment(desc->b.b.format, desc->b.b.nr_samples, - desc->microtile, R300_BUFFER_TILED, dim); + desc->microtile, R300_BUFFER_TILED, dim, 0); if (dim == DIM_WIDTH) { texdim = u_minify(desc->width0, level); } else { @@ -113,6 +122,9 @@ static unsigned r300_texture_get_stride(struct r300_screen *screen, unsigned level) { unsigned tile_width, width, stride; + boolean is_rs690 = (screen->caps.family == CHIP_FAMILY_RS600 || + screen->caps.family == CHIP_FAMILY_RS690 || + screen->caps.family == CHIP_FAMILY_RS740); if (desc->stride_in_bytes_override) return desc->stride_in_bytes_override; @@ -131,38 +143,14 @@ static unsigned r300_texture_get_stride(struct r300_screen *screen, desc->b.b.nr_samples, desc->microtile, desc->macrotile[level], - DIM_WIDTH); + DIM_WIDTH, is_rs690); width = align(width, tile_width); stride = util_format_get_stride(desc->b.b.format, width); - - /* Some IGPs need a minimum stride of 64 bytes, hmm... */ - if (!desc->macrotile[level] && - (screen->caps.family == CHIP_FAMILY_RS600 || - screen->caps.family == CHIP_FAMILY_RS690 || - screen->caps.family == CHIP_FAMILY_RS740)) { - unsigned min_stride; - - if (desc->microtile) { - unsigned tile_height = - r300_get_pixel_alignment(desc->b.b.format, - desc->b.b.nr_samples, - desc->microtile, - desc->macrotile[level], - DIM_HEIGHT); - - min_stride = 64 / tile_height; - } else { - min_stride = 64; - } - - return stride < min_stride ? min_stride : stride; - } - /* The alignment to 32 bytes is sort of implied by the layout... */ return stride; } else { - return align(util_format_get_stride(desc->b.b.format, width), 32); + return align(util_format_get_stride(desc->b.b.format, width), is_rs690 ? 64 : 32); } } @@ -179,7 +167,7 @@ static unsigned r300_texture_get_nblocksy(struct r300_texture_desc *desc, desc->b.b.nr_samples, desc->microtile, desc->macrotile[level], - DIM_HEIGHT); + DIM_HEIGHT, 0); height = align(height, tile_height); /* This is needed for the kernel checker, unfortunately. */ diff --git a/src/gallium/drivers/r300/r300_texture_desc.h b/src/gallium/drivers/r300/r300_texture_desc.h index 44d88794a12..121d215b4cb 100644 --- a/src/gallium/drivers/r300/r300_texture_desc.h +++ b/src/gallium/drivers/r300/r300_texture_desc.h @@ -41,7 +41,7 @@ unsigned r300_get_pixel_alignment(enum pipe_format format, unsigned num_samples, enum r300_buffer_tiling microtile, enum r300_buffer_tiling macrotile, - enum r300_dim dim); + enum r300_dim dim, boolean is_rs690); boolean r300_texture_desc_init(struct r300_screen *rscreen, struct r300_texture_desc *desc, diff --git a/src/gallium/drivers/r300/r300_winsys.h b/src/gallium/drivers/r300/r300_winsys.h index 0dd330d101f..b8324afe511 100644 --- a/src/gallium/drivers/r300/r300_winsys.h +++ b/src/gallium/drivers/r300/r300_winsys.h @@ -51,8 +51,9 @@ enum r300_value_id { R300_VID_GB_PIPES, R300_VID_Z_PIPES, R300_VID_SQUARE_TILING_SUPPORT, - R300_VID_DRM_2_3_0, - R300_VID_DRM_2_6_0, + R300_VID_DRM_2_3_0, /* R500 VAP regs, MSPOS regs, fixed tex3D size checking */ + R300_VID_DRM_2_6_0, /* Hyper-Z, GB_Z_PEQ_CONFIG on rv350->r4xx, R500 FG_ALPHA_VALUE */ + R300_VID_DRM_2_8_0, /* R500 US_FORMAT regs, R500 ARGB2101010 colorbuffer */ R300_CAN_HYPERZ, }; diff --git a/src/gallium/drivers/r600/evergreen_state.c b/src/gallium/drivers/r600/evergreen_state.c index a9d4a862c32..c6f3669c9a3 100644 --- a/src/gallium/drivers/r600/evergreen_state.c +++ b/src/gallium/drivers/r600/evergreen_state.c @@ -770,8 +770,6 @@ static void evergreen_set_framebuffer_state(struct pipe_context *ctx, util_copy_framebuffer_state(&rctx->framebuffer, state); - rctx->pframebuffer = &rctx->framebuffer; - /* build states */ for (int i = 0; i < state->nr_cbufs; i++) { evergreen_cb(rctx, rstate, state, i); @@ -1295,11 +1293,6 @@ void evergreen_vertex_buffer_update(struct r600_pipe_context *rctx) if (rctx->vertex_elements == NULL || !rctx->nvertex_buffer) return; - /* delete previous translated vertex elements */ - if (rctx->tran.new_velems) { - r600_end_vertex_translate(rctx); - } - if (rctx->vertex_elements->incompatible_layout) { /* translate rebind new vertex elements so * return once translated @@ -1332,16 +1325,16 @@ void evergreen_vertex_buffer_update(struct r600_pipe_context *rctx) vbuffer_index = rctx->vertex_elements->elements[i].vertex_buffer_index; vertex_buffer = &rctx->vertex_buffer[vbuffer_index]; rbuffer = (struct r600_resource*)vertex_buffer->buffer; - offset = rctx->vertex_elements->vbuffer_offset[i] + - vertex_buffer->buffer_offset + - r600_bo_offset(rbuffer->bo); + offset = rctx->vertex_elements->vbuffer_offset[i]; } else { /* bind vertex buffer once */ vertex_buffer = &rctx->vertex_buffer[i]; rbuffer = (struct r600_resource*)vertex_buffer->buffer; - offset = vertex_buffer->buffer_offset + - r600_bo_offset(rbuffer->bo); + offset = 0; } + if (vertex_buffer == NULL || rbuffer == NULL) + continue; + offset += vertex_buffer->buffer_offset + r600_bo_offset(rbuffer->bo); r600_pipe_state_add_reg(rstate, R_030000_RESOURCE0_WORD0, offset, 0xFFFFFFFF, rbuffer->bo); @@ -1364,7 +1357,7 @@ void evergreen_vertex_buffer_update(struct r600_pipe_context *rctx) 0x00000000, 0xFFFFFFFF, NULL); r600_pipe_state_add_reg(rstate, R_03001C_RESOURCE0_WORD7, 0xC0000000, 0xFFFFFFFF, NULL); - evergreen_fs_resource_set(&rctx->ctx, rstate, i); + evergreen_context_pipe_state_set_fs_resource(&rctx->ctx, rstate, i); } } diff --git a/src/gallium/drivers/r600/r600.h b/src/gallium/drivers/r600/r600.h index aa456d493f7..85e29665053 100644 --- a/src/gallium/drivers/r600/r600.h +++ b/src/gallium/drivers/r600/r600.h @@ -284,10 +284,6 @@ void r600_context_queries_resume(struct r600_context *ctx); int evergreen_context_init(struct r600_context *ctx, struct radeon *radeon); void evergreen_context_draw(struct r600_context *ctx, const struct r600_draw *draw); -void evergreen_ps_resource_set(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid); -void evergreen_vs_resource_set(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid); -void evergreen_fs_resource_set(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid); - void evergreen_context_pipe_state_set_ps_resource(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid); void evergreen_context_pipe_state_set_vs_resource(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid); void evergreen_context_pipe_state_set_fs_resource(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid); diff --git a/src/gallium/drivers/r600/r600_buffer.c b/src/gallium/drivers/r600/r600_buffer.c index 7d29f760a5d..a17c54d6eeb 100644 --- a/src/gallium/drivers/r600/r600_buffer.c +++ b/src/gallium/drivers/r600/r600_buffer.c @@ -103,7 +103,7 @@ static void r600_buffer_destroy(struct pipe_screen *screen, { struct r600_resource_buffer *rbuffer = r600_buffer(buf); - if (!rbuffer->uploaded && rbuffer->r.bo) { + if (rbuffer->r.bo) { r600_bo_reference((struct radeon*)screen->winsys, &rbuffer->r.bo, NULL); } rbuffer->r.bo = NULL; diff --git a/src/gallium/drivers/r600/r600_pipe.h b/src/gallium/drivers/r600/r600_pipe.h index 43dbee99b0f..d9c35a44f18 100644 --- a/src/gallium/drivers/r600/r600_pipe.h +++ b/src/gallium/drivers/r600/r600_pipe.h @@ -136,7 +136,6 @@ struct r600_upload; struct r600_pipe_context { struct pipe_context context; struct blitter_context *blitter; - struct pipe_framebuffer_state *pframebuffer; unsigned family; void *custom_dsa_flush; struct r600_screen *screen; diff --git a/src/gallium/drivers/r600/r600_shader.c b/src/gallium/drivers/r600/r600_shader.c index 80972b04ab0..2bfe1be5447 100644 --- a/src/gallium/drivers/r600/r600_shader.c +++ b/src/gallium/drivers/r600/r600_shader.c @@ -1421,7 +1421,7 @@ static int tgsi_pow(struct r600_shader_ctx *ctx) return r; /* b * LOG2(a) */ memset(&alu, 0, sizeof(struct r600_bc_alu)); - alu.inst = CTX_INST(V_SQ_ALU_WORD1_OP2_SQ_OP2_INST_MUL_IEEE); + alu.inst = CTX_INST(V_SQ_ALU_WORD1_OP2_SQ_OP2_INST_MUL); r = tgsi_src(ctx, &inst->Src[1], &alu.src[0]); if (r) return r; diff --git a/src/gallium/drivers/r600/r600_state.c b/src/gallium/drivers/r600/r600_state.c index 63c54200d4e..8b001e11f4a 100644 --- a/src/gallium/drivers/r600/r600_state.c +++ b/src/gallium/drivers/r600/r600_state.c @@ -135,11 +135,6 @@ void r600_vertex_buffer_update(struct r600_pipe_context *rctx) if (rctx->vertex_elements == NULL || !rctx->nvertex_buffer) return; - /* delete previous translated vertex elements */ - if (rctx->tran.new_velems) { - r600_end_vertex_translate(rctx); - } - if (rctx->vertex_elements->incompatible_layout) { /* translate rebind new vertex elements so * return once translated @@ -172,16 +167,16 @@ void r600_vertex_buffer_update(struct r600_pipe_context *rctx) vbuffer_index = rctx->vertex_elements->elements[i].vertex_buffer_index; vertex_buffer = &rctx->vertex_buffer[vbuffer_index]; rbuffer = (struct r600_resource*)vertex_buffer->buffer; - offset = rctx->vertex_elements->vbuffer_offset[i] + - vertex_buffer->buffer_offset + - r600_bo_offset(rbuffer->bo); + offset = rctx->vertex_elements->vbuffer_offset[i]; } else { /* bind vertex buffer once */ vertex_buffer = &rctx->vertex_buffer[i]; rbuffer = (struct r600_resource*)vertex_buffer->buffer; - offset = vertex_buffer->buffer_offset + - r600_bo_offset(rbuffer->bo); + offset = 0; } + if (vertex_buffer == NULL || rbuffer == NULL) + continue; + offset += vertex_buffer->buffer_offset + r600_bo_offset(rbuffer->bo); r600_pipe_state_add_reg(rstate, R_038000_RESOURCE0_WORD0, offset, 0xFFFFFFFF, rbuffer->bo); @@ -280,7 +275,6 @@ void r600_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info *info) { struct r600_pipe_context *rctx = (struct r600_pipe_context *)ctx; struct r600_drawl draw; - boolean translate = FALSE; memset(&draw, 0, sizeof(struct r600_drawl)); draw.ctx = ctx; @@ -312,9 +306,6 @@ void r600_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info *info) } r600_draw_common(&draw); - if (translate) - r600_end_vertex_translate(rctx); - pipe_resource_reference(&draw.index_buffer, NULL); } @@ -1039,8 +1030,6 @@ static void r600_set_framebuffer_state(struct pipe_context *ctx, util_copy_framebuffer_state(&rctx->framebuffer, state); - rctx->pframebuffer = &rctx->framebuffer; - /* build states */ for (int i = 0; i < state->nr_cbufs; i++) { r600_cb(rctx, rstate, state, i); diff --git a/src/gallium/drivers/r600/r600_state_common.c b/src/gallium/drivers/r600/r600_state_common.c index c647e77b373..3603376f738 100644 --- a/src/gallium/drivers/r600/r600_state_common.c +++ b/src/gallium/drivers/r600/r600_state_common.c @@ -119,6 +119,11 @@ void r600_bind_vertex_elements(struct pipe_context *ctx, void *state) struct r600_pipe_context *rctx = (struct r600_pipe_context *)ctx; struct r600_vertex_element *v = (struct r600_vertex_element*)state; + /* delete previous translated vertex elements */ + if (rctx->tran.new_velems) { + r600_end_vertex_translate(rctx); + } + rctx->vertex_elements = v; if (v) { rctx->states[v->rstate.id] = &v->rstate; @@ -174,8 +179,16 @@ void r600_set_vertex_buffers(struct pipe_context *ctx, unsigned count, struct pipe_vertex_buffer *vbo; unsigned max_index = (unsigned)-1; - for (int i = 0; i < rctx->nvertex_buffer; i++) { - pipe_resource_reference(&rctx->vertex_buffer[i].buffer, NULL); + if (rctx->family >= CHIP_CEDAR) { + for (int i = 0; i < rctx->nvertex_buffer; i++) { + pipe_resource_reference(&rctx->vertex_buffer[i].buffer, NULL); + evergreen_context_pipe_state_set_fs_resource(&rctx->ctx, NULL, i); + } + } else { + for (int i = 0; i < rctx->nvertex_buffer; i++) { + pipe_resource_reference(&rctx->vertex_buffer[i].buffer, NULL); + r600_context_pipe_state_set_fs_resource(&rctx->ctx, NULL, i); + } } memcpy(rctx->vertex_buffer, buffers, sizeof(struct pipe_vertex_buffer) * count); @@ -183,15 +196,19 @@ void r600_set_vertex_buffers(struct pipe_context *ctx, unsigned count, vbo = (struct pipe_vertex_buffer*)&buffers[i]; rctx->vertex_buffer[i].buffer = NULL; + if (buffers[i].buffer == NULL) + continue; if (r600_buffer_is_user_buffer(buffers[i].buffer)) rctx->any_user_vbs = TRUE; pipe_resource_reference(&rctx->vertex_buffer[i].buffer, buffers[i].buffer); + /* The stride of zero means we will be fetching only the first + * vertex, so don't care about max_index. */ + if (!vbo->stride) + continue; + if (vbo->max_index == ~0) { - if (!vbo->stride) - vbo->max_index = 1; - else - vbo->max_index = (vbo->buffer->width0 - vbo->buffer_offset) / vbo->stride; + vbo->max_index = (vbo->buffer->width0 - vbo->buffer_offset) / vbo->stride; } max_index = MIN2(vbo->max_index, max_index); } diff --git a/src/gallium/drivers/r600/r600_translate.c b/src/gallium/drivers/r600/r600_translate.c index 1c227d32151..f80fa7af941 100644 --- a/src/gallium/drivers/r600/r600_translate.c +++ b/src/gallium/drivers/r600/r600_translate.c @@ -42,6 +42,7 @@ void r600_begin_vertex_translate(struct r600_pipe_context *rctx) struct pipe_resource *out_buffer; unsigned i, num_verts; struct pipe_vertex_element new_velems[PIPE_MAX_ATTRIBS]; + void *tmp; /* Initialize the translate key, i.e. the recipe how vertices should be * translated. */ @@ -124,12 +125,11 @@ void r600_begin_vertex_translate(struct r600_pipe_context *rctx) /* Unmap all buffers. */ for (i = 0; i < rctx->nvertex_buffer; i++) { if (vb_translated[i]) { - pipe_buffer_unmap(pipe, rctx->vertex_buffer[i].buffer, - vb_transfer[i]); + pipe_buffer_unmap(pipe, vb_transfer[i]); } } - pipe_buffer_unmap(pipe, out_buffer, out_transfer); + pipe_buffer_unmap(pipe, out_transfer); /* Setup the new vertex buffer in the first free slot. */ for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { @@ -159,8 +159,9 @@ void r600_begin_vertex_translate(struct r600_pipe_context *rctx) } } - rctx->tran.new_velems = pipe->create_vertex_elements_state(pipe, ve->count, new_velems); - pipe->bind_vertex_elements_state(pipe, rctx->tran.new_velems); + tmp = pipe->create_vertex_elements_state(pipe, ve->count, new_velems); + pipe->bind_vertex_elements_state(pipe, tmp); + rctx->tran.new_velems = tmp; pipe_resource_reference(&out_buffer, NULL); } @@ -173,15 +174,11 @@ void r600_end_vertex_translate(struct r600_pipe_context *rctx) return; } /* Restore vertex elements. */ - if (rctx->vertex_elements == rctx->tran.new_velems) { - pipe->bind_vertex_elements_state(pipe, NULL); - } pipe->delete_vertex_elements_state(pipe, rctx->tran.new_velems); rctx->tran.new_velems = NULL; /* Delete the now-unused VBO. */ - pipe_resource_reference(&rctx->vertex_buffer[rctx->tran.vb_slot].buffer, - NULL); + pipe_resource_reference(&rctx->vertex_buffer[rctx->tran.vb_slot].buffer, NULL); } void r600_translate_index_buffer(struct r600_pipe_context *r600, diff --git a/src/gallium/drivers/r600/r600_upload.c b/src/gallium/drivers/r600/r600_upload.c index ac72854ab16..44102ff55b6 100644 --- a/src/gallium/drivers/r600/r600_upload.c +++ b/src/gallium/drivers/r600/r600_upload.c @@ -69,6 +69,7 @@ void r600_upload_flush(struct r600_upload *upload) upload->default_size = MAX2(upload->total_alloc_size, upload->default_size); upload->total_alloc_size = 0; upload->size = 0; + upload->offset = 0; upload->ptr = NULL; upload->buffer = NULL; } @@ -105,7 +106,8 @@ int r600_upload_buffer(struct r600_upload *upload, unsigned offset, memcpy(upload->ptr + upload->offset, (uint8_t *) in_ptr + offset, size); *out_offset = upload->offset; *out_size = upload->size; - *out_buffer = upload->buffer; + *out_buffer = NULL; + r600_bo_reference(upload->rctx->radeon, out_buffer, upload->buffer); upload->offset += alloc_size; return 0; diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h index 9361a3df09e..903574b7e19 100644 --- a/src/gallium/drivers/softpipe/sp_context.h +++ b/src/gallium/drivers/softpipe/sp_context.h @@ -154,9 +154,9 @@ struct softpipe_context { /** TGSI exec things */ struct { - struct sp_sampler_varient *geom_samplers_list[PIPE_MAX_GEOMETRY_SAMPLERS]; - struct sp_sampler_varient *vert_samplers_list[PIPE_MAX_VERTEX_SAMPLERS]; - struct sp_sampler_varient *frag_samplers_list[PIPE_MAX_SAMPLERS]; + struct sp_sampler_variant *geom_samplers_list[PIPE_MAX_GEOMETRY_SAMPLERS]; + struct sp_sampler_variant *vert_samplers_list[PIPE_MAX_VERTEX_SAMPLERS]; + struct sp_sampler_variant *frag_samplers_list[PIPE_MAX_SAMPLERS]; } tgsi; struct tgsi_exec_machine *fs_machine; @@ -192,7 +192,7 @@ softpipe_context( struct pipe_context *pipe ) } void -softpipe_reset_sampler_varients(struct softpipe_context *softpipe); +softpipe_reset_sampler_variants(struct softpipe_context *softpipe); struct pipe_context * softpipe_create_context( struct pipe_screen *, void *priv ); diff --git a/src/gallium/drivers/softpipe/sp_state_derived.c b/src/gallium/drivers/softpipe/sp_state_derived.c index 3ba4d934fd2..bf4c12701af 100644 --- a/src/gallium/drivers/softpipe/sp_state_derived.c +++ b/src/gallium/drivers/softpipe/sp_state_derived.c @@ -197,7 +197,7 @@ update_tgsi_samplers( struct softpipe_context *softpipe ) { unsigned i; - softpipe_reset_sampler_varients( softpipe ); + softpipe_reset_sampler_variants( softpipe ); for (i = 0; i < PIPE_MAX_SAMPLERS; i++) { struct softpipe_tex_tile_cache *tc = softpipe->tex_cache[i]; diff --git a/src/gallium/drivers/softpipe/sp_state_sampler.c b/src/gallium/drivers/softpipe/sp_state_sampler.c index b59fbc33ed6..cfa211b60a0 100644 --- a/src/gallium/drivers/softpipe/sp_state_sampler.c +++ b/src/gallium/drivers/softpipe/sp_state_sampler.c @@ -43,8 +43,8 @@ struct sp_sampler { struct pipe_sampler_state base; - struct sp_sampler_varient *varients; - struct sp_sampler_varient *current; + struct sp_sampler_variant *variants; + struct sp_sampler_variant *current; }; static struct sp_sampler *sp_sampler( struct pipe_sampler_state *sampler ) @@ -60,7 +60,7 @@ softpipe_create_sampler_state(struct pipe_context *pipe, struct sp_sampler *sp_sampler = CALLOC_STRUCT(sp_sampler); sp_sampler->base = *sampler; - sp_sampler->varients = NULL; + sp_sampler->variants = NULL; return (void *)sp_sampler; } @@ -277,23 +277,24 @@ softpipe_set_geometry_sampler_views(struct pipe_context *pipe, /** - * Find/create an sp_sampler_varient object for sampling the given texture, + * Find/create an sp_sampler_variant 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 + * variant for multiple texture units because the sampler variant contains * the texture object pointer. If the texture object pointer were stored - * somewhere outside the sampler varient, we could re-use samplers for + * somewhere outside the sampler variant, we could re-use samplers for * multiple texture units. */ -static struct sp_sampler_varient * -get_sampler_varient( unsigned unit, +static struct sp_sampler_variant * +get_sampler_variant( unsigned unit, struct sp_sampler *sampler, + struct pipe_sampler_view *view, struct pipe_resource *resource, unsigned processor ) { struct softpipe_resource *sp_texture = softpipe_resource(resource); - struct sp_sampler_varient *v = NULL; + struct sp_sampler_variant *v = NULL; union sp_sampler_key key; /* if this fails, widen the key.unit field and update this assertion */ @@ -303,6 +304,10 @@ get_sampler_varient( unsigned unit, key.bits.is_pot = sp_texture->pot; key.bits.processor = processor; key.bits.unit = unit; + key.bits.swizzle_r = view->swizzle_r; + key.bits.swizzle_g = view->swizzle_g; + key.bits.swizzle_b = view->swizzle_b; + key.bits.swizzle_a = view->swizzle_a; key.bits.pad = 0; if (sampler->current && @@ -311,14 +316,14 @@ get_sampler_varient( unsigned unit, } if (v == NULL) { - for (v = sampler->varients; v; v = v->next) + for (v = sampler->variants; 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; + v = sp_create_sampler_variant( &sampler->base, key ); + v->next = sampler->variants; + sampler->variants = v; } } @@ -328,7 +333,7 @@ get_sampler_varient( unsigned unit, void -softpipe_reset_sampler_varients(struct softpipe_context *softpipe) +softpipe_reset_sampler_variants(struct softpipe_context *softpipe) { int i; @@ -345,12 +350,13 @@ softpipe_reset_sampler_varients(struct softpipe_context *softpipe) } softpipe->tgsi.vert_samplers_list[i] = - get_sampler_varient( i, + get_sampler_variant( i, sp_sampler(softpipe->vertex_samplers[i]), + softpipe->vertex_sampler_views[i], texture, TGSI_PROCESSOR_VERTEX ); - sp_sampler_varient_bind_texture( softpipe->tgsi.vert_samplers_list[i], + sp_sampler_variant_bind_texture( softpipe->tgsi.vert_samplers_list[i], softpipe->vertex_tex_cache[i], texture ); } @@ -366,13 +372,14 @@ softpipe_reset_sampler_varients(struct softpipe_context *softpipe) } softpipe->tgsi.geom_samplers_list[i] = - get_sampler_varient( + get_sampler_variant( i, sp_sampler(softpipe->geometry_samplers[i]), + softpipe->geometry_sampler_views[i], texture, TGSI_PROCESSOR_GEOMETRY ); - sp_sampler_varient_bind_texture( + sp_sampler_variant_bind_texture( softpipe->tgsi.geom_samplers_list[i], softpipe->geometry_tex_cache[i], texture ); @@ -389,12 +396,13 @@ softpipe_reset_sampler_varients(struct softpipe_context *softpipe) } softpipe->tgsi.frag_samplers_list[i] = - get_sampler_varient( i, + get_sampler_variant( i, sp_sampler(softpipe->sampler[i]), + softpipe->sampler_views[i], texture, TGSI_PROCESSOR_FRAGMENT ); - sp_sampler_varient_bind_texture( softpipe->tgsi.frag_samplers_list[i], + sp_sampler_variant_bind_texture( softpipe->tgsi.frag_samplers_list[i], softpipe->tex_cache[i], texture ); } @@ -406,11 +414,11 @@ 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; + struct sp_sampler_variant *v, *tmp; - for (v = sp_sampler->varients; v; v = tmp) { + for (v = sp_sampler->variants; v; v = tmp) { tmp = v->next; - sp_sampler_varient_destroy(v); + sp_sampler_variant_destroy(v); } FREE( sampler ); diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c index 2eac4c7a82b..cbc40d4b446 100644 --- a/src/gallium/drivers/softpipe/sp_tex_sample.c +++ b/src/gallium/drivers/softpipe/sp_tex_sample.c @@ -545,7 +545,7 @@ wrap_linear_unorm_clamp_to_edge(const float s[4], unsigned size, * derivatives w.r.t X and Y, then compute lambda (level of detail). */ static float -compute_lambda_1d(const struct sp_sampler_varient *samp, +compute_lambda_1d(const struct sp_sampler_variant *samp, const float s[QUAD_SIZE], const float t[QUAD_SIZE], const float p[QUAD_SIZE]) @@ -560,7 +560,7 @@ compute_lambda_1d(const struct sp_sampler_varient *samp, static float -compute_lambda_2d(const struct sp_sampler_varient *samp, +compute_lambda_2d(const struct sp_sampler_variant *samp, const float s[QUAD_SIZE], const float t[QUAD_SIZE], const float p[QUAD_SIZE]) @@ -579,7 +579,7 @@ compute_lambda_2d(const struct sp_sampler_varient *samp, static float -compute_lambda_3d(const struct sp_sampler_varient *samp, +compute_lambda_3d(const struct sp_sampler_variant *samp, const float s[QUAD_SIZE], const float t[QUAD_SIZE], const float p[QUAD_SIZE]) @@ -608,7 +608,7 @@ compute_lambda_3d(const struct sp_sampler_varient *samp, * Since there aren't derivatives to use, just return 0. */ static float -compute_lambda_vert(const struct sp_sampler_varient *samp, +compute_lambda_vert(const struct sp_sampler_variant *samp, const float s[QUAD_SIZE], const float t[QUAD_SIZE], const float p[QUAD_SIZE]) @@ -634,7 +634,7 @@ compute_lambda_vert(const struct sp_sampler_varient *samp, static INLINE const float * -get_texel_2d_no_border(const struct sp_sampler_varient *samp, +get_texel_2d_no_border(const struct sp_sampler_variant *samp, union tex_tile_address addr, int x, int y) { const struct softpipe_tex_cached_tile *tile; @@ -651,7 +651,7 @@ get_texel_2d_no_border(const struct sp_sampler_varient *samp, static INLINE const float * -get_texel_2d(const struct sp_sampler_varient *samp, +get_texel_2d(const struct sp_sampler_variant *samp, union tex_tile_address addr, int x, int y) { const struct pipe_resource *texture = samp->texture; @@ -659,8 +659,7 @@ get_texel_2d(const struct sp_sampler_varient *samp, if (x < 0 || x >= (int) u_minify(texture->width0, level) || y < 0 || y >= (int) u_minify(texture->height0, level)) { - return sp_tex_tile_cache_border_color(samp->cache, - samp->sampler->border_color); + return samp->sampler->border_color; } else { return get_texel_2d_no_border( samp, addr, x, y ); @@ -671,7 +670,7 @@ get_texel_2d(const struct sp_sampler_varient *samp, /* 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, +get_texel_quad_2d_no_border_single_tile(const struct sp_sampler_variant *samp, union tex_tile_address addr, unsigned x, unsigned y, const float *out[4]) @@ -695,7 +694,7 @@ get_texel_quad_2d_no_border_single_tile(const struct sp_sampler_varient *samp, /* Gather a quad of potentially non-adjacent texels: */ static INLINE void -get_texel_quad_2d_no_border(const struct sp_sampler_varient *samp, +get_texel_quad_2d_no_border(const struct sp_sampler_variant *samp, union tex_tile_address addr, int x0, int y0, int x1, int y1, @@ -710,7 +709,7 @@ get_texel_quad_2d_no_border(const struct sp_sampler_varient *samp, /* Can involve a lot of unnecessary checks for border color: */ static INLINE void -get_texel_quad_2d(const struct sp_sampler_varient *samp, +get_texel_quad_2d(const struct sp_sampler_variant *samp, union tex_tile_address addr, int x0, int y0, int x1, int y1, @@ -724,10 +723,10 @@ get_texel_quad_2d(const struct sp_sampler_varient *samp, -/* 3d varients: +/* 3d variants: */ static INLINE const float * -get_texel_3d_no_border(const struct sp_sampler_varient *samp, +get_texel_3d_no_border(const struct sp_sampler_variant *samp, union tex_tile_address addr, int x, int y, int z) { const struct softpipe_tex_cached_tile *tile; @@ -745,7 +744,7 @@ get_texel_3d_no_border(const struct sp_sampler_varient *samp, static INLINE const float * -get_texel_3d(const struct sp_sampler_varient *samp, +get_texel_3d(const struct sp_sampler_variant *samp, union tex_tile_address addr, int x, int y, int z) { const struct pipe_resource *texture = samp->texture; @@ -754,8 +753,7 @@ get_texel_3d(const struct sp_sampler_varient *samp, if (x < 0 || x >= (int) u_minify(texture->width0, level) || y < 0 || y >= (int) u_minify(texture->height0, level) || z < 0 || z >= (int) u_minify(texture->depth0, level)) { - return sp_tex_tile_cache_border_color(samp->cache, - samp->sampler->border_color); + return samp->sampler->border_color; } else { return get_texel_3d_no_border( samp, addr, x, y, z ); @@ -800,7 +798,7 @@ img_filter_2d_linear_repeat_POT(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); unsigned j; unsigned level = samp->level; unsigned xpot = pot_level_size(samp->xpot, level); @@ -863,7 +861,7 @@ img_filter_2d_nearest_repeat_POT(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); unsigned j; unsigned level = samp->level; unsigned xpot = pot_level_size(samp->xpot, level); @@ -907,7 +905,7 @@ img_filter_2d_nearest_clamp_POT(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); unsigned j; unsigned level = samp->level; unsigned xpot = pot_level_size(samp->xpot, level); @@ -960,7 +958,7 @@ img_filter_1d_nearest(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; unsigned level0, j; int width; @@ -1000,7 +998,7 @@ img_filter_2d_nearest(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; unsigned level0, j; int width, height; @@ -1052,7 +1050,7 @@ img_filter_cube_nearest(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; const unsigned *faces = samp->faces; /* zero when not cube-mapping */ unsigned level0, j; @@ -1096,7 +1094,7 @@ img_filter_3d_nearest(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; unsigned level0, j; int width, height, depth; @@ -1138,7 +1136,7 @@ img_filter_1d_linear(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; unsigned level0, j; int width; @@ -1178,7 +1176,7 @@ img_filter_2d_linear(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; unsigned level0, j; int width, height; @@ -1225,7 +1223,7 @@ img_filter_cube_linear(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; const unsigned *faces = samp->faces; /* zero when not cube-mapping */ unsigned level0, j; @@ -1274,7 +1272,7 @@ img_filter_3d_linear(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - const struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + const struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; unsigned level0, j; int width, height, depth; @@ -1350,7 +1348,7 @@ mip_filter_linear(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; int level0; float lambda; @@ -1417,7 +1415,7 @@ mip_filter_nearest(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; float lambda; float lod[QUAD_SIZE]; @@ -1460,7 +1458,7 @@ mip_filter_none(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); float lambda; float lod[QUAD_SIZE]; @@ -1501,7 +1499,7 @@ mip_filter_linear_2d_linear_repeat_POT( enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_resource *texture = samp->texture; int level0; float lambda; @@ -1569,7 +1567,7 @@ sample_compare(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); const struct pipe_sampler_state *sampler = samp->sampler; int j, k0, k1, k2, k3; float val; @@ -1656,7 +1654,7 @@ sample_cube(struct tgsi_sampler *tgsi_sampler, enum tgsi_sampler_control control, float rgba[NUM_CHANNELS][QUAD_SIZE]) { - struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); + struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); unsigned j; float ssss[4], tttt[4]; @@ -1731,6 +1729,86 @@ sample_cube(struct tgsi_sampler *tgsi_sampler, } +static void +sample_swizzle(struct tgsi_sampler *tgsi_sampler, + const float s[QUAD_SIZE], + const float t[QUAD_SIZE], + const float p[QUAD_SIZE], + const float c0[QUAD_SIZE], + enum tgsi_sampler_control control, + float rgba[NUM_CHANNELS][QUAD_SIZE]) +{ + struct sp_sampler_variant *samp = sp_sampler_variant(tgsi_sampler); + float rgba_temp[NUM_CHANNELS][QUAD_SIZE]; + const unsigned swizzle_r = samp->key.bits.swizzle_r; + const unsigned swizzle_g = samp->key.bits.swizzle_g; + const unsigned swizzle_b = samp->key.bits.swizzle_b; + const unsigned swizzle_a = samp->key.bits.swizzle_a; + unsigned j; + + samp->sample_target(tgsi_sampler, s, t, p, c0, control, rgba_temp); + + switch (swizzle_r) { + case PIPE_SWIZZLE_ZERO: + for (j = 0; j < 4; j++) + rgba[0][j] = 0.0f; + break; + case PIPE_SWIZZLE_ONE: + for (j = 0; j < 4; j++) + rgba[0][j] = 1.0f; + break; + default: + assert(swizzle_r < 4); + for (j = 0; j < 4; j++) + rgba[0][j] = rgba_temp[swizzle_r][j]; + } + + switch (swizzle_g) { + case PIPE_SWIZZLE_ZERO: + for (j = 0; j < 4; j++) + rgba[1][j] = 0.0f; + break; + case PIPE_SWIZZLE_ONE: + for (j = 0; j < 4; j++) + rgba[1][j] = 1.0f; + break; + default: + assert(swizzle_g < 4); + for (j = 0; j < 4; j++) + rgba[1][j] = rgba_temp[swizzle_g][j]; + } + + switch (swizzle_b) { + case PIPE_SWIZZLE_ZERO: + for (j = 0; j < 4; j++) + rgba[2][j] = 0.0f; + break; + case PIPE_SWIZZLE_ONE: + for (j = 0; j < 4; j++) + rgba[2][j] = 1.0f; + break; + default: + assert(swizzle_b < 4); + for (j = 0; j < 4; j++) + rgba[2][j] = rgba_temp[swizzle_b][j]; + } + + switch (swizzle_a) { + case PIPE_SWIZZLE_ZERO: + for (j = 0; j < 4; j++) + rgba[3][j] = 0.0f; + break; + case PIPE_SWIZZLE_ONE: + for (j = 0; j < 4; j++) + rgba[3][j] = 1.0f; + break; + default: + assert(swizzle_a < 4); + for (j = 0; j < 4; j++) + rgba[3][j] = rgba_temp[swizzle_a][j]; + } +} + static wrap_nearest_func get_nearest_unorm_wrap(unsigned mode) @@ -1909,10 +1987,10 @@ get_img_filter(const union sp_sampler_key key, /** - * Bind the given texture object and texture cache to the sampler varient. + * Bind the given texture object and texture cache to the sampler variant. */ void -sp_sampler_varient_bind_texture( struct sp_sampler_varient *samp, +sp_sampler_variant_bind_texture( struct sp_sampler_variant *samp, struct softpipe_tex_tile_cache *tex_cache, const struct pipe_resource *texture ) { @@ -1927,20 +2005,20 @@ sp_sampler_varient_bind_texture( struct sp_sampler_varient *samp, void -sp_sampler_varient_destroy( struct sp_sampler_varient *samp ) +sp_sampler_variant_destroy( struct sp_sampler_variant *samp ) { FREE(samp); } /** - * Create a sampler varient for a given set of non-orthogonal state. + * Create a sampler variant for a given set of non-orthogonal state. */ -struct sp_sampler_varient * -sp_create_sampler_varient( const struct pipe_sampler_state *sampler, +struct sp_sampler_variant * +sp_create_sampler_variant( const struct pipe_sampler_state *sampler, const union sp_sampler_key key ) { - struct sp_sampler_varient *samp = CALLOC_STRUCT(sp_sampler_varient); + struct sp_sampler_variant *samp = CALLOC_STRUCT(sp_sampler_variant); if (!samp) return NULL; @@ -2015,7 +2093,7 @@ sp_create_sampler_varient( const struct pipe_sampler_state *sampler, } if (key.bits.target == PIPE_TEXTURE_CUBE) { - samp->base.get_samples = sample_cube; + samp->sample_target = sample_cube; } else { samp->faces[0] = 0; @@ -2026,7 +2104,17 @@ sp_create_sampler_varient( const struct pipe_sampler_state *sampler, /* Skip cube face determination by promoting the compare * function pointer: */ - samp->base.get_samples = samp->compare; + samp->sample_target = samp->compare; + } + + if (key.bits.swizzle_r != PIPE_SWIZZLE_RED || + key.bits.swizzle_g != PIPE_SWIZZLE_GREEN || + key.bits.swizzle_b != PIPE_SWIZZLE_BLUE || + key.bits.swizzle_a != PIPE_SWIZZLE_ALPHA) { + samp->base.get_samples = sample_swizzle; + } + else { + samp->base.get_samples = samp->sample_target; } return samp; diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.h b/src/gallium/drivers/softpipe/sp_tex_sample.h index 6114acf7371..ed99006ab02 100644 --- a/src/gallium/drivers/softpipe/sp_tex_sample.h +++ b/src/gallium/drivers/softpipe/sp_tex_sample.h @@ -32,7 +32,7 @@ #include "tgsi/tgsi_exec.h" -struct sp_sampler_varient; +struct sp_sampler_variant; typedef void (*wrap_nearest_func)(const float s[4], unsigned size, @@ -44,7 +44,7 @@ typedef void (*wrap_linear_func)(const float s[4], int icoord1[4], float w[4]); -typedef float (*compute_lambda_func)(const struct sp_sampler_varient *sampler, +typedef float (*compute_lambda_func)(const struct sp_sampler_variant *sampler, const float s[QUAD_SIZE], const float t[QUAD_SIZE], const float p[QUAD_SIZE]); @@ -64,7 +64,11 @@ union sp_sampler_key { unsigned is_pot:1; unsigned processor:2; unsigned unit:4; - unsigned pad:22; + unsigned swizzle_r:3; + unsigned swizzle_g:3; + unsigned swizzle_b:3; + unsigned swizzle_a:3; + unsigned pad:10; } bits; unsigned value; }; @@ -72,7 +76,7 @@ union sp_sampler_key { /** * Subclass of tgsi_sampler */ -struct sp_sampler_varient +struct sp_sampler_variant { struct tgsi_sampler base; /**< base class */ @@ -113,32 +117,33 @@ struct sp_sampler_varient filter_func mip_filter; filter_func compare; + filter_func sample_target; /* Linked list: */ - struct sp_sampler_varient *next; + struct sp_sampler_variant *next; }; struct sp_sampler; -/* Create a sampler varient for a given set of non-orthogonal state. Currently the +/* Create a sampler variant for a given set of non-orthogonal state. Currently the */ -struct sp_sampler_varient * -sp_create_sampler_varient( const struct pipe_sampler_state *sampler, +struct sp_sampler_variant * +sp_create_sampler_variant( const struct pipe_sampler_state *sampler, const union sp_sampler_key key ); -void sp_sampler_varient_bind_texture( struct sp_sampler_varient *varient, +void sp_sampler_variant_bind_texture( struct sp_sampler_variant *variant, struct softpipe_tex_tile_cache *tex_cache, const struct pipe_resource *tex ); -void sp_sampler_varient_destroy( struct sp_sampler_varient * ); +void sp_sampler_variant_destroy( struct sp_sampler_variant * ); -static INLINE struct sp_sampler_varient * -sp_sampler_varient(const struct tgsi_sampler *sampler) +static INLINE struct sp_sampler_variant * +sp_sampler_variant(const struct tgsi_sampler *sampler) { - return (struct sp_sampler_varient *) sampler; + return (struct sp_sampler_variant *) sampler; } extern void diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c index 1393164150e..e42015ad498 100644 --- a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c +++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c @@ -278,45 +278,26 @@ sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, tc->tex_z = addr.bits.z; } - /* get tile from the transfer (view into texture) */ + /* get tile from the transfer (view into texture) + * Note we're using the swizzle version of this fuction only because + * we need to pass the texture cache's format explicitly. + */ pipe_get_tile_swizzle(tc->pipe, tc->tex_trans, addr.bits.x * TILE_SIZE, addr.bits.y * TILE_SIZE, TILE_SIZE, TILE_SIZE, - tc->swizzle_r, - tc->swizzle_g, - tc->swizzle_b, - tc->swizzle_a, + PIPE_SWIZZLE_RED, + PIPE_SWIZZLE_GREEN, + PIPE_SWIZZLE_BLUE, + PIPE_SWIZZLE_ALPHA, tc->format, (float *) tile->data.color); + tile->addr = addr; } tc->last_tile = tile; return tile; } - - - -/** - * Return the swizzled border color. - */ -const float * -sp_tex_tile_cache_border_color(struct softpipe_tex_tile_cache *tc, - const float border_color[4]) -{ - float rgba01[6]; - - COPY_4V(rgba01, border_color); - rgba01[PIPE_SWIZZLE_ZERO] = 0.0f; - rgba01[PIPE_SWIZZLE_ONE] = 1.0f; - - tc->swz_border_color[0] = rgba01[tc->swizzle_r]; - tc->swz_border_color[1] = rgba01[tc->swizzle_g]; - tc->swz_border_color[2] = rgba01[tc->swizzle_b]; - tc->swz_border_color[3] = rgba01[tc->swizzle_a]; - - return tc->swz_border_color; -} diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.h b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h index e0b66bf3f7c..2220955b715 100644 --- a/src/gallium/drivers/softpipe/sp_tex_tile_cache.h +++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h @@ -95,8 +95,6 @@ struct softpipe_tex_tile_cache unsigned format; struct softpipe_tex_cached_tile *last_tile; /**< most recently retrieved tile */ - - float swz_border_color[4]; /**< swizzled border color */ }; @@ -161,10 +159,5 @@ sp_get_cached_tile_tex(struct softpipe_tex_tile_cache *tc, } -const float * -sp_tex_tile_cache_border_color(struct softpipe_tex_tile_cache *tc, - const float border_color[4]); - - #endif /* SP_TEX_TILE_CACHE_H */ diff --git a/src/gallium/drivers/svga/svga_draw_arrays.c b/src/gallium/drivers/svga/svga_draw_arrays.c index da33fae62f1..be0e7abe21b 100644 --- a/src/gallium/drivers/svga/svga_draw_arrays.c +++ b/src/gallium/drivers/svga/svga_draw_arrays.c @@ -65,14 +65,14 @@ static enum pipe_error generate_indices( struct svga_hwtnl *hwtnl, generate( nr, dst_map ); - pipe_buffer_unmap( pipe, dst, transfer ); + pipe_buffer_unmap( pipe, transfer ); *out_buf = dst; return PIPE_OK; fail: if (dst_map) - pipe_buffer_unmap( pipe, dst, transfer ); + pipe_buffer_unmap( pipe, transfer ); if (dst) pipe->screen->resource_destroy( pipe->screen, dst ); diff --git a/src/gallium/drivers/svga/svga_draw_elements.c b/src/gallium/drivers/svga/svga_draw_elements.c index c4579177b77..c7ea014bba5 100644 --- a/src/gallium/drivers/svga/svga_draw_elements.c +++ b/src/gallium/drivers/svga/svga_draw_elements.c @@ -72,18 +72,18 @@ translate_indices( struct svga_hwtnl *hwtnl, nr, dst_map ); - pipe_buffer_unmap( pipe, src, src_transfer ); - pipe_buffer_unmap( pipe, dst, dst_transfer ); + pipe_buffer_unmap( pipe, src_transfer ); + pipe_buffer_unmap( pipe, dst_transfer ); *out_buf = dst; return PIPE_OK; fail: if (src_map) - pipe_buffer_unmap( pipe, src, src_transfer ); + pipe_buffer_unmap( pipe, src_transfer ); if (dst_map) - pipe_buffer_unmap( pipe, dst, dst_transfer ); + pipe_buffer_unmap( pipe, dst_transfer ); if (dst) pipe->screen->resource_destroy( pipe->screen, dst ); diff --git a/src/gallium/drivers/svga/svga_pipe_rasterizer.c b/src/gallium/drivers/svga/svga_pipe_rasterizer.c index 660eb0757a6..e97b4e57415 100644 --- a/src/gallium/drivers/svga/svga_pipe_rasterizer.c +++ b/src/gallium/drivers/svga/svga_pipe_rasterizer.c @@ -68,7 +68,7 @@ svga_create_rasterizer_state(struct pipe_context *pipe, /* need this for draw module. */ rast->templ = *templ; - /* light_twoside - XXX: need fragment shader varient */ + /* light_twoside - XXX: need fragment shader variant */ /* poly_smooth - XXX: no fallback available */ /* poly_stipple_enable - draw module */ /* sprite_coord_enable - ? */ diff --git a/src/gallium/drivers/svga/svga_screen.c b/src/gallium/drivers/svga/svga_screen.c index 078190342a1..d0f42c614c9 100644 --- a/src/gallium/drivers/svga/svga_screen.c +++ b/src/gallium/drivers/svga/svga_screen.c @@ -35,7 +35,6 @@ #include "svga_resource_texture.h" #include "svga_resource.h" #include "svga_debug.h" -#include "svga_surface.h" #include "svga3d_shaderdefs.h" diff --git a/src/gallium/drivers/svga/svga_state_constants.c b/src/gallium/drivers/svga/svga_state_constants.c index 97c818cd379..daf1024fd02 100644 --- a/src/gallium/drivers/svga/svga_state_constants.c +++ b/src/gallium/drivers/svga/svga_state_constants.c @@ -110,7 +110,7 @@ static int emit_consts( struct svga_context *svga, done: if (data) - pipe_buffer_unmap(&svga->pipe, svga->curr.cb[unit], transfer); + pipe_buffer_unmap(&svga->pipe, transfer); return ret; } diff --git a/src/gallium/drivers/svga/svga_state_fs.c b/src/gallium/drivers/svga/svga_state_fs.c index ad6f2947137..9c04adec8ee 100644 --- a/src/gallium/drivers/svga/svga_state_fs.c +++ b/src/gallium/drivers/svga/svga_state_fs.c @@ -136,7 +136,7 @@ static int make_fs_key( const struct svga_context *svga, /* The blend workaround for simulating logicop xor behaviour * requires that the incoming fragment color be white. This change - * achieves that by creating a varient of the current fragment + * achieves that by creating a variant of the current fragment * shader that overrides all output colors with 1,1,1,1 * * This will work for most shaders, including those containing diff --git a/src/gallium/drivers/svga/svga_state_vs.c b/src/gallium/drivers/svga/svga_state_vs.c index 5133c70593c..6682a1efe66 100644 --- a/src/gallium/drivers/svga/svga_state_vs.c +++ b/src/gallium/drivers/svga/svga_state_vs.c @@ -233,9 +233,7 @@ static int update_zero_stride( struct svga_context *svga, translate->run(translate, 0, 1, 0, svga->curr.zero_stride_constants); - pipe_buffer_unmap(&svga->pipe, - vbuffer->buffer, - transfer); + pipe_buffer_unmap(&svga->pipe, transfer); translate->release(translate); } diff --git a/src/gallium/drivers/svga/svga_swtnl_backend.c b/src/gallium/drivers/svga/svga_swtnl_backend.c index ff3da842729..24646b48f62 100644 --- a/src/gallium/drivers/svga/svga_swtnl_backend.c +++ b/src/gallium/drivers/svga/svga_swtnl_backend.c @@ -141,7 +141,7 @@ svga_vbuf_render_unmap_vertices( struct vbuf_render *render, pipe_buffer_flush_mapped_range(&svga->pipe, svga_render->vbuf_transfer, offset, length); - pipe_buffer_unmap(&svga->pipe, svga_render->vbuf, svga_render->vbuf_transfer); + pipe_buffer_unmap(&svga->pipe, svga_render->vbuf_transfer); svga_render->min_index = min_index; svga_render->max_index = max_index; svga_render->vbuf_used = MAX2(svga_render->vbuf_used, used); diff --git a/src/gallium/drivers/svga/svga_swtnl_draw.c b/src/gallium/drivers/svga/svga_swtnl_draw.c index 814e8edd70f..d9039845819 100644 --- a/src/gallium/drivers/svga/svga_swtnl_draw.c +++ b/src/gallium/drivers/svga/svga_swtnl_draw.c @@ -106,20 +106,17 @@ svga_swtnl_draw_vbo(struct svga_context *svga, * unmap vertex/index buffers */ for (i = 0; i < svga->curr.num_vertex_buffers; i++) { - pipe_buffer_unmap(&svga->pipe, svga->curr.vb[i].buffer, - vb_transfer[i]); + pipe_buffer_unmap(&svga->pipe, vb_transfer[i]); draw_set_mapped_vertex_buffer(draw, i, NULL); } if (ib_transfer) { - pipe_buffer_unmap(&svga->pipe, svga->curr.ib.buffer, ib_transfer); + pipe_buffer_unmap(&svga->pipe, ib_transfer); draw_set_mapped_index_buffer(draw, NULL); } if (svga->curr.cb[PIPE_SHADER_VERTEX]) { - pipe_buffer_unmap(&svga->pipe, - svga->curr.cb[PIPE_SHADER_VERTEX], - cb_transfer); + pipe_buffer_unmap(&svga->pipe, cb_transfer); } return ret; diff --git a/src/gallium/state_trackers/egl/common/egl_g3d.c b/src/gallium/state_trackers/egl/common/egl_g3d.c index a3750ac56fb..f2b137a674a 100644 --- a/src/gallium/state_trackers/egl/common/egl_g3d.c +++ b/src/gallium/state_trackers/egl/common/egl_g3d.c @@ -183,17 +183,21 @@ init_config_attributes(_EGLConfig *conf, const struct native_config *nconf, } surface_type = 0x0; - if (nconf->window_bit) - surface_type |= EGL_WINDOW_BIT; - if (nconf->pixmap_bit) - surface_type |= EGL_PIXMAP_BIT; + /* pixmap surfaces should be EGL_SINGLE_BUFFER */ + if (nconf->buffer_mask & (1 << NATIVE_ATTACHMENT_FRONT_LEFT)) { + if (nconf->pixmap_bit) + surface_type |= EGL_PIXMAP_BIT; + } + /* the others surfaces should be EGL_BACK_BUFFER (or settable) */ + if (nconf->buffer_mask & (1 << NATIVE_ATTACHMENT_BACK_LEFT)) { + if (nconf->window_bit) + surface_type |= EGL_WINDOW_BIT; #ifdef EGL_MESA_screen_surface - if (nconf->scanout_bit) - surface_type |= EGL_SCREEN_BIT_MESA; + if (nconf->scanout_bit) + surface_type |= EGL_SCREEN_BIT_MESA; #endif - - if (nconf->buffer_mask & (1 << NATIVE_ATTACHMENT_BACK_LEFT)) surface_type |= EGL_PBUFFER_BIT; + } conf->Conformant = api_mask; conf->RenderableType = api_mask; @@ -226,11 +230,6 @@ init_config_attributes(_EGLConfig *conf, const struct native_config *nconf, } conf->Level = nconf->level; - conf->Samples = nconf->samples; - conf->SampleBuffers = 0; - - if (nconf->slow_config) - conf->ConfigCaveat = EGL_SLOW_CONFIG; if (nconf->transparent_rgb) { conf->TransparentType = EGL_TRANSPARENT_RGB; @@ -257,13 +256,9 @@ egl_g3d_init_config(_EGLDriver *drv, _EGLDisplay *dpy, int preserve_buffer, int max_swap_interval) { struct egl_g3d_config *gconf = egl_g3d_config(conf); - EGLint buffer_mask, api_mask; + EGLint buffer_mask; EGLBoolean valid; - /* skip single-buffered configs */ - if (!(nconf->buffer_mask & (1 << NATIVE_ATTACHMENT_BACK_LEFT))) - return EGL_FALSE; - buffer_mask = 0x0; if (nconf->buffer_mask & (1 << NATIVE_ATTACHMENT_FRONT_LEFT)) buffer_mask |= ST_ATTACHMENT_FRONT_LEFT_MASK; @@ -278,24 +273,14 @@ egl_g3d_init_config(_EGLDriver *drv, _EGLDisplay *dpy, gconf->stvis.color_format = nconf->color_format; gconf->stvis.depth_stencil_format = depth_stencil_format; gconf->stvis.accum_format = PIPE_FORMAT_NONE; - gconf->stvis.samples = nconf->samples; + gconf->stvis.samples = 0; + /* will be overridden per surface */ gconf->stvis.render_buffer = (buffer_mask & ST_ATTACHMENT_BACK_LEFT_MASK) ? ST_ATTACHMENT_BACK_LEFT : ST_ATTACHMENT_FRONT_LEFT; - api_mask = dpy->ClientAPIsMask; - /* this is required by EGL, not by OpenGL ES */ - if (nconf->window_bit && - gconf->stvis.render_buffer != ST_ATTACHMENT_BACK_LEFT) - api_mask &= ~(EGL_OPENGL_ES_BIT | EGL_OPENGL_ES2_BIT); - - if (!api_mask) { - _eglLog(_EGL_DEBUG, "no state tracker supports config 0x%x", - nconf->native_visual_id); - } - valid = init_config_attributes(&gconf->base, - nconf, api_mask, depth_stencil_format, + nconf, dpy->ClientAPIsMask, depth_stencil_format, preserve_buffer, max_swap_interval); if (!valid) { _eglLog(_EGL_DEBUG, "skip invalid config 0x%x", nconf->native_visual_id); diff --git a/src/gallium/state_trackers/egl/common/egl_g3d_api.c b/src/gallium/state_trackers/egl/common/egl_g3d_api.c index 8e53e1dccba..bce901ac9a9 100644 --- a/src/gallium/state_trackers/egl/common/egl_g3d_api.c +++ b/src/gallium/state_trackers/egl/common/egl_g3d_api.c @@ -324,7 +324,8 @@ egl_g3d_create_surface(_EGLDriver *drv, _EGLDisplay *dpy, _EGLConfig *conf, } gsurf->stvis = gconf->stvis; - if (gsurf->base.RenderBuffer == EGL_SINGLE_BUFFER) + if (gsurf->base.RenderBuffer == EGL_SINGLE_BUFFER && + gconf->stvis.buffer_mask & ST_ATTACHMENT_FRONT_LEFT_MASK) gsurf->stvis.render_buffer = ST_ATTACHMENT_FRONT_LEFT; gsurf->stfbi = egl_g3d_create_st_framebuffer(&gsurf->base); @@ -402,7 +403,6 @@ egl_g3d_create_pbuffer_surface(_EGLDriver *drv, _EGLDisplay *dpy, _EGLConfig *conf, const EGLint *attribs) { struct egl_g3d_surface *gsurf; - struct pipe_resource *ptex = NULL; gsurf = create_pbuffer_surface(dpy, conf, attribs, "eglCreatePbufferSurface"); @@ -411,13 +411,6 @@ egl_g3d_create_pbuffer_surface(_EGLDriver *drv, _EGLDisplay *dpy, gsurf->client_buffer_type = EGL_NONE; - if (!gsurf->stfbi->validate(gsurf->stfbi, - &gsurf->stvis.render_buffer, 1, &ptex)) { - egl_g3d_destroy_st_framebuffer(gsurf->stfbi); - FREE(gsurf); - return NULL; - } - return &gsurf->base; } @@ -477,12 +470,14 @@ egl_g3d_create_pbuffer_from_client_buffer(_EGLDriver *drv, _EGLDisplay *dpy, gsurf->client_buffer_type = buftype; gsurf->client_buffer = buffer; + /* validate now so that it fails if the client buffer is invalid */ if (!gsurf->stfbi->validate(gsurf->stfbi, &gsurf->stvis.render_buffer, 1, &ptex)) { egl_g3d_destroy_st_framebuffer(gsurf->stfbi); FREE(gsurf); return NULL; } + pipe_resource_reference(&ptex, NULL); return &gsurf->base; } @@ -643,19 +638,13 @@ egl_g3d_copy_buffers(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surf, struct egl_g3d_display *gdpy = egl_g3d_display(dpy); struct egl_g3d_surface *gsurf = egl_g3d_surface(surf); _EGLContext *ctx = _eglGetCurrentContext(); - struct egl_g3d_config *gconf; struct native_surface *nsurf; struct pipe_resource *ptex; if (!gsurf->render_texture) return EGL_TRUE; - gconf = egl_g3d_config(egl_g3d_find_pixmap_config(dpy, target)); - if (!gconf) - return _eglError(EGL_BAD_NATIVE_PIXMAP, "eglCopyBuffers"); - - nsurf = gdpy->native->create_pixmap_surface(gdpy->native, - target, gconf->native); + nsurf = gdpy->native->create_pixmap_surface(gdpy->native, target, NULL); if (!nsurf) return _eglError(EGL_BAD_NATIVE_PIXMAP, "eglCopyBuffers"); @@ -676,14 +665,13 @@ egl_g3d_copy_buffers(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surf, ptex = get_pipe_resource(gdpy->native, nsurf, NATIVE_ATTACHMENT_FRONT_LEFT); if (ptex) { - struct pipe_resource *psrc = gsurf->render_texture; struct pipe_box src_box; + u_box_origin_2d(ptex->width0, ptex->height0, &src_box); - if (psrc) { - gdpy->pipe->resource_copy_region(gdpy->pipe, ptex, 0, 0, 0, 0, - gsurf->render_texture, 0, &src_box); - nsurf->present(nsurf, NATIVE_ATTACHMENT_FRONT_LEFT, FALSE, 0); - } + gdpy->pipe->resource_copy_region(gdpy->pipe, ptex, 0, 0, 0, 0, + gsurf->render_texture, 0, &src_box); + gdpy->pipe->flush(gdpy->pipe, PIPE_FLUSH_RENDER_CACHE, NULL); + nsurf->present(nsurf, NATIVE_ATTACHMENT_FRONT_LEFT, FALSE, 0); pipe_resource_reference(&ptex, NULL); } @@ -888,25 +876,6 @@ egl_g3d_show_screen_surface(_EGLDriver *drv, _EGLDisplay *dpy, #endif /* EGL_MESA_screen_surface */ -/** - * Find a config that supports the pixmap. - */ -_EGLConfig * -egl_g3d_find_pixmap_config(_EGLDisplay *dpy, EGLNativePixmapType pix) -{ - struct egl_g3d_display *gdpy = egl_g3d_display(dpy); - struct egl_g3d_config *gconf; - EGLint i; - - for (i = 0; i < dpy->Configs->Size; i++) { - gconf = egl_g3d_config((_EGLConfig *) dpy->Configs->Elements[i]); - if (gdpy->native->is_pixmap_supported(gdpy->native, pix, gconf->native)) - break; - } - - return (i < dpy->Configs->Size) ? &gconf->base : NULL; -} - void egl_g3d_init_driver_api(_EGLDriver *drv) { diff --git a/src/gallium/state_trackers/egl/common/egl_g3d_api.h b/src/gallium/state_trackers/egl/common/egl_g3d_api.h index d5196c12fe9..17fd7958aa6 100644 --- a/src/gallium/state_trackers/egl/common/egl_g3d_api.h +++ b/src/gallium/state_trackers/egl/common/egl_g3d_api.h @@ -31,7 +31,4 @@ void egl_g3d_init_driver_api(_EGLDriver *drv); -_EGLConfig * -egl_g3d_find_pixmap_config(_EGLDisplay *dpy, EGLNativePixmapType pix); - #endif /* _EGL_G3D_API_H_ */ diff --git a/src/gallium/state_trackers/egl/common/egl_g3d_image.c b/src/gallium/state_trackers/egl/common/egl_g3d_image.c index b2d6b433c5e..e6ff100de04 100644 --- a/src/gallium/state_trackers/egl/common/egl_g3d_image.c +++ b/src/gallium/state_trackers/egl/common/egl_g3d_image.c @@ -48,17 +48,11 @@ static struct pipe_resource * egl_g3d_reference_native_pixmap(_EGLDisplay *dpy, EGLNativePixmapType pix) { struct egl_g3d_display *gdpy = egl_g3d_display(dpy); - struct egl_g3d_config *gconf; struct native_surface *nsurf; struct pipe_resource *textures[NUM_NATIVE_ATTACHMENTS]; enum native_attachment natt; - gconf = egl_g3d_config(egl_g3d_find_pixmap_config(dpy, pix)); - if (!gconf) - return NULL; - - nsurf = gdpy->native->create_pixmap_surface(gdpy->native, - pix, gconf->native); + nsurf = gdpy->native->create_pixmap_surface(gdpy->native, pix, NULL); if (!nsurf) return NULL; diff --git a/src/gallium/state_trackers/egl/common/native.h b/src/gallium/state_trackers/egl/common/native.h index 3886ca20562..654f445fca6 100644 --- a/src/gallium/state_trackers/egl/common/native.h +++ b/src/gallium/state_trackers/egl/common/native.h @@ -126,8 +126,6 @@ struct native_config { int native_visual_id; int native_visual_type; int level; - int samples; - boolean slow_config; boolean transparent_rgb; int transparent_rgb_values[3]; }; @@ -185,7 +183,9 @@ struct native_display { const struct native_config *nconf); /** - * Create a pixmap surface. Required unless no config has pixmap_bit set. + * Create a pixmap surface. The native config may be NULL. In that case, a + * "best config" will be picked. Required unless no config has pixmap_bit + * set. */ struct native_surface *(*create_pixmap_surface)(struct native_display *ndpy, EGLNativePixmapType pix, diff --git a/src/gallium/state_trackers/egl/gdi/native_gdi.c b/src/gallium/state_trackers/egl/gdi/native_gdi.c index d259e6edc89..2d0450604c6 100644 --- a/src/gallium/state_trackers/egl/gdi/native_gdi.c +++ b/src/gallium/state_trackers/egl/gdi/native_gdi.c @@ -319,7 +319,6 @@ gdi_display_get_configs(struct native_display *ndpy, int *num_configs) nconf->color_format = formats[i]; nconf->window_bit = TRUE; - nconf->slow_config = TRUE; } gdpy->num_configs = count; diff --git a/src/gallium/state_trackers/egl/x11/native_dri2.c b/src/gallium/state_trackers/egl/x11/native_dri2.c index 8108ce45865..adcc3b39d68 100644 --- a/src/gallium/state_trackers/egl/x11/native_dri2.c +++ b/src/gallium/state_trackers/egl/x11/native_dri2.c @@ -40,11 +40,6 @@ #ifdef GLX_DIRECT_RENDERING -enum dri2_surface_type { - DRI2_SURFACE_TYPE_WINDOW, - DRI2_SURFACE_TYPE_PIXMAP, -}; - struct dri2_display { struct native_display base; Display *dpy; @@ -66,7 +61,6 @@ struct dri2_display { struct dri2_surface { struct native_surface base; Drawable drawable; - enum dri2_surface_type type; enum pipe_format color_format; struct dri2_display *dri2dpy; @@ -439,12 +433,10 @@ dri2_surface_destroy(struct native_surface *nsurf) static struct dri2_surface * dri2_display_create_surface(struct native_display *ndpy, - enum dri2_surface_type type, Drawable drawable, - const struct native_config *nconf) + enum pipe_format color_format) { struct dri2_display *dri2dpy = dri2_display(ndpy); - struct dri2_config *dri2conf = dri2_config(nconf); struct dri2_surface *dri2surf; dri2surf = CALLOC_STRUCT(dri2_surface); @@ -452,9 +444,8 @@ dri2_display_create_surface(struct native_display *ndpy, return NULL; dri2surf->dri2dpy = dri2dpy; - dri2surf->type = type; dri2surf->drawable = drawable; - dri2surf->color_format = dri2conf->base.color_format; + dri2surf->color_format = color_format; dri2surf->base.destroy = dri2_surface_destroy; dri2surf->base.present = dri2_surface_present; @@ -480,8 +471,8 @@ dri2_display_create_window_surface(struct native_display *ndpy, { struct dri2_surface *dri2surf; - dri2surf = dri2_display_create_surface(ndpy, DRI2_SURFACE_TYPE_WINDOW, - (Drawable) win, nconf); + dri2surf = dri2_display_create_surface(ndpy, + (Drawable) win, nconf->color_format); return (dri2surf) ? &dri2surf->base : NULL; } @@ -492,8 +483,29 @@ dri2_display_create_pixmap_surface(struct native_display *ndpy, { struct dri2_surface *dri2surf; - dri2surf = dri2_display_create_surface(ndpy, DRI2_SURFACE_TYPE_PIXMAP, - (Drawable) pix, nconf); + if (!nconf) { + struct dri2_display *dri2dpy = dri2_display(ndpy); + uint depth, nconf_depth; + int i; + + depth = x11_drawable_get_depth(dri2dpy->xscr, (Drawable) pix); + for (i = 0; i < dri2dpy->num_configs; i++) { + nconf_depth = util_format_get_blocksizebits( + dri2dpy->configs[i].base.color_format); + /* simple depth match for now */ + if (depth == nconf_depth || + (depth == 24 && depth + 8 == nconf_depth)) { + nconf = &dri2dpy->configs[i].base; + break; + } + } + + if (!nconf) + return NULL; + } + + dri2surf = dri2_display_create_surface(ndpy, + (Drawable) pix, nconf->color_format); return (dri2surf) ? &dri2surf->base : NULL; } @@ -548,6 +560,10 @@ dri2_display_convert_config(struct native_display *ndpy, if (!mode->xRenderable || !mode->drawableType) return FALSE; + /* fast/slow configs are probably not relevant */ + if (mode->visualRating == GLX_SLOW_CONFIG) + return FALSE; + nconf->buffer_mask = 1 << NATIVE_ATTACHMENT_FRONT_LEFT; if (mode->doubleBufferMode) nconf->buffer_mask |= 1 << NATIVE_ATTACHMENT_BACK_LEFT; @@ -568,17 +584,33 @@ dri2_display_convert_config(struct native_display *ndpy, if (nconf->color_format == PIPE_FORMAT_NONE) return FALSE; - if (mode->drawableType & GLX_WINDOW_BIT) + if ((mode->drawableType & GLX_WINDOW_BIT) && mode->visualID) nconf->window_bit = TRUE; if (mode->drawableType & GLX_PIXMAP_BIT) nconf->pixmap_bit = TRUE; nconf->native_visual_id = mode->visualID; - nconf->native_visual_type = mode->visualType; + switch (mode->visualType) { + case GLX_TRUE_COLOR: + nconf->native_visual_type = TrueColor; + break; + case GLX_DIRECT_COLOR: + nconf->native_visual_type = DirectColor; + break; + case GLX_PSEUDO_COLOR: + nconf->native_visual_type = PseudoColor; + break; + case GLX_STATIC_COLOR: + nconf->native_visual_type = StaticColor; + break; + case GLX_GRAY_SCALE: + nconf->native_visual_type = GrayScale; + break; + case GLX_STATIC_GRAY: + nconf->native_visual_type = StaticGray; + break; + } nconf->level = mode->level; - nconf->samples = mode->samples; - - nconf->slow_config = (mode->visualRating == GLX_SLOW_CONFIG); if (mode->transparentPixel == GLX_TRANSPARENT_RGB) { nconf->transparent_rgb = TRUE; @@ -614,8 +646,17 @@ dri2_display_get_configs(struct native_display *ndpy, int *num_configs) count = 0; for (i = 0; i < num_modes; i++) { struct native_config *nconf = &dri2dpy->configs[count].base; - if (dri2_display_convert_config(&dri2dpy->base, modes, nconf)) - count++; + + if (dri2_display_convert_config(&dri2dpy->base, modes, nconf)) { + int j; + /* look for duplicates */ + for (j = 0; j < count; j++) { + if (memcmp(&dri2dpy->configs[j], nconf, sizeof(*nconf)) == 0) + break; + } + if (j == count) + count++; + } modes = modes->next; } diff --git a/src/gallium/state_trackers/egl/x11/native_ximage.c b/src/gallium/state_trackers/egl/x11/native_ximage.c index 84811fb6e14..d4f4dd04df4 100644 --- a/src/gallium/state_trackers/egl/x11/native_ximage.c +++ b/src/gallium/state_trackers/egl/x11/native_ximage.c @@ -38,11 +38,6 @@ #include "native_x11.h" #include "x11_screen.h" -enum ximage_surface_type { - XIMAGE_SURFACE_TYPE_WINDOW, - XIMAGE_SURFACE_TYPE_PIXMAP, -}; - struct ximage_display { struct native_display base; Display *dpy; @@ -60,7 +55,6 @@ struct ximage_display { struct ximage_surface { struct native_surface base; Drawable drawable; - enum ximage_surface_type type; enum pipe_format color_format; XVisualInfo visual; struct ximage_display *xdpy; @@ -245,7 +239,6 @@ ximage_surface_destroy(struct native_surface *nsurf) static struct ximage_surface * ximage_display_create_surface(struct native_display *ndpy, - enum ximage_surface_type type, Drawable drawable, const struct native_config *nconf) { @@ -258,7 +251,6 @@ ximage_display_create_surface(struct native_display *ndpy, return NULL; xsurf->xdpy = xdpy; - xsurf->type = type; xsurf->color_format = xconf->base.color_format; xsurf->drawable = drawable; @@ -297,11 +289,37 @@ ximage_display_create_window_surface(struct native_display *ndpy, { struct ximage_surface *xsurf; - xsurf = ximage_display_create_surface(ndpy, XIMAGE_SURFACE_TYPE_WINDOW, - (Drawable) win, nconf); + xsurf = ximage_display_create_surface(ndpy, (Drawable) win, nconf); return (xsurf) ? &xsurf->base : NULL; } +static enum pipe_format +get_pixmap_format(struct native_display *ndpy, EGLNativePixmapType pix) +{ + struct ximage_display *xdpy = ximage_display(ndpy); + enum pipe_format fmt; + uint depth; + + depth = x11_drawable_get_depth(xdpy->xscr, (Drawable) pix); + + switch (depth) { + case 32: + fmt = PIPE_FORMAT_B8G8R8A8_UNORM; + break; + case 24: + fmt = PIPE_FORMAT_B8G8R8X8_UNORM; + break; + case 16: + fmt = PIPE_FORMAT_B5G6R5_UNORM; + break; + default: + fmt = PIPE_FORMAT_NONE; + break; + } + + return fmt; +} + static struct native_surface * ximage_display_create_pixmap_surface(struct native_display *ndpy, EGLNativePixmapType pix, @@ -309,8 +327,26 @@ ximage_display_create_pixmap_surface(struct native_display *ndpy, { struct ximage_surface *xsurf; - xsurf = ximage_display_create_surface(ndpy, XIMAGE_SURFACE_TYPE_PIXMAP, - (Drawable) pix, nconf); + /* find the config */ + if (!nconf) { + struct ximage_display *xdpy = ximage_display(ndpy); + enum pipe_format fmt = get_pixmap_format(&xdpy->base, pix); + int i; + + if (fmt != PIPE_FORMAT_NONE) { + for (i = 0; i < xdpy->num_configs; i++) { + if (xdpy->configs[i].base.color_format == fmt) { + nconf = &xdpy->configs[i].base; + break; + } + } + } + + if (!nconf) + return NULL; + } + + xsurf = ximage_display_create_surface(ndpy, (Drawable) pix, nconf); return (xsurf) ? &xsurf->base : NULL; } @@ -384,8 +420,6 @@ ximage_display_get_configs(struct native_display *ndpy, int *num_configs) xconf->base.native_visual_type = xconf->visual->class; #endif - xconf->base.slow_config = TRUE; - count++; } @@ -408,24 +442,7 @@ ximage_display_is_pixmap_supported(struct native_display *ndpy, const struct native_config *nconf) { struct ximage_display *xdpy = ximage_display(ndpy); - enum pipe_format fmt; - uint depth; - - depth = x11_drawable_get_depth(xdpy->xscr, (Drawable) pix); - switch (depth) { - case 32: - fmt = PIPE_FORMAT_B8G8R8A8_UNORM; - break; - case 24: - fmt = PIPE_FORMAT_B8G8R8X8_UNORM; - break; - case 16: - fmt = PIPE_FORMAT_B5G6R5_UNORM; - break; - default: - fmt = PIPE_FORMAT_NONE; - break; - } + enum pipe_format fmt = get_pixmap_format(&xdpy->base, pix); return (fmt == nconf->color_format); } diff --git a/src/gallium/state_trackers/python/p_context.i b/src/gallium/state_trackers/python/p_context.i index 5bdeaa8e4ea..d694651eef7 100644 --- a/src/gallium/state_trackers/python/p_context.i +++ b/src/gallium/state_trackers/python/p_context.i @@ -360,7 +360,7 @@ struct st_context { if (!map) goto error2; memcpy(map, vertices, size); - pipe_buffer_unmap(pipe, vbuf, transfer); + pipe_buffer_unmap(pipe, transfer); cso_save_vertex_elements($self->cso); diff --git a/src/gallium/state_trackers/vega/api_masks.c b/src/gallium/state_trackers/vega/api_masks.c index beb15c33a59..d96afe66387 100644 --- a/src/gallium/state_trackers/vega/api_masks.c +++ b/src/gallium/state_trackers/vega/api_masks.c @@ -32,10 +32,6 @@ #include "vg_context.h" #include "pipe/p_context.h" -#include "util/u_inlines.h" - -#include "util/u_pack_color.h" -#include "util/u_draw_quad.h" void vegaMask(VGHandle mask, VGMaskOperation operation, VGint x, VGint y, diff --git a/src/gallium/state_trackers/vega/renderer.c b/src/gallium/state_trackers/vega/renderer.c index 7871c516c41..e42bad76492 100644 --- a/src/gallium/state_trackers/vega/renderer.c +++ b/src/gallium/state_trackers/vega/renderer.c @@ -28,7 +28,6 @@ #include "renderer.h" #include "vg_context.h" -#include "image.h" #include "pipe/p_context.h" #include "pipe/p_state.h" diff --git a/src/gallium/state_trackers/vega/shader.c b/src/gallium/state_trackers/vega/shader.c index 0ed721376cf..2a6bae8a630 100644 --- a/src/gallium/state_trackers/vega/shader.c +++ b/src/gallium/state_trackers/vega/shader.c @@ -34,9 +34,7 @@ #include "renderer.h" #include "pipe/p_context.h" -#include "pipe/p_screen.h" #include "pipe/p_state.h" -#include "util/u_inlines.h" #include "util/u_memory.h" #include "util/u_math.h" #include "util/u_format.h" diff --git a/src/gallium/state_trackers/vega/text.c b/src/gallium/state_trackers/vega/text.c index 01ff602f58a..6714ee9ad35 100644 --- a/src/gallium/state_trackers/vega/text.c +++ b/src/gallium/state_trackers/vega/text.c @@ -30,7 +30,6 @@ #include "text.h" #include "image.h" #include "path.h" -#include "api.h" #ifdef OPENVG_VERSION_1_1 diff --git a/src/gallium/state_trackers/vega/vg_context.c b/src/gallium/state_trackers/vega/vg_context.c index 0844012cc3b..f36f55d6c89 100644 --- a/src/gallium/state_trackers/vega/vg_context.c +++ b/src/gallium/state_trackers/vega/vg_context.c @@ -30,7 +30,6 @@ #include "renderer.h" #include "shaders_cache.h" #include "shader.h" -#include "asm_util.h" #include "vg_manager.h" #include "api.h" #include "mask.h" @@ -40,7 +39,6 @@ #include "cso_cache/cso_context.h" -#include "util/u_simple_shaders.h" #include "util/u_memory.h" #include "util/u_blit.h" #include "util/u_sampler.h" diff --git a/src/gallium/state_trackers/vega/vg_manager.c b/src/gallium/state_trackers/vega/vg_manager.c index de935768b22..ec713b7fb1d 100644 --- a/src/gallium/state_trackers/vega/vg_manager.c +++ b/src/gallium/state_trackers/vega/vg_manager.c @@ -33,14 +33,12 @@ #include "pipe/p_screen.h" #include "util/u_memory.h" #include "util/u_inlines.h" -#include "util/u_sampler.h" #include "util/u_box.h" #include "util/u_surface.h" #include "vg_api.h" #include "vg_manager.h" #include "vg_context.h" -#include "image.h" #include "api.h" static boolean diff --git a/src/gallium/winsys/r600/drm/Makefile b/src/gallium/winsys/r600/drm/Makefile index 91c65012c83..7310734f051 100644 --- a/src/gallium/winsys/r600/drm/Makefile +++ b/src/gallium/winsys/r600/drm/Makefile @@ -9,7 +9,6 @@ C_SOURCES = \ evergreen_hw_context.c \ radeon_bo.c \ radeon_pciid.c \ - r600.c \ r600_bo.c \ r600_drm.c \ r600_hw_context.c \ diff --git a/src/gallium/winsys/r600/drm/SConscript b/src/gallium/winsys/r600/drm/SConscript index dac0097f144..f97434e995d 100644 --- a/src/gallium/winsys/r600/drm/SConscript +++ b/src/gallium/winsys/r600/drm/SConscript @@ -7,7 +7,6 @@ r600_sources = [ 'evergreen_hw_context.c', 'radeon_bo.c', 'radeon_pciid.c', - 'r600.c', 'r600_bo.c', 'r600_drm.c', 'r600_hw_context.c', diff --git a/src/gallium/winsys/r600/drm/evergreen_hw_context.c b/src/gallium/winsys/r600/drm/evergreen_hw_context.c index 2175d578ec7..47d73c2e094 100644 --- a/src/gallium/winsys/r600/drm/evergreen_hw_context.c +++ b/src/gallium/winsys/r600/drm/evergreen_hw_context.c @@ -880,59 +880,3 @@ void evergreen_context_draw(struct r600_context *ctx, const struct r600_draw *dr ctx->pm4_dirty_cdwords = 0; } -static inline void evergreen_resource_set(struct r600_context *ctx, struct r600_pipe_state *state, unsigned offset) -{ - struct r600_range *range; - struct r600_block *block; - - range = &ctx->range[CTX_RANGE_ID(ctx, offset)]; - block = range->blocks[CTX_BLOCK_ID(ctx, offset)]; - block->reg[0] = state->regs[0].value; - block->reg[1] = state->regs[1].value; - block->reg[2] = state->regs[2].value; - block->reg[3] = state->regs[3].value; - block->reg[4] = state->regs[4].value; - block->reg[5] = state->regs[5].value; - block->reg[6] = state->regs[6].value; - block->reg[7] = state->regs[7].value; - r600_bo_reference(ctx->radeon, &block->reloc[1].bo, NULL); - r600_bo_reference(ctx->radeon , &block->reloc[2].bo, NULL); - if (state->regs[0].bo) { - /* VERTEX RESOURCE, we preted there is 2 bo to relocate so - * we have single case btw VERTEX & TEXTURE resource - */ - r600_bo_reference(ctx->radeon, &block->reloc[1].bo, state->regs[0].bo); - r600_bo_reference(ctx->radeon, &block->reloc[2].bo, state->regs[0].bo); - } else { - /* TEXTURE RESOURCE */ - r600_bo_reference(ctx->radeon, &block->reloc[1].bo, state->regs[2].bo); - r600_bo_reference(ctx->radeon, &block->reloc[2].bo, state->regs[3].bo); - } - if (!(block->status & R600_BLOCK_STATUS_DIRTY)) { - block->status |= R600_BLOCK_STATUS_ENABLED; - block->status |= R600_BLOCK_STATUS_DIRTY; - ctx->pm4_dirty_cdwords += block->pm4_ndwords + block->pm4_flush_ndwords; - LIST_ADDTAIL(&block->list,&ctx->dirty); - } -} - -void evergreen_ps_resource_set(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid) -{ - unsigned offset = R_030000_RESOURCE0_WORD0 + 0x20 * rid; - - evergreen_resource_set(ctx, state, offset); -} - -void evergreen_vs_resource_set(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid) -{ - unsigned offset = R_030000_RESOURCE0_WORD0 + 0x1600 + 0x20 * rid; - - evergreen_resource_set(ctx, state, offset); -} - -void evergreen_fs_resource_set(struct r600_context *ctx, struct r600_pipe_state *state, unsigned rid) -{ - unsigned offset = R_030000_RESOURCE0_WORD0 + 0x7C00 + 0x20 * rid; - - evergreen_resource_set(ctx, state, offset); -} diff --git a/src/gallium/winsys/r600/drm/r600.c b/src/gallium/winsys/r600/drm/r600.c deleted file mode 100644 index b88733f80f1..00000000000 --- a/src/gallium/winsys/r600/drm/r600.c +++ /dev/null @@ -1,174 +0,0 @@ -/* - * Copyright 2010 Jerome Glisse <[email protected]> - * - * 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 - * on 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 - * THE AUTHOR(S) AND/OR THEIR 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. - * - * Authors: - * Jerome Glisse - */ -#include "xf86drm.h" -#include "radeon_drm.h" -#include "pipe/p_compiler.h" -#include "util/u_inlines.h" -#include "r600_priv.h" - -enum radeon_family r600_get_family(struct radeon *r600) -{ - return r600->family; -} - -enum chip_class r600_get_family_class(struct radeon *radeon) -{ - return radeon->chip_class; -} - -struct r600_tiling_info *r600_get_tiling_info(struct radeon *radeon) -{ - return &radeon->tiling_info; -} - -static int r600_get_device(struct radeon *r600) -{ - struct drm_radeon_info info; - - r600->device = 0; - info.request = RADEON_INFO_DEVICE_ID; - info.value = (uintptr_t)&r600->device; - return drmCommandWriteRead(r600->fd, DRM_RADEON_INFO, &info, sizeof(struct drm_radeon_info)); -} - -struct radeon *r600_new(int fd, unsigned device) -{ - struct radeon *r600; - int r; - - r600 = calloc(1, sizeof(*r600)); - if (r600 == NULL) { - return NULL; - } - r600->fd = fd; - r600->device = device; - if (fd >= 0) { - r = r600_get_device(r600); - if (r) { - R600_ERR("Failed to get device id\n"); - r600_delete(r600); - return NULL; - } - } - r600->family = radeon_family_from_device(r600->device); - if (r600->family == CHIP_UNKNOWN) { - R600_ERR("Unknown chipset 0x%04X\n", r600->device); - r600_delete(r600); - return NULL; - } - switch (r600->family) { - case CHIP_R600: - case CHIP_RV610: - case CHIP_RV630: - case CHIP_RV670: - case CHIP_RV620: - case CHIP_RV635: - case CHIP_RS780: - case CHIP_RS880: - case CHIP_RV770: - case CHIP_RV730: - case CHIP_RV710: - case CHIP_RV740: - case CHIP_CEDAR: - case CHIP_REDWOOD: - case CHIP_JUNIPER: - case CHIP_CYPRESS: - case CHIP_HEMLOCK: - case CHIP_PALM: - break; - case CHIP_R100: - case CHIP_RV100: - case CHIP_RS100: - case CHIP_RV200: - case CHIP_RS200: - case CHIP_R200: - case CHIP_RV250: - case CHIP_RS300: - case CHIP_RV280: - case CHIP_R300: - case CHIP_R350: - case CHIP_RV350: - case CHIP_RV380: - case CHIP_R420: - case CHIP_R423: - case CHIP_RV410: - case CHIP_RS400: - case CHIP_RS480: - case CHIP_RS600: - case CHIP_RS690: - case CHIP_RS740: - case CHIP_RV515: - case CHIP_R520: - case CHIP_RV530: - case CHIP_RV560: - case CHIP_RV570: - case CHIP_R580: - default: - R600_ERR("unknown or unsupported chipset 0x%04X\n", r600->device); - break; - } - - /* setup class */ - switch (r600->family) { - case CHIP_R600: - case CHIP_RV610: - case CHIP_RV630: - case CHIP_RV670: - case CHIP_RV620: - case CHIP_RV635: - case CHIP_RS780: - case CHIP_RS880: - r600->chip_class = R600; - break; - case CHIP_RV770: - case CHIP_RV730: - case CHIP_RV710: - case CHIP_RV740: - r600->chip_class = R700; - break; - case CHIP_CEDAR: - case CHIP_REDWOOD: - case CHIP_JUNIPER: - case CHIP_CYPRESS: - case CHIP_HEMLOCK: - case CHIP_PALM: - r600->chip_class = EVERGREEN; - break; - default: - R600_ERR("unknown or unsupported chipset 0x%04X\n", r600->device); - break; - } - - return r600; -} - -void r600_delete(struct radeon *r600) -{ - if (r600 == NULL) - return; - drmClose(r600->fd); - free(r600); -} diff --git a/src/gallium/winsys/r600/drm/r600_drm.c b/src/gallium/winsys/r600/drm/r600_drm.c index 3cbbf91878d..58aacb77c9e 100644 --- a/src/gallium/winsys/r600/drm/r600_drm.c +++ b/src/gallium/winsys/r600/drm/r600_drm.c @@ -42,6 +42,21 @@ static struct radeon *radeon_new(int fd, unsigned device); +enum radeon_family r600_get_family(struct radeon *r600) +{ + return r600->family; +} + +enum chip_class r600_get_family_class(struct radeon *radeon) +{ + return radeon->chip_class; +} + +struct r600_tiling_info *r600_get_tiling_info(struct radeon *radeon) +{ + return &radeon->tiling_info; +} + static int radeon_get_device(struct radeon *radeon) { struct drm_radeon_info info; @@ -134,59 +149,6 @@ static struct radeon *radeon_new(int fd, unsigned device) fprintf(stderr, "Unknown chipset 0x%04X\n", radeon->device); return radeon_decref(radeon); } - switch (radeon->family) { - case CHIP_R600: - case CHIP_RV610: - case CHIP_RV630: - case CHIP_RV670: - case CHIP_RV620: - case CHIP_RV635: - case CHIP_RS780: - case CHIP_RS880: - case CHIP_RV770: - case CHIP_RV730: - case CHIP_RV710: - case CHIP_RV740: - case CHIP_CEDAR: - case CHIP_REDWOOD: - case CHIP_JUNIPER: - case CHIP_CYPRESS: - case CHIP_HEMLOCK: - case CHIP_PALM: - break; - case CHIP_R100: - case CHIP_RV100: - case CHIP_RS100: - case CHIP_RV200: - case CHIP_RS200: - case CHIP_R200: - case CHIP_RV250: - case CHIP_RS300: - case CHIP_RV280: - case CHIP_R300: - case CHIP_R350: - case CHIP_RV350: - case CHIP_RV380: - case CHIP_R420: - case CHIP_R423: - case CHIP_RV410: - case CHIP_RS400: - case CHIP_RS480: - case CHIP_RS600: - case CHIP_RS690: - case CHIP_RS740: - case CHIP_RV515: - case CHIP_R520: - case CHIP_RV530: - case CHIP_RV560: - case CHIP_RV570: - case CHIP_R580: - default: - fprintf(stderr, "%s unknown or unsupported chipset 0x%04X\n", - __func__, radeon->device); - break; - } - /* setup class */ switch (radeon->family) { case CHIP_R600: diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_common.c b/src/gallium/winsys/radeon/drm/radeon_drm_common.c index 6bc6244115c..f38ab6e3b43 100644 --- a/src/gallium/winsys/radeon/drm/radeon_drm_common.c +++ b/src/gallium/winsys/radeon/drm/radeon_drm_common.c @@ -122,6 +122,10 @@ static void do_ioctls(struct radeon_drm_winsys *winsys) (version->version_major == 2 && version->version_minor >= 6); + winsys->drm_2_8_0 = version->version_major > 2 || + (version->version_major == 2 && + version->version_minor >= 8); + info.request = RADEON_INFO_DEVICE_ID; retval = drmCommandWriteRead(winsys->fd, DRM_RADEON_INFO, &info, sizeof(info)); if (retval) { diff --git a/src/gallium/winsys/radeon/drm/radeon_r300.c b/src/gallium/winsys/radeon/drm/radeon_r300.c index 9f59b3de461..ae7020a0639 100644 --- a/src/gallium/winsys/radeon/drm/radeon_r300.c +++ b/src/gallium/winsys/radeon/drm/radeon_r300.c @@ -211,6 +211,8 @@ static uint32_t radeon_get_value(struct r300_winsys_screen *rws, return ws->drm_2_3_0; case R300_VID_DRM_2_6_0: return ws->drm_2_6_0; + case R300_VID_DRM_2_8_0: + return ws->drm_2_8_0; case R300_CAN_HYPERZ: return ws->hyperz; } diff --git a/src/gallium/winsys/radeon/drm/radeon_winsys.h b/src/gallium/winsys/radeon/drm/radeon_winsys.h index 81da1a25e0f..6f232143f6a 100644 --- a/src/gallium/winsys/radeon/drm/radeon_winsys.h +++ b/src/gallium/winsys/radeon/drm/radeon_winsys.h @@ -49,8 +49,10 @@ struct radeon_drm_winsys { boolean squaretiling; /* Square tiling support. */ /* DRM 2.3.0 (R500 VAP regs, MSPOS regs, fixed tex3D size checking) */ boolean drm_2_3_0; - /* DRM 2.6.0 (Hyper-Z, GB_Z_PEQ_CONFIG allowed on rv350->r4xx) */ + /* DRM 2.6.0 (Hyper-Z, GB_Z_PEQ_CONFIG allowed on rv350->r4xx, FG_ALPHA_VALUE) */ boolean drm_2_6_0; + /* DRM 2.8.0 (US_FORMAT regs, ARGB2101010 colorbuffer) */ + boolean drm_2_8_0; /* Hyper-Z user */ boolean hyperz; diff --git a/src/glsl/Makefile b/src/glsl/Makefile index 2674c6ec485..86a577e0aeb 100644 --- a/src/glsl/Makefile +++ b/src/glsl/Makefile @@ -16,6 +16,7 @@ GLCPP_SOURCES = \ glcpp/glcpp.c C_SOURCES = \ + strtod.c \ $(LIBGLCPP_SOURCES) CXX_SOURCES = \ diff --git a/src/glsl/SConscript b/src/glsl/SConscript index b5b1728beef..f179721d527 100644 --- a/src/glsl/SConscript +++ b/src/glsl/SConscript @@ -76,6 +76,7 @@ sources = [ 'opt_swizzle_swizzle.cpp', 'opt_tree_grafting.cpp', 's_expression.cpp', + 'strtod.c', ] glsl = env.ConvenienceLibrary( diff --git a/src/glsl/ast_to_hir.cpp b/src/glsl/ast_to_hir.cpp index 1f4972cfca2..741cd19e9d3 100644 --- a/src/glsl/ast_to_hir.cpp +++ b/src/glsl/ast_to_hir.cpp @@ -598,17 +598,15 @@ ir_rvalue * validate_assignment(struct _mesa_glsl_parse_state *state, const glsl_type *lhs_type, ir_rvalue *rhs) { - const glsl_type *rhs_type = rhs->type; - /* If there is already some error in the RHS, just return it. Anything * else will lead to an avalanche of error message back to the user. */ - if (rhs_type->is_error()) + if (rhs->type->is_error()) return rhs; /* If the types are identical, the assignment can trivially proceed. */ - if (rhs_type == lhs_type) + if (rhs->type == lhs_type) return rhs; /* If the array element types are the same and the size of the LHS is zero, @@ -625,8 +623,7 @@ validate_assignment(struct _mesa_glsl_parse_state *state, /* Check for implicit conversion in GLSL 1.20 */ if (apply_implicit_conversion(lhs_type, rhs, state)) { - rhs_type = rhs->type; - if (rhs_type == lhs_type) + if (rhs->type == lhs_type) return rhs; } @@ -2242,6 +2239,17 @@ ast_declarator_list::hir(exec_list *instructions, if (this->type->qualifier.flags.q.constant) var->read_only = false; + /* Never emit code to initialize a uniform. + */ + const glsl_type *initializer_type; + if (!this->type->qualifier.flags.q.uniform) { + result = do_assignment(&initializer_instructions, state, + lhs, rhs, + this->get_location()); + initializer_type = result->type; + } else + initializer_type = rhs->type; + /* If the declared variable is an unsized array, it must inherrit * its full type from the initializer. A declaration such as * @@ -2256,16 +2264,14 @@ ast_declarator_list::hir(exec_list *instructions, * * If the declared variable is not an array, the types must * already match exactly. As a result, the type assignment - * here can be done unconditionally. + * here can be done unconditionally. For non-uniforms the call + * to do_assignment can change the type of the initializer (via + * the implicit conversion rules). For uniforms the initializer + * must be a constant expression, and the type of that expression + * was validated above. */ - var->type = rhs->type; + var->type = initializer_type; - /* Never emit code to initialize a uniform. - */ - if (!this->type->qualifier.flags.q.uniform) - result = do_assignment(&initializer_instructions, state, - lhs, rhs, - this->get_location()); var->read_only = temp; } } diff --git a/src/glsl/builtin_function.cpp b/src/glsl/builtin_function.cpp index 631aeac12ce..1c6d59d5a21 100644 --- a/src/glsl/builtin_function.cpp +++ b/src/glsl/builtin_function.cpp @@ -3010,40 +3010,26 @@ static const char builtin_smoothstep[] = " (declare (in) float edge1)\n" " (declare (in) float x))\n" " ((declare () float t)\n" - "\n" " (assign (constant bool (1)) (x) (var_ref t)\n" " (expression float max\n" " (expression float min\n" " (expression float / (expression float - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" " (constant float (1.0)))\n" " (constant float (0.0))))\n" - " (return (expression float * (var_ref t) (expression float * (var_ref t) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (var_ref t))))))))\n" - "\n" + " (return (expression float * (var_ref t) (expression float * (var_ref t) (expression float - (constant float (3.0)) (expression float * (constant float (2.0)) (var_ref t))))))))\n" " (signature vec2\n" " (parameters\n" " (declare (in) float edge0)\n" " (declare (in) float edge1)\n" " (declare (in) vec2 x))\n" " ((declare () vec2 t)\n" - " (declare () vec2 retval)\n" - "\n" - " (assign (constant bool (1)) (x) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" - " (constant float (1.0)))\n" - " (constant float (0.0))))\n" - " (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t)))))))\n" - "\n" - " (assign (constant bool (1)) (y) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" + " (assign (constant bool (1)) (xy) (var_ref t)\n" + " (expression vec2 max\n" + " (expression vec2 min\n" + " (expression vec2 / (expression vec2 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" " (constant float (1.0)))\n" " (constant float (0.0))))\n" - " (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t)))))))\n" - " (return (var_ref retval))\n" - " ))\n" + " (return (expression vec2 * (var_ref t) (expression vec2 * (var_ref t) (expression vec2 - (constant float (3.0)) (expression vec2 * (constant float (2.0)) (var_ref t))))))))\n" "\n" " (signature vec3\n" " (parameters\n" @@ -3051,33 +3037,13 @@ static const char builtin_smoothstep[] = " (declare (in) float edge1)\n" " (declare (in) vec3 x))\n" " ((declare () vec3 t)\n" - " (declare () vec3 retval)\n" - "\n" - " (assign (constant bool (1)) (x) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" - " (constant float (1.0)))\n" - " (constant float (0.0))))\n" - " (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t)))))))\n" - "\n" - " (assign (constant bool (1)) (y) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" - " (constant float (1.0)))\n" - " (constant float (0.0))))\n" - " (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t)))))))\n" - "\n" - " (assign (constant bool (1)) (z) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz z (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" + " (assign (constant bool (1)) (xyz) (var_ref t)\n" + " (expression vec3 max\n" + " (expression vec3 min\n" + " (expression vec3 / (expression vec3 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" " (constant float (1.0)))\n" " (constant float (0.0))))\n" - " (assign (constant bool (1)) (z) (var_ref retval) (expression float * (swiz z (var_ref t)) (expression float * (swiz z (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz z (var_ref t)))))))\n" - " (return (var_ref retval))\n" - " ))\n" + " (return (expression vec3 * (var_ref t) (expression vec3 * (var_ref t) (expression vec3 - (constant float (3.0)) (expression vec3 * (constant float (2.0)) (var_ref t))))))))\n" "\n" "\n" " (signature vec4\n" @@ -3086,74 +3052,55 @@ static const char builtin_smoothstep[] = " (declare (in) float edge1)\n" " (declare (in) vec4 x))\n" " ((declare () vec4 t)\n" - " (declare () vec4 retval)\n" - "\n" - " (assign (constant bool (1)) (x) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" - " (constant float (1.0)))\n" - " (constant float (0.0))))\n" - " (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t)))))))\n" - "\n" - " (assign (constant bool (1)) (y) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" - " (constant float (1.0)))\n" - " (constant float (0.0))))\n" - " (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t)))))))\n" - "\n" - " (assign (constant bool (1)) (z) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz z (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" - " (constant float (1.0)))\n" - " (constant float (0.0))))\n" - " (assign (constant bool (1)) (z) (var_ref retval) (expression float * (swiz z (var_ref t)) (expression float * (swiz z (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz z (var_ref t)))))))\n" - "\n" - " (assign (constant bool (1)) (w) (var_ref t)\n" - " (expression float max\n" - " (expression float min\n" - " (expression float / (expression float - (swiz w (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" + " (assign (constant bool (1)) (xyzw) (var_ref t)\n" + " (expression vec4 max\n" + " (expression vec4 min\n" + " (expression vec4 / (expression vec4 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))\n" " (constant float (1.0)))\n" " (constant float (0.0))))\n" - " (assign (constant bool (1)) (w) (var_ref retval) (expression float * (swiz w (var_ref t)) (expression float * (swiz w (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz w (var_ref t)))))))\n" - " (return (var_ref retval))\n" - " ))\n" + " (return (expression vec4 * (var_ref t) (expression vec4 * (var_ref t) (expression vec4 - (constant float (3.0)) (expression vec4 * (constant float (2.0)) (var_ref t))))))))\n" "\n" " (signature vec2\n" " (parameters\n" " (declare (in) vec2 edge0)\n" " (declare (in) vec2 edge1)\n" " (declare (in) vec2 x))\n" - " ((return (expression vec2 max\n" + " ((declare () vec2 t)\n" + " (assign (constant bool (1)) (xy) (var_ref t)\n" + " (expression vec2 max\n" " (expression vec2 min\n" " (expression vec2 / (expression vec2 - (var_ref x) (var_ref edge0)) (expression vec2 - (var_ref edge1) (var_ref edge0)))\n" - " (constant vec2 (1.0 1.0)))\n" - " (constant vec2 (0.0 0.0))))))\n" + " (constant float (1.0)))\n" + " (constant float (0.0))))\n" + " (return (expression vec2 * (var_ref t) (expression vec2 * (var_ref t) (expression vec2 - (constant float (3.0)) (expression vec2 * (constant float (2.0)) (var_ref t))))))))\n" "\n" " (signature vec3\n" " (parameters\n" " (declare (in) vec3 edge0)\n" " (declare (in) vec3 edge1)\n" " (declare (in) vec3 x))\n" - " ((return (expression vec3 max\n" + " ((declare () vec3 t)\n" + " (assign (constant bool (1)) (xyz) (var_ref t)\n" + " (expression vec3 max\n" " (expression vec3 min\n" " (expression vec3 / (expression vec3 - (var_ref x) (var_ref edge0)) (expression vec3 - (var_ref edge1) (var_ref edge0)))\n" - " (constant vec3 (1.0 1.0 1.0)))\n" - " (constant vec3 (0.0 0.0 0.0))))))\n" + " (constant float (1.0)))\n" + " (constant float (0.0))))\n" + " (return (expression vec3 * (var_ref t) (expression vec3 * (var_ref t) (expression vec3 - (constant float (3.0)) (expression vec3 * (constant float (2.0)) (var_ref t))))))))\n" "\n" " (signature vec4\n" " (parameters\n" " (declare (in) vec4 edge0)\n" " (declare (in) vec4 edge1)\n" " (declare (in) vec4 x))\n" - " ((return (expression vec4 max\n" + " ((declare () vec4 t)\n" + " (assign (constant bool (1)) (xyzw) (var_ref t)\n" + " (expression vec4 max\n" " (expression vec4 min\n" " (expression vec4 / (expression vec4 - (var_ref x) (var_ref edge0)) (expression vec4 - (var_ref edge1) (var_ref edge0)))\n" - " (constant vec4 (1.0 1.0 1.0 1.0)))\n" - " (constant vec4 (0.0 0.0 0.0 0.0))))))\n" + " (constant float (1.0)))\n" + " (constant float (0.0))))\n" + " (return (expression vec4 * (var_ref t) (expression vec4 * (var_ref t) (expression vec4 - (constant float (3.0)) (expression vec4 * (constant float (2.0)) (var_ref t))))))))\n" "))\n" "\n" "" diff --git a/src/glsl/builtin_types.h b/src/glsl/builtin_types.h index 443ae1606ed..8ccbf6e312f 100644 --- a/src/glsl/builtin_types.h +++ b/src/glsl/builtin_types.h @@ -24,10 +24,11 @@ const glsl_type glsl_type::_error_type = glsl_type(GL_INVALID_ENUM, GLSL_TYPE_ERROR, 0, 0, ""); -const glsl_type glsl_type::void_type = +const glsl_type glsl_type::_void_type = glsl_type(GL_INVALID_ENUM, GLSL_TYPE_VOID, 0, 0, "void"); const glsl_type *const glsl_type::error_type = & glsl_type::_error_type; +const glsl_type *const glsl_type::void_type = & glsl_type::_void_type; /** \name Core built-in types * diff --git a/src/glsl/builtins/ir/smoothstep b/src/glsl/builtins/ir/smoothstep index 0164219a056..b283f73d8d3 100644 --- a/src/glsl/builtins/ir/smoothstep +++ b/src/glsl/builtins/ir/smoothstep @@ -5,40 +5,26 @@ (declare (in) float edge1) (declare (in) float x)) ((declare () float t) - (assign (constant bool (1)) (x) (var_ref t) (expression float max (expression float min (expression float / (expression float - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) (constant float (1.0))) (constant float (0.0)))) - (return (expression float * (var_ref t) (expression float * (var_ref t) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (var_ref t)))))))) - + (return (expression float * (var_ref t) (expression float * (var_ref t) (expression float - (constant float (3.0)) (expression float * (constant float (2.0)) (var_ref t)))))))) (signature vec2 (parameters (declare (in) float edge0) (declare (in) float edge1) (declare (in) vec2 x)) ((declare () vec2 t) - (declare () vec2 retval) - - (assign (constant bool (1)) (x) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) + (assign (constant bool (1)) (xy) (var_ref t) + (expression vec2 max + (expression vec2 min + (expression vec2 / (expression vec2 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) (constant float (1.0))) (constant float (0.0)))) - (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t))))))) - - (assign (constant bool (1)) (y) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) - (constant float (1.0))) - (constant float (0.0)))) - (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t))))))) - (return (var_ref retval)) - )) + (return (expression vec2 * (var_ref t) (expression vec2 * (var_ref t) (expression vec2 - (constant float (3.0)) (expression vec2 * (constant float (2.0)) (var_ref t)))))))) (signature vec3 (parameters @@ -46,33 +32,13 @@ (declare (in) float edge1) (declare (in) vec3 x)) ((declare () vec3 t) - (declare () vec3 retval) - - (assign (constant bool (1)) (x) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) - (constant float (1.0))) - (constant float (0.0)))) - (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t))))))) - - (assign (constant bool (1)) (y) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) + (assign (constant bool (1)) (xyz) (var_ref t) + (expression vec3 max + (expression vec3 min + (expression vec3 / (expression vec3 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) (constant float (1.0))) (constant float (0.0)))) - (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t))))))) - - (assign (constant bool (1)) (z) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz z (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) - (constant float (1.0))) - (constant float (0.0)))) - (assign (constant bool (1)) (z) (var_ref retval) (expression float * (swiz z (var_ref t)) (expression float * (swiz z (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz z (var_ref t))))))) - (return (var_ref retval)) - )) + (return (expression vec3 * (var_ref t) (expression vec3 * (var_ref t) (expression vec3 - (constant float (3.0)) (expression vec3 * (constant float (2.0)) (var_ref t)))))))) (signature vec4 @@ -81,73 +47,54 @@ (declare (in) float edge1) (declare (in) vec4 x)) ((declare () vec4 t) - (declare () vec4 retval) - - (assign (constant bool (1)) (x) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) - (constant float (1.0))) - (constant float (0.0)))) - (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t))))))) - - (assign (constant bool (1)) (y) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) + (assign (constant bool (1)) (xyzw) (var_ref t) + (expression vec4 max + (expression vec4 min + (expression vec4 / (expression vec4 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) (constant float (1.0))) (constant float (0.0)))) - (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t))))))) - - (assign (constant bool (1)) (z) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz z (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) - (constant float (1.0))) - (constant float (0.0)))) - (assign (constant bool (1)) (z) (var_ref retval) (expression float * (swiz z (var_ref t)) (expression float * (swiz z (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz z (var_ref t))))))) - - (assign (constant bool (1)) (w) (var_ref t) - (expression float max - (expression float min - (expression float / (expression float - (swiz w (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0))) - (constant float (1.0))) - (constant float (0.0)))) - (assign (constant bool (1)) (w) (var_ref retval) (expression float * (swiz w (var_ref t)) (expression float * (swiz w (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz w (var_ref t))))))) - (return (var_ref retval)) - )) + (return (expression vec4 * (var_ref t) (expression vec4 * (var_ref t) (expression vec4 - (constant float (3.0)) (expression vec4 * (constant float (2.0)) (var_ref t)))))))) (signature vec2 (parameters (declare (in) vec2 edge0) (declare (in) vec2 edge1) (declare (in) vec2 x)) - ((return (expression vec2 max + ((declare () vec2 t) + (assign (constant bool (1)) (xy) (var_ref t) + (expression vec2 max (expression vec2 min (expression vec2 / (expression vec2 - (var_ref x) (var_ref edge0)) (expression vec2 - (var_ref edge1) (var_ref edge0))) - (constant vec2 (1.0 1.0))) - (constant vec2 (0.0 0.0)))))) + (constant float (1.0))) + (constant float (0.0)))) + (return (expression vec2 * (var_ref t) (expression vec2 * (var_ref t) (expression vec2 - (constant float (3.0)) (expression vec2 * (constant float (2.0)) (var_ref t)))))))) (signature vec3 (parameters (declare (in) vec3 edge0) (declare (in) vec3 edge1) (declare (in) vec3 x)) - ((return (expression vec3 max + ((declare () vec3 t) + (assign (constant bool (1)) (xyz) (var_ref t) + (expression vec3 max (expression vec3 min (expression vec3 / (expression vec3 - (var_ref x) (var_ref edge0)) (expression vec3 - (var_ref edge1) (var_ref edge0))) - (constant vec3 (1.0 1.0 1.0))) - (constant vec3 (0.0 0.0 0.0)))))) + (constant float (1.0))) + (constant float (0.0)))) + (return (expression vec3 * (var_ref t) (expression vec3 * (var_ref t) (expression vec3 - (constant float (3.0)) (expression vec3 * (constant float (2.0)) (var_ref t)))))))) (signature vec4 (parameters (declare (in) vec4 edge0) (declare (in) vec4 edge1) (declare (in) vec4 x)) - ((return (expression vec4 max + ((declare () vec4 t) + (assign (constant bool (1)) (xyzw) (var_ref t) + (expression vec4 max (expression vec4 min (expression vec4 / (expression vec4 - (var_ref x) (var_ref edge0)) (expression vec4 - (var_ref edge1) (var_ref edge0))) - (constant vec4 (1.0 1.0 1.0 1.0))) - (constant vec4 (0.0 0.0 0.0 0.0)))))) + (constant float (1.0))) + (constant float (0.0)))) + (return (expression vec4 * (var_ref t) (expression vec4 * (var_ref t) (expression vec4 - (constant float (3.0)) (expression vec4 * (constant float (2.0)) (var_ref t)))))))) )) diff --git a/src/glsl/glsl_lexer.cpp b/src/glsl/glsl_lexer.cpp index e3d0a3cfc9d..39c119001fd 100644 --- a/src/glsl/glsl_lexer.cpp +++ b/src/glsl/glsl_lexer.cpp @@ -1001,6 +1001,7 @@ static yyconst flex_int16_t yy_chk[1255] = * DEALINGS IN THE SOFTWARE. */ #include <ctype.h> +#include "strtod.h" #include "ast.h" #include "glsl_parser_extras.h" #include "glsl_parser.h" @@ -1054,7 +1055,7 @@ static yyconst flex_int16_t yy_chk[1255] = */ #define ES yyextra->es_shader -#line 1058 "glsl_lexer.cpp" +#line 1059 "glsl_lexer.cpp" #define INITIAL 0 #define PP 1 @@ -1141,10 +1142,6 @@ int _mesa_glsl_get_lineno (yyscan_t yyscanner ); void _mesa_glsl_set_lineno (int line_number ,yyscan_t yyscanner ); -int _mesa_glsl_get_column (yyscan_t yyscanner ); - -void _mesa_glsl_set_column (int column_no ,yyscan_t yyscanner ); - YYSTYPE * _mesa_glsl_get_lval (yyscan_t yyscanner ); void _mesa_glsl_set_lval (YYSTYPE * yylval_param ,yyscan_t yyscanner ); @@ -1292,10 +1289,10 @@ YY_DECL register int yy_act; struct yyguts_t * yyg = (struct yyguts_t*)yyscanner; -#line 94 "glsl_lexer.lpp" +#line 95 "glsl_lexer.lpp" -#line 1299 "glsl_lexer.cpp" +#line 1296 "glsl_lexer.cpp" yylval = yylval_param; @@ -1381,7 +1378,7 @@ do_action: /* This label is used only to access EOF actions. */ case 1: YY_RULE_SETUP -#line 96 "glsl_lexer.lpp" +#line 97 "glsl_lexer.lpp" ; YY_BREAK /* Preprocessor tokens. */ @@ -1390,17 +1387,17 @@ case 2: yyg->yy_c_buf_p = yy_cp -= 1; YY_DO_BEFORE_ACTION; /* set up yytext again */ YY_RULE_SETUP -#line 99 "glsl_lexer.lpp" +#line 100 "glsl_lexer.lpp" ; YY_BREAK case 3: YY_RULE_SETUP -#line 100 "glsl_lexer.lpp" +#line 101 "glsl_lexer.lpp" { BEGIN PP; return VERSION; } YY_BREAK case 4: YY_RULE_SETUP -#line 101 "glsl_lexer.lpp" +#line 102 "glsl_lexer.lpp" { BEGIN PP; return EXTENSION; } YY_BREAK case 5: @@ -1408,7 +1405,7 @@ case 5: yyg->yy_c_buf_p = yy_cp -= 1; YY_DO_BEFORE_ACTION; /* set up yytext again */ YY_RULE_SETUP -#line 102 "glsl_lexer.lpp" +#line 103 "glsl_lexer.lpp" { /* Eat characters until the first digit is * encountered @@ -1430,7 +1427,7 @@ case 6: yyg->yy_c_buf_p = yy_cp -= 1; YY_DO_BEFORE_ACTION; /* set up yytext again */ YY_RULE_SETUP -#line 117 "glsl_lexer.lpp" +#line 118 "glsl_lexer.lpp" { /* Eat characters until the first digit is * encountered @@ -1448,7 +1445,7 @@ YY_RULE_SETUP YY_BREAK case 7: YY_RULE_SETUP -#line 131 "glsl_lexer.lpp" +#line 132 "glsl_lexer.lpp" { BEGIN PP; return PRAGMA_DEBUG_ON; @@ -1456,7 +1453,7 @@ YY_RULE_SETUP YY_BREAK case 8: YY_RULE_SETUP -#line 135 "glsl_lexer.lpp" +#line 136 "glsl_lexer.lpp" { BEGIN PP; return PRAGMA_DEBUG_OFF; @@ -1464,7 +1461,7 @@ YY_RULE_SETUP YY_BREAK case 9: YY_RULE_SETUP -#line 139 "glsl_lexer.lpp" +#line 140 "glsl_lexer.lpp" { BEGIN PP; return PRAGMA_OPTIMIZE_ON; @@ -1472,7 +1469,7 @@ YY_RULE_SETUP YY_BREAK case 10: YY_RULE_SETUP -#line 143 "glsl_lexer.lpp" +#line 144 "glsl_lexer.lpp" { BEGIN PP; return PRAGMA_OPTIMIZE_OFF; @@ -1480,38 +1477,38 @@ YY_RULE_SETUP YY_BREAK case 11: YY_RULE_SETUP -#line 147 "glsl_lexer.lpp" +#line 148 "glsl_lexer.lpp" { BEGIN PRAGMA; } YY_BREAK case 12: /* rule 12 can match eol */ YY_RULE_SETUP -#line 149 "glsl_lexer.lpp" +#line 150 "glsl_lexer.lpp" { BEGIN 0; yylineno++; yycolumn = 0; } YY_BREAK case 13: YY_RULE_SETUP -#line 150 "glsl_lexer.lpp" +#line 151 "glsl_lexer.lpp" { } YY_BREAK case 14: YY_RULE_SETUP -#line 152 "glsl_lexer.lpp" +#line 153 "glsl_lexer.lpp" { } YY_BREAK case 15: YY_RULE_SETUP -#line 153 "glsl_lexer.lpp" +#line 154 "glsl_lexer.lpp" { } YY_BREAK case 16: YY_RULE_SETUP -#line 154 "glsl_lexer.lpp" +#line 155 "glsl_lexer.lpp" return COLON; YY_BREAK case 17: YY_RULE_SETUP -#line 155 "glsl_lexer.lpp" +#line 156 "glsl_lexer.lpp" { yylval->identifier = strdup(yytext); return IDENTIFIER; @@ -1519,7 +1516,7 @@ YY_RULE_SETUP YY_BREAK case 18: YY_RULE_SETUP -#line 159 "glsl_lexer.lpp" +#line 160 "glsl_lexer.lpp" { yylval->n = strtol(yytext, NULL, 10); return INTCONSTANT; @@ -1528,388 +1525,388 @@ YY_RULE_SETUP case 19: /* rule 19 can match eol */ YY_RULE_SETUP -#line 163 "glsl_lexer.lpp" +#line 164 "glsl_lexer.lpp" { BEGIN 0; yylineno++; yycolumn = 0; return EOL; } YY_BREAK case 20: /* rule 20 can match eol */ YY_RULE_SETUP -#line 165 "glsl_lexer.lpp" +#line 166 "glsl_lexer.lpp" { yylineno++; yycolumn = 0; } YY_BREAK case 21: YY_RULE_SETUP -#line 167 "glsl_lexer.lpp" +#line 168 "glsl_lexer.lpp" return ATTRIBUTE; YY_BREAK case 22: YY_RULE_SETUP -#line 168 "glsl_lexer.lpp" +#line 169 "glsl_lexer.lpp" return CONST_TOK; YY_BREAK case 23: YY_RULE_SETUP -#line 169 "glsl_lexer.lpp" +#line 170 "glsl_lexer.lpp" return BOOL_TOK; YY_BREAK case 24: YY_RULE_SETUP -#line 170 "glsl_lexer.lpp" +#line 171 "glsl_lexer.lpp" return FLOAT_TOK; YY_BREAK case 25: YY_RULE_SETUP -#line 171 "glsl_lexer.lpp" +#line 172 "glsl_lexer.lpp" return INT_TOK; YY_BREAK case 26: YY_RULE_SETUP -#line 172 "glsl_lexer.lpp" +#line 173 "glsl_lexer.lpp" KEYWORD(130, 130, UINT_TOK); YY_BREAK case 27: YY_RULE_SETUP -#line 174 "glsl_lexer.lpp" +#line 175 "glsl_lexer.lpp" return BREAK; YY_BREAK case 28: YY_RULE_SETUP -#line 175 "glsl_lexer.lpp" +#line 176 "glsl_lexer.lpp" return CONTINUE; YY_BREAK case 29: YY_RULE_SETUP -#line 176 "glsl_lexer.lpp" +#line 177 "glsl_lexer.lpp" return DO; YY_BREAK case 30: YY_RULE_SETUP -#line 177 "glsl_lexer.lpp" +#line 178 "glsl_lexer.lpp" return WHILE; YY_BREAK case 31: YY_RULE_SETUP -#line 178 "glsl_lexer.lpp" +#line 179 "glsl_lexer.lpp" return ELSE; YY_BREAK case 32: YY_RULE_SETUP -#line 179 "glsl_lexer.lpp" +#line 180 "glsl_lexer.lpp" return FOR; YY_BREAK case 33: YY_RULE_SETUP -#line 180 "glsl_lexer.lpp" +#line 181 "glsl_lexer.lpp" return IF; YY_BREAK case 34: YY_RULE_SETUP -#line 181 "glsl_lexer.lpp" +#line 182 "glsl_lexer.lpp" return DISCARD; YY_BREAK case 35: YY_RULE_SETUP -#line 182 "glsl_lexer.lpp" +#line 183 "glsl_lexer.lpp" return RETURN; YY_BREAK case 36: YY_RULE_SETUP -#line 184 "glsl_lexer.lpp" +#line 185 "glsl_lexer.lpp" return BVEC2; YY_BREAK case 37: YY_RULE_SETUP -#line 185 "glsl_lexer.lpp" +#line 186 "glsl_lexer.lpp" return BVEC3; YY_BREAK case 38: YY_RULE_SETUP -#line 186 "glsl_lexer.lpp" +#line 187 "glsl_lexer.lpp" return BVEC4; YY_BREAK case 39: YY_RULE_SETUP -#line 187 "glsl_lexer.lpp" +#line 188 "glsl_lexer.lpp" return IVEC2; YY_BREAK case 40: YY_RULE_SETUP -#line 188 "glsl_lexer.lpp" +#line 189 "glsl_lexer.lpp" return IVEC3; YY_BREAK case 41: YY_RULE_SETUP -#line 189 "glsl_lexer.lpp" +#line 190 "glsl_lexer.lpp" return IVEC4; YY_BREAK case 42: YY_RULE_SETUP -#line 190 "glsl_lexer.lpp" +#line 191 "glsl_lexer.lpp" KEYWORD(130, 130, UVEC2); YY_BREAK case 43: YY_RULE_SETUP -#line 191 "glsl_lexer.lpp" +#line 192 "glsl_lexer.lpp" KEYWORD(130, 130, UVEC3); YY_BREAK case 44: YY_RULE_SETUP -#line 192 "glsl_lexer.lpp" +#line 193 "glsl_lexer.lpp" KEYWORD(130, 130, UVEC4); YY_BREAK case 45: YY_RULE_SETUP -#line 193 "glsl_lexer.lpp" +#line 194 "glsl_lexer.lpp" return VEC2; YY_BREAK case 46: YY_RULE_SETUP -#line 194 "glsl_lexer.lpp" +#line 195 "glsl_lexer.lpp" return VEC3; YY_BREAK case 47: YY_RULE_SETUP -#line 195 "glsl_lexer.lpp" +#line 196 "glsl_lexer.lpp" return VEC4; YY_BREAK case 48: YY_RULE_SETUP -#line 196 "glsl_lexer.lpp" +#line 197 "glsl_lexer.lpp" return MAT2X2; YY_BREAK case 49: YY_RULE_SETUP -#line 197 "glsl_lexer.lpp" +#line 198 "glsl_lexer.lpp" return MAT3X3; YY_BREAK case 50: YY_RULE_SETUP -#line 198 "glsl_lexer.lpp" +#line 199 "glsl_lexer.lpp" return MAT4X4; YY_BREAK case 51: YY_RULE_SETUP -#line 199 "glsl_lexer.lpp" +#line 200 "glsl_lexer.lpp" KEYWORD(120, 120, MAT2X2); YY_BREAK case 52: YY_RULE_SETUP -#line 200 "glsl_lexer.lpp" +#line 201 "glsl_lexer.lpp" KEYWORD(120, 120, MAT2X3); YY_BREAK case 53: YY_RULE_SETUP -#line 201 "glsl_lexer.lpp" +#line 202 "glsl_lexer.lpp" KEYWORD(120, 120, MAT2X4); YY_BREAK case 54: YY_RULE_SETUP -#line 202 "glsl_lexer.lpp" +#line 203 "glsl_lexer.lpp" KEYWORD(120, 120, MAT3X2); YY_BREAK case 55: YY_RULE_SETUP -#line 203 "glsl_lexer.lpp" +#line 204 "glsl_lexer.lpp" KEYWORD(120, 120, MAT3X3); YY_BREAK case 56: YY_RULE_SETUP -#line 204 "glsl_lexer.lpp" +#line 205 "glsl_lexer.lpp" KEYWORD(120, 120, MAT3X4); YY_BREAK case 57: YY_RULE_SETUP -#line 205 "glsl_lexer.lpp" +#line 206 "glsl_lexer.lpp" KEYWORD(120, 120, MAT4X2); YY_BREAK case 58: YY_RULE_SETUP -#line 206 "glsl_lexer.lpp" +#line 207 "glsl_lexer.lpp" KEYWORD(120, 120, MAT4X3); YY_BREAK case 59: YY_RULE_SETUP -#line 207 "glsl_lexer.lpp" +#line 208 "glsl_lexer.lpp" KEYWORD(120, 120, MAT4X4); YY_BREAK case 60: YY_RULE_SETUP -#line 209 "glsl_lexer.lpp" +#line 210 "glsl_lexer.lpp" return IN_TOK; YY_BREAK case 61: YY_RULE_SETUP -#line 210 "glsl_lexer.lpp" +#line 211 "glsl_lexer.lpp" return OUT_TOK; YY_BREAK case 62: YY_RULE_SETUP -#line 211 "glsl_lexer.lpp" +#line 212 "glsl_lexer.lpp" return INOUT_TOK; YY_BREAK case 63: YY_RULE_SETUP -#line 212 "glsl_lexer.lpp" +#line 213 "glsl_lexer.lpp" return UNIFORM; YY_BREAK case 64: YY_RULE_SETUP -#line 213 "glsl_lexer.lpp" +#line 214 "glsl_lexer.lpp" return VARYING; YY_BREAK case 65: YY_RULE_SETUP -#line 214 "glsl_lexer.lpp" +#line 215 "glsl_lexer.lpp" KEYWORD(120, 120, CENTROID); YY_BREAK case 66: YY_RULE_SETUP -#line 215 "glsl_lexer.lpp" +#line 216 "glsl_lexer.lpp" KEYWORD(120 || ES, 120 || ES, INVARIANT); YY_BREAK case 67: YY_RULE_SETUP -#line 216 "glsl_lexer.lpp" +#line 217 "glsl_lexer.lpp" KEYWORD(130 || ES, 130, FLAT); YY_BREAK case 68: YY_RULE_SETUP -#line 217 "glsl_lexer.lpp" +#line 218 "glsl_lexer.lpp" KEYWORD(130, 130, SMOOTH); YY_BREAK case 69: YY_RULE_SETUP -#line 218 "glsl_lexer.lpp" +#line 219 "glsl_lexer.lpp" KEYWORD(130, 130, NOPERSPECTIVE); YY_BREAK case 70: YY_RULE_SETUP -#line 220 "glsl_lexer.lpp" +#line 221 "glsl_lexer.lpp" return SAMPLER1D; YY_BREAK case 71: YY_RULE_SETUP -#line 221 "glsl_lexer.lpp" +#line 222 "glsl_lexer.lpp" return SAMPLER2D; YY_BREAK case 72: YY_RULE_SETUP -#line 222 "glsl_lexer.lpp" +#line 223 "glsl_lexer.lpp" return SAMPLER3D; YY_BREAK case 73: YY_RULE_SETUP -#line 223 "glsl_lexer.lpp" +#line 224 "glsl_lexer.lpp" return SAMPLERCUBE; YY_BREAK case 74: YY_RULE_SETUP -#line 224 "glsl_lexer.lpp" +#line 225 "glsl_lexer.lpp" KEYWORD(130, 130, SAMPLER1DARRAY); YY_BREAK case 75: YY_RULE_SETUP -#line 225 "glsl_lexer.lpp" +#line 226 "glsl_lexer.lpp" KEYWORD(130, 130, SAMPLER2DARRAY); YY_BREAK case 76: YY_RULE_SETUP -#line 226 "glsl_lexer.lpp" +#line 227 "glsl_lexer.lpp" return SAMPLER1DSHADOW; YY_BREAK case 77: YY_RULE_SETUP -#line 227 "glsl_lexer.lpp" +#line 228 "glsl_lexer.lpp" return SAMPLER2DSHADOW; YY_BREAK case 78: YY_RULE_SETUP -#line 228 "glsl_lexer.lpp" +#line 229 "glsl_lexer.lpp" KEYWORD(130, 130, SAMPLERCUBESHADOW); YY_BREAK case 79: YY_RULE_SETUP -#line 229 "glsl_lexer.lpp" +#line 230 "glsl_lexer.lpp" KEYWORD(130, 130, SAMPLER1DARRAYSHADOW); YY_BREAK case 80: YY_RULE_SETUP -#line 230 "glsl_lexer.lpp" +#line 231 "glsl_lexer.lpp" KEYWORD(130, 130, SAMPLER2DARRAYSHADOW); YY_BREAK case 81: YY_RULE_SETUP -#line 231 "glsl_lexer.lpp" +#line 232 "glsl_lexer.lpp" KEYWORD(130, 130, ISAMPLER1D); YY_BREAK case 82: YY_RULE_SETUP -#line 232 "glsl_lexer.lpp" +#line 233 "glsl_lexer.lpp" KEYWORD(130, 130, ISAMPLER2D); YY_BREAK case 83: YY_RULE_SETUP -#line 233 "glsl_lexer.lpp" +#line 234 "glsl_lexer.lpp" KEYWORD(130, 130, ISAMPLER3D); YY_BREAK case 84: YY_RULE_SETUP -#line 234 "glsl_lexer.lpp" +#line 235 "glsl_lexer.lpp" KEYWORD(130, 130, ISAMPLERCUBE); YY_BREAK case 85: YY_RULE_SETUP -#line 235 "glsl_lexer.lpp" +#line 236 "glsl_lexer.lpp" KEYWORD(130, 130, ISAMPLER1DARRAY); YY_BREAK case 86: YY_RULE_SETUP -#line 236 "glsl_lexer.lpp" +#line 237 "glsl_lexer.lpp" KEYWORD(130, 130, ISAMPLER2DARRAY); YY_BREAK case 87: YY_RULE_SETUP -#line 237 "glsl_lexer.lpp" +#line 238 "glsl_lexer.lpp" KEYWORD(130, 130, USAMPLER1D); YY_BREAK case 88: YY_RULE_SETUP -#line 238 "glsl_lexer.lpp" +#line 239 "glsl_lexer.lpp" KEYWORD(130, 130, USAMPLER2D); YY_BREAK case 89: YY_RULE_SETUP -#line 239 "glsl_lexer.lpp" +#line 240 "glsl_lexer.lpp" KEYWORD(130, 130, USAMPLER3D); YY_BREAK case 90: YY_RULE_SETUP -#line 240 "glsl_lexer.lpp" +#line 241 "glsl_lexer.lpp" KEYWORD(130, 130, USAMPLERCUBE); YY_BREAK case 91: YY_RULE_SETUP -#line 241 "glsl_lexer.lpp" +#line 242 "glsl_lexer.lpp" KEYWORD(130, 130, USAMPLER1DARRAY); YY_BREAK case 92: YY_RULE_SETUP -#line 242 "glsl_lexer.lpp" +#line 243 "glsl_lexer.lpp" KEYWORD(130, 130, USAMPLER2DARRAY); YY_BREAK case 93: YY_RULE_SETUP -#line 245 "glsl_lexer.lpp" +#line 246 "glsl_lexer.lpp" return STRUCT; YY_BREAK case 94: YY_RULE_SETUP -#line 246 "glsl_lexer.lpp" +#line 247 "glsl_lexer.lpp" return VOID_TOK; YY_BREAK case 95: YY_RULE_SETUP -#line 248 "glsl_lexer.lpp" +#line 249 "glsl_lexer.lpp" { if ((yyextra->language_version >= 140) || yyextra->ARB_explicit_attrib_location_enable @@ -1923,112 +1920,112 @@ YY_RULE_SETUP YY_BREAK case 96: YY_RULE_SETUP -#line 259 "glsl_lexer.lpp" +#line 260 "glsl_lexer.lpp" return INC_OP; YY_BREAK case 97: YY_RULE_SETUP -#line 260 "glsl_lexer.lpp" +#line 261 "glsl_lexer.lpp" return DEC_OP; YY_BREAK case 98: YY_RULE_SETUP -#line 261 "glsl_lexer.lpp" +#line 262 "glsl_lexer.lpp" return LE_OP; YY_BREAK case 99: YY_RULE_SETUP -#line 262 "glsl_lexer.lpp" +#line 263 "glsl_lexer.lpp" return GE_OP; YY_BREAK case 100: YY_RULE_SETUP -#line 263 "glsl_lexer.lpp" +#line 264 "glsl_lexer.lpp" return EQ_OP; YY_BREAK case 101: YY_RULE_SETUP -#line 264 "glsl_lexer.lpp" +#line 265 "glsl_lexer.lpp" return NE_OP; YY_BREAK case 102: YY_RULE_SETUP -#line 265 "glsl_lexer.lpp" +#line 266 "glsl_lexer.lpp" return AND_OP; YY_BREAK case 103: YY_RULE_SETUP -#line 266 "glsl_lexer.lpp" +#line 267 "glsl_lexer.lpp" return OR_OP; YY_BREAK case 104: YY_RULE_SETUP -#line 267 "glsl_lexer.lpp" +#line 268 "glsl_lexer.lpp" return XOR_OP; YY_BREAK case 105: YY_RULE_SETUP -#line 268 "glsl_lexer.lpp" +#line 269 "glsl_lexer.lpp" return LEFT_OP; YY_BREAK case 106: YY_RULE_SETUP -#line 269 "glsl_lexer.lpp" +#line 270 "glsl_lexer.lpp" return RIGHT_OP; YY_BREAK case 107: YY_RULE_SETUP -#line 271 "glsl_lexer.lpp" +#line 272 "glsl_lexer.lpp" return MUL_ASSIGN; YY_BREAK case 108: YY_RULE_SETUP -#line 272 "glsl_lexer.lpp" +#line 273 "glsl_lexer.lpp" return DIV_ASSIGN; YY_BREAK case 109: YY_RULE_SETUP -#line 273 "glsl_lexer.lpp" +#line 274 "glsl_lexer.lpp" return ADD_ASSIGN; YY_BREAK case 110: YY_RULE_SETUP -#line 274 "glsl_lexer.lpp" +#line 275 "glsl_lexer.lpp" return MOD_ASSIGN; YY_BREAK case 111: YY_RULE_SETUP -#line 275 "glsl_lexer.lpp" +#line 276 "glsl_lexer.lpp" return LEFT_ASSIGN; YY_BREAK case 112: YY_RULE_SETUP -#line 276 "glsl_lexer.lpp" +#line 277 "glsl_lexer.lpp" return RIGHT_ASSIGN; YY_BREAK case 113: YY_RULE_SETUP -#line 277 "glsl_lexer.lpp" +#line 278 "glsl_lexer.lpp" return AND_ASSIGN; YY_BREAK case 114: YY_RULE_SETUP -#line 278 "glsl_lexer.lpp" +#line 279 "glsl_lexer.lpp" return XOR_ASSIGN; YY_BREAK case 115: YY_RULE_SETUP -#line 279 "glsl_lexer.lpp" +#line 280 "glsl_lexer.lpp" return OR_ASSIGN; YY_BREAK case 116: YY_RULE_SETUP -#line 280 "glsl_lexer.lpp" +#line 281 "glsl_lexer.lpp" return SUB_ASSIGN; YY_BREAK case 117: YY_RULE_SETUP -#line 282 "glsl_lexer.lpp" +#line 283 "glsl_lexer.lpp" { yylval->n = strtol(yytext, NULL, 10); return IS_UINT ? UINTCONSTANT : INTCONSTANT; @@ -2036,7 +2033,7 @@ YY_RULE_SETUP YY_BREAK case 118: YY_RULE_SETUP -#line 286 "glsl_lexer.lpp" +#line 287 "glsl_lexer.lpp" { yylval->n = strtol(yytext + 2, NULL, 16); return IS_UINT ? UINTCONSTANT : INTCONSTANT; @@ -2044,7 +2041,7 @@ YY_RULE_SETUP YY_BREAK case 119: YY_RULE_SETUP -#line 290 "glsl_lexer.lpp" +#line 291 "glsl_lexer.lpp" { yylval->n = strtol(yytext, NULL, 8); return IS_UINT ? UINTCONSTANT : INTCONSTANT; @@ -2052,47 +2049,47 @@ YY_RULE_SETUP YY_BREAK case 120: YY_RULE_SETUP -#line 295 "glsl_lexer.lpp" +#line 296 "glsl_lexer.lpp" { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } YY_BREAK case 121: YY_RULE_SETUP -#line 299 "glsl_lexer.lpp" +#line 300 "glsl_lexer.lpp" { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } YY_BREAK case 122: YY_RULE_SETUP -#line 303 "glsl_lexer.lpp" +#line 304 "glsl_lexer.lpp" { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } YY_BREAK case 123: YY_RULE_SETUP -#line 307 "glsl_lexer.lpp" +#line 308 "glsl_lexer.lpp" { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } YY_BREAK case 124: YY_RULE_SETUP -#line 311 "glsl_lexer.lpp" +#line 312 "glsl_lexer.lpp" { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } YY_BREAK case 125: YY_RULE_SETUP -#line 316 "glsl_lexer.lpp" +#line 317 "glsl_lexer.lpp" { yylval->n = 1; return BOOLCONSTANT; @@ -2100,7 +2097,7 @@ YY_RULE_SETUP YY_BREAK case 126: YY_RULE_SETUP -#line 320 "glsl_lexer.lpp" +#line 321 "glsl_lexer.lpp" { yylval->n = 0; return BOOLCONSTANT; @@ -2109,409 +2106,409 @@ YY_RULE_SETUP /* Reserved words in GLSL 1.10. */ case 127: YY_RULE_SETUP -#line 327 "glsl_lexer.lpp" +#line 328 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, ASM); YY_BREAK case 128: YY_RULE_SETUP -#line 328 "glsl_lexer.lpp" +#line 329 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, CLASS); YY_BREAK case 129: YY_RULE_SETUP -#line 329 "glsl_lexer.lpp" +#line 330 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, UNION); YY_BREAK case 130: YY_RULE_SETUP -#line 330 "glsl_lexer.lpp" +#line 331 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, ENUM); YY_BREAK case 131: YY_RULE_SETUP -#line 331 "glsl_lexer.lpp" +#line 332 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, TYPEDEF); YY_BREAK case 132: YY_RULE_SETUP -#line 332 "glsl_lexer.lpp" +#line 333 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, TEMPLATE); YY_BREAK case 133: YY_RULE_SETUP -#line 333 "glsl_lexer.lpp" +#line 334 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, THIS); YY_BREAK case 134: YY_RULE_SETUP -#line 334 "glsl_lexer.lpp" +#line 335 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, PACKED_TOK); YY_BREAK case 135: YY_RULE_SETUP -#line 335 "glsl_lexer.lpp" +#line 336 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, GOTO); YY_BREAK case 136: YY_RULE_SETUP -#line 336 "glsl_lexer.lpp" +#line 337 "glsl_lexer.lpp" KEYWORD(110 || ES, 130, SWITCH); YY_BREAK case 137: YY_RULE_SETUP -#line 337 "glsl_lexer.lpp" +#line 338 "glsl_lexer.lpp" KEYWORD(110 || ES, 130, DEFAULT); YY_BREAK case 138: YY_RULE_SETUP -#line 338 "glsl_lexer.lpp" +#line 339 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, INLINE_TOK); YY_BREAK case 139: YY_RULE_SETUP -#line 339 "glsl_lexer.lpp" +#line 340 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, NOINLINE); YY_BREAK case 140: YY_RULE_SETUP -#line 340 "glsl_lexer.lpp" +#line 341 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, VOLATILE); YY_BREAK case 141: YY_RULE_SETUP -#line 341 "glsl_lexer.lpp" +#line 342 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, PUBLIC_TOK); YY_BREAK case 142: YY_RULE_SETUP -#line 342 "glsl_lexer.lpp" +#line 343 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, STATIC); YY_BREAK case 143: YY_RULE_SETUP -#line 343 "glsl_lexer.lpp" +#line 344 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, EXTERN); YY_BREAK case 144: YY_RULE_SETUP -#line 344 "glsl_lexer.lpp" +#line 345 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, EXTERNAL); YY_BREAK case 145: YY_RULE_SETUP -#line 345 "glsl_lexer.lpp" +#line 346 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, INTERFACE); YY_BREAK case 146: YY_RULE_SETUP -#line 346 "glsl_lexer.lpp" +#line 347 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, LONG_TOK); YY_BREAK case 147: YY_RULE_SETUP -#line 347 "glsl_lexer.lpp" +#line 348 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, SHORT_TOK); YY_BREAK case 148: YY_RULE_SETUP -#line 348 "glsl_lexer.lpp" +#line 349 "glsl_lexer.lpp" KEYWORD(110 || ES, 400, DOUBLE_TOK); YY_BREAK case 149: YY_RULE_SETUP -#line 349 "glsl_lexer.lpp" +#line 350 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, HALF); YY_BREAK case 150: YY_RULE_SETUP -#line 350 "glsl_lexer.lpp" +#line 351 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, FIXED_TOK); YY_BREAK case 151: YY_RULE_SETUP -#line 351 "glsl_lexer.lpp" +#line 352 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, UNSIGNED); YY_BREAK case 152: YY_RULE_SETUP -#line 352 "glsl_lexer.lpp" +#line 353 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, INPUT_TOK); YY_BREAK case 153: YY_RULE_SETUP -#line 353 "glsl_lexer.lpp" +#line 354 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, OUTPUT); YY_BREAK case 154: YY_RULE_SETUP -#line 354 "glsl_lexer.lpp" +#line 355 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, HVEC2); YY_BREAK case 155: YY_RULE_SETUP -#line 355 "glsl_lexer.lpp" +#line 356 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, HVEC3); YY_BREAK case 156: YY_RULE_SETUP -#line 356 "glsl_lexer.lpp" +#line 357 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, HVEC4); YY_BREAK case 157: YY_RULE_SETUP -#line 357 "glsl_lexer.lpp" +#line 358 "glsl_lexer.lpp" KEYWORD(110 || ES, 400, DVEC2); YY_BREAK case 158: YY_RULE_SETUP -#line 358 "glsl_lexer.lpp" +#line 359 "glsl_lexer.lpp" KEYWORD(110 || ES, 400, DVEC3); YY_BREAK case 159: YY_RULE_SETUP -#line 359 "glsl_lexer.lpp" +#line 360 "glsl_lexer.lpp" KEYWORD(110 || ES, 400, DVEC4); YY_BREAK case 160: YY_RULE_SETUP -#line 360 "glsl_lexer.lpp" +#line 361 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, FVEC2); YY_BREAK case 161: YY_RULE_SETUP -#line 361 "glsl_lexer.lpp" +#line 362 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, FVEC3); YY_BREAK case 162: YY_RULE_SETUP -#line 362 "glsl_lexer.lpp" +#line 363 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, FVEC4); YY_BREAK case 163: YY_RULE_SETUP -#line 363 "glsl_lexer.lpp" +#line 364 "glsl_lexer.lpp" return SAMPLER2DRECT; YY_BREAK case 164: YY_RULE_SETUP -#line 364 "glsl_lexer.lpp" +#line 365 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, SAMPLER3DRECT); YY_BREAK case 165: YY_RULE_SETUP -#line 365 "glsl_lexer.lpp" +#line 366 "glsl_lexer.lpp" return SAMPLER2DRECTSHADOW; YY_BREAK case 166: YY_RULE_SETUP -#line 366 "glsl_lexer.lpp" +#line 367 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, SIZEOF); YY_BREAK case 167: YY_RULE_SETUP -#line 367 "glsl_lexer.lpp" +#line 368 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, CAST); YY_BREAK case 168: YY_RULE_SETUP -#line 368 "glsl_lexer.lpp" +#line 369 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, NAMESPACE); YY_BREAK case 169: YY_RULE_SETUP -#line 369 "glsl_lexer.lpp" +#line 370 "glsl_lexer.lpp" KEYWORD(110 || ES, 999, USING); YY_BREAK /* Additional reserved words in GLSL 1.20. */ case 170: YY_RULE_SETUP -#line 372 "glsl_lexer.lpp" +#line 373 "glsl_lexer.lpp" KEYWORD(120, 130 || ES, LOWP); YY_BREAK case 171: YY_RULE_SETUP -#line 373 "glsl_lexer.lpp" +#line 374 "glsl_lexer.lpp" KEYWORD(120, 130 || ES, MEDIUMP); YY_BREAK case 172: YY_RULE_SETUP -#line 374 "glsl_lexer.lpp" +#line 375 "glsl_lexer.lpp" KEYWORD(120, 130 || ES, HIGHP); YY_BREAK case 173: YY_RULE_SETUP -#line 375 "glsl_lexer.lpp" +#line 376 "glsl_lexer.lpp" KEYWORD(120, 130 || ES, PRECISION); YY_BREAK /* Additional reserved words in GLSL 1.30. */ case 174: YY_RULE_SETUP -#line 378 "glsl_lexer.lpp" +#line 379 "glsl_lexer.lpp" KEYWORD(130, 130, CASE); YY_BREAK case 175: YY_RULE_SETUP -#line 379 "glsl_lexer.lpp" +#line 380 "glsl_lexer.lpp" KEYWORD(130, 999, COMMON); YY_BREAK case 176: YY_RULE_SETUP -#line 380 "glsl_lexer.lpp" +#line 381 "glsl_lexer.lpp" KEYWORD(130, 999, PARTITION); YY_BREAK case 177: YY_RULE_SETUP -#line 381 "glsl_lexer.lpp" +#line 382 "glsl_lexer.lpp" KEYWORD(130, 999, ACTIVE); YY_BREAK case 178: YY_RULE_SETUP -#line 382 "glsl_lexer.lpp" +#line 383 "glsl_lexer.lpp" KEYWORD(130 || ES, 999, SUPERP); YY_BREAK case 179: YY_RULE_SETUP -#line 383 "glsl_lexer.lpp" +#line 384 "glsl_lexer.lpp" KEYWORD(130, 140, SAMPLERBUFFER); YY_BREAK case 180: YY_RULE_SETUP -#line 384 "glsl_lexer.lpp" +#line 385 "glsl_lexer.lpp" KEYWORD(130, 999, FILTER); YY_BREAK case 181: YY_RULE_SETUP -#line 385 "glsl_lexer.lpp" +#line 386 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE1D); YY_BREAK case 182: YY_RULE_SETUP -#line 386 "glsl_lexer.lpp" +#line 387 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE2D); YY_BREAK case 183: YY_RULE_SETUP -#line 387 "glsl_lexer.lpp" +#line 388 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE3D); YY_BREAK case 184: YY_RULE_SETUP -#line 388 "glsl_lexer.lpp" +#line 389 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGECUBE); YY_BREAK case 185: YY_RULE_SETUP -#line 389 "glsl_lexer.lpp" +#line 390 "glsl_lexer.lpp" KEYWORD(130, 999, IIMAGE1D); YY_BREAK case 186: YY_RULE_SETUP -#line 390 "glsl_lexer.lpp" +#line 391 "glsl_lexer.lpp" KEYWORD(130, 999, IIMAGE2D); YY_BREAK case 187: YY_RULE_SETUP -#line 391 "glsl_lexer.lpp" +#line 392 "glsl_lexer.lpp" KEYWORD(130, 999, IIMAGE3D); YY_BREAK case 188: YY_RULE_SETUP -#line 392 "glsl_lexer.lpp" +#line 393 "glsl_lexer.lpp" KEYWORD(130, 999, IIMAGECUBE); YY_BREAK case 189: YY_RULE_SETUP -#line 393 "glsl_lexer.lpp" +#line 394 "glsl_lexer.lpp" KEYWORD(130, 999, UIMAGE1D); YY_BREAK case 190: YY_RULE_SETUP -#line 394 "glsl_lexer.lpp" +#line 395 "glsl_lexer.lpp" KEYWORD(130, 999, UIMAGE2D); YY_BREAK case 191: YY_RULE_SETUP -#line 395 "glsl_lexer.lpp" +#line 396 "glsl_lexer.lpp" KEYWORD(130, 999, UIMAGE3D); YY_BREAK case 192: YY_RULE_SETUP -#line 396 "glsl_lexer.lpp" +#line 397 "glsl_lexer.lpp" KEYWORD(130, 999, UIMAGECUBE); YY_BREAK case 193: YY_RULE_SETUP -#line 397 "glsl_lexer.lpp" +#line 398 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE1DARRAY); YY_BREAK case 194: YY_RULE_SETUP -#line 398 "glsl_lexer.lpp" +#line 399 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE2DARRAY); YY_BREAK case 195: YY_RULE_SETUP -#line 399 "glsl_lexer.lpp" +#line 400 "glsl_lexer.lpp" KEYWORD(130, 999, IIMAGE1DARRAY); YY_BREAK case 196: YY_RULE_SETUP -#line 400 "glsl_lexer.lpp" +#line 401 "glsl_lexer.lpp" KEYWORD(130, 999, IIMAGE2DARRAY); YY_BREAK case 197: YY_RULE_SETUP -#line 401 "glsl_lexer.lpp" +#line 402 "glsl_lexer.lpp" KEYWORD(130, 999, UIMAGE1DARRAY); YY_BREAK case 198: YY_RULE_SETUP -#line 402 "glsl_lexer.lpp" +#line 403 "glsl_lexer.lpp" KEYWORD(130, 999, UIMAGE2DARRAY); YY_BREAK case 199: YY_RULE_SETUP -#line 403 "glsl_lexer.lpp" +#line 404 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE1DSHADOW); YY_BREAK case 200: YY_RULE_SETUP -#line 404 "glsl_lexer.lpp" +#line 405 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE2DSHADOW); YY_BREAK case 201: YY_RULE_SETUP -#line 405 "glsl_lexer.lpp" +#line 406 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE1DARRAYSHADOW); YY_BREAK case 202: YY_RULE_SETUP -#line 406 "glsl_lexer.lpp" +#line 407 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGE2DARRAYSHADOW); YY_BREAK case 203: YY_RULE_SETUP -#line 407 "glsl_lexer.lpp" +#line 408 "glsl_lexer.lpp" KEYWORD(130, 999, IMAGEBUFFER); YY_BREAK case 204: YY_RULE_SETUP -#line 408 "glsl_lexer.lpp" +#line 409 "glsl_lexer.lpp" KEYWORD(130, 999, IIMAGEBUFFER); YY_BREAK case 205: YY_RULE_SETUP -#line 409 "glsl_lexer.lpp" +#line 410 "glsl_lexer.lpp" KEYWORD(130, 999, UIMAGEBUFFER); YY_BREAK case 206: YY_RULE_SETUP -#line 410 "glsl_lexer.lpp" +#line 411 "glsl_lexer.lpp" KEYWORD(130, 999, ROW_MAJOR); YY_BREAK case 207: YY_RULE_SETUP -#line 412 "glsl_lexer.lpp" +#line 413 "glsl_lexer.lpp" { struct _mesa_glsl_parse_state *state = yyextra; void *ctx = state; @@ -2521,15 +2518,15 @@ YY_RULE_SETUP YY_BREAK case 208: YY_RULE_SETUP -#line 419 "glsl_lexer.lpp" +#line 420 "glsl_lexer.lpp" { return yytext[0]; } YY_BREAK case 209: YY_RULE_SETUP -#line 421 "glsl_lexer.lpp" +#line 422 "glsl_lexer.lpp" ECHO; YY_BREAK -#line 2533 "glsl_lexer.cpp" +#line 2530 "glsl_lexer.cpp" case YY_STATE_EOF(INITIAL): case YY_STATE_EOF(PP): case YY_STATE_EOF(PRAGMA): @@ -3672,7 +3669,7 @@ void _mesa_glsl_free (void * ptr , yyscan_t yyscanner) #define YYTABLES_NAME "yytables" -#line 421 "glsl_lexer.lpp" +#line 422 "glsl_lexer.lpp" diff --git a/src/glsl/glsl_lexer.lpp b/src/glsl/glsl_lexer.lpp index 7a3f1a67e66..15742ac3636 100644 --- a/src/glsl/glsl_lexer.lpp +++ b/src/glsl/glsl_lexer.lpp @@ -22,6 +22,7 @@ * DEALINGS IN THE SOFTWARE. */ #include <ctype.h> +#include "strtod.h" #include "ast.h" #include "glsl_parser_extras.h" #include "glsl_parser.h" @@ -293,23 +294,23 @@ layout { } [0-9]+\.[0-9]+([eE][+-]?[0-9]+)?[fF]? { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } \.[0-9]+([eE][+-]?[0-9]+)?[fF]? { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } [0-9]+\.([eE][+-]?[0-9]+)?[fF]? { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } [0-9]+[eE][+-]?[0-9]+[fF]? { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } [0-9]+[fF] { - yylval->real = strtod(yytext, NULL); + yylval->real = glsl_strtod(yytext, NULL); return FLOATCONSTANT; } diff --git a/src/glsl/glsl_types.cpp b/src/glsl/glsl_types.cpp index 8c33cfa20d0..95b85926482 100644 --- a/src/glsl/glsl_types.cpp +++ b/src/glsl/glsl_types.cpp @@ -120,7 +120,7 @@ glsl_type::generate_100ES_types(glsl_symbol_table *symtab) add_types_to_symbol_table(symtab, builtin_structure_types, Elements(builtin_structure_types), false); - add_types_to_symbol_table(symtab, &void_type, 1, false); + add_types_to_symbol_table(symtab, void_type, 1, false); } void @@ -279,7 +279,7 @@ const glsl_type * glsl_type::get_instance(unsigned base_type, unsigned rows, unsigned columns) { if (base_type == GLSL_TYPE_VOID) - return &void_type; + return void_type; if ((rows < 1) || (rows > 4) || (columns < 1) || (columns > 4)) return error_type; diff --git a/src/glsl/glsl_types.h b/src/glsl/glsl_types.h index d62d493c8e8..57e7b09d98c 100644 --- a/src/glsl/glsl_types.h +++ b/src/glsl/glsl_types.h @@ -142,6 +142,7 @@ struct glsl_type { */ /*@{*/ static const glsl_type *const error_type; + static const glsl_type *const void_type; static const glsl_type *const int_type; static const glsl_type *const ivec4_type; static const glsl_type *const uint_type; @@ -424,7 +425,7 @@ private: */ /*@{*/ static const glsl_type _error_type; - static const glsl_type void_type; + static const glsl_type _void_type; static const glsl_type builtin_core_types[]; static const glsl_type builtin_structure_types[]; static const glsl_type builtin_110_deprecated_structure_types[]; diff --git a/src/glsl/linker.cpp b/src/glsl/linker.cpp index e62fe6d7e9f..d8c42ac4624 100644 --- a/src/glsl/linker.cpp +++ b/src/glsl/linker.cpp @@ -487,14 +487,35 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog, /* Check that the types match between stages. */ if (input->type != output->type) { - linker_error_printf(prog, - "%s shader output `%s' declared as " - "type `%s', but %s shader input declared " - "as type `%s'\n", - producer_stage, output->name, - output->type->name, - consumer_stage, input->type->name); - return false; + /* There is a bit of a special case for gl_TexCoord. This + * built-in is unsized by default. Appliations that variable + * access it must redeclare it with a size. There is some + * language in the GLSL spec that implies the fragment shader + * and vertex shader do not have to agree on this size. Other + * driver behave this way, and one or two applications seem to + * rely on it. + * + * Neither declaration needs to be modified here because the array + * sizes are fixed later when update_array_sizes is called. + * + * From page 48 (page 54 of the PDF) of the GLSL 1.10 spec: + * + * "Unlike user-defined varying variables, the built-in + * varying variables don't have a strict one-to-one + * correspondence between the vertex language and the + * fragment language." + */ + if (!output->type->is_array() + || (strncmp("gl_", output->name, 3) != 0)) { + linker_error_printf(prog, + "%s shader output `%s' declared as " + "type `%s', but %s shader input declared " + "as type `%s'\n", + producer_stage, output->name, + output->type->name, + consumer_stage, input->type->name); + return false; + } } /* Check that all of the qualifiers match between stages. diff --git a/src/glsl/s_expression.cpp b/src/glsl/s_expression.cpp index 4c8829fea9a..e420cd6e7ec 100644 --- a/src/glsl/s_expression.cpp +++ b/src/glsl/s_expression.cpp @@ -62,7 +62,7 @@ read_atom(void *ctx, const char *& src) // Check if the atom is a number. char *float_end = NULL; - double f = strtod(src, &float_end); + double f = glsl_strtod(src, &float_end); if (float_end != src) { char *int_end = NULL; int i = strtol(src, &int_end, 10); diff --git a/src/glsl/s_expression.h b/src/glsl/s_expression.h index aa22475a1bf..29d800e394a 100644 --- a/src/glsl/s_expression.h +++ b/src/glsl/s_expression.h @@ -26,6 +26,7 @@ #ifndef S_EXPRESSION_H #define S_EXPRESSION_H +#include "strtod.h" #include "list.h" #define SX_AS_(t,x) ((x) && ((s_expression*) x)->is_##t()) ? ((s_##t*) (x)) \ diff --git a/src/glsl/strtod.c b/src/glsl/strtod.c new file mode 100644 index 00000000000..ff34591488c --- /dev/null +++ b/src/glsl/strtod.c @@ -0,0 +1,56 @@ +/* + * Copyright 2010 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. + */ + + +#include <stdlib.h> + +#ifdef _GNU_SOURCE +#include <locale.h> +#ifdef __APPLE__ +#include <xlocale.h> +#endif +#endif + +#include "strtod.h" + + + +/** + * Wrapper around strtod which uses the "C" locale so the decimal + * point is always '.' + */ +double +glsl_strtod(const char *s, char **end) +{ +#if defined(_GNU_SOURCE) && !defined(__CYGWIN__) && !defined(__FreeBSD__) + static locale_t loc = NULL; + if (!loc) { + loc = newlocale(LC_CTYPE_MASK, "C", NULL); + } + return strtod_l(s, end, loc); +#else + return strtod(s, end); +#endif +} diff --git a/src/glsl/strtod.h b/src/glsl/strtod.h new file mode 100644 index 00000000000..0cf6409d425 --- /dev/null +++ b/src/glsl/strtod.h @@ -0,0 +1,43 @@ +/* + * Copyright 2010 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 STRTOD_H +#define STRTOD_H + +#ifdef __cplusplus +extern "C" { +#endif + +extern double +glsl_strtod(const char *s, char **end); + + +#ifdef __cplusplus +} +#endif + + +#endif diff --git a/src/mapi/es1api/Makefile b/src/mapi/es1api/Makefile index 576ac5afdc3..3659d6486a8 100644 --- a/src/mapi/es1api/Makefile +++ b/src/mapi/es1api/Makefile @@ -44,10 +44,10 @@ GLAPI_ASM_OBJECTS := $(GLAPI_ASM_SOURCES:.S=.o) GLAPI_ASM_SOURCES := $(addprefix $(GEN)/, $(GLAPI_ASM_SOURCES)) include $(MAPI)/sources.mak -MAPI_GLAPI_OBJECTS := $(MAPI_GLAPI_SOURCES:.c=.o) -MAPI_GLAPI_SOURCES := $(addprefix $(MAPI)/, $(MAPI_GLAPI_SOURCES)) +MAPI_UTIL_OBJECTS := $(MAPI_UTIL_SOURCES:.c=.o) +MAPI_UTIL_SOURCES := $(addprefix $(MAPI)/, $(MAPI_UTIL_SOURCES)) -ESAPI_OBJECTS = $(GLAPI_OBJECTS) $(GLAPI_ASM_OBJECTS) $(MAPI_GLAPI_OBJECTS) +ESAPI_OBJECTS = $(GLAPI_OBJECTS) $(GLAPI_ASM_OBJECTS) $(MAPI_UTIL_OBJECTS) INCLUDE_DIRS = \ -I$(TOP)/include \ @@ -72,13 +72,13 @@ lib$(ESAPI).a: $(ESAPI_OBJECTS) @$(MKLIB) -o $(ESAPI) -static $(ESAPI_OBJECTS) $(GLAPI_OBJECTS): %.o: $(GLAPI)/%.c - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $< -o $@ + $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) -DMAPI_MODE_UTIL $< -o $@ $(GLAPI_ASM_OBJECTS): %.o: $(GEN)/%.S $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $< -o $@ -$(MAPI_GLAPI_OBJECTS): %.o: $(MAPI)/%.c - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) -DMAPI_GLAPI_CURRENT $< -o $@ +$(MAPI_UTIL_OBJECTS): %.o: $(MAPI)/%.c + $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) -DMAPI_MODE_UTIL $< -o $@ $(GLAPI_SOURCES) $(GLAPI_ASM_SOURCES): | glapi-stamp @@ -135,10 +135,10 @@ install: default install-headers install-pc $(MINSTALL) $(TOP)/$(LIB_DIR)/$(GLES_LIB_GLOB) \ $(DESTDIR)$(INSTALL_LIB_DIR) -depend: $(GLAPI_SOURCES) $(MAPI_GLAPI_SOURCES) +depend: $(GLAPI_SOURCES) $(MAPI_UTIL_SOURCES) @echo "running $(MKDEP)" @touch depend @$(MKDEP) $(MKDEP_OPTIONS) -f- $(DEFINES) $(INCLUDE_DIRS) \ - -DMAPI_GLAPI_CURRENT $(GLAPI_SOURCES) $(MAPI_GLAPI_SOURCES) \ + -DMAPI_MODE_UTIL $(GLAPI_SOURCES) $(MAPI_UTIL_SOURCES) \ 2>/dev/null | sed -e 's,^$(GLAPI)/,,' -e 's,^$(MAPI)/,,' \ > depend diff --git a/src/mapi/glapi/Makefile b/src/mapi/glapi/Makefile index ca9381d4253..a9ab07903c1 100644 --- a/src/mapi/glapi/Makefile +++ b/src/mapi/glapi/Makefile @@ -12,10 +12,10 @@ GLAPI_OBJECTS = $(GLAPI_SOURCES:.c=.o) GLAPI_ASM_OBJECTS = $(GLAPI_ASM_SOURCES:.S=.o) include $(MAPI)/sources.mak -MAPI_GLAPI_OBJECTS := $(MAPI_GLAPI_SOURCES:.c=.o) -MAPI_GLAPI_SOURCES := $(addprefix $(MAPI)/, $(MAPI_GLAPI_SOURCES)) +MAPI_UTIL_OBJECTS := $(MAPI_UTIL_SOURCES:.c=.o) +MAPI_UTIL_SOURCES := $(addprefix $(MAPI)/, $(MAPI_UTIL_SOURCES)) -TARGET_OBJECTS = $(GLAPI_OBJECTS) $(GLAPI_ASM_OBJECTS) $(MAPI_GLAPI_OBJECTS) +TARGET_OBJECTS = $(GLAPI_OBJECTS) $(GLAPI_ASM_OBJECTS) $(MAPI_UTIL_OBJECTS) INCLUDE_DIRS = \ -I$(TOP)/include \ @@ -28,13 +28,13 @@ lib$(TARGET).a: $(TARGET_OBJECTS) @$(MKLIB) -o $(TARGET) -static $(TARGET_OBJECTS) $(GLAPI_OBJECTS): %.o: %.c - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $< -o $@ + $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) -DMAPI_MODE_UTIL $< -o $@ $(GLAPI_ASM_OBJECTS): %.o: %.S $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $< -o $@ -$(MAPI_GLAPI_OBJECTS): %.o: $(MAPI)/%.c - $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) -DMAPI_GLAPI_CURRENT $< -o $@ +$(MAPI_UTIL_OBJECTS): %.o: $(MAPI)/%.c + $(CC) -c $(INCLUDE_DIRS) $(CFLAGS) -DMAPI_MODE_UTIL $< -o $@ install: @@ -43,11 +43,11 @@ clean: -rm -f lib$(TARGET).a -rm -f depend depend.bak -depend: $(GLAPI_SOURCES) $(MAPI_GLAPI_SOURCES) +depend: $(GLAPI_SOURCES) $(MAPI_UTIL_SOURCES) @ echo "running $(MKDEP)" @ touch depend @$(MKDEP) $(MKDEP_OPTIONS) -f- $(DEFINES) $(INCLUDE_DIRS) \ - -DMAPI_GLAPI_CURRENT $(GLAPI_SOURCES) $(MAPI_GLAPI_SOURCES) \ + -DMAPI_MODE_UTIL $(GLAPI_SOURCES) $(MAPI_UTIL_SOURCES) \ 2>/dev/null | sed -e 's,^$(MAPI)/,,' > depend -include depend diff --git a/src/mapi/glapi/SConscript b/src/mapi/glapi/SConscript index 4ef855fc350..4057991ec8d 100644 --- a/src/mapi/glapi/SConscript +++ b/src/mapi/glapi/SConscript @@ -9,7 +9,7 @@ if env['platform'] != 'winddk': env = env.Clone() env.Append(CPPDEFINES = [ - 'MAPI_GLAPI_CURRENT', + 'MAPI_MODE_UTIL', ]) if env['platform'] == 'windows': @@ -29,6 +29,7 @@ if env['platform'] != 'winddk': 'glapi_getproc.c', 'glapi_nop.c', 'glthread.c', + 'glapi.c', ] mapi_sources = [ diff --git a/src/mapi/glapi/glapi.c b/src/mapi/glapi/glapi.c new file mode 100644 index 00000000000..f7655c558c4 --- /dev/null +++ b/src/mapi/glapi/glapi.c @@ -0,0 +1,65 @@ +/* + * Mesa 3-D graphics library + * Version: 7.9 + * + * Copyright (C) 2010 LunarG Inc. + * + * 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 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: + * Chia-I Wu <[email protected]> + */ + +#include "glapi/glapi.h" +#include "mapi/u_current.h" + +/* + * Global variables, _glapi_get_context, and _glapi_get_dispatch are defined in + * u_current.c. + */ + +#ifdef GLX_USE_TLS +/* not used, but defined for compatibility */ +const struct _glapi_table *_glapi_Dispatch; +const void *_glapi_Context; +#endif /* GLX_USE_TLS */ + +void +_glapi_destroy_multithread(void) +{ + u_current_destroy(); +} + +void +_glapi_check_multithread(void) +{ + u_current_init(); +} + +void +_glapi_set_context(void *context) +{ + u_current_set_user((const void *) context); +} + +void +_glapi_set_dispatch(struct _glapi_table *dispatch) +{ + u_current_set((const struct mapi_table *) dispatch); +} diff --git a/src/mapi/glapi/sources.mak b/src/mapi/glapi/sources.mak index cdcfa36e51f..0b4d8cf65df 100644 --- a/src/mapi/glapi/sources.mak +++ b/src/mapi/glapi/sources.mak @@ -5,7 +5,8 @@ GLAPI_SOURCES = \ glapi_entrypoint.c \ glapi_getproc.c \ glapi_nop.c \ - glthread.c + glthread.c \ + glapi.c X86_API = \ glapi_x86.S diff --git a/src/mapi/mapi/entry_x86-64_tls.h b/src/mapi/mapi/entry_x86-64_tls.h index 2fbe73b5b32..2c02933802f 100644 --- a/src/mapi/mapi/entry_x86-64_tls.h +++ b/src/mapi/mapi/entry_x86-64_tls.h @@ -30,10 +30,23 @@ #include "u_execmem.h" #include "u_macros.h" +#ifdef __linux__ +__asm__(".section .note.ABI-tag, \"a\"\n\t" + ".p2align 2\n\t" + ".long 1f - 0f\n\t" /* name length */ + ".long 3f - 2f\n\t" /* data length */ + ".long 1\n\t" /* note length */ + "0: .asciz \"GNU\"\n\t" /* vendor name */ + "1: .p2align 2\n\t" + "2: .long 0\n\t" /* note data: the ABI tag */ + ".long 2,4,20\n\t" /* Minimum kernel version w/TLS */ + "3: .p2align 2\n\t"); /* pad out section */ +#endif /* __linux__ */ + __asm__(".text"); __asm__("x86_64_current_tls:\n\t" - "movq u_current_table_tls@GOTTPOFF(%rip), %rax\n\t" + "movq u_current_table@GOTTPOFF(%rip), %rax\n\t" "ret"); #define STUB_ASM_ENTRY(func) \ @@ -43,7 +56,7 @@ __asm__("x86_64_current_tls:\n\t" func ":" #define STUB_ASM_CODE(slot) \ - "movq u_current_table_tls@GOTTPOFF(%rip), %rax\n\t" \ + "movq u_current_table@GOTTPOFF(%rip), %rax\n\t" \ "movq %fs:(%rax), %r11\n\t" \ "jmp *(8 * " slot ")(%r11)" diff --git a/src/mapi/mapi/entry_x86_tls.h b/src/mapi/mapi/entry_x86_tls.h index d4f7d98cf12..3d0b7caffe1 100644 --- a/src/mapi/mapi/entry_x86_tls.h +++ b/src/mapi/mapi/entry_x86_tls.h @@ -30,6 +30,19 @@ #include "u_execmem.h" #include "u_macros.h" +#ifdef __linux__ +__asm__(".section .note.ABI-tag, \"a\"\n\t" + ".p2align 2\n\t" + ".long 1f - 0f\n\t" /* name length */ + ".long 3f - 2f\n\t" /* data length */ + ".long 1\n\t" /* note length */ + "0: .asciz \"GNU\"\n\t" /* vendor name */ + "1: .p2align 2\n\t" + "2: .long 0\n\t" /* note data: the ABI tag */ + ".long 2,4,20\n\t" /* Minimum kernel version w/TLS */ + "3: .p2align 2\n\t"); /* pad out section */ +#endif /* __linux__ */ + __asm__(".text"); __asm__("x86_current_tls:\n\t" @@ -37,7 +50,7 @@ __asm__("x86_current_tls:\n\t" "1:\n\t" "popl %eax\n\t" "addl $_GLOBAL_OFFSET_TABLE_+[.-1b], %eax\n\t" - "movl u_current_table_tls@GOTNTPOFF(%eax), %eax\n\t" + "movl u_current_table@GOTNTPOFF(%eax), %eax\n\t" "ret"); #ifndef GLX_X86_READONLY_TEXT diff --git a/src/mapi/mapi/sources.mak b/src/mapi/mapi/sources.mak index 5f327f3df62..ed36e1af486 100644 --- a/src/mapi/mapi/sources.mak +++ b/src/mapi/mapi/sources.mak @@ -1,12 +1,14 @@ # src/mapi/mapi/sources.mak # -# When MAPI_GLAPI_CURRENT is defined, MAPI_GLAPI_SOURCES can be built without -# MAPI_SOURCES and it is used by glapi. +# mapi may be used in several ways # -# Otherwise, MAPI_ABI_HEADER must be defined. It should expand to the header -# generated by mapi_abi.py. +# - In default mode, mapi implements the interface defined by mapi.h. To use +# this mode, compile MAPI_SOURCES. +# +# - In util mode, mapi provides utility functions for use with glapi. To use +# this mode, compile MAPI_UTIL_SOURCES with MAPI_MODE_UTIL defined. -MAPI_GLAPI_SOURCES = \ +MAPI_UTIL_SOURCES = \ u_current.c \ u_execmem.c \ u_thread.c @@ -16,4 +18,4 @@ MAPI_SOURCES = \ mapi.c \ stub.c \ table.c \ - $(MAPI_GLAPI_SOURCES) + $(MAPI_UTIL_SOURCES) diff --git a/src/mapi/mapi/u_current.c b/src/mapi/mapi/u_current.c index 77a593b330c..7946d2a2741 100644 --- a/src/mapi/mapi/u_current.c +++ b/src/mapi/mapi/u_current.c @@ -51,7 +51,7 @@ #include "u_current.h" #include "u_thread.h" -#ifndef MAPI_GLAPI_CURRENT +#ifndef MAPI_MODE_UTIL #include "table.h" #include "stub.h" @@ -99,16 +99,13 @@ extern void (*__glapi_noop_table[])(void); /*@{*/ #if defined(GLX_USE_TLS) -__thread struct mapi_table *u_current_table_tls +__thread struct mapi_table *u_current_table __attribute__((tls_model("initial-exec"))) = (struct mapi_table *) table_noop_array; -__thread void *u_current_user_tls +__thread void *u_current_user __attribute__((tls_model("initial-exec"))); -const struct mapi_table *u_current_table; -const void *u_current_user; - #else struct mapi_table *u_current_table = @@ -179,8 +176,8 @@ u_current_init(void) } else if (knownID != u_thread_self()) { ThreadSafe = 1; - u_current_set_internal(NULL); - u_current_set_user_internal(NULL); + u_current_set(NULL); + u_current_set_user(NULL); } CHECK_MULTITHREAD_UNLOCK(); } @@ -202,17 +199,17 @@ u_current_init(void) * void from the real context pointer type. */ void -u_current_set_user_internal(void *ptr) +u_current_set_user(const void *ptr) { u_current_init(); #if defined(GLX_USE_TLS) - u_current_user_tls = ptr; + u_current_user = (void *) ptr; #elif defined(THREADS) - u_tsd_set(&u_current_user_tsd, ptr); - u_current_user = (ThreadSafe) ? NULL : ptr; + u_tsd_set(&u_current_user_tsd, (void *) ptr); + u_current_user = (ThreadSafe) ? NULL : (void *) ptr; #else - u_current_user = ptr; + u_current_user = (void *) ptr; #endif } @@ -225,7 +222,7 @@ void * u_current_get_user_internal(void) { #if defined(GLX_USE_TLS) - return u_current_user_tls; + return u_current_user; #elif defined(THREADS) return (ThreadSafe) ? u_tsd_get(&u_current_user_tsd) @@ -241,22 +238,22 @@ u_current_get_user_internal(void) * table (__glapi_noop_table). */ void -u_current_set_internal(struct mapi_table *tbl) +u_current_set(const struct mapi_table *tbl) { u_current_init(); stub_init_once(); if (!tbl) - tbl = (struct mapi_table *) table_noop_array; + tbl = (const struct mapi_table *) table_noop_array; #if defined(GLX_USE_TLS) - u_current_table_tls = tbl; + u_current_table = (struct mapi_table *) tbl; #elif defined(THREADS) u_tsd_set(&u_current_table_tsd, (void *) tbl); - u_current_table = (ThreadSafe) ? NULL : tbl; + u_current_table = (ThreadSafe) ? NULL : (void *) tbl; #else - u_current_table = tbl; + u_current_table = (struct mapi_table *) tbl; #endif } @@ -267,7 +264,7 @@ struct mapi_table * u_current_get_internal(void) { #if defined(GLX_USE_TLS) - return u_current_table_tls; + return u_current_table; #elif defined(THREADS) return (struct mapi_table *) ((ThreadSafe) ? u_tsd_get(&u_current_table_tsd) : (void *) u_current_table); diff --git a/src/mapi/mapi/u_current.h b/src/mapi/mapi/u_current.h index 62e54c6c93d..061fccc9495 100644 --- a/src/mapi/mapi/u_current.h +++ b/src/mapi/mapi/u_current.h @@ -1,28 +1,27 @@ #ifndef _U_CURRENT_H_ #define _U_CURRENT_H_ -#ifdef MAPI_GLAPI_CURRENT +#ifdef MAPI_MODE_UTIL #include "glapi/glapi.h" /* ugly renames to match glapi.h */ #define mapi_table _glapi_table -#define u_current_table_tls _glapi_tls_Dispatch -#define u_current_user_tls _glapi_tls_Context +#ifdef GLX_USE_TLS +#define u_current_table _glapi_tls_Dispatch +#define u_current_user _glapi_tls_Context +#else #define u_current_table _glapi_Dispatch #define u_current_user _glapi_Context +#endif -#define u_current_destroy _glapi_destroy_multithread -#define u_current_init _glapi_check_multithread -#define u_current_set_internal _glapi_set_dispatch #define u_current_get_internal _glapi_get_dispatch -#define u_current_set_user_internal _glapi_set_context #define u_current_get_user_internal _glapi_get_context #define u_current_table_tsd _gl_DispatchTSD -#else /* MAPI_GLAPI_CURRENT */ +#else /* MAPI_MODE_UTIL */ #include "u_compiler.h" @@ -30,15 +29,12 @@ struct mapi_table; #ifdef GLX_USE_TLS -extern __thread struct mapi_table *u_current_table_tls +extern __thread struct mapi_table *u_current_table __attribute__((tls_model("initial-exec"))); -extern __thread void *u_current_user_tls +extern __thread void *u_current_user __attribute__((tls_model("initial-exec"))); -extern const struct mapi_table *u_current_table; -extern const void *u_current_user; - #else /* GLX_USE_TLS */ extern struct mapi_table *u_current_table; @@ -46,6 +42,8 @@ extern void *u_current_user; #endif /* GLX_USE_TLS */ +#endif /* MAPI_MODE_UTIL */ + void u_current_init(void); @@ -53,50 +51,36 @@ void u_current_destroy(void); void -u_current_set_internal(struct mapi_table *tbl); +u_current_set(const struct mapi_table *tbl); struct mapi_table * u_current_get_internal(void); void -u_current_set_user_internal(void *ptr); +u_current_set_user(const void *ptr); void * u_current_get_user_internal(void); -static INLINE void -u_current_set(const struct mapi_table *tbl) -{ - u_current_set_internal((struct mapi_table *) tbl); -} - static INLINE const struct mapi_table * u_current_get(void) { #ifdef GLX_USE_TLS - return (const struct mapi_table *) u_current_table_tls; + return u_current_table; #else return (likely(u_current_table) ? - (const struct mapi_table *) u_current_table : u_current_get_internal()); + u_current_table : u_current_get_internal()); #endif } -static INLINE void -u_current_set_user(void *ptr) -{ - u_current_set_internal(ptr); -} - -static INLINE void * +static INLINE const void * u_current_get_user(void) { #ifdef GLX_USE_TLS - return u_current_user_tls; + return u_current_user; #else return likely(u_current_user) ? u_current_user : u_current_get_user_internal(); #endif } -#endif /* MAPI_GLAPI_CURRENT */ - #endif /* _U_CURRENT_H_ */ diff --git a/src/mapi/mapi/u_thread.c b/src/mapi/mapi/u_thread.c index e9eae55364b..138db47b5cd 100644 --- a/src/mapi/mapi/u_thread.c +++ b/src/mapi/mapi/u_thread.c @@ -222,7 +222,7 @@ u_tsd_set(struct u_tsd *tsd, void *ptr) */ unsigned long -_glthread_GetID(void) +u_thread_self(void) { return 0; } diff --git a/src/mesa/drivers/dri/common/spantmp2.h b/src/mesa/drivers/dri/common/spantmp2.h index abd79562f98..f436d1398c2 100644 --- a/src/mesa/drivers/dri/common/spantmp2.h +++ b/src/mesa/drivers/dri/common/spantmp2.h @@ -48,6 +48,15 @@ #define HW_WRITE_CLIPLOOP() HW_CLIPLOOP() #endif +#ifdef SPANTMP_MESA_FMT +#define SPANTMP_PIXEL_FMT GL_NONE +#define SPANTMP_PIXEL_TYPE GL_NONE +#endif + +#ifndef SPANTMP_MESA_FMT +#define SPANTMP_MESA_FMT MESA_FORMAT_COUNT +#endif + #if (SPANTMP_PIXEL_FMT == GL_RGB) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_5_6_5) /** @@ -445,6 +454,118 @@ rgba[3] = p; \ } while (0) +#elif (SPANTMP_MESA_FMT == MESA_FORMAT_R8) + +#ifndef GET_VALUE +#ifndef GET_PTR +#define GET_PTR(_x, _y) ( buf + (_x) + (_y) * pitch) +#endif + +#define GET_VALUE(_x, _y) *(volatile GLubyte *)(GET_PTR(_x, _y)) +#define PUT_VALUE(_x, _y, _v) *(volatile GLubyte *)(GET_PTR(_x, _y)) = (_v) +#endif /* GET_VALUE */ + +# define INIT_MONO_PIXEL(p, color) \ + p = color[0] + +# define WRITE_RGBA(_x, _y, r, g, b, a) \ + PUT_VALUE(_x, _y, r) + +#define WRITE_PIXEL(_x, _y, p) PUT_VALUE(_x, _y, p) + +#define READ_RGBA( rgba, _x, _y ) \ + do { \ + GLubyte p = GET_VALUE(_x, _y); \ + rgba[0] = p; \ + rgba[1] = 0; \ + rgba[2] = 0; \ + rgba[3] = 0; \ + } while (0) + +#elif (SPANTMP_MESA_FMT == MESA_FORMAT_RG88) + +#ifndef GET_VALUE +#ifndef GET_PTR +#define GET_PTR(_x, _y) ( buf + (_x) * 2 + (_y) * pitch) +#endif + +#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y)) +#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v) +#endif /* GET_VALUE */ + +# define INIT_MONO_PIXEL(p, color) \ + PACK_COLOR_8888(color[0], color[1], 0, 0) + +# define WRITE_RGBA(_x, _y, r, g, b, a) \ + PUT_VALUE(_x, _y, r) + +#define WRITE_PIXEL(_x, _y, p) PUT_VALUE(_x, _y, p) + +#define READ_RGBA( rgba, _x, _y ) \ + do { \ + GLushort p = GET_VALUE(_x, _y); \ + rgba[0] = p & 0xff; \ + rgba[1] = (p >> 8) & 0xff; \ + rgba[2] = 0; \ + rgba[3] = 0; \ + } while (0) + +#elif (SPANTMP_MESA_FMT == MESA_FORMAT_R16) + +#ifndef GET_VALUE +#ifndef GET_PTR +#define GET_PTR(_x, _y) ( buf + (_x) * 2 + (_y) * pitch) +#endif + +#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y)) +#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v) +#endif /* GET_VALUE */ + +# define INIT_MONO_PIXEL(p, color) \ + p = color[0] + +# define WRITE_RGBA(_x, _y, r, g, b, a) \ + PUT_VALUE(_x, _y, r) + +#define WRITE_PIXEL(_x, _y, p) PUT_VALUE(_x, _y, p) + +#define READ_RGBA( rgba, _x, _y ) \ + do { \ + GLushort p = GET_VALUE(_x, _y); \ + rgba[0] = p; \ + rgba[1] = 0; \ + rgba[2] = 0; \ + rgba[3] = 0; \ + } while (0) + +#elif (SPANTMP_MESA_FMT == MESA_FORMAT_RG1616) + +#ifndef GET_VALUE +#ifndef GET_PTR +#define GET_PTR(_x, _y) ( buf + (_x) * 4 + (_y) * pitch) +#endif + +#define GET_VALUE(_x, _y) *(volatile GLuint *)(GET_PTR(_x, _y)) +#define PUT_VALUE(_x, _y, _v) *(volatile GLuint *)(GET_PTR(_x, _y)) = (_v) +#endif /* GET_VALUE */ + +# define INIT_MONO_PIXEL(p, color) \ + ((color[1] << 16) | (color[0])) + +# define WRITE_RGBA(_x, _y, r, g, b, a) \ + PUT_VALUE(_x, _y, r) + +#define WRITE_PIXEL(_x, _y, p) PUT_VALUE(_x, _y, p) + +#define READ_RGBA( rgba, _x, _y ) \ + do { \ + GLuint p = GET_VALUE(_x, _y); \ + rgba[0] = p & 0xffff; \ + rgba[1] = (p >> 16) & 0xffff; \ + rgba[2] = 0; \ + rgba[3] = 0; \ + } while (0) + #else #error SPANTMP_PIXEL_FMT must be set to a valid value! #endif @@ -914,3 +1035,4 @@ static void TAG(InitPointers)(struct gl_renderbuffer *rb) #undef GET_PTR #undef SPANTMP_PIXEL_FMT #undef SPANTMP_PIXEL_TYPE +#undef SPANTMP_MESA_FMT diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c index f7fdb78d059..1621c9544ac 100644 --- a/src/mesa/drivers/dri/i915/i830_vtbl.c +++ b/src/mesa/drivers/dri/i915/i830_vtbl.c @@ -364,7 +364,7 @@ i830_emit_invarient_state(struct intel_context *intel) #define emit( intel, state, size ) \ - intel_batchbuffer_data(intel->batch, state, size ) + intel_batchbuffer_data(intel->batch, state, size, false) static GLuint get_dirty(struct i830_hw_state *state) @@ -429,7 +429,8 @@ i830_emit_state(struct intel_context *intel) * batchbuffer fills up. */ intel_batchbuffer_require_space(intel->batch, - get_state_size(state) + INTEL_PRIM_EMIT_SIZE); + get_state_size(state) + INTEL_PRIM_EMIT_SIZE, + false); count = 0; again: aper_count = 0; diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c index 7a9fb7f088b..1c6e9845172 100644 --- a/src/mesa/drivers/dri/i915/i915_fragprog.c +++ b/src/mesa/drivers/dri/i915/i915_fragprog.c @@ -1162,11 +1162,6 @@ translate_program(struct i915_fragment_program *p) fixup_depth_write(p); i915_fini_program(p); - if (INTEL_DEBUG & DEBUG_WM) { - printf("i915:\n"); - i915_disassemble_program(i915->state.Program, i915->state.ProgramSize); - } - p->translated = 1; } @@ -1427,6 +1422,11 @@ i915ValidateFragmentProgram(struct i915_context *i915) if (!p->on_hardware) i915_upload_program(i915, p); + + if (INTEL_DEBUG & DEBUG_WM) { + printf("i915:\n"); + i915_disassemble_program(i915->state.Program, i915->state.ProgramSize); + } } void diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c index 59dfe085632..8d9020f5ef3 100644 --- a/src/mesa/drivers/dri/i915/i915_vtbl.c +++ b/src/mesa/drivers/dri/i915/i915_vtbl.c @@ -217,7 +217,7 @@ i915_emit_invarient_state(struct intel_context *intel) #define emit(intel, state, size ) \ - intel_batchbuffer_data(intel->batch, state, size) + intel_batchbuffer_data(intel->batch, state, size, false) static GLuint get_dirty(struct i915_hw_state *state) @@ -300,7 +300,8 @@ i915_emit_state(struct intel_context *intel) * batchbuffer fills up. */ intel_batchbuffer_require_space(intel->batch, - get_state_size(state) + INTEL_PRIM_EMIT_SIZE); + get_state_size(state) + INTEL_PRIM_EMIT_SIZE, + false); count = 0; again: aper_count = 0; diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h index 7f3e4986808..b48a30d6be9 100644 --- a/src/mesa/drivers/dri/i965/brw_defines.h +++ b/src/mesa/drivers/dri/i965/brw_defines.h @@ -906,6 +906,8 @@ # define GEN6_VS_VECTOR_MASK_ENABLE (1 << 30) # define GEN6_VS_SAMPLER_COUNT_SHIFT 27 # define GEN6_VS_BINDING_TABLE_ENTRY_COUNT_SHIFT 18 +# define GEN6_VS_FLOATING_POINT_MODE_IEEE_754 (0 << 16) +# define GEN6_VS_FLOATING_POINT_MODE_ALT (1 << 16) /* DW4 */ # define GEN6_VS_DISPATCH_START_GRF_SHIFT 20 # define GEN6_VS_URB_READ_LENGTH_SHIFT 11 @@ -1048,6 +1050,8 @@ # define GEN6_WM_VECTOR_MASK_ENABLE (1 << 30) # define GEN6_WM_SAMPLER_COUNT_SHIFT 27 # define GEN6_WM_BINDING_TABLE_ENTRY_COUNT_SHIFT 18 +# define GEN6_WM_FLOATING_POINT_MODE_IEEE_754 (0 << 16) +# define GEN6_WM_FLOATING_POINT_MODE_ALT (1 << 16) /* DW3: scratch space */ /* DW4 */ # define GEN6_WM_STATISTICS_ENABLE (1 << 31) diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c index a1f403ca4e6..7eb16b71f4a 100644 --- a/src/mesa/drivers/dri/i965/brw_draw.c +++ b/src/mesa/drivers/dri/i965/brw_draw.c @@ -159,7 +159,7 @@ static void brw_emit_prim(struct brw_context *brw, } if (prim_packet.verts_per_instance) { intel_batchbuffer_data( brw->intel.batch, &prim_packet, - sizeof(prim_packet)); + sizeof(prim_packet), false); } if (intel->always_flush_cache) { intel_batchbuffer_emit_mi_flush(intel->batch); @@ -351,7 +351,8 @@ static GLboolean brw_try_draw_prims( struct gl_context *ctx, * an upper bound of how much we might emit in a single * brw_try_draw_prims(). */ - intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4); + intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4, + false); hw_prim = brw_set_prim(brw, &prim[i]); diff --git a/src/mesa/drivers/dri/i965/brw_eu.h b/src/mesa/drivers/dri/i965/brw_eu.h index 4dbdc522100..119ffc72370 100644 --- a/src/mesa/drivers/dri/i965/brw_eu.h +++ b/src/mesa/drivers/dri/i965/brw_eu.h @@ -861,7 +861,8 @@ void brw_fb_WRITE(struct brw_compile *p, GLuint binding_table_index, GLuint msg_length, GLuint response_length, - GLboolean eot); + GLboolean eot, + GLboolean header_present); void brw_SAMPLE(struct brw_compile *p, struct brw_reg dest, diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c index 9c764fe779d..6d48ca0e46d 100644 --- a/src/mesa/drivers/dri/i965/brw_eu_emit.c +++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c @@ -1812,12 +1812,12 @@ void brw_fb_WRITE(struct brw_compile *p, GLuint binding_table_index, GLuint msg_length, GLuint response_length, - GLboolean eot) + GLboolean eot, + GLboolean header_present) { struct intel_context *intel = &p->brw->intel; struct brw_instruction *insn; GLuint msg_control, msg_type; - GLboolean header_present = GL_TRUE; if (intel->gen >= 6 && binding_table_index == 0) { insn = next_insn(p, BRW_OPCODE_SENDC); @@ -1829,9 +1829,6 @@ void brw_fb_WRITE(struct brw_compile *p, insn->header.compression_control = BRW_COMPRESSION_NONE; if (intel->gen >= 6) { - if (msg_length == 4) - header_present = GL_FALSE; - /* headerless version, just submit color payload */ src0 = brw_message_reg(msg_reg_nr); @@ -1936,7 +1933,8 @@ void brw_SAMPLE(struct brw_compile *p, brw_set_compression_control(p, BRW_COMPRESSION_NONE); brw_set_mask_control(p, BRW_MASK_DISABLE); - brw_MOV(p, m1, brw_vec8_grf(0,0)); + brw_MOV(p, retype(m1, BRW_REGISTER_TYPE_UD), + retype(brw_vec8_grf(0,0), BRW_REGISTER_TYPE_UD)); brw_MOV(p, get_element_ud(m1, 2), brw_imm_ud(newmask << 12)); brw_pop_insn_state(p); @@ -1997,7 +1995,8 @@ void brw_SAMPLE(struct brw_compile *p, */ brw_push_insn_state(p); brw_set_compression_control(p, BRW_COMPRESSION_NONE); - brw_MOV(p, reg, reg); + brw_MOV(p, retype(reg, BRW_REGISTER_TYPE_UD), + retype(reg, BRW_REGISTER_TYPE_UD)); brw_pop_insn_state(p); } @@ -2029,7 +2028,8 @@ void brw_urb_WRITE(struct brw_compile *p, if (intel->gen >= 6) { brw_push_insn_state(p); brw_set_mask_control( p, BRW_MASK_DISABLE ); - brw_MOV(p, brw_message_reg(msg_reg_nr), src0); + brw_MOV(p, retype(brw_message_reg(msg_reg_nr), BRW_REGISTER_TYPE_UD), + retype(src0, BRW_REGISTER_TYPE_UD)); brw_pop_insn_state(p); src0 = brw_message_reg(msg_reg_nr); } diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp index c3cbe0df618..2de81b28371 100644 --- a/src/mesa/drivers/dri/i965/brw_fs.cpp +++ b/src/mesa/drivers/dri/i965/brw_fs.cpp @@ -474,8 +474,13 @@ fs_visitor::emit_fragcoord_interpolation(ir_variable *ir) wpos.reg_offset++; /* gl_FragCoord.z */ - emit(fs_inst(FS_OPCODE_LINTERP, wpos, this->delta_x, this->delta_y, - interp_reg(FRAG_ATTRIB_WPOS, 2))); + if (intel->gen >= 6) { + emit(fs_inst(BRW_OPCODE_MOV, wpos, + fs_reg(brw_vec8_grf(c->source_depth_reg, 0)))); + } else { + emit(fs_inst(FS_OPCODE_LINTERP, wpos, this->delta_x, this->delta_y, + interp_reg(FRAG_ATTRIB_WPOS, 2))); + } wpos.reg_offset++; /* gl_FragCoord.w: Already set up in emit_interpolation */ @@ -2158,7 +2163,8 @@ fs_visitor::generate_fb_write(fs_inst *inst) inst->target, inst->mlen, 0, - eot); + eot, + inst->header_present); } void diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h index 3beed16945b..4bb93e73369 100644 --- a/src/mesa/drivers/dri/i965/brw_state.h +++ b/src/mesa/drivers/dri/i965/brw_state.h @@ -164,7 +164,8 @@ void brw_destroy_caches( struct brw_context *brw ); /*********************************************************************** * brw_state_batch.c */ -#define BRW_BATCH_STRUCT(brw, s) intel_batchbuffer_data( brw->intel.batch, (s), sizeof(*(s))) +#define BRW_BATCH_STRUCT(brw, s) intel_batchbuffer_data(brw->intel.batch, (s), \ + sizeof(*(s)), false) #define BRW_CACHED_BATCH_STRUCT(brw, s) brw_cached_batch_struct( brw, (s), sizeof(*(s)) ) GLboolean brw_cached_batch_struct( struct brw_context *brw, diff --git a/src/mesa/drivers/dri/i965/brw_state_batch.c b/src/mesa/drivers/dri/i965/brw_state_batch.c index be3989eb7db..a21af13caa3 100644 --- a/src/mesa/drivers/dri/i965/brw_state_batch.c +++ b/src/mesa/drivers/dri/i965/brw_state_batch.c @@ -48,7 +48,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw, struct header *newheader = (struct header *)data; if (brw->emit_state_always) { - intel_batchbuffer_data(brw->intel.batch, data, sz); + intel_batchbuffer_data(brw->intel.batch, data, sz, false); return GL_TRUE; } @@ -75,7 +75,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw, emit: memcpy(item->header, newheader, sz); - intel_batchbuffer_data(brw->intel.batch, data, sz); + intel_batchbuffer_data(brw->intel.batch, data, sz, false); return GL_TRUE; } diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index 09887dae95d..326bb1e562f 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -658,6 +658,22 @@ static void emit_min( struct brw_compile *p, } } +static void emit_arl(struct brw_compile *p, + struct brw_reg dst, + struct brw_reg src) +{ + struct intel_context *intel = &p->brw->intel; + + if (intel->gen >= 6) { + struct brw_reg dst_f = retype(dst, BRW_REGISTER_TYPE_F); + + brw_RNDD(p, dst_f, src); + brw_MOV(p, dst, dst_f); + } else { + brw_RNDD(p, dst, src); + } +} + static void emit_math1_gen4(struct brw_vs_compile *c, GLuint function, struct brw_reg dst, @@ -1963,7 +1979,7 @@ void brw_vs_emit(struct brw_vs_compile *c ) emit_math1(c, BRW_MATH_FUNCTION_EXP, dst, args[0], BRW_MATH_PRECISION_FULL); break; case OPCODE_ARL: - brw_RNDD(p, dst, args[0]); + emit_arl(p, dst, args[0]); break; case OPCODE_FLR: brw_RNDD(p, dst, args[0]); diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c index a0e86034e1e..56725c0d471 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_emit.c +++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c @@ -219,43 +219,45 @@ void emit_wpos_xy(struct brw_wm_compile *c, const struct brw_reg *arg0) { struct brw_compile *p = &c->func; + struct intel_context *intel = &p->brw->intel; + struct brw_reg delta_x = retype(arg0[0], BRW_REGISTER_TYPE_W); + struct brw_reg delta_y = retype(arg0[1], BRW_REGISTER_TYPE_W); if (mask & WRITEMASK_X) { + if (intel->gen >= 6) { + struct brw_reg delta_x_f = retype(delta_x, BRW_REGISTER_TYPE_F); + brw_MOV(p, delta_x_f, delta_x); + delta_x = delta_x_f; + } + if (c->fp->program.PixelCenterInteger) { /* X' = X */ - brw_MOV(p, - dst[0], - retype(arg0[0], BRW_REGISTER_TYPE_W)); + brw_MOV(p, dst[0], delta_x); } else { /* X' = X + 0.5 */ - brw_ADD(p, - dst[0], - retype(arg0[0], BRW_REGISTER_TYPE_W), - brw_imm_f(0.5)); + brw_ADD(p, dst[0], delta_x, brw_imm_f(0.5)); } } if (mask & WRITEMASK_Y) { + if (intel->gen >= 6) { + struct brw_reg delta_y_f = retype(delta_y, BRW_REGISTER_TYPE_F); + brw_MOV(p, delta_y_f, delta_y); + delta_y = delta_y_f; + } + if (c->fp->program.OriginUpperLeft) { if (c->fp->program.PixelCenterInteger) { /* Y' = Y */ - brw_MOV(p, - dst[1], - retype(arg0[1], BRW_REGISTER_TYPE_W)); + brw_MOV(p, dst[1], delta_y); } else { - /* Y' = Y + 0.5 */ - brw_ADD(p, - dst[1], - retype(arg0[1], BRW_REGISTER_TYPE_W), - brw_imm_f(0.5)); + brw_ADD(p, dst[1], delta_y, brw_imm_f(0.5)); } } else { float center_offset = c->fp->program.PixelCenterInteger ? 0.0 : 0.5; /* Y' = (height - 1) - Y + center */ - brw_ADD(p, - dst[1], - negate(retype(arg0[1], BRW_REGISTER_TYPE_W)), + brw_ADD(p, dst[1], negate(delta_y), brw_imm_f(c->key.drawable_height - 1 + center_offset)); } } @@ -971,34 +973,23 @@ void emit_math2(struct brw_wm_compile *c, struct brw_reg temp_dst = dst[dst_chan]; if (arg0[0].hstride == BRW_HORIZONTAL_STRIDE_0) { - if (arg1[0].hstride == BRW_HORIZONTAL_STRIDE_0) { - /* Both scalar arguments. Do scalar calc. */ - src0.hstride = BRW_HORIZONTAL_STRIDE_1; - src1.hstride = BRW_HORIZONTAL_STRIDE_1; - temp_dst.hstride = BRW_HORIZONTAL_STRIDE_1; - temp_dst.width = BRW_WIDTH_1; - - if (arg0[0].subnr != 0) { - brw_MOV(p, temp_dst, src0); - src0 = temp_dst; - - /* Ouch. We've used the temp as a dst, and we still - * need a temp to store arg1 in, because src and dst - * offsets have to be equal. Leaving this up to - * glsl2-965 to handle correctly. - */ - assert(arg1[0].subnr == 0); - } else if (arg1[0].subnr != 0) { - brw_MOV(p, temp_dst, src1); - src1 = temp_dst; - } - } else { - brw_MOV(p, temp_dst, src0); - src0 = temp_dst; - } - } else if (arg1[0].hstride == BRW_HORIZONTAL_STRIDE_0) { - brw_MOV(p, temp_dst, src1); - src1 = temp_dst; + brw_MOV(p, temp_dst, src0); + src0 = temp_dst; + } + + if (arg1[0].hstride == BRW_HORIZONTAL_STRIDE_0) { + /* This is a heinous hack to get a temporary register for use + * in case both arg0 and arg1 are constants. Why you're + * doing exponentiation on constant values in the shader, we + * don't know. + * + * max_wm_grf is almost surely less than the maximum GRF, and + * gen6 doesn't care about the number of GRFs used in a + * shader like pre-gen6 did. + */ + struct brw_reg temp = brw_vec8_grf(c->max_wm_grf, 0); + brw_MOV(p, temp, src1); + src1 = temp; } brw_set_saturate(p, (mask & SATURATE) ? 1 : 0); @@ -1016,14 +1007,6 @@ void emit_math2(struct brw_wm_compile *c, sechalf(src0), sechalf(src1)); } - - /* Splat a scalar result into all the channels. */ - if (arg0[0].hstride == BRW_HORIZONTAL_STRIDE_0 && - arg1[0].hstride == BRW_HORIZONTAL_STRIDE_0) { - temp_dst.hstride = BRW_HORIZONTAL_STRIDE_0; - temp_dst.vstride = BRW_VERTICAL_STRIDE_0; - brw_MOV(p, dst[dst_chan], temp_dst); - } } else { GLuint saturate = ((mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : @@ -1373,7 +1356,8 @@ static void fire_fb_write( struct brw_wm_compile *c, target, nr, 0, - eot); + eot, + GL_TRUE); } @@ -1518,7 +1502,8 @@ void emit_fb_write(struct brw_wm_compile *c, */ brw_push_insn_state(p); brw_set_mask_control(p, BRW_MASK_DISABLE); - brw_MOV(p, brw_message_reg(0), brw_vec8_grf(0, 0)); + brw_MOV(p, retype(brw_message_reg(0), BRW_REGISTER_TYPE_UD), + retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UD)); brw_pop_insn_state(p); if (target != 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 ad744044c70..1cd736a1119 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c +++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c @@ -68,71 +68,43 @@ static GLuint translate_tex_target( GLenum target ) } } +static uint32_t brw_format_for_mesa_format[MESA_FORMAT_COUNT] = +{ + [MESA_FORMAT_L8] = BRW_SURFACEFORMAT_L8_UNORM, + [MESA_FORMAT_I8] = BRW_SURFACEFORMAT_I8_UNORM, + [MESA_FORMAT_A8] = BRW_SURFACEFORMAT_A8_UNORM, + [MESA_FORMAT_AL88] = BRW_SURFACEFORMAT_L8A8_UNORM, + [MESA_FORMAT_AL1616] = BRW_SURFACEFORMAT_L16A16_UNORM, + [MESA_FORMAT_R8] = BRW_SURFACEFORMAT_R8_UNORM, + [MESA_FORMAT_R16] = BRW_SURFACEFORMAT_R16_UNORM, + [MESA_FORMAT_RG88] = BRW_SURFACEFORMAT_R8G8_UNORM, + [MESA_FORMAT_RG1616] = BRW_SURFACEFORMAT_R16G16_UNORM, + [MESA_FORMAT_ARGB8888] = BRW_SURFACEFORMAT_B8G8R8A8_UNORM, + [MESA_FORMAT_XRGB8888] = BRW_SURFACEFORMAT_B8G8R8X8_UNORM, + [MESA_FORMAT_RGB565] = BRW_SURFACEFORMAT_B5G6R5_UNORM, + [MESA_FORMAT_ARGB1555] = BRW_SURFACEFORMAT_B5G5R5A1_UNORM, + [MESA_FORMAT_ARGB4444] = BRW_SURFACEFORMAT_B4G4R4A4_UNORM, + [MESA_FORMAT_YCBCR_REV] = BRW_SURFACEFORMAT_YCRCB_NORMAL, + [MESA_FORMAT_YCBCR] = BRW_SURFACEFORMAT_YCRCB_SWAPUVY, + [MESA_FORMAT_RGB_FXT1] = BRW_SURFACEFORMAT_FXT1, + [MESA_FORMAT_RGBA_FXT1] = BRW_SURFACEFORMAT_FXT1, + [MESA_FORMAT_RGB_DXT1] = BRW_SURFACEFORMAT_DXT1_RGB, + [MESA_FORMAT_RGBA_DXT1] = BRW_SURFACEFORMAT_BC1_UNORM, + [MESA_FORMAT_RGBA_DXT3] = BRW_SURFACEFORMAT_BC2_UNORM, + [MESA_FORMAT_RGBA_DXT5] = BRW_SURFACEFORMAT_BC3_UNORM, + [MESA_FORMAT_SRGB_DXT1] = BRW_SURFACEFORMAT_BC1_UNORM_SRGB, + [MESA_FORMAT_SARGB8] = BRW_SURFACEFORMAT_B8G8R8A8_UNORM_SRGB, + [MESA_FORMAT_SLA8] = BRW_SURFACEFORMAT_L8A8_UNORM_SRGB, + [MESA_FORMAT_SL8] = BRW_SURFACEFORMAT_L8_UNORM_SRGB, + [MESA_FORMAT_DUDV8] = BRW_SURFACEFORMAT_R8G8_SNORM, + [MESA_FORMAT_SIGNED_RGBA8888_REV] = BRW_SURFACEFORMAT_R8G8B8A8_SNORM, +}; static GLuint translate_tex_format( gl_format mesa_format, GLenum internal_format, GLenum depth_mode ) { switch( mesa_format ) { - case MESA_FORMAT_L8: - return BRW_SURFACEFORMAT_L8_UNORM; - - case MESA_FORMAT_I8: - return BRW_SURFACEFORMAT_I8_UNORM; - - case MESA_FORMAT_A8: - return BRW_SURFACEFORMAT_A8_UNORM; - - case MESA_FORMAT_AL88: - return BRW_SURFACEFORMAT_L8A8_UNORM; - - case MESA_FORMAT_AL1616: - return BRW_SURFACEFORMAT_L16A16_UNORM; - - case MESA_FORMAT_R8: - return BRW_SURFACEFORMAT_R8_UNORM; - - case MESA_FORMAT_R16: - return BRW_SURFACEFORMAT_R16_UNORM; - - case MESA_FORMAT_RG88: - return BRW_SURFACEFORMAT_R8G8_UNORM; - - case MESA_FORMAT_RG1616: - return BRW_SURFACEFORMAT_R16G16_UNORM; - - case MESA_FORMAT_RGB888: - assert(0); /* not supported for sampling */ - return BRW_SURFACEFORMAT_R8G8B8_UNORM; - - case MESA_FORMAT_ARGB8888: - return BRW_SURFACEFORMAT_B8G8R8A8_UNORM; - - case MESA_FORMAT_XRGB8888: - return BRW_SURFACEFORMAT_B8G8R8X8_UNORM; - - case MESA_FORMAT_RGBA8888_REV: - _mesa_problem(NULL, "unexpected format in i965:translate_tex_format()"); - return BRW_SURFACEFORMAT_R8G8B8A8_UNORM; - - case MESA_FORMAT_RGB565: - return BRW_SURFACEFORMAT_B5G6R5_UNORM; - - case MESA_FORMAT_ARGB1555: - return BRW_SURFACEFORMAT_B5G5R5A1_UNORM; - - case MESA_FORMAT_ARGB4444: - return BRW_SURFACEFORMAT_B4G4R4A4_UNORM; - - case MESA_FORMAT_YCBCR_REV: - return BRW_SURFACEFORMAT_YCRCB_NORMAL; - - case MESA_FORMAT_YCBCR: - return BRW_SURFACEFORMAT_YCRCB_SWAPUVY; - - case MESA_FORMAT_RGB_FXT1: - case MESA_FORMAT_RGBA_FXT1: - return BRW_SURFACEFORMAT_FXT1; case MESA_FORMAT_Z16: if (depth_mode == GL_INTENSITY) @@ -144,30 +116,6 @@ static GLuint translate_tex_format( gl_format mesa_format, else return BRW_SURFACEFORMAT_L16_UNORM; - case MESA_FORMAT_RGB_DXT1: - return BRW_SURFACEFORMAT_DXT1_RGB; - - case MESA_FORMAT_RGBA_DXT1: - return BRW_SURFACEFORMAT_BC1_UNORM; - - case MESA_FORMAT_RGBA_DXT3: - return BRW_SURFACEFORMAT_BC2_UNORM; - - case MESA_FORMAT_RGBA_DXT5: - return BRW_SURFACEFORMAT_BC3_UNORM; - - case MESA_FORMAT_SARGB8: - return BRW_SURFACEFORMAT_B8G8R8A8_UNORM_SRGB; - - case MESA_FORMAT_SLA8: - return BRW_SURFACEFORMAT_L8A8_UNORM_SRGB; - - case MESA_FORMAT_SL8: - return BRW_SURFACEFORMAT_L8_UNORM_SRGB; - - case MESA_FORMAT_SRGB_DXT1: - return BRW_SURFACEFORMAT_BC1_UNORM_SRGB; - case MESA_FORMAT_S8_Z24: /* XXX: these different surface formats don't seem to * make any difference for shadow sampler/compares. @@ -181,15 +129,9 @@ static GLuint translate_tex_format( gl_format mesa_format, else return BRW_SURFACEFORMAT_L24X8_UNORM; - case MESA_FORMAT_DUDV8: - return BRW_SURFACEFORMAT_R8G8_SNORM; - - case MESA_FORMAT_SIGNED_RGBA8888_REV: - return BRW_SURFACEFORMAT_R8G8B8A8_SNORM; - default: - assert(0); - return 0; + assert(brw_format_for_mesa_format[mesa_format] != 0); + return brw_format_for_mesa_format[mesa_format]; } } @@ -448,45 +390,19 @@ brw_update_renderbuffer_surface(struct brw_context *brw, key.surface_type = BRW_SURFACE_2D; switch (irb->Base.Format) { - /* XRGB and ARGB are treated the same here because the chips in this - * family cannot render to XRGB targets. This means that we have to - * mask writes to alpha (ala glColorMask) and reconfigure the alpha - * blending hardware to use GL_ONE (or GL_ZERO) for cases where - * GL_DST_ALPHA (or GL_ONE_MINUS_DST_ALPHA) is used. - */ - case MESA_FORMAT_ARGB8888: case MESA_FORMAT_XRGB8888: + /* XRGB is handled as ARGB because the chips in this family + * cannot render to XRGB targets. This means that we have to + * mask writes to alpha (ala glColorMask) and reconfigure the + * alpha blending hardware to use GL_ONE (or GL_ZERO) for + * cases where GL_DST_ALPHA (or GL_ONE_MINUS_DST_ALPHA) is + * used. + */ key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM; break; - case MESA_FORMAT_SARGB8: - key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM_SRGB; - break; - case MESA_FORMAT_RGB565: - key.surface_format = BRW_SURFACEFORMAT_B5G6R5_UNORM; - break; - case MESA_FORMAT_ARGB1555: - key.surface_format = BRW_SURFACEFORMAT_B5G5R5A1_UNORM; - break; - case MESA_FORMAT_ARGB4444: - key.surface_format = BRW_SURFACEFORMAT_B4G4R4A4_UNORM; - break; - case MESA_FORMAT_A8: - key.surface_format = BRW_SURFACEFORMAT_A8_UNORM; - break; - case MESA_FORMAT_R8: - key.surface_format = BRW_SURFACEFORMAT_R8_UNORM; - break; - case MESA_FORMAT_R16: - key.surface_format = BRW_SURFACEFORMAT_R16_UNORM; - break; - case MESA_FORMAT_RG88: - key.surface_format = BRW_SURFACEFORMAT_R8G8_UNORM; - break; - case MESA_FORMAT_RG1616: - key.surface_format = BRW_SURFACEFORMAT_R16G16_UNORM; - break; default: - _mesa_problem(ctx, "Bad renderbuffer format: %d\n", irb->Base.Format); + key.surface_format = brw_format_for_mesa_format[irb->Base.Format]; + assert(key.surface_format != 0); } key.tiling = region->tiling; key.width = rb->Width; diff --git a/src/mesa/drivers/dri/i965/gen6_vs_state.c b/src/mesa/drivers/dri/i965/gen6_vs_state.c index 4ef9e2e6072..ed132bdbd93 100644 --- a/src/mesa/drivers/dri/i965/gen6_vs_state.c +++ b/src/mesa/drivers/dri/i965/gen6_vs_state.c @@ -130,6 +130,7 @@ upload_vs_state(struct brw_context *brw) OUT_BATCH(CMD_3D_VS_STATE << 16 | (6 - 2)); OUT_RELOC(brw->vs.prog_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0); OUT_BATCH((0 << GEN6_VS_SAMPLER_COUNT_SHIFT) | + GEN6_VS_FLOATING_POINT_MODE_ALT | (brw->vs.nr_surfaces << GEN6_VS_BINDING_TABLE_ENTRY_COUNT_SHIFT)); OUT_BATCH(0); /* scratch space base offset */ OUT_BATCH((1 << GEN6_VS_DISPATCH_START_GRF_SHIFT) | diff --git a/src/mesa/drivers/dri/i965/gen6_wm_state.c b/src/mesa/drivers/dri/i965/gen6_wm_state.c index d80df4e254b..2ae0c093ebe 100644 --- a/src/mesa/drivers/dri/i965/gen6_wm_state.c +++ b/src/mesa/drivers/dri/i965/gen6_wm_state.c @@ -133,6 +133,9 @@ upload_wm_state(struct brw_context *brw) dw5 |= GEN6_WM_LINE_AA_WIDTH_1_0; dw5 |= GEN6_WM_LINE_END_CAP_AA_WIDTH_0_5; + /* OpenGL non-ieee floating point mode */ + dw2 |= GEN6_WM_FLOATING_POINT_MODE_ALT; + /* BRW_NEW_NR_WM_SURFACES */ dw2 |= brw->wm.nr_surfaces << GEN6_WM_BINDING_TABLE_ENTRY_COUNT_SHIFT; diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.c b/src/mesa/drivers/dri/intel/intel_batchbuffer.c index 21fc9ece886..20574ab5462 100644 --- a/src/mesa/drivers/dri/intel/intel_batchbuffer.c +++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.c @@ -93,8 +93,16 @@ do_flush_locked(struct intel_batchbuffer *batch, GLuint used) batch->ptr = NULL; if (!intel->intelScreen->no_hw) { - drm_intel_bo_exec(batch->buf, used, NULL, 0, - (x_off & 0xffff) | (y_off << 16)); + int ring; + + if (intel->gen < 6 || !intel->batch->is_blit) { + ring = I915_EXEC_RENDER; + } else { + ring = I915_EXEC_BLT; + } + + drm_intel_bo_mrb_exec(batch->buf, used, NULL, 0, + (x_off & 0xffff) | (y_off << 16), ring); } if (unlikely(INTEL_DEBUG & DEBUG_BATCH)) { @@ -242,10 +250,10 @@ intel_batchbuffer_emit_reloc_fenced(struct intel_batchbuffer *batch, void intel_batchbuffer_data(struct intel_batchbuffer *batch, - const void *data, GLuint bytes) + const void *data, GLuint bytes, bool is_blit) { assert((bytes & 3) == 0); - intel_batchbuffer_require_space(batch, bytes); + intel_batchbuffer_require_space(batch, bytes, is_blit); __memcpy(batch->ptr, data, bytes); batch->ptr += bytes; } @@ -262,22 +270,29 @@ intel_batchbuffer_emit_mi_flush(struct intel_batchbuffer *batch) struct intel_context *intel = batch->intel; if (intel->gen >= 6) { - BEGIN_BATCH(8); - - /* XXX workaround: issue any post sync != 0 before write cache flush = 1 */ - OUT_BATCH(_3DSTATE_PIPE_CONTROL); - OUT_BATCH(PIPE_CONTROL_WRITE_IMMEDIATE); - OUT_BATCH(0); /* write address */ - OUT_BATCH(0); /* write data */ - - OUT_BATCH(_3DSTATE_PIPE_CONTROL); - OUT_BATCH(PIPE_CONTROL_INSTRUCTION_FLUSH | - PIPE_CONTROL_WRITE_FLUSH | - PIPE_CONTROL_DEPTH_CACHE_FLUSH | - PIPE_CONTROL_NO_WRITE); - OUT_BATCH(0); /* write address */ - OUT_BATCH(0); /* write data */ - ADVANCE_BATCH(); + if (intel->batch->is_blit) { + BEGIN_BATCH_BLT(1); + OUT_BATCH(MI_FLUSH); + ADVANCE_BATCH(); + } else { + BEGIN_BATCH(8); + /* XXX workaround: issue any post sync != 0 before write + * cache flush = 1 + */ + OUT_BATCH(_3DSTATE_PIPE_CONTROL); + OUT_BATCH(PIPE_CONTROL_WRITE_IMMEDIATE); + OUT_BATCH(0); /* write address */ + OUT_BATCH(0); /* write data */ + + OUT_BATCH(_3DSTATE_PIPE_CONTROL); + OUT_BATCH(PIPE_CONTROL_INSTRUCTION_FLUSH | + PIPE_CONTROL_WRITE_FLUSH | + PIPE_CONTROL_DEPTH_CACHE_FLUSH | + PIPE_CONTROL_NO_WRITE); + OUT_BATCH(0); /* write address */ + OUT_BATCH(0); /* write data */ + ADVANCE_BATCH(); + } } else if (intel->gen >= 4) { BEGIN_BATCH(4); OUT_BATCH(_3DSTATE_PIPE_CONTROL | diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.h b/src/mesa/drivers/dri/intel/intel_batchbuffer.h index 428c027c2f1..635708587a6 100644 --- a/src/mesa/drivers/dri/intel/intel_batchbuffer.h +++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.h @@ -31,6 +31,7 @@ struct intel_batchbuffer } emit; #endif + bool is_blit; GLuint dirty_state; GLuint reserved_space; }; @@ -55,7 +56,7 @@ void intel_batchbuffer_reset(struct intel_batchbuffer *batch); * intel_buffer_dword() calls. */ void intel_batchbuffer_data(struct intel_batchbuffer *batch, - const void *data, GLuint bytes); + const void *data, GLuint bytes, bool is_blit); void intel_batchbuffer_release_space(struct intel_batchbuffer *batch, GLuint bytes); @@ -114,8 +115,16 @@ intel_batchbuffer_emit_float(struct intel_batchbuffer *batch, float f) static INLINE void intel_batchbuffer_require_space(struct intel_batchbuffer *batch, - GLuint sz) + GLuint sz, int is_blit) { + + if (batch->intel->gen >= 6 && batch->is_blit != is_blit && + batch->ptr != batch->map) { + intel_batchbuffer_flush(batch); + } + + batch->is_blit = is_blit; + #ifdef DEBUG assert(sz < batch->size - 8); #endif @@ -124,9 +133,10 @@ intel_batchbuffer_require_space(struct intel_batchbuffer *batch, } static INLINE void -intel_batchbuffer_begin(struct intel_batchbuffer *batch, int n) +intel_batchbuffer_begin(struct intel_batchbuffer *batch, int n, bool is_blit) { - intel_batchbuffer_require_space(batch, n * 4); + intel_batchbuffer_require_space(batch, n * 4, is_blit); + #ifdef DEBUG assert(batch->map); assert(batch->emit.start_ptr == NULL); @@ -154,7 +164,8 @@ intel_batchbuffer_advance(struct intel_batchbuffer *batch) */ #define BATCH_LOCALS -#define BEGIN_BATCH(n) intel_batchbuffer_begin(intel->batch, n) +#define BEGIN_BATCH(n) intel_batchbuffer_begin(intel->batch, n, false) +#define BEGIN_BATCH_BLT(n) intel_batchbuffer_begin(intel->batch, n, true) #define OUT_BATCH(d) intel_batchbuffer_emit_dword(intel->batch, d) #define OUT_BATCH_F(f) intel_batchbuffer_emit_float(intel->batch,f) #define OUT_RELOC(buf, read_domains, write_domain, delta) do { \ diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c index c2917e9b07e..a2822b11d96 100644 --- a/src/mesa/drivers/dri/intel/intel_blit.c +++ b/src/mesa/drivers/dri/intel/intel_blit.c @@ -38,6 +38,8 @@ #include "intel_reg.h" #include "intel_regions.h" #include "intel_batchbuffer.h" +#include "intel_tex.h" +#include "intel_mipmap_tree.h" #define FILE_DEBUG_FLAG DEBUG_BLIT @@ -107,10 +109,6 @@ intelEmitCopyBlit(struct intel_context *intel, drm_intel_bo *aper_array[3]; BATCH_LOCALS; - /* Blits are in a different ringbuffer so we don't use them. */ - if (intel->gen >= 6) - return GL_FALSE; - if (dst_tiling != I915_TILING_NONE) { if (dst_offset & 4095) return GL_FALSE; @@ -140,7 +138,7 @@ intelEmitCopyBlit(struct intel_context *intel, if (pass >= 2) return GL_FALSE; - intel_batchbuffer_require_space(intel->batch, 8 * 4); + intel_batchbuffer_require_space(intel->batch, 8 * 4, true); DBG("%s src:buf(%p)/%d+%d %d,%d dst:buf(%p)/%d+%d %d,%d sz:%dx%d\n", __FUNCTION__, src_buffer, src_pitch, src_offset, src_x, src_y, @@ -181,7 +179,7 @@ intelEmitCopyBlit(struct intel_context *intel, assert(dst_x < dst_x2); assert(dst_y < dst_y2); - BEGIN_BATCH(8); + BEGIN_BATCH_BLT(8); OUT_BATCH(CMD); OUT_BATCH(BR13 | (uint16_t)dst_pitch); OUT_BATCH((dst_y << 16) | dst_x); @@ -219,9 +217,6 @@ intelClearWithBlit(struct gl_context *ctx, GLbitfield mask) GLint cx, cy, cw, ch; BATCH_LOCALS; - /* Blits are in a different ringbuffer so we don't use them. */ - assert(intel->gen < 6); - /* * Compute values for clearing the buffers. */ @@ -356,7 +351,7 @@ intelClearWithBlit(struct gl_context *ctx, GLbitfield mask) intel_batchbuffer_flush(intel->batch); } - BEGIN_BATCH(6); + BEGIN_BATCH_BLT(6); OUT_BATCH(CMD); OUT_BATCH(BR13); OUT_BATCH((y1 << 16) | x1); @@ -393,10 +388,6 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel, int dwords = ALIGN(src_size, 8) / 4; uint32_t opcode, br13, blit_cmd; - /* Blits are in a different ringbuffer so we don't use them. */ - if (intel->gen >= 6) - return GL_FALSE; - if (dst_tiling != I915_TILING_NONE) { if (dst_offset & 4095) return GL_FALSE; @@ -420,7 +411,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel, intel_batchbuffer_require_space( intel->batch, (8 * 4) + (3 * 4) + - dwords * 4 ); + dwords * 4, true); opcode = XY_SETUP_BLT_CMD; if (cpp == 4) @@ -439,7 +430,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel, if (dst_tiling != I915_TILING_NONE) blit_cmd |= XY_DST_TILED; - BEGIN_BATCH(8 + 3); + BEGIN_BATCH_BLT(8 + 3); OUT_BATCH(opcode); OUT_BATCH(br13); OUT_BATCH((0 << 16) | 0); /* clip x1, y1 */ @@ -456,9 +447,9 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel, OUT_BATCH(((y + h) << 16) | (x + w)); ADVANCE_BATCH(); - intel_batchbuffer_data( intel->batch, - src_bits, - dwords * 4 ); + intel_batchbuffer_data(intel->batch, + src_bits, + dwords * 4, true); intel_batchbuffer_emit_mi_flush(intel->batch); @@ -480,9 +471,6 @@ intel_emit_linear_blit(struct intel_context *intel, GLuint pitch, height; GLboolean ok; - /* Blits are in a different ringbuffer so we don't use them. */ - assert(intel->gen < 6); - /* The pitch given to the GPU must be DWORD aligned, and * we want width to match pitch. Max width is (1 << 15 - 1), * rounding that down to the nearest DWORD is 1 << 15 - 4 @@ -514,3 +502,81 @@ intel_emit_linear_blit(struct intel_context *intel, assert(ok); } } + +/** + * Used to initialize the alpha value of an ARGB8888 teximage after + * loading it from an XRGB8888 source. + * + * This is very common with glCopyTexImage2D(). + */ +void +intel_set_teximage_alpha_to_one(struct gl_context *ctx, + struct intel_texture_image *intel_image) +{ + struct intel_context *intel = intel_context(ctx); + unsigned int image_x, image_y; + uint32_t x1, y1, x2, y2; + uint32_t BR13, CMD; + int pitch, cpp; + drm_intel_bo *aper_array[2]; + struct intel_region *region = intel_image->mt->region; + BATCH_LOCALS; + + assert(intel_image->base.TexFormat == MESA_FORMAT_ARGB8888); + + /* get dest x/y in destination texture */ + intel_miptree_get_image_offset(intel_image->mt, + intel_image->level, + intel_image->face, + 0, + &image_x, &image_y); + + x1 = image_x; + y1 = image_y; + x2 = image_x + intel_image->base.Width; + y2 = image_y + intel_image->base.Height; + + pitch = region->pitch; + cpp = region->cpp; + + DBG("%s dst:buf(%p)/%d %d,%d sz:%dx%d\n", + __FUNCTION__, + intel_image->mt->region->buffer, (pitch * region->cpp), + x1, y1, x2 - x1, y2 - y1); + + BR13 = br13_for_cpp(region->cpp) | 0xf0 << 16; + CMD = XY_COLOR_BLT_CMD; + CMD |= XY_BLT_WRITE_ALPHA; + + assert(region->tiling != I915_TILING_Y); + +#ifndef I915 + if (region->tiling != I915_TILING_NONE) { + CMD |= XY_DST_TILED; + pitch /= 4; + } +#endif + BR13 |= (pitch * region->cpp); + + /* do space check before going any further */ + aper_array[0] = intel->batch->buf; + aper_array[1] = region->buffer; + + if (drm_intel_bufmgr_check_aperture_space(aper_array, + ARRAY_SIZE(aper_array)) != 0) { + intel_batchbuffer_flush(intel->batch); + } + + BEGIN_BATCH_BLT(6); + OUT_BATCH(CMD); + OUT_BATCH(BR13); + OUT_BATCH((y1 << 16) | x1); + OUT_BATCH((y2 << 16) | x2); + OUT_RELOC_FENCED(region->buffer, + I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER, + 0); + OUT_BATCH(0xffffffff); /* white, but only alpha gets written */ + ADVANCE_BATCH(); + + intel_batchbuffer_emit_mi_flush(intel->batch); +} diff --git a/src/mesa/drivers/dri/intel/intel_blit.h b/src/mesa/drivers/dri/intel/intel_blit.h index 01631465735..ff69e4f8f8f 100644 --- a/src/mesa/drivers/dri/intel/intel_blit.h +++ b/src/mesa/drivers/dri/intel/intel_blit.h @@ -69,5 +69,7 @@ void intel_emit_linear_blit(struct intel_context *intel, drm_intel_bo *src_bo, unsigned int src_offset, unsigned int size); +void intel_set_teximage_alpha_to_one(struct gl_context *ctx, + struct intel_texture_image *intel_image); #endif diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h index 96493c0f2bb..53a11ba9a7e 100644 --- a/src/mesa/drivers/dri/intel/intel_context.h +++ b/src/mesa/drivers/dri/intel/intel_context.h @@ -29,7 +29,7 @@ #define INTELCONTEXT_INC - +#include <stdbool.h> #include "main/mtypes.h" #include "main/mm.h" #include "dri_metaops.h" diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c index 18e796a1186..c3f528c2ae5 100644 --- a/src/mesa/drivers/dri/intel/intel_fbo.c +++ b/src/mesa/drivers/dri/intel/intel_fbo.c @@ -42,6 +42,8 @@ #include "intel_fbo.h" #include "intel_mipmap_tree.h" #include "intel_regions.h" +#include "intel_tex.h" +#include "intel_span.h" #define FILE_DEBUG_FLAG DEBUG_FBO @@ -107,79 +109,27 @@ intel_alloc_renderbuffer_storage(struct gl_context * ctx, struct gl_renderbuffer ASSERT(rb->Name != 0); switch (internalFormat) { - case GL_RED: - case GL_R8: - rb->Format = MESA_FORMAT_R8; - rb->DataType = GL_UNSIGNED_BYTE; - break; - case GL_R16: - rb->Format = MESA_FORMAT_R16; - rb->DataType = GL_UNSIGNED_SHORT; - break; - case GL_RG: - case GL_RG8: - rb->Format = MESA_FORMAT_RG88; - rb->DataType = GL_UNSIGNED_BYTE; - break; - case GL_RG16: - rb->Format = MESA_FORMAT_RG1616; - rb->DataType = GL_UNSIGNED_SHORT; - break; - case GL_R3_G3_B2: - case GL_RGB4: - case GL_RGB5: - rb->Format = MESA_FORMAT_RGB565; - rb->DataType = GL_UNSIGNED_BYTE; - break; - case GL_RGB: - case GL_RGB8: - case GL_RGB10: - case GL_RGB12: - case GL_RGB16: - rb->Format = MESA_FORMAT_XRGB8888; - rb->DataType = GL_UNSIGNED_BYTE; - break; - case GL_RGBA: - case GL_RGBA2: - case GL_RGBA4: - case GL_RGB5_A1: - case GL_RGBA8: - case GL_RGB10_A2: - case GL_RGBA12: - case GL_RGBA16: - rb->Format = MESA_FORMAT_ARGB8888; - rb->DataType = GL_UNSIGNED_BYTE; - break; - case GL_ALPHA: - case GL_ALPHA8: - rb->Format = MESA_FORMAT_A8; - rb->DataType = GL_UNSIGNED_BYTE; - break; - case GL_DEPTH_COMPONENT16: - rb->Format = MESA_FORMAT_Z16; - rb->DataType = GL_UNSIGNED_SHORT; + default: + /* Use the same format-choice logic as for textures. + * Renderbuffers aren't any different from textures for us, + * except they're less useful because you can't texture with + * them. + */ + rb->Format = intelChooseTextureFormat(ctx, internalFormat, + GL_NONE, GL_NONE); break; case GL_STENCIL_INDEX: case GL_STENCIL_INDEX1_EXT: case GL_STENCIL_INDEX4_EXT: case GL_STENCIL_INDEX8_EXT: case GL_STENCIL_INDEX16_EXT: - case GL_DEPTH_COMPONENT: - case GL_DEPTH_COMPONENT24: - case GL_DEPTH_COMPONENT32: - case GL_DEPTH_STENCIL_EXT: - case GL_DEPTH24_STENCIL8_EXT: - /* alloc a depth+stencil buffer */ + /* These aren't actual texture formats, so force them here. */ rb->Format = MESA_FORMAT_S8_Z24; - rb->DataType = GL_UNSIGNED_INT_24_8_EXT; break; - default: - _mesa_problem(ctx, - "Unexpected format in intel_alloc_renderbuffer_storage"); - return GL_FALSE; } rb->_BaseFormat = _mesa_base_fbo_format(ctx, internalFormat); + rb->DataType = intel_mesa_format_to_rb_datatype(rb->Format); cpp = _mesa_get_format_bytes(rb->Format); intel_flush(ctx); @@ -338,39 +288,30 @@ intel_create_renderbuffer(gl_format format) switch (format) { case MESA_FORMAT_RGB565: irb->Base._BaseFormat = GL_RGB; - irb->Base.DataType = GL_UNSIGNED_BYTE; break; case MESA_FORMAT_XRGB8888: irb->Base._BaseFormat = GL_RGB; - irb->Base.DataType = GL_UNSIGNED_BYTE; break; case MESA_FORMAT_ARGB8888: irb->Base._BaseFormat = GL_RGBA; - irb->Base.DataType = GL_UNSIGNED_BYTE; break; case MESA_FORMAT_Z16: irb->Base._BaseFormat = GL_DEPTH_COMPONENT; - irb->Base.DataType = GL_UNSIGNED_SHORT; break; case MESA_FORMAT_X8_Z24: irb->Base._BaseFormat = GL_DEPTH_COMPONENT; - irb->Base.DataType = GL_UNSIGNED_INT; break; case MESA_FORMAT_S8_Z24: irb->Base._BaseFormat = GL_DEPTH_STENCIL; - irb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT; break; case MESA_FORMAT_A8: irb->Base._BaseFormat = GL_ALPHA; - irb->Base.DataType = GL_UNSIGNED_BYTE; break; case MESA_FORMAT_R8: irb->Base._BaseFormat = GL_RED; - irb->Base.DataType = GL_UNSIGNED_BYTE; break; case MESA_FORMAT_RG88: irb->Base._BaseFormat = GL_RG; - irb->Base.DataType = GL_UNSIGNED_BYTE; break; default: _mesa_problem(NULL, @@ -381,6 +322,7 @@ intel_create_renderbuffer(gl_format format) irb->Base.Format = format; irb->Base.InternalFormat = irb->Base._BaseFormat; + irb->Base.DataType = intel_mesa_format_to_rb_datatype(format); /* intel-specific methods */ irb->Base.Delete = intel_delete_renderbuffer; @@ -457,70 +399,16 @@ static GLboolean intel_update_wrapper(struct gl_context *ctx, struct intel_renderbuffer *irb, struct gl_texture_image *texImage) { - if (texImage->TexFormat == MESA_FORMAT_ARGB8888) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to RGBA8 texture OK\n"); - } - else if (texImage->TexFormat == MESA_FORMAT_XRGB8888) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to XGBA8 texture OK\n"); - } -#ifndef I915 - else if (texImage->TexFormat == MESA_FORMAT_SARGB8) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to SARGB8 texture OK\n"); - } -#endif - else if (texImage->TexFormat == MESA_FORMAT_RGB565) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to RGB5 texture OK\n"); - } - else if (texImage->TexFormat == MESA_FORMAT_ARGB1555) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to ARGB1555 texture OK\n"); - } - else if (texImage->TexFormat == MESA_FORMAT_ARGB4444) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to ARGB4444 texture OK\n"); - } -#ifndef I915 - else if (texImage->TexFormat == MESA_FORMAT_A8) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to A8 texture OK\n"); - } - else if (texImage->TexFormat == MESA_FORMAT_R8) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to R8 texture OK\n"); - } - else if (texImage->TexFormat == MESA_FORMAT_RG88) { - irb->Base.DataType = GL_UNSIGNED_BYTE; - DBG("Render to RG88 texture OK\n"); - } - else if (texImage->TexFormat == MESA_FORMAT_R16) { - irb->Base.DataType = GL_UNSIGNED_SHORT; - DBG("Render to R8 texture OK\n"); - } - else if (texImage->TexFormat == MESA_FORMAT_RG1616) { - irb->Base.DataType = GL_UNSIGNED_SHORT; - DBG("Render to RG88 texture OK\n"); - } -#endif - else if (texImage->TexFormat == MESA_FORMAT_Z16) { - irb->Base.DataType = GL_UNSIGNED_SHORT; - DBG("Render to DEPTH16 texture OK\n"); - } - else if (texImage->TexFormat == MESA_FORMAT_S8_Z24) { - irb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT; - DBG("Render to DEPTH_STENCIL texture OK\n"); - } - else { + if (!intel_span_supports_format(texImage->TexFormat)) { DBG("Render to texture BAD FORMAT %s\n", _mesa_get_format_name(texImage->TexFormat)); return GL_FALSE; + } else { + DBG("Render to texture %s\n", _mesa_get_format_name(texImage->TexFormat)); } irb->Base.Format = texImage->TexFormat; - + irb->Base.DataType = intel_mesa_format_to_rb_datatype(texImage->TexFormat); irb->Base.InternalFormat = texImage->InternalFormat; irb->Base._BaseFormat = _mesa_base_fbo_format(ctx, irb->Base.InternalFormat); irb->Base.Width = texImage->Width; @@ -659,7 +547,8 @@ intel_finish_render_texture(struct gl_context * ctx, _glthread_GetID(), att->Texture->Name); /* Flag that this image may now be validated into the object's miptree. */ - intel_image->used_as_render_target = GL_FALSE; + if (intel_image) + intel_image->used_as_render_target = GL_FALSE; /* Since we've (probably) rendered to the texture and will (likely) use * it in the texture domain later on in this batchbuffer, flush the @@ -711,22 +600,9 @@ intel_validate_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb) continue; } - switch (irb->Base.Format) { - case MESA_FORMAT_ARGB8888: - case MESA_FORMAT_XRGB8888: - case MESA_FORMAT_RGB565: - case MESA_FORMAT_ARGB1555: - case MESA_FORMAT_ARGB4444: -#ifndef I915 - case MESA_FORMAT_SARGB8: - case MESA_FORMAT_A8: - case MESA_FORMAT_R8: - case MESA_FORMAT_R16: - case MESA_FORMAT_RG88: - case MESA_FORMAT_RG1616: -#endif - break; - default: + if (!intel_span_supports_format(irb->Base.Format)) { + DBG("Unsupported texture/renderbuffer format attached: %s\n", + _mesa_get_format_name(irb->Base.Format)); fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT; } } diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c index 3f13589a214..d683e675328 100644 --- a/src/mesa/drivers/dri/intel/intel_screen.c +++ b/src/mesa/drivers/dri/intel/intel_screen.c @@ -35,19 +35,6 @@ #include "utils.h" #include "xmlpool.h" -#include "intel_batchbuffer.h" -#include "intel_buffers.h" -#include "intel_bufmgr.h" -#include "intel_chipset.h" -#include "intel_fbo.h" -#include "intel_screen.h" -#include "intel_tex.h" -#include "intel_regions.h" - -#include "i915_drm.h" - -#define DRI_CONF_TEXTURE_TILING(def) \ - PUBLIC const char __driConfigOptions[] = DRI_CONF_BEGIN DRI_CONF_SECTION_PERFORMANCE @@ -92,6 +79,17 @@ DRI_CONF_END; const GLuint __driNConfigOptions = 11; +#include "intel_batchbuffer.h" +#include "intel_buffers.h" +#include "intel_bufmgr.h" +#include "intel_chipset.h" +#include "intel_fbo.h" +#include "intel_screen.h" +#include "intel_tex.h" +#include "intel_regions.h" + +#include "i915_drm.h" + #ifdef USE_NEW_INTERFACE static PFNGLXCREATECONTEXTMODES create_context_modes = NULL; #endif /*USE_NEW_INTERFACE */ diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c index 104cadf0f9e..1f41518535c 100644 --- a/src/mesa/drivers/dri/intel/intel_span.c +++ b/src/mesa/drivers/dri/intel/intel_span.c @@ -25,6 +25,7 @@ * **************************************************************************/ +#include <stdbool.h> #include "main/glheader.h" #include "main/macros.h" #include "main/mtypes.h" @@ -113,6 +114,26 @@ intel_set_span_functions(struct intel_context *intel, #define TAG2(x,y) intel_##x##y##_A8 #include "spantmp2.h" +#define SPANTMP_MESA_FMT MESA_FORMAT_R8 +#define TAG(x) intel_##x##_R8 +#define TAG2(x,y) intel_##x##y##_R8 +#include "spantmp2.h" + +#define SPANTMP_MESA_FMT MESA_FORMAT_RG88 +#define TAG(x) intel_##x##_RG88 +#define TAG2(x,y) intel_##x##y##_RG88 +#include "spantmp2.h" + +#define SPANTMP_MESA_FMT MESA_FORMAT_R16 +#define TAG(x) intel_##x##_R16 +#define TAG2(x,y) intel_##x##y##_R16 +#include "spantmp2.h" + +#define SPANTMP_MESA_FMT MESA_FORMAT_RG1616 +#define TAG(x) intel_##x##_RG1616 +#define TAG2(x,y) intel_##x##y##_RG1616 +#include "spantmp2.h" + #define LOCAL_DEPTH_VARS \ struct intel_renderbuffer *irb = intel_renderbuffer(rb); \ const GLint yScale = rb->Name ? 1 : -1; \ @@ -339,6 +360,32 @@ intel_unmap_vertex_shader_textures(struct gl_context *ctx) } } +typedef void (*span_init_func)(struct gl_renderbuffer *rb); + +static span_init_func intel_span_init_funcs[MESA_FORMAT_COUNT] = +{ + [MESA_FORMAT_A8] = intel_InitPointers_A8, + [MESA_FORMAT_RGB565] = intel_InitPointers_RGB565, + [MESA_FORMAT_ARGB4444] = intel_InitPointers_ARGB4444, + [MESA_FORMAT_ARGB1555] = intel_InitPointers_ARGB1555, + [MESA_FORMAT_XRGB8888] = intel_InitPointers_xRGB8888, + [MESA_FORMAT_ARGB8888] = intel_InitPointers_ARGB8888, + [MESA_FORMAT_SARGB8] = intel_InitPointers_ARGB8888, + [MESA_FORMAT_Z16] = intel_InitDepthPointers_z16, + [MESA_FORMAT_X8_Z24] = intel_InitDepthPointers_z24_s8, + [MESA_FORMAT_S8_Z24] = intel_InitDepthPointers_z24_s8, + [MESA_FORMAT_R8] = intel_InitPointers_R8, + [MESA_FORMAT_RG88] = intel_InitPointers_RG88, + [MESA_FORMAT_R16] = intel_InitPointers_R16, + [MESA_FORMAT_RG1616] = intel_InitPointers_RG1616, +}; + +bool +intel_span_supports_format(gl_format format) +{ + return intel_span_init_funcs[format] != NULL; +} + /** * Plug in appropriate span read/write functions for the given renderbuffer. * These are used for the software fallbacks. @@ -349,37 +396,6 @@ intel_set_span_functions(struct intel_context *intel, { struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb; - switch (irb->Base.Format) { - case MESA_FORMAT_A8: - intel_InitPointers_A8(rb); - break; - case MESA_FORMAT_RGB565: - intel_InitPointers_RGB565(rb); - break; - case MESA_FORMAT_ARGB4444: - intel_InitPointers_ARGB4444(rb); - break; - case MESA_FORMAT_ARGB1555: - intel_InitPointers_ARGB1555(rb); - break; - case MESA_FORMAT_XRGB8888: - intel_InitPointers_xRGB8888(rb); - break; - case MESA_FORMAT_ARGB8888: - case MESA_FORMAT_SARGB8: - intel_InitPointers_ARGB8888(rb); - break; - case MESA_FORMAT_Z16: - intel_InitDepthPointers_z16(rb); - break; - case MESA_FORMAT_X8_Z24: - case MESA_FORMAT_S8_Z24: - intel_InitDepthPointers_z24_s8(rb); - break; - default: - _mesa_problem(NULL, - "Unexpected MesaFormat %d in intelSetSpanFunctions", - irb->Base.Format); - break; - } + assert(intel_span_init_funcs[irb->Base.Format]); + intel_span_init_funcs[irb->Base.Format](rb); } diff --git a/src/mesa/drivers/dri/intel/intel_span.h b/src/mesa/drivers/dri/intel/intel_span.h index aa8d08e843a..5a4c4e8e52a 100644 --- a/src/mesa/drivers/dri/intel/intel_span.h +++ b/src/mesa/drivers/dri/intel/intel_span.h @@ -28,6 +28,9 @@ #ifndef _INTEL_SPAN_H #define _INTEL_SPAN_H +#include "main/formats.h" +#include <stdbool.h> + extern void intelInitSpanFuncs(struct gl_context * ctx); extern void intelSpanRenderFinish(struct gl_context * ctx); @@ -38,5 +41,6 @@ void intel_renderbuffer_unmap(struct intel_context *intel, struct gl_renderbuffer *rb); void intel_map_vertex_shader_textures(struct gl_context *ctx); void intel_unmap_vertex_shader_textures(struct gl_context *ctx); +bool intel_span_supports_format(gl_format format); #endif diff --git a/src/mesa/drivers/dri/intel/intel_tex.h b/src/mesa/drivers/dri/intel/intel_tex.h index 7906554e453..b638628c711 100644 --- a/src/mesa/drivers/dri/intel/intel_tex.h +++ b/src/mesa/drivers/dri/intel/intel_tex.h @@ -42,6 +42,7 @@ void intelInitTextureCopyImageFuncs(struct dd_function_table *functions); gl_format intelChooseTextureFormat(struct gl_context *ctx, GLint internalFormat, GLenum format, GLenum type); +GLenum intel_mesa_format_to_rb_datatype(gl_format format); void intelSetTexBuffer(__DRIcontext *pDRICtx, GLint target, __DRIdrawable *pDraw); diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c index 87b31bf078c..c6bc3d962ab 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_copy.c +++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c @@ -50,44 +50,20 @@ * Do the best we can using the blitter. A future project is to use * the texture engine and fragment programs for these copies. */ -static const struct intel_region * -get_teximage_source(struct intel_context *intel, GLenum internalFormat) +static struct intel_renderbuffer * +get_teximage_readbuffer(struct intel_context *intel, GLenum internalFormat) { - struct intel_renderbuffer *irb; - DBG("%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(internalFormat)); switch (internalFormat) { case GL_DEPTH_COMPONENT: case GL_DEPTH_COMPONENT16: - irb = intel_get_renderbuffer(intel->ctx.ReadBuffer, BUFFER_DEPTH); - if (irb && irb->region && irb->region->cpp == 2) - return irb->region; - return NULL; case GL_DEPTH24_STENCIL8_EXT: case GL_DEPTH_STENCIL_EXT: - irb = intel_get_renderbuffer(intel->ctx.ReadBuffer, BUFFER_DEPTH); - if (irb && irb->region && irb->region->cpp == 4) - return irb->region; - return NULL; - case 4: - case GL_RGBA: - case GL_RGBA8: - irb = intel_renderbuffer(intel->ctx.ReadBuffer->_ColorReadBuffer); - /* We're required to set alpha to 1.0 in this case, but we can't - * do that with the blitter, so fall back. We could use the 3D - * engine or do two passes with the blitter, but it doesn't seem - * worth it for this case. */ - if (irb->Base._BaseFormat == GL_RGB) - return NULL; - return irb->region; - case 3: - case GL_RGB: - case GL_RGB8: - return intel_readbuf_region(intel); + return intel_get_renderbuffer(intel->ctx.ReadBuffer, BUFFER_DEPTH); default: - return NULL; + return intel_renderbuffer(intel->ctx.ReadBuffer->_ColorReadBuffer); } } @@ -101,23 +77,34 @@ do_copy_texsubimage(struct intel_context *intel, GLint x, GLint y, GLsizei width, GLsizei height) { struct gl_context *ctx = &intel->ctx; - const struct intel_region *src = get_teximage_source(intel, internalFormat); + struct intel_renderbuffer *irb; + bool copy_supported_with_alpha_override = false; + + intel_prepare_render(intel); - if (!intelImage->mt || !src || !src->buffer) { + irb = get_teximage_readbuffer(intel, internalFormat); + if (!intelImage->mt || !irb || !irb->region) { if (unlikely(INTEL_DEBUG & DEBUG_FALLBACKS)) fprintf(stderr, "%s fail %p %p (0x%08x)\n", - __FUNCTION__, intelImage->mt, src, internalFormat); + __FUNCTION__, intelImage->mt, irb, internalFormat); return GL_FALSE; } - if (intelImage->mt->cpp != src->cpp) { - fallback_debug("%s fail %d vs %d cpp\n", - __FUNCTION__, intelImage->mt->cpp, src->cpp); + if (irb->Base.Format == MESA_FORMAT_XRGB8888 && + intelImage->base.TexFormat == MESA_FORMAT_ARGB8888) { + copy_supported_with_alpha_override = true; + } + + if (intelImage->base.TexFormat != irb->Base.Format && + !copy_supported_with_alpha_override) { + if (unlikely(INTEL_DEBUG & DEBUG_FALLBACKS)) + fprintf(stderr, "%s mismatched formats %s, %s\n", + __FUNCTION__, + _mesa_get_format_name(intelImage->base.TexFormat), + _mesa_get_format_name(irb->Base.Format)); return GL_FALSE; } - /* intel_flush(ctx); */ - intel_prepare_render(intel); { drm_intel_bo *dst_bo = intel_region_buffer(intel, intelImage->mt->region, @@ -140,24 +127,24 @@ do_copy_texsubimage(struct intel_context *intel, if (ctx->ReadBuffer->Name == 0) { /* Flip vertical orientation for system framebuffers */ y = ctx->ReadBuffer->Height - (y + height); - src_pitch = -src->pitch; + src_pitch = -irb->region->pitch; } else { /* reading from a FBO, y is already oriented the way we like */ - src_pitch = src->pitch; + src_pitch = irb->region->pitch; } /* blit from src buffer to texture */ if (!intelEmitCopyBlit(intel, intelImage->mt->cpp, src_pitch, - src->buffer, + irb->region->buffer, 0, - src->tiling, + irb->region->tiling, intelImage->mt->region->pitch, dst_bo, 0, intelImage->mt->region->tiling, - src->draw_x + x, src->draw_y + y, + irb->region->draw_x + x, irb->region->draw_y + y, image_x + dstx, image_y + dsty, width, height, GL_COPY)) { @@ -165,6 +152,9 @@ do_copy_texsubimage(struct intel_context *intel, } } + if (copy_supported_with_alpha_override) + intel_set_teximage_alpha_to_one(ctx, intelImage); + return GL_TRUE; } diff --git a/src/mesa/drivers/dri/intel/intel_tex_format.c b/src/mesa/drivers/dri/intel/intel_tex_format.c index f8316ae2f8d..c9763c9ae16 100644 --- a/src/mesa/drivers/dri/intel/intel_tex_format.c +++ b/src/mesa/drivers/dri/intel/intel_tex_format.c @@ -4,6 +4,39 @@ #include "main/formats.h" /** + * Returns the renderbuffer DataType for a MESA_FORMAT. + */ +GLenum +intel_mesa_format_to_rb_datatype(gl_format format) +{ + switch (format) { + case MESA_FORMAT_ARGB8888: + case MESA_FORMAT_XRGB8888: + case MESA_FORMAT_SARGB8: + case MESA_FORMAT_R8: + case MESA_FORMAT_RG88: + case MESA_FORMAT_A8: + case MESA_FORMAT_AL88: + case MESA_FORMAT_RGB565: + case MESA_FORMAT_ARGB1555: + case MESA_FORMAT_ARGB4444: + return GL_UNSIGNED_BYTE; + case MESA_FORMAT_R16: + case MESA_FORMAT_RG1616: + case MESA_FORMAT_Z16: + return GL_UNSIGNED_SHORT; + case MESA_FORMAT_X8_Z24: + return GL_UNSIGNED_INT; + case MESA_FORMAT_S8_Z24: + return GL_UNSIGNED_INT_24_8_EXT; + default: + _mesa_problem(NULL, "unexpected MESA_FORMAT for renderbuffer"); + return GL_UNSIGNED_BYTE; + } +} + + +/** * Choose hardware texture format given the user's glTexImage parameters. * * It works out that this function is fine for all the supported diff --git a/src/mesa/drivers/dri/nouveau/nouveau_driver.h b/src/mesa/drivers/dri/nouveau/nouveau_driver.h index 8036b18edc0..c5ac1282d0d 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_driver.h +++ b/src/mesa/drivers/dri/nouveau/nouveau_driver.h @@ -38,7 +38,6 @@ #include <assert.h> #include "nouveau_device.h" -#include "nouveau_pushbuf.h" #include "nouveau_grobj.h" #include "nouveau_channel.h" #include "nouveau_bo.h" @@ -46,6 +45,7 @@ #include "nouveau_screen.h" #include "nouveau_state.h" #include "nouveau_surface.h" +#include "nv04_pushbuf.h" #define DRIVER_DATE "20091015" #define DRIVER_AUTHOR "Nouveau" diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_swizzles.c b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_swizzles.c index a0f7bd81741..133a9f72ec7 100644 --- a/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_swizzles.c +++ b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_swizzles.c @@ -56,6 +56,7 @@ static void rewrite_source(struct radeon_compiler * c, mov->U.I.DstReg.Index = tempreg; mov->U.I.DstReg.WriteMask = split.Phase[phase]; mov->U.I.SrcReg[0] = inst->U.I.SrcReg[src]; + mov->U.I.PreSub = inst->U.I.PreSub; phase_refmask = 0; for(unsigned int chan = 0; chan < 4; ++chan) { diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c index 58977a40c7c..c8063171b81 100644 --- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c +++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c @@ -689,11 +689,12 @@ static void transform_r300_vertex_fix_LIT(struct radeon_compiler* c, &constant_swizzle); /* MOV dst, src */ + dst.WriteMask = RC_MASK_XYZW; emit1(c, inst->Prev, RC_OPCODE_MOV, 0, dst, inst->U.I.SrcReg[0]); - /* MAX dst.z, src, 0.00...001 */ + /* MAX dst.y, src, 0.00...001 */ emit2(c, inst->Prev, RC_OPCODE_MAX, 0, dstregtmpmask(dst.Index, RC_MASK_Y), srcreg(RC_FILE_TEMPORARY, dst.Index), diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c index 1fa559cec1a..bee9c3bc6d3 100644 --- a/src/mesa/drivers/dri/r600/r700_assembler.c +++ b/src/mesa/drivers/dri/r600/r700_assembler.c @@ -1134,7 +1134,7 @@ GLboolean EG_assemble_vfetch_instruction(r700_AssemblerBase* pAsm, EG_VTX_WORD1__DST_SEL_W_shift, EG_VTX_WORD1__DST_SEL_W_mask); - SETfield(vfetch_instruction_ptr->m_Word1.val, 0, /* use format here, in r6/r7, format used set in const, need to use same */ + SETfield(vfetch_instruction_ptr->m_Word1.val, 1, EG_VTX_WORD1__UCF_shift, EG_VTX_WORD1__UCF_bit); SETfield(vfetch_instruction_ptr->m_Word1.val, data_format, diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c index b71afdd61f3..fd5b4e915cd 100644 --- a/src/mesa/main/extensions.c +++ b/src/mesa/main/extensions.c @@ -249,7 +249,8 @@ _mesa_enable_sw_extensions(struct gl_context *ctx) #if FEATURE_ARB_framebuffer_object ctx->Extensions.ARB_framebuffer_object = GL_TRUE; #endif -#if FEATURE_ARB_geometry_shader4 +#if FEATURE_ARB_geometry_shader4 && 0 + /* XXX re-enable when GLSL compiler again supports geometry shaders */ ctx->Extensions.ARB_geometry_shader4 = GL_TRUE; #endif ctx->Extensions.ARB_half_float_pixel = GL_TRUE; @@ -258,6 +259,7 @@ _mesa_enable_sw_extensions(struct gl_context *ctx) ctx->Extensions.ARB_multitexture = GL_TRUE; #if FEATURE_queryobj ctx->Extensions.ARB_occlusion_query = GL_TRUE; + ctx->Extensions.ARB_occlusion_query2 = GL_TRUE; #endif ctx->Extensions.ARB_point_sprite = GL_TRUE; #if FEATURE_ARB_shader_objects @@ -857,8 +859,6 @@ make_extension_string_es2(const struct gl_context *ctx, GLubyte *str) { size_t len = 0; - len += append_extension(&str, "GL_OES_compressed_paletted_texture"); - if (ctx->Extensions.ARB_framebuffer_object) { len += append_extension(&str, "GL_OES_depth24"); len += append_extension(&str, "GL_OES_depth32"); diff --git a/src/mesa/main/formats.c b/src/mesa/main/formats.c index cd9eb81852f..42f70ca232b 100644 --- a/src/mesa/main/formats.c +++ b/src/mesa/main/formats.c @@ -863,7 +863,6 @@ const char * _mesa_get_format_name(gl_format format) { const struct gl_format_info *info = _mesa_get_format_info(format); - ASSERT(info->BytesPerBlock); return info->StrName; } diff --git a/src/mesa/main/queryobj.c b/src/mesa/main/queryobj.c index 88743977206..66fb5980589 100644 --- a/src/mesa/main/queryobj.c +++ b/src/mesa/main/queryobj.c @@ -143,6 +143,11 @@ get_query_binding_point(struct gl_context *ctx, GLenum target) return &ctx->Query.CurrentOcclusionObject; else return NULL; + case GL_ANY_SAMPLES_PASSED: + if (ctx->Extensions.ARB_occlusion_query2) + return &ctx->Query.CurrentOcclusionObject; + else + return NULL; case GL_TIME_ELAPSED_EXT: if (ctx->Extensions.EXT_timer_query) return &ctx->Query.CurrentTimerObject; @@ -378,11 +383,18 @@ _mesa_GetQueryObjectivARB(GLuint id, GLenum pname, GLint *params) if (!q->Ready) ctx->Driver.WaitQuery(ctx, q); /* if result is too large for returned type, clamp to max value */ - if (q->Result > 0x7fffffff) { - *params = 0x7fffffff; - } - else { - *params = (GLint)q->Result; + if (q->Target == GL_ANY_SAMPLES_PASSED) { + if (q->Result) + *params = GL_TRUE; + else + *params = GL_FALSE; + } else { + if (q->Result > 0x7fffffff) { + *params = 0x7fffffff; + } + else { + *params = (GLint)q->Result; + } } break; case GL_QUERY_RESULT_AVAILABLE_ARB: @@ -418,11 +430,18 @@ _mesa_GetQueryObjectuivARB(GLuint id, GLenum pname, GLuint *params) if (!q->Ready) ctx->Driver.WaitQuery(ctx, q); /* if result is too large for returned type, clamp to max value */ - if (q->Result > 0xffffffff) { - *params = 0xffffffff; - } - else { - *params = (GLuint)q->Result; + if (q->Target == GL_ANY_SAMPLES_PASSED) { + if (q->Result) + *params = GL_TRUE; + else + *params = GL_FALSE; + } else { + if (q->Result > 0xffffffff) { + *params = 0xffffffff; + } + else { + *params = (GLuint)q->Result; + } } break; case GL_QUERY_RESULT_AVAILABLE_ARB: diff --git a/src/mesa/program/ir_to_mesa.cpp b/src/mesa/program/ir_to_mesa.cpp index b274a961b28..490c4cab7ab 100644 --- a/src/mesa/program/ir_to_mesa.cpp +++ b/src/mesa/program/ir_to_mesa.cpp @@ -1569,7 +1569,13 @@ ir_to_mesa_visitor::visit(ir_dereference_record *ir) break; offset += type_size(struct_type->fields.structure[i].type); } - this->result.swizzle = swizzle_for_size(ir->type->vector_elements); + + /* If the type is smaller than a vec4, replicate the last channel out. */ + if (ir->type->is_scalar() || ir->type->is_vector()) + this->result.swizzle = swizzle_for_size(ir->type->vector_elements); + else + this->result.swizzle = SWIZZLE_NOOP; + this->result.index += offset; } diff --git a/src/mesa/program/ir_to_mesa.h b/src/mesa/program/ir_to_mesa.h index 7197615f949..7410e149735 100644 --- a/src/mesa/program/ir_to_mesa.h +++ b/src/mesa/program/ir_to_mesa.h @@ -25,8 +25,11 @@ extern "C" { #endif -#include "main/config.h" -#include "main/mtypes.h" +#include "main/glheader.h" + +struct gl_context; +struct gl_shader; +struct gl_shader_program; void _mesa_glsl_compile_shader(struct gl_context *ctx, struct gl_shader *sh); void _mesa_glsl_link_shader(struct gl_context *ctx, struct gl_shader_program *prog); diff --git a/src/mesa/program/nvfragparse.h b/src/mesa/program/nvfragparse.h index 3e85dd2c30b..088e7527d5b 100644 --- a/src/mesa/program/nvfragparse.h +++ b/src/mesa/program/nvfragparse.h @@ -30,7 +30,10 @@ #ifndef NVFRAGPARSE_H #define NVFRAGPARSE_H -#include "main/mtypes.h" +#include "main/glheader.h" + +struct gl_context; +struct gl_fragment_program; extern void _mesa_parse_nv_fragment_program(struct gl_context *ctx, GLenum target, diff --git a/src/mesa/program/nvvertparse.h b/src/mesa/program/nvvertparse.h index e98e867320f..7318e149416 100644 --- a/src/mesa/program/nvvertparse.h +++ b/src/mesa/program/nvvertparse.h @@ -29,7 +29,10 @@ #ifndef NVVERTPARSE_H #define NVVERTPARSE_H -#include "main/mtypes.h" +#include "main/glheader.h" + +struct gl_context; +struct gl_vertex_program; extern void _mesa_parse_nv_vertex_program(struct gl_context *ctx, GLenum target, diff --git a/src/mesa/program/prog_cache.h b/src/mesa/program/prog_cache.h index 4907ae3030e..01673348279 100644 --- a/src/mesa/program/prog_cache.h +++ b/src/mesa/program/prog_cache.h @@ -30,8 +30,9 @@ #define PROG_CACHE_H -#include "main/mtypes.h" +#include "main/glheader.h" +struct gl_context; /** Opaque type */ struct gl_program_cache; diff --git a/src/mesa/program/prog_execute.c b/src/mesa/program/prog_execute.c index 1d97a077f52..dd15e9a1ccd 100644 --- a/src/mesa/program/prog_execute.c +++ b/src/mesa/program/prog_execute.c @@ -1670,6 +1670,18 @@ _mesa_execute_program(struct gl_context * ctx, fetch_texel(ctx, machine, inst, texcoord, lodBias, color); + if (DEBUG_PROG) { + printf("TXB (%g, %g, %g, %g) = texture[%d][%g %g %g %g]" + " bias %g\n", + color[0], color[1], color[2], color[3], + inst->TexSrcUnit, + texcoord[0], + texcoord[1], + texcoord[2], + texcoord[3], + lodBias); + } + store_vector4(inst, machine, color); } break; diff --git a/src/mesa/program/prog_optimize.h b/src/mesa/program/prog_optimize.h index 00f1080449b..463f5fc51c4 100644 --- a/src/mesa/program/prog_optimize.h +++ b/src/mesa/program/prog_optimize.h @@ -27,9 +27,10 @@ #include "main/config.h" -#include "main/mtypes.h" +#include "main/glheader.h" +struct gl_context; struct gl_program; struct prog_instruction; diff --git a/src/mesa/program/prog_statevars.h b/src/mesa/program/prog_statevars.h index 009ebde0012..f2407af9c87 100644 --- a/src/mesa/program/prog_statevars.h +++ b/src/mesa/program/prog_statevars.h @@ -25,8 +25,10 @@ #ifndef PROG_STATEVARS_H #define PROG_STATEVARS_H -#include "main/mtypes.h" +#include "main/glheader.h" +struct gl_context; +struct gl_program_parameter_list; /** * Number of STATE_* values we need to address any GL state. diff --git a/src/mesa/state_tracker/st_atom_shader.c b/src/mesa/state_tracker/st_atom_shader.c index 05442ef91b5..c311d043931 100644 --- a/src/mesa/state_tracker/st_atom_shader.c +++ b/src/mesa/state_tracker/st_atom_shader.c @@ -50,99 +50,6 @@ #include "st_program.h" - -/** - * Translate fragment program if needed. - */ -static void -translate_fp(struct st_context *st, - struct st_fragment_program *stfp) -{ - if (!stfp->tgsi.tokens) { - assert(stfp->Base.Base.NumInstructions > 0); - - st_translate_fragment_program(st, stfp); - } -} - -/* - * Translate geometry program if needed. - */ -static void -translate_gp(struct st_context *st, - struct st_geometry_program *stgp) -{ - if (!stgp->tgsi.tokens) { - assert(stgp->Base.Base.NumInstructions > 1); - - st_translate_geometry_program(st, stgp); - } -} - -/** - * Find a translated vertex program that corresponds to stvp and - * has outputs matched to stfp's inputs. - * This performs vertex and fragment translation (to TGSI) when needed. - */ -static struct st_vp_varient * -find_translated_vp(struct st_context *st, - struct st_vertex_program *stvp ) -{ - struct st_vp_varient *vpv; - struct st_vp_varient_key key; - - /* Nothing in our key yet. This will change: - */ - memset(&key, 0, sizeof key); - - /* When this is true, we will add an extra input to the vertex - * shader translation (for edgeflags), an extra output with - * edgeflag semantics, and extend the vertex shader to pass through - * the input to the output. We'll need to use similar logic to set - * up the extra vertex_element input for edgeflags. - * _NEW_POLYGON, ST_NEW_EDGEFLAGS_DATA - */ - key.passthrough_edgeflags = (st->vertdata_edgeflags && ( - st->ctx->Polygon.FrontMode != GL_FILL || - st->ctx->Polygon.BackMode != GL_FILL)); - - - /* Do we need to throw away old translations after a change in the - * GL program string? - */ - if (stvp->serialNo != stvp->lastSerialNo) { - /* These may have changed if the program string changed. - */ - st_prepare_vertex_program( st, stvp ); - - /* We are now up-to-date: - */ - stvp->lastSerialNo = stvp->serialNo; - } - - /* See if we've got a translated vertex program whose outputs match - * the fragment program's inputs. - */ - for (vpv = stvp->varients; vpv; vpv = vpv->next) { - if (memcmp(&vpv->key, &key, sizeof key) == 0) { - break; - } - } - - /* No? Perform new translation here. */ - if (!vpv) { - vpv = st_translate_vertex_program(st, stvp, &key); - if (!vpv) - return NULL; - - vpv->next = stvp->varients; - stvp->varients = vpv; - } - - return vpv; -} - - /** * Return pointer to a pass-through fragment shader. * This shader is used when a texture is missing/incomplete. @@ -167,12 +74,16 @@ static void update_fp( struct st_context *st ) { struct st_fragment_program *stfp; + struct st_fp_variant_key key; assert(st->ctx->FragmentProgram._Current); stfp = st_fragment_program(st->ctx->FragmentProgram._Current); assert(stfp->Base.Base.Target == GL_FRAGMENT_PROGRAM_ARB); - translate_fp(st, stfp); + memset(&key, 0, sizeof(key)); + key.st = st; + + st->fp_variant = st_get_fp_variant(st, stfp, &key); st_reference_fragprog(st, &st->fp, stfp); @@ -182,7 +93,8 @@ update_fp( struct st_context *st ) cso_set_fragment_shader_handle(st->cso_context, fs); } else { - cso_set_fragment_shader_handle(st->cso_context, stfp->driver_shader); + cso_set_fragment_shader_handle(st->cso_context, + st->fp_variant->driver_shader); } } @@ -206,6 +118,7 @@ static void update_vp( struct st_context *st ) { struct st_vertex_program *stvp; + struct st_vp_variant_key key; /* find active shader and params -- Should be covered by * ST_NEW_VERTEX_PROGRAM @@ -214,12 +127,26 @@ update_vp( struct st_context *st ) stvp = st_vertex_program(st->ctx->VertexProgram._Current); assert(stvp->Base.Base.Target == GL_VERTEX_PROGRAM_ARB); - st->vp_varient = find_translated_vp(st, stvp); + memset(&key, 0, sizeof key); + key.st = st; /* variants are per-context */ + + /* When this is true, we will add an extra input to the vertex + * shader translation (for edgeflags), an extra output with + * edgeflag semantics, and extend the vertex shader to pass through + * the input to the output. We'll need to use similar logic to set + * up the extra vertex_element input for edgeflags. + * _NEW_POLYGON, ST_NEW_EDGEFLAGS_DATA + */ + key.passthrough_edgeflags = (st->vertdata_edgeflags && ( + st->ctx->Polygon.FrontMode != GL_FILL || + st->ctx->Polygon.BackMode != GL_FILL)); + + st->vp_variant = st_get_vp_variant(st, stvp, &key); st_reference_vertprog(st, &st->vp, stvp); cso_set_vertex_shader_handle(st->cso_context, - st->vp_varient->driver_shader); + st->vp_variant->driver_shader); st->vertex_result_to_slot = stvp->result_to_output; } @@ -231,14 +158,16 @@ const struct st_tracked_state st_update_vp = { _NEW_POLYGON, /* mesa */ ST_NEW_VERTEX_PROGRAM | ST_NEW_EDGEFLAGS_DATA /* st */ }, - update_vp /* update */ + update_vp /* update */ }; + + static void update_gp( struct st_context *st ) { - struct st_geometry_program *stgp; + struct st_gp_variant_key key; if (!st->ctx->GeometryProgram._Current) { cso_set_geometry_shader_handle(st->cso_context, NULL); @@ -248,18 +177,22 @@ update_gp( struct st_context *st ) stgp = st_geometry_program(st->ctx->GeometryProgram._Current); assert(stgp->Base.Base.Target == MESA_GEOMETRY_PROGRAM); - translate_gp(st, stgp); + memset(&key, 0, sizeof(key)); + key.st = st; + + st->gp_variant = st_get_gp_variant(st, stgp, &key); st_reference_geomprog(st, &st->gp, stgp); - cso_set_geometry_shader_handle(st->cso_context, stgp->driver_shader); + cso_set_geometry_shader_handle(st->cso_context, + st->gp_variant->driver_shader); } const struct st_tracked_state st_update_gp = { - "st_update_gp", /* name */ - { /* dirty */ - 0, /* mesa */ - ST_NEW_GEOMETRY_PROGRAM /* st */ + "st_update_gp", /* name */ + { /* dirty */ + 0, /* mesa */ + ST_NEW_GEOMETRY_PROGRAM /* st */ }, - update_gp /* update */ + update_gp /* update */ }; diff --git a/src/mesa/state_tracker/st_cb_accum.h b/src/mesa/state_tracker/st_cb_accum.h index b8c9c350031..050a21483e4 100644 --- a/src/mesa/state_tracker/st_cb_accum.h +++ b/src/mesa/state_tracker/st_cb_accum.h @@ -30,7 +30,11 @@ #define ST_CB_ACCUM_H -#include "main/mtypes.h" +#include "main/mfeatures.h" + +struct dd_function_table; +struct gl_context; +struct gl_renderbuffer; #if FEATURE_accum diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c index f08697fe23b..516346c8c7a 100644 --- a/src/mesa/state_tracker/st_cb_bitmap.c +++ b/src/mesa/state_tracker/st_cb_bitmap.c @@ -185,48 +185,47 @@ find_free_bit(uint bitfield) /** * Combine basic bitmap fragment program with the user-defined program. + * \param st current context + * \param fpIn the incoming fragment program + * \param fpOut the new fragment program which does fragment culling + * \param bitmap_sampler sampler number for the bitmap texture */ -static struct st_fragment_program * -combined_bitmap_fragment_program(struct gl_context *ctx) +void +st_make_bitmap_fragment_program(struct st_context *st, + struct gl_fragment_program *fpIn, + struct gl_fragment_program **fpOut, + GLuint *bitmap_sampler) { - struct st_context *st = st_context(ctx); - struct st_fragment_program *stfp = st->fp; - - if (!stfp->bitmap_program) { - /* - * Generate new program which is the user-defined program prefixed - * with the bitmap sampler/kill instructions. - */ - struct st_fragment_program *bitmap_prog; - uint sampler; - - sampler = find_free_bit(st->fp->Base.Base.SamplersUsed); - bitmap_prog = make_bitmap_fragment_program(ctx, sampler); + struct st_fragment_program *bitmap_prog; + struct gl_program *newProg; + uint sampler; - stfp->bitmap_program = (struct st_fragment_program *) - _mesa_combine_programs(ctx, - &bitmap_prog->Base.Base, &stfp->Base.Base); - stfp->bitmap_program->bitmap_sampler = sampler; + /* + * Generate new program which is the user-defined program prefixed + * with the bitmap sampler/kill instructions. + */ + sampler = find_free_bit(fpIn->Base.SamplersUsed); + bitmap_prog = make_bitmap_fragment_program(st->ctx, sampler); - /* done with this after combining */ - st_reference_fragprog(st, &bitmap_prog, NULL); + newProg = _mesa_combine_programs(st->ctx, + &bitmap_prog->Base.Base, + &fpIn->Base); + /* done with this after combining */ + st_reference_fragprog(st, &bitmap_prog, NULL); #if 0 - { - struct gl_program *p = &stfp->bitmap_program->Base.Base; - printf("Combined bitmap program:\n"); - _mesa_print_program(p); - printf("InputsRead: 0x%x\n", p->InputsRead); - printf("OutputsWritten: 0x%x\n", p->OutputsWritten); - _mesa_print_parameter_list(p->Parameters); - } -#endif - - /* translate to TGSI tokens */ - st_translate_fragment_program(st, stfp->bitmap_program); + { + printf("Combined bitmap program:\n"); + _mesa_print_program(newProg); + printf("InputsRead: 0x%x\n", newProg->InputsRead); + printf("OutputsWritten: 0x%x\n", newProg->OutputsWritten); + _mesa_print_parameter_list(newProg->Parameters); } +#endif - return stfp->bitmap_program; + /* return results */ + *fpOut = (struct gl_fragment_program *) newProg; + *bitmap_sampler = sampler; } @@ -349,7 +348,8 @@ setup_bitmap_vertex_data(struct st_context *st, bool normalized, if (!st->bitmap.vbuf) { st->bitmap.vbuf = pipe_buffer_create(pipe->screen, PIPE_BIND_VERTEX_BUFFER, - max_slots * sizeof(st->bitmap.vertices)); + max_slots * + sizeof(st->bitmap.vertices)); } /* Positions are in clip coords since we need to do clipping in case @@ -389,10 +389,11 @@ setup_bitmap_vertex_data(struct st_context *st, bool normalized, /* put vertex data into vbuf */ pipe_buffer_write_nooverlap(st->pipe, - st->bitmap.vbuf, - st->bitmap.vbuf_slot * sizeof st->bitmap.vertices, - sizeof st->bitmap.vertices, - st->bitmap.vertices); + st->bitmap.vbuf, + st->bitmap.vbuf_slot + * sizeof(st->bitmap.vertices), + sizeof st->bitmap.vertices, + st->bitmap.vertices); return st->bitmap.vbuf_slot++ * sizeof st->bitmap.vertices; } @@ -411,11 +412,16 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, struct st_context *st = st_context(ctx); struct pipe_context *pipe = st->pipe; struct cso_context *cso = st->cso_context; - struct st_fragment_program *stfp; + struct st_fp_variant *fpv; + struct st_fp_variant_key key; GLuint maxSize; GLuint offset; - stfp = combined_bitmap_fragment_program(ctx); + memset(&key, 0, sizeof(key)); + key.st = st; + key.bitmap = GL_TRUE; + + fpv = st_get_fp_variant(st, st->fp, &key); /* As an optimization, Mesa's fragment programs will sometimes get the * primary color from a statevar/constant rather than a varying variable. @@ -428,7 +434,7 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, GLfloat colorSave[4]; COPY_4V(colorSave, ctx->Current.Attrib[VERT_ATTRIB_COLOR0]); COPY_4V(ctx->Current.Attrib[VERT_ATTRIB_COLOR0], color); - st_upload_constants(st, stfp->Base.Base.Parameters, PIPE_SHADER_FRAGMENT); + st_upload_constants(st, fpv->parameters, PIPE_SHADER_FRAGMENT); COPY_4V(ctx->Current.Attrib[VERT_ATTRIB_COLOR0], colorSave); } @@ -437,7 +443,8 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, /* XXX if the bitmap is larger than the max texture size, break * it up into chunks. */ - maxSize = 1 << (pipe->screen->get_param(pipe->screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1); + maxSize = 1 << (pipe->screen->get_param(pipe->screen, + PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1); assert(width <= (GLsizei)maxSize); assert(height <= (GLsizei)maxSize); @@ -454,7 +461,7 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, cso_set_rasterizer(cso, &st->bitmap.rasterizer); /* fragment shader state: TEX lookup program */ - cso_set_fragment_shader_handle(cso, stfp->driver_shader); + cso_set_fragment_shader_handle(cso, fpv->driver_shader); /* vertex shader state: position + texcoord pass-through */ cso_set_vertex_shader_handle(cso, st->bitmap.vs); @@ -462,21 +469,22 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, /* user samplers, plus our bitmap sampler */ { struct pipe_sampler_state *samplers[PIPE_MAX_SAMPLERS]; - uint num = MAX2(stfp->bitmap_sampler + 1, st->state.num_samplers); + uint num = MAX2(fpv->bitmap_sampler + 1, st->state.num_samplers); uint i; for (i = 0; i < st->state.num_samplers; i++) { samplers[i] = &st->state.samplers[i]; } - samplers[stfp->bitmap_sampler] = &st->bitmap.samplers[sv->texture->target != PIPE_TEXTURE_RECT]; + samplers[fpv->bitmap_sampler] = + &st->bitmap.samplers[sv->texture->target != PIPE_TEXTURE_RECT]; cso_set_samplers(cso, num, (const struct pipe_sampler_state **) samplers); } /* user textures, plus the bitmap texture */ { struct pipe_sampler_view *sampler_views[PIPE_MAX_SAMPLERS]; - uint num = MAX2(stfp->bitmap_sampler + 1, st->state.num_textures); + uint num = MAX2(fpv->bitmap_sampler + 1, st->state.num_textures); memcpy(sampler_views, st->state.sampler_views, sizeof(sampler_views)); - sampler_views[stfp->bitmap_sampler] = sv; + sampler_views[fpv->bitmap_sampler] = sv; cso_set_fragment_sampler_views(cso, num, sampler_views); } @@ -504,7 +512,9 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z, z = z * 2.0 - 1.0; /* draw textured quad */ - offset = setup_bitmap_vertex_data(st, sv->texture->target != PIPE_TEXTURE_RECT, x, y, width, height, z, color); + offset = setup_bitmap_vertex_data(st, + sv->texture->target != PIPE_TEXTURE_RECT, + x, y, width, height, z, color); util_draw_vertex_buffer(pipe, st->bitmap.vbuf, offset, PIPE_PRIM_TRIANGLE_FAN, @@ -573,6 +583,9 @@ print_cache(const struct bitmap_cache *cache) } +/** + * Create gallium pipe_transfer object for the bitmap cache. + */ static void create_cache_trans(struct st_context *st) { @@ -651,7 +664,9 @@ st_flush_bitmap_cache(struct st_context *st) } } -/* Flush bitmap cache and release vertex buffer. + +/** + * Flush bitmap cache and release vertex buffer. */ void st_flush_bitmap( struct st_context *st ) @@ -737,7 +752,8 @@ accum_bitmap(struct st_context *st, * Called via ctx->Driver.Bitmap() */ static void -st_Bitmap(struct gl_context *ctx, GLint x, GLint y, GLsizei width, GLsizei height, +st_Bitmap(struct gl_context *ctx, GLint x, GLint y, + GLsizei width, GLsizei height, const struct gl_pixelstore_attrib *unpack, const GLubyte *bitmap ) { struct st_context *st = st_context(ctx); @@ -764,7 +780,8 @@ st_Bitmap(struct gl_context *ctx, GLint x, GLint y, GLsizei width, GLsizei heigh pt = make_bitmap_texture(ctx, width, height, unpack, bitmap); if (pt) { - struct pipe_sampler_view *sv = st_create_texture_sampler_view(st->pipe, pt); + struct pipe_sampler_view *sv = + st_create_texture_sampler_view(st->pipe, pt); assert(pt->target == PIPE_TEXTURE_2D || pt->target == PIPE_TEXTURE_RECT); @@ -814,15 +831,18 @@ st_init_bitmap(struct st_context *st) st->bitmap.rasterizer.gl_rasterization_rules = 1; /* find a usable texture format */ - if (screen->is_format_supported(screen, PIPE_FORMAT_I8_UNORM, PIPE_TEXTURE_2D, 0, + if (screen->is_format_supported(screen, PIPE_FORMAT_I8_UNORM, + PIPE_TEXTURE_2D, 0, PIPE_BIND_SAMPLER_VIEW, 0)) { st->bitmap.tex_format = PIPE_FORMAT_I8_UNORM; } - else if (screen->is_format_supported(screen, PIPE_FORMAT_A8_UNORM, PIPE_TEXTURE_2D, 0, + else if (screen->is_format_supported(screen, PIPE_FORMAT_A8_UNORM, + PIPE_TEXTURE_2D, 0, PIPE_BIND_SAMPLER_VIEW, 0)) { st->bitmap.tex_format = PIPE_FORMAT_A8_UNORM; } - else if (screen->is_format_supported(screen, PIPE_FORMAT_L8_UNORM, PIPE_TEXTURE_2D, 0, + else if (screen->is_format_supported(screen, PIPE_FORMAT_L8_UNORM, + PIPE_TEXTURE_2D, 0, PIPE_BIND_SAMPLER_VIEW, 0)) { st->bitmap.tex_format = PIPE_FORMAT_L8_UNORM; } @@ -845,8 +865,6 @@ st_destroy_bitmap(struct st_context *st) struct pipe_context *pipe = st->pipe; struct bitmap_cache *cache = st->bitmap.cache; - - if (st->bitmap.vs) { cso_delete_vertex_shader(st->cso_context, st->bitmap.vs); st->bitmap.vs = NULL; diff --git a/src/mesa/state_tracker/st_cb_bitmap.h b/src/mesa/state_tracker/st_cb_bitmap.h index d04b2b67795..2bd63b9b741 100644 --- a/src/mesa/state_tracker/st_cb_bitmap.h +++ b/src/mesa/state_tracker/st_cb_bitmap.h @@ -34,6 +34,8 @@ struct dd_function_table; struct st_context; +struct gl_fragment_program; +struct st_fragment_program; #if FEATURE_drawpix @@ -47,6 +49,12 @@ extern void st_destroy_bitmap(struct st_context *st); extern void +st_make_bitmap_fragment_program(struct st_context *st, + struct gl_fragment_program *fpIn, + struct gl_fragment_program **fpOut, + GLuint *bitmap_sampler); + +extern void st_flush_bitmap_cache(struct st_context *st); /* Flush bitmap cache and release vertex buffer. Needed at end of diff --git a/src/mesa/state_tracker/st_cb_bufferobjects.c b/src/mesa/state_tracker/st_cb_bufferobjects.c index 8b60f9040d0..d4d9af4ada3 100644 --- a/src/mesa/state_tracker/st_cb_bufferobjects.c +++ b/src/mesa/state_tracker/st_cb_bufferobjects.c @@ -360,7 +360,7 @@ st_bufferobj_unmap(struct gl_context *ctx, GLenum target, struct gl_buffer_objec struct st_buffer_object *st_obj = st_buffer_object(obj); if (obj->Length) - pipe_buffer_unmap(pipe, st_obj->buffer, st_obj->transfer); + pipe_buffer_unmap(pipe, st_obj->transfer); st_obj->transfer = NULL; obj->Pointer = NULL; @@ -409,8 +409,8 @@ st_copy_buffer_subdata(struct gl_context *ctx, if (srcPtr && dstPtr) memcpy(dstPtr + writeOffset, srcPtr + readOffset, size); - pipe_buffer_unmap(pipe, srcObj->buffer, src_transfer); - pipe_buffer_unmap(pipe, dstObj->buffer, dst_transfer); + pipe_buffer_unmap(pipe, src_transfer); + pipe_buffer_unmap(pipe, dst_transfer); } diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index c9786024575..1fc47688e4a 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -94,88 +94,61 @@ is_passthrough_program(const struct gl_fragment_program *prog) /** * Make fragment shader for glDraw/CopyPixels. This shader is made * by combining the pixel transfer shader with the user-defined shader. - * \return pointer to Gallium driver fragment shader + * \param fpIn the current/incoming fragment program + * \param fpOut returns the combined fragment program */ -static void * -combined_drawpix_fragment_program(struct gl_context *ctx) +void +st_make_drawpix_fragment_program(struct st_context *st, + struct gl_fragment_program *fpIn, + struct gl_fragment_program **fpOut) { - struct st_context *st = st_context(ctx); - struct st_fragment_program *stfp; + struct gl_program *newProg; - if (st->pixel_xfer.program->serialNo == st->pixel_xfer.xfer_prog_sn - && st->fp->serialNo == st->pixel_xfer.user_prog_sn) { - /* the pixel tranfer program has not changed and the user-defined - * program has not changed, so re-use the combined program. - */ - stfp = st->pixel_xfer.combined_prog; + if (is_passthrough_program(fpIn)) { + newProg = (struct gl_program *) _mesa_clone_fragment_program(st->ctx, + &st->pixel_xfer.program->Base); } else { - /* Concatenate the pixel transfer program with the current user- - * defined program. - */ - if (is_passthrough_program(&st->fp->Base)) { - stfp = (struct st_fragment_program *) - _mesa_clone_fragment_program(ctx, &st->pixel_xfer.program->Base); - } - else { #if 0 - printf("Base program:\n"); - _mesa_print_program(&st->fp->Base.Base); - printf("DrawPix program:\n"); - _mesa_print_program(&st->pixel_xfer.program->Base.Base); + /* debug */ + printf("Base program:\n"); + _mesa_print_program(&fpIn->Base); + printf("DrawPix program:\n"); + _mesa_print_program(&st->pixel_xfer.program->Base.Base); #endif - stfp = (struct st_fragment_program *) - _mesa_combine_programs(ctx, - &st->pixel_xfer.program->Base.Base, - &st->fp->Base.Base); - } + newProg = _mesa_combine_programs(st->ctx, + &st->pixel_xfer.program->Base.Base, + &fpIn->Base); + } #if 0 - { - struct gl_program *p = &stfp->Base.Base; - printf("Combined DrawPixels program:\n"); - _mesa_print_program(p); - printf("InputsRead: 0x%x\n", p->InputsRead); - printf("OutputsWritten: 0x%x\n", p->OutputsWritten); - _mesa_print_parameter_list(p->Parameters); - } + /* debug */ + printf("Combined DrawPixels program:\n"); + _mesa_print_program(newProg); + printf("InputsRead: 0x%x\n", newProg->InputsRead); + printf("OutputsWritten: 0x%x\n", newProg->OutputsWritten); + _mesa_print_parameter_list(newProg->Parameters); #endif - /* translate to TGSI tokens */ - st_translate_fragment_program(st, stfp); - - /* save new program, update serial numbers */ - st->pixel_xfer.xfer_prog_sn = st->pixel_xfer.program->serialNo; - st->pixel_xfer.user_prog_sn = st->fp->serialNo; - st->pixel_xfer.combined_prog_sn = stfp->serialNo; - /* can't reference new program directly, already have a reference on it */ - st_reference_fragprog(st, &st->pixel_xfer.combined_prog, NULL); - st->pixel_xfer.combined_prog = stfp; - } - - /* Ideally we'd have updated the pipe constants during the normal - * st/atom mechanism. But we can't since this is specific to glDrawPixels. - */ - st_upload_constants(st, stfp->Base.Base.Parameters, PIPE_SHADER_FRAGMENT); - - return stfp->driver_shader; + *fpOut = (struct gl_fragment_program *) newProg; } /** - * Create fragment shader that does a TEX() instruction to get a Z and/or + * Create fragment program that does a TEX() instruction to get a Z and/or * stencil value value, then writes to FRAG_RESULT_DEPTH/FRAG_RESULT_STENCIL. * Used for glDrawPixels(GL_DEPTH_COMPONENT / GL_STENCIL_INDEX). * Pass fragment color through as-is. - * \return pointer to the Gallium driver fragment shader + * \return pointer to the gl_fragment program */ -static void * -make_fragment_shader_z_stencil(struct st_context *st, GLboolean write_depth, - GLboolean write_stencil) +struct gl_fragment_program * +st_make_drawpix_z_stencil_program(struct st_context *st, + GLboolean write_depth, + GLboolean write_stencil) { struct gl_context *ctx = st->ctx; struct gl_program *p; - struct st_fragment_program *stp; + struct gl_fragment_program *fp; GLuint ic = 0; const GLuint shaderIndex = write_depth * 2 + write_stencil; @@ -183,7 +156,7 @@ make_fragment_shader_z_stencil(struct st_context *st, GLboolean write_depth, if (st->drawpix.shaders[shaderIndex]) { /* already have the proper shader */ - return st->drawpix.shaders[shaderIndex]->driver_shader; + return st->drawpix.shaders[shaderIndex]; } /* @@ -245,18 +218,15 @@ make_fragment_shader_z_stencil(struct st_context *st, GLboolean write_depth, if (write_stencil) p->SamplersUsed |= 1 << 1; - stp = st_fragment_program((struct gl_fragment_program *) p); + fp = (struct gl_fragment_program *) p; /* save the new shader */ - st->drawpix.shaders[shaderIndex] = stp; + st->drawpix.shaders[shaderIndex] = fp; - st_translate_fragment_program(st, stp); - - return stp->driver_shader; + return fp; } - /** * Create a simple vertex shader that just passes through the * vertex position and texcoord (and optionally, color). @@ -873,6 +843,61 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y, /** + * Get fragment program variant for a glDrawPixels or glCopyPixels + * command for RGBA data. + */ +static struct st_fp_variant * +get_color_fp_variant(struct st_context *st) +{ + struct gl_context *ctx = st->ctx; + struct st_fp_variant_key key; + struct st_fp_variant *fpv; + + memset(&key, 0, sizeof(key)); + + key.st = st; + key.drawpixels = 1; + key.scaleAndBias = (ctx->Pixel.RedBias != 0.0 || + ctx->Pixel.RedScale != 1.0 || + ctx->Pixel.GreenBias != 0.0 || + ctx->Pixel.GreenScale != 1.0 || + ctx->Pixel.BlueBias != 0.0 || + ctx->Pixel.BlueScale != 1.0 || + ctx->Pixel.AlphaBias != 0.0 || + ctx->Pixel.AlphaScale != 1.0); + key.pixelMaps = ctx->Pixel.MapColorFlag; + + fpv = st_get_fp_variant(st, st->fp, &key); + + return fpv; +} + + +/** + * Get fragment program variant for a glDrawPixels or glCopyPixels + * command for depth/stencil data. + */ +static struct st_fp_variant * +get_depth_stencil_fp_variant(struct st_context *st, GLboolean write_depth, + GLboolean write_stencil) +{ + struct st_fp_variant_key key; + struct st_fp_variant *fpv; + + memset(&key, 0, sizeof(key)); + + key.st = st; + key.drawpixels = 1; + key.drawpixels_z = write_depth; + key.drawpixels_stencil = write_stencil; + + fpv = st_get_fp_variant(st, st->fp, &key); + + return fpv; +} + + +/** * Called via ctx->Driver.DrawPixels() */ static void @@ -889,6 +914,7 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, struct pipe_sampler_view *sv[2]; int num_sampler_view = 1; enum pipe_format stencil_format = PIPE_FORMAT_NONE; + struct st_fp_variant *fpv; if (format == GL_DEPTH_STENCIL) write_stencil = write_depth = GL_TRUE; @@ -921,14 +947,25 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, st_validate_state(st); + /* + * Get vertex/fragment shaders + */ if (write_depth || write_stencil) { - driver_fp = make_fragment_shader_z_stencil(st, write_depth, write_stencil); + fpv = get_depth_stencil_fp_variant(st, write_depth, write_stencil); + + driver_fp = fpv->driver_shader; + driver_vp = make_passthrough_vertex_shader(st, GL_TRUE); + color = ctx->Current.RasterColor; } else { - driver_fp = combined_drawpix_fragment_program(ctx); + fpv = get_color_fp_variant(st); + + driver_fp = fpv->driver_shader; + driver_vp = make_passthrough_vertex_shader(st, GL_FALSE); + color = NULL; if (st->pixel_xfer.pixelmap_enabled) { sv[1] = st->pixel_xfer.pixelmap_sampler_view; @@ -936,6 +973,9 @@ st_DrawPixels(struct gl_context *ctx, GLint x, GLint y, } } + /* update fragment program constants */ + st_upload_constants(st, fpv->parameters, PIPE_SHADER_FRAGMENT); + /* draw with textured quad */ { struct pipe_resource *pt @@ -1116,6 +1156,7 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, GLint readX, readY, readW, readH; GLuint sample_count; struct gl_pixelstore_attrib pack = ctx->DefaultPacking; + struct st_fp_variant *fpv; st_validate_state(st); @@ -1125,11 +1166,18 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, return; } + /* + * Get vertex/fragment shaders + */ if (type == GL_COLOR) { rbRead = st_get_color_read_renderbuffer(ctx); color = NULL; - driver_fp = combined_drawpix_fragment_program(ctx); + + fpv = get_color_fp_variant(st); + driver_fp = fpv->driver_shader; + driver_vp = make_passthrough_vertex_shader(st, GL_FALSE); + if (st->pixel_xfer.pixelmap_enabled) { sv[1] = st->pixel_xfer.pixelmap_sampler_view; num_sampler_view++; @@ -1139,10 +1187,17 @@ st_CopyPixels(struct gl_context *ctx, GLint srcx, GLint srcy, assert(type == GL_DEPTH); rbRead = st_renderbuffer(ctx->ReadBuffer->_DepthBuffer); color = ctx->Current.Attrib[VERT_ATTRIB_COLOR0]; - driver_fp = make_fragment_shader_z_stencil(st, GL_TRUE, GL_FALSE); + + fpv = get_depth_stencil_fp_variant(st, GL_TRUE, GL_FALSE); + driver_fp = fpv->driver_shader; + driver_vp = make_passthrough_vertex_shader(st, GL_TRUE); } + /* update fragment program constants */ + st_upload_constants(st, fpv->parameters, PIPE_SHADER_FRAGMENT); + + if (rbRead->Base.Wrapped) rbRead = st_renderbuffer(rbRead->Base.Wrapped); @@ -1294,7 +1349,7 @@ st_destroy_drawpix(struct st_context *st) for (i = 0; i < Elements(st->drawpix.shaders); i++) { if (st->drawpix.shaders[i]) - st_reference_fragprog(st, &st->drawpix.shaders[i], NULL); + _mesa_reference_fragprog(st->ctx, &st->drawpix.shaders[i], NULL); } st_reference_fragprog(st, &st->pixel_xfer.combined_prog, NULL); diff --git a/src/mesa/state_tracker/st_cb_drawpixels.h b/src/mesa/state_tracker/st_cb_drawpixels.h index 575f169e08e..8f73e626240 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.h +++ b/src/mesa/state_tracker/st_cb_drawpixels.h @@ -42,6 +42,16 @@ extern void st_init_drawpixels_functions(struct dd_function_table *functions); extern void st_destroy_drawpix(struct st_context *st); +extern void +st_make_drawpix_fragment_program(struct st_context *st, + struct gl_fragment_program *fpIn, + struct gl_fragment_program **fpOut); + +extern struct gl_fragment_program * +st_make_drawpix_z_stencil_program(struct st_context *st, + GLboolean write_depth, + GLboolean write_stencil); + #else static INLINE void diff --git a/src/mesa/state_tracker/st_cb_drawtex.c b/src/mesa/state_tracker/st_cb_drawtex.c index 6cad7d3216e..22a5ed425e5 100644 --- a/src/mesa/state_tracker/st_cb_drawtex.c +++ b/src/mesa/state_tracker/st_cb_drawtex.c @@ -220,7 +220,7 @@ st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z, } } - pipe_buffer_unmap(pipe, vbuffer, vbuffer_transfer); + pipe_buffer_unmap(pipe, vbuffer_transfer); #undef SET_ATTRIB } diff --git a/src/mesa/state_tracker/st_cb_program.c b/src/mesa/state_tracker/st_cb_program.c index 4d83fcc6ccb..32694975d17 100644 --- a/src/mesa/state_tracker/st_cb_program.c +++ b/src/mesa/state_tracker/st_cb_program.c @@ -46,16 +46,13 @@ #include "st_cb_program.h" -static GLuint SerialNo = 1; - /** * Called via ctx->Driver.BindProgram() to bind an ARB vertex or * fragment program. */ -static void st_bind_program( struct gl_context *ctx, - GLenum target, - struct gl_program *prog ) +static void +st_bind_program(struct gl_context *ctx, GLenum target, struct gl_program *prog) { struct st_context *st = st_context(ctx); @@ -77,7 +74,8 @@ static void st_bind_program( struct gl_context *ctx, * Called via ctx->Driver.UseProgram() to bind a linked GLSL program * (vertex shader + fragment shader). */ -static void st_use_program( struct gl_context *ctx, struct gl_shader_program *shProg) +static void +st_use_program(struct gl_context *ctx, struct gl_shader_program *shProg) { struct st_context *st = st_context(ctx); @@ -87,48 +85,28 @@ static void st_use_program( struct gl_context *ctx, struct gl_shader_program *sh } - /** * Called via ctx->Driver.NewProgram() to allocate a new vertex or * fragment program. */ -static struct gl_program *st_new_program( struct gl_context *ctx, - GLenum target, - GLuint id ) +static struct gl_program * +st_new_program(struct gl_context *ctx, GLenum target, GLuint id) { switch (target) { case GL_VERTEX_PROGRAM_ARB: { struct st_vertex_program *prog = ST_CALLOC_STRUCT(st_vertex_program); - - prog->serialNo = SerialNo++; - - return _mesa_init_vertex_program( ctx, - &prog->Base, - target, - id ); + return _mesa_init_vertex_program(ctx, &prog->Base, target, id); } case GL_FRAGMENT_PROGRAM_ARB: case GL_FRAGMENT_PROGRAM_NV: { struct st_fragment_program *prog = ST_CALLOC_STRUCT(st_fragment_program); - - prog->serialNo = SerialNo++; - - return _mesa_init_fragment_program( ctx, - &prog->Base, - target, - id ); + return _mesa_init_fragment_program(ctx, &prog->Base, target, id); } case MESA_GEOMETRY_PROGRAM: { struct st_geometry_program *prog = ST_CALLOC_STRUCT(st_geometry_program); - - prog->serialNo = SerialNo++; - - return _mesa_init_geometry_program( ctx, - &prog->Base, - target, - id ); + return _mesa_init_geometry_program(ctx, &prog->Base, target, id); } default: @@ -138,7 +116,10 @@ static struct gl_program *st_new_program( struct gl_context *ctx, } -void +/** + * Called via ctx->Driver.DeleteProgram() + */ +static void st_delete_program(struct gl_context *ctx, struct gl_program *prog) { struct st_context *st = st_context(ctx); @@ -147,17 +128,15 @@ st_delete_program(struct gl_context *ctx, struct gl_program *prog) case GL_VERTEX_PROGRAM_ARB: { struct st_vertex_program *stvp = (struct st_vertex_program *) prog; - st_vp_release_varients( st, stvp ); + st_release_vp_variants( st, stvp ); } break; case MESA_GEOMETRY_PROGRAM: { - struct st_geometry_program *stgp = (struct st_geometry_program *) prog; + struct st_geometry_program *stgp = + (struct st_geometry_program *) prog; - if (stgp->driver_shader) { - cso_delete_geometry_shader(st->cso_context, stgp->driver_shader); - stgp->driver_shader = NULL; - } + st_release_gp_variants(st, stgp); if (stgp->tgsi.tokens) { st_free_tokens((void *) stgp->tgsi.tokens); @@ -167,23 +146,15 @@ st_delete_program(struct gl_context *ctx, struct gl_program *prog) break; case GL_FRAGMENT_PROGRAM_ARB: { - struct st_fragment_program *stfp = (struct st_fragment_program *) prog; + struct st_fragment_program *stfp = + (struct st_fragment_program *) prog; - if (stfp->driver_shader) { - cso_delete_fragment_shader(st->cso_context, stfp->driver_shader); - stfp->driver_shader = NULL; - } + st_release_fp_variants(st, stfp); if (stfp->tgsi.tokens) { st_free_tokens(stfp->tgsi.tokens); stfp->tgsi.tokens = NULL; } - - if (stfp->bitmap_program) { - struct gl_program *prg = &stfp->bitmap_program->Base.Base; - _mesa_reference_program(ctx, &prg, NULL); - stfp->bitmap_program = NULL; - } } break; default: @@ -195,15 +166,25 @@ st_delete_program(struct gl_context *ctx, struct gl_program *prog) } -static GLboolean st_is_program_native( struct gl_context *ctx, - GLenum target, - struct gl_program *prog ) +/** + * Called via ctx->Driver.IsProgramNative() + */ +static GLboolean +st_is_program_native(struct gl_context *ctx, + GLenum target, + struct gl_program *prog) { return GL_TRUE; } -static GLboolean st_program_string_notify( struct gl_context *ctx, +/** + * Called via ctx->Driver.ProgramStringNotify() + * Called when the program's text/code is changed. We have to free + * all shader variants and corresponding gallium shaders when this happens. + */ +static GLboolean +st_program_string_notify( struct gl_context *ctx, GLenum target, struct gl_program *prog ) { @@ -212,12 +193,7 @@ static GLboolean st_program_string_notify( struct gl_context *ctx, if (target == GL_FRAGMENT_PROGRAM_ARB) { struct st_fragment_program *stfp = (struct st_fragment_program *) prog; - stfp->serialNo++; - - if (stfp->driver_shader) { - cso_delete_fragment_shader(st->cso_context, stfp->driver_shader); - stfp->driver_shader = NULL; - } + st_release_fp_variants(st, stfp); if (stfp->tgsi.tokens) { st_free_tokens(stfp->tgsi.tokens); @@ -230,12 +206,7 @@ static GLboolean st_program_string_notify( struct gl_context *ctx, else if (target == MESA_GEOMETRY_PROGRAM) { struct st_geometry_program *stgp = (struct st_geometry_program *) prog; - stgp->serialNo++; - - if (stgp->driver_shader) { - cso_delete_geometry_shader(st->cso_context, stgp->driver_shader); - stgp->driver_shader = NULL; - } + st_release_gp_variants(st, stgp); if (stgp->tgsi.tokens) { st_free_tokens((void *) stgp->tgsi.tokens); @@ -248,9 +219,7 @@ static GLboolean st_program_string_notify( struct gl_context *ctx, else if (target == GL_VERTEX_PROGRAM_ARB) { struct st_vertex_program *stvp = (struct st_vertex_program *) prog; - stvp->serialNo++; - - st_vp_release_varients( st, stvp ); + st_release_vp_variants( st, stvp ); if (st->vp == stvp) st->dirty.st |= ST_NEW_VERTEX_PROGRAM; @@ -261,8 +230,11 @@ static GLboolean st_program_string_notify( struct gl_context *ctx, } - -void st_init_program_functions(struct dd_function_table *functions) +/** + * Plug in the program and shader-related device driver functions. + */ +void +st_init_program_functions(struct dd_function_table *functions) { functions->BindProgram = st_bind_program; functions->UseProgram = st_use_program; diff --git a/src/mesa/state_tracker/st_cb_program.h b/src/mesa/state_tracker/st_cb_program.h index 004afb6d812..091a4439c48 100644 --- a/src/mesa/state_tracker/st_cb_program.h +++ b/src/mesa/state_tracker/st_cb_program.h @@ -29,15 +29,10 @@ #define ST_CB_PROGRAM_H -#include "main/mtypes.h" - struct dd_function_table; extern void st_init_program_functions(struct dd_function_table *functions); -extern void -st_delete_program(struct gl_context *ctx, struct gl_program *prog); - #endif diff --git a/src/mesa/state_tracker/st_cb_readpixels.h b/src/mesa/state_tracker/st_cb_readpixels.h index 83c9b659e3d..9622ae6feea 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.h +++ b/src/mesa/state_tracker/st_cb_readpixels.h @@ -29,9 +29,11 @@ #ifndef ST_CB_READPIXELS_H #define ST_CB_READPIXELS_H -#include "main/mtypes.h" +#include "main/glheader.h" struct dd_function_table; +struct gl_context; +struct gl_pixelstore_attrib; extern struct st_renderbuffer * st_get_color_read_renderbuffer(struct gl_context *ctx); diff --git a/src/mesa/state_tracker/st_cb_texture.h b/src/mesa/state_tracker/st_cb_texture.h index 60987055eb1..f1502bda788 100644 --- a/src/mesa/state_tracker/st_cb_texture.h +++ b/src/mesa/state_tracker/st_cb_texture.h @@ -31,9 +31,10 @@ #include "main/glheader.h" -#include "main/mtypes.h" struct dd_function_table; +struct gl_context; +struct gl_texture_object; struct pipe_context; struct st_context; diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c index 6ec9c699a26..21bb91f47a8 100644 --- a/src/mesa/state_tracker/st_context.c +++ b/src/mesa/state_tracker/st_context.c @@ -252,6 +252,8 @@ void st_destroy_context( struct st_context *st ) _vbo_DestroyContext(st->ctx); + st_destroy_program_variants(st); + _mesa_free_context_data(ctx); st_destroy_context_priv(st); diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h index d342c0cff1e..1fd80053c1b 100644 --- a/src/mesa/state_tracker/st_context.h +++ b/src/mesa/state_tracker/st_context.h @@ -1,4 +1,3 @@ -//struct dd_function_table; /************************************************************************** * * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas. @@ -130,7 +129,9 @@ struct st_context struct st_fragment_program *fp; /**< Currently bound fragment program */ struct st_geometry_program *gp; /**< Currently bound geometry program */ - struct st_vp_varient *vp_varient; + struct st_vp_variant *vp_variant; + struct st_fp_variant *fp_variant; + struct st_gp_variant *gp_variant; struct gl_texture_object *default_texture; @@ -160,7 +161,7 @@ struct st_context /** for glDraw/CopyPixels */ struct { - struct st_fragment_program *shaders[4]; + struct gl_fragment_program *shaders[4]; void *vert_shaders[2]; /**< ureg shaders */ } drawpix; diff --git a/src/mesa/state_tracker/st_debug.c b/src/mesa/state_tracker/st_debug.c index df32491d044..e1e373d07d4 100644 --- a/src/mesa/state_tracker/st_debug.c +++ b/src/mesa/state_tracker/st_debug.c @@ -89,8 +89,8 @@ st_print_current(void) } #endif - if (st->vp->varients) - tgsi_dump( st->vp->varients[0].tgsi.tokens, 0 ); + if (st->vp->variants) + tgsi_dump( st->vp->variants[0].tgsi.tokens, 0 ); if (st->vp->Base.Base.Parameters) _mesa_print_parameter_list(st->vp->Base.Base.Parameters); diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index 61a0e1b0877..f4bf1802390 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -241,7 +241,7 @@ st_pipe_vertex_format(GLenum type, GLuint size, GLenum format, */ static GLboolean is_interleaved_arrays(const struct st_vertex_program *vp, - const struct st_vp_varient *vpv, + const struct st_vp_variant *vpv, const struct gl_client_array **arrays, GLboolean *userSpace) { @@ -297,7 +297,7 @@ is_interleaved_arrays(const struct st_vertex_program *vp, */ static void get_arrays_bounds(const struct st_vertex_program *vp, - const struct st_vp_varient *vpv, + const struct st_vp_variant *vpv, const struct gl_client_array **arrays, GLuint max_index, const GLubyte **low, const GLubyte **high) @@ -343,7 +343,7 @@ get_arrays_bounds(const struct st_vertex_program *vp, static void setup_interleaved_attribs(struct gl_context *ctx, const struct st_vertex_program *vp, - const struct st_vp_varient *vpv, + const struct st_vp_variant *vpv, const struct gl_client_array **arrays, GLuint max_index, GLboolean userSpace, @@ -409,7 +409,7 @@ setup_interleaved_attribs(struct gl_context *ctx, static void setup_non_interleaved_attribs(struct gl_context *ctx, const struct st_vertex_program *vp, - const struct st_vp_varient *vpv, + const struct st_vp_variant *vpv, const struct gl_client_array **arrays, GLuint max_index, GLboolean *userSpace, @@ -617,7 +617,7 @@ st_draw_vbo(struct gl_context *ctx, struct st_context *st = st_context(ctx); struct pipe_context *pipe = st->pipe; const struct st_vertex_program *vp; - const struct st_vp_varient *vpv; + const struct st_vp_variant *vpv; struct pipe_vertex_buffer vbuffer[PIPE_MAX_SHADER_INPUTS]; GLuint attr; struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS]; @@ -650,7 +650,7 @@ st_draw_vbo(struct gl_context *ctx, /* must get these after state validation! */ vp = st->vp; - vpv = st->vp_varient; + vpv = st->vp_variant; #if 0 if (MESA_VERBOSE & VERBOSE_GLSL) { diff --git a/src/mesa/state_tracker/st_draw.h b/src/mesa/state_tracker/st_draw.h index 2e4c468cff5..5d3c278228f 100644 --- a/src/mesa/state_tracker/st_draw.h +++ b/src/mesa/state_tracker/st_draw.h @@ -36,10 +36,11 @@ #include "main/compiler.h" #include "main/glheader.h" -#include "main/mtypes.h" struct _mesa_index_buffer; struct _mesa_prim; +struct gl_client_array; +struct gl_context; struct st_context; void st_init_draw( struct st_context *st ); diff --git a/src/mesa/state_tracker/st_draw_feedback.c b/src/mesa/state_tracker/st_draw_feedback.c index 7f392fc4916..da67c713710 100644 --- a/src/mesa/state_tracker/st_draw_feedback.c +++ b/src/mesa/state_tracker/st_draw_feedback.c @@ -123,10 +123,10 @@ st_feedback_draw_vbo(struct gl_context *ctx, /* must get these after state validation! */ vp = st->vp; - vs = &st->vp_varient->tgsi; + vs = &st->vp_variant->tgsi; - if (!st->vp_varient->draw_shader) { - st->vp_varient->draw_shader = draw_create_vertex_shader(draw, vs); + if (!st->vp_variant->draw_shader) { + st->vp_variant->draw_shader = draw_create_vertex_shader(draw, vs); } /* @@ -139,7 +139,7 @@ st_feedback_draw_vbo(struct gl_context *ctx, draw_set_viewport_state(draw, &st->state.viewport); draw_set_clip_state(draw, &st->state.clip); draw_set_rasterizer_state(draw, &st->state.rasterizer, NULL); - draw_bind_vertex_shader(draw, st->vp_varient->draw_shader); + draw_bind_vertex_shader(draw, st->vp_variant->draw_shader); set_feedback_vertex_format(ctx); /* loop over TGSI shader inputs to determine vertex buffer @@ -259,16 +259,14 @@ st_feedback_draw_vbo(struct gl_context *ctx, /* unmap constant buffers */ - pipe_buffer_unmap(pipe, st->state.constants[PIPE_SHADER_VERTEX], - cb_transfer); + pipe_buffer_unmap(pipe, cb_transfer); /* * unmap vertex/index buffers */ for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { if (draw->pt.vertex_buffer[i].buffer) { - pipe_buffer_unmap(pipe, draw->pt.vertex_buffer[i].buffer, - vb_transfer[i]); + pipe_buffer_unmap(pipe, vb_transfer[i]); pipe_resource_reference(&draw->pt.vertex_buffer[i].buffer, NULL); draw_set_mapped_vertex_buffer(draw, i, NULL); } @@ -279,7 +277,7 @@ st_feedback_draw_vbo(struct gl_context *ctx, draw_set_index_buffer(draw, NULL); if (ib_transfer) - pipe_buffer_unmap(pipe, ibuffer.buffer, ib_transfer); + pipe_buffer_unmap(pipe, ib_transfer); pipe_resource_reference(&ibuffer.buffer, NULL); } } diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c index 930b60ade2d..62c9ce7273d 100644 --- a/src/mesa/state_tracker/st_extensions.c +++ b/src/mesa/state_tracker/st_extensions.c @@ -343,6 +343,7 @@ void st_init_extensions(struct st_context *st) if (screen->get_param(screen, PIPE_CAP_OCCLUSION_QUERY)) { ctx->Extensions.ARB_occlusion_query = GL_TRUE; + ctx->Extensions.ARB_occlusion_query2 = GL_TRUE; } if (screen->get_param(screen, PIPE_CAP_TIMER_QUERY)) { ctx->Extensions.EXT_timer_query = GL_TRUE; @@ -442,7 +443,9 @@ void st_init_extensions(struct st_context *st) #endif if (screen->get_shader_param(screen, PIPE_SHADER_GEOMETRY, PIPE_SHADER_CAP_MAX_INSTRUCTIONS) > 0) { +#if 0 /* XXX re-enable when GLSL compiler again supports geometry shaders */ ctx->Extensions.ARB_geometry_shader4 = GL_TRUE; +#endif } if (screen->get_param(screen, PIPE_CAP_PRIMITIVE_RESTART)) { diff --git a/src/mesa/state_tracker/st_format.h b/src/mesa/state_tracker/st_format.h index 43fa59b1006..fe195c1069b 100644 --- a/src/mesa/state_tracker/st_format.h +++ b/src/mesa/state_tracker/st_format.h @@ -31,11 +31,12 @@ #define ST_FORMAT_H #include "main/formats.h" -#include "main/mtypes.h" +#include "main/glheader.h" #include "pipe/p_defines.h" #include "pipe/p_format.h" +struct gl_context; struct pipe_screen; extern GLenum diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c index aae2913c202..cfdc96b9dbe 100644 --- a/src/mesa/state_tracker/st_program.c +++ b/src/mesa/state_tracker/st_program.c @@ -32,7 +32,9 @@ #include "main/imports.h" +#include "main/hash.h" #include "main/mtypes.h" +#include "program/prog_parameter.h" #include "program/prog_print.h" #include "program/programopt.h" @@ -44,6 +46,8 @@ #include "tgsi/tgsi_ureg.h" #include "st_debug.h" +#include "st_cb_bitmap.h" +#include "st_cb_drawpixels.h" #include "st_context.h" #include "st_program.h" #include "st_mesa_to_tgsi.h" @@ -52,34 +56,109 @@ /** + * Delete a vertex program variant. Note the caller must unlink + * the variant from the linked list. + */ +static void +delete_vp_variant(struct st_context *st, struct st_vp_variant *vpv) +{ + if (vpv->driver_shader) + cso_delete_vertex_shader(st->cso_context, vpv->driver_shader); + +#if FEATURE_feedback || FEATURE_rastpos + if (vpv->draw_shader) + draw_delete_vertex_shader( st->draw, vpv->draw_shader ); +#endif + + if (vpv->tgsi.tokens) + st_free_tokens(vpv->tgsi.tokens); + + FREE( vpv ); +} + + + +/** * Clean out any old compilations: */ void -st_vp_release_varients( struct st_context *st, +st_release_vp_variants( struct st_context *st, struct st_vertex_program *stvp ) { - struct st_vp_varient *vpv; + struct st_vp_variant *vpv; - for (vpv = stvp->varients; vpv; ) { - struct st_vp_varient *next = vpv->next; + for (vpv = stvp->variants; vpv; ) { + struct st_vp_variant *next = vpv->next; + delete_vp_variant(st, vpv); + vpv = next; + } - if (vpv->driver_shader) - cso_delete_vertex_shader(st->cso_context, vpv->driver_shader); - -#if FEATURE_feedback || FEATURE_rastpos - if (vpv->draw_shader) - draw_delete_vertex_shader( st->draw, vpv->draw_shader ); -#endif + stvp->variants = NULL; +} + + + +/** + * Delete a fragment program variant. Note the caller must unlink + * the variant from the linked list. + */ +static void +delete_fp_variant(struct st_context *st, struct st_fp_variant *fpv) +{ + if (fpv->driver_shader) + cso_delete_fragment_shader(st->cso_context, fpv->driver_shader); - if (vpv->tgsi.tokens) - st_free_tokens(vpv->tgsi.tokens); + FREE(fpv); +} + + +/** + * Free all variants of a fragment program. + */ +void +st_release_fp_variants(struct st_context *st, struct st_fragment_program *stfp) +{ + struct st_fp_variant *fpv; + + for (fpv = stfp->variants; fpv; ) { + struct st_fp_variant *next = fpv->next; + delete_fp_variant(st, fpv); + fpv = next; + } + + stfp->variants = NULL; +} + + +/** + * Delete a geometry program variant. Note the caller must unlink + * the variant from the linked list. + */ +static void +delete_gp_variant(struct st_context *st, struct st_gp_variant *gpv) +{ + if (gpv->driver_shader) + cso_delete_geometry_shader(st->cso_context, gpv->driver_shader); - FREE( vpv ); + FREE(gpv); +} - vpv = next; + +/** + * Free all variants of a geometry program. + */ +void +st_release_gp_variants(struct st_context *st, struct st_geometry_program *stgp) +{ + struct st_gp_variant *gpv; + + for (gpv = stgp->variants; gpv; ) { + struct st_gp_variant *next = gpv->next; + delete_gp_variant(st, gpv); + gpv = next; } - stvp->varients = NULL; + stgp->variants = NULL; } @@ -92,7 +171,7 @@ st_vp_release_varients( struct st_context *st, * \param tokensOut destination for TGSI tokens * \return pointer to cached pipe_shader object. */ -void +static void st_prepare_vertex_program(struct st_context *st, struct st_vertex_program *stvp) { @@ -196,17 +275,22 @@ st_prepare_vertex_program(struct st_context *st, } -struct st_vp_varient * +/** + * Translate a vertex program to create a new variant. + */ +static struct st_vp_variant * st_translate_vertex_program(struct st_context *st, struct st_vertex_program *stvp, - const struct st_vp_varient_key *key) + const struct st_vp_variant_key *key) { - struct st_vp_varient *vpv = CALLOC_STRUCT(st_vp_varient); + struct st_vp_variant *vpv = CALLOC_STRUCT(st_vp_variant); struct pipe_context *pipe = st->pipe; struct ureg_program *ureg; enum pipe_error error; unsigned num_outputs; + st_prepare_vertex_program( st, stvp ); + _mesa_remove_output_reads(&stvp->Base.Base, PROGRAM_OUTPUT); _mesa_remove_output_reads(&stvp->Base.Base, PROGRAM_VARYING); @@ -231,23 +315,22 @@ st_translate_vertex_program(struct st_context *st, debug_printf("\n"); } - error = - st_translate_mesa_program(st->ctx, - TGSI_PROCESSOR_VERTEX, - ureg, - &stvp->Base.Base, - /* inputs */ - vpv->num_inputs, - stvp->input_to_index, - NULL, /* input semantic name */ - NULL, /* input semantic index */ - NULL, - /* outputs */ - num_outputs, - stvp->result_to_output, - stvp->output_semantic_name, - stvp->output_semantic_index, - key->passthrough_edgeflags ); + error = st_translate_mesa_program(st->ctx, + TGSI_PROCESSOR_VERTEX, + ureg, + &stvp->Base.Base, + /* inputs */ + vpv->num_inputs, + stvp->input_to_index, + NULL, /* input semantic name */ + NULL, /* input semantic index */ + NULL, + /* outputs */ + num_outputs, + stvp->result_to_output, + stvp->output_semantic_name, + stvp->output_semantic_index, + key->passthrough_edgeflags ); if (error) goto fail; @@ -277,201 +360,310 @@ fail: } +/** + * Find/create a vertex program variant. + */ +struct st_vp_variant * +st_get_vp_variant(struct st_context *st, + struct st_vertex_program *stvp, + const struct st_vp_variant_key *key) +{ + struct st_vp_variant *vpv; + + /* Search for existing variant */ + for (vpv = stvp->variants; vpv; vpv = vpv->next) { + if (memcmp(&vpv->key, key, sizeof(*key)) == 0) { + break; + } + } + + if (!vpv) { + /* create now */ + vpv = st_translate_vertex_program(st, stvp, key); + if (vpv) { + /* insert into list */ + vpv->next = stvp->variants; + stvp->variants = vpv; + } + } + + return vpv; +} + /** - * Translate a Mesa fragment shader into a TGSI shader. - * \return pointer to cached pipe_shader object. + * Translate a Mesa fragment shader into a TGSI shader using extra info in + * the key. + * \return new fragment program variant */ -void +static struct st_fp_variant * st_translate_fragment_program(struct st_context *st, - struct st_fragment_program *stfp ) + struct st_fragment_program *stfp, + const struct st_fp_variant_key *key) { struct pipe_context *pipe = st->pipe; - GLuint outputMapping[FRAG_RESULT_MAX]; - GLuint inputMapping[FRAG_ATTRIB_MAX]; - GLuint interpMode[PIPE_MAX_SHADER_INPUTS]; /* XXX size? */ - GLuint attr; - enum pipe_error error; - const GLbitfield inputsRead = stfp->Base.Base.InputsRead; - struct ureg_program *ureg; + struct st_fp_variant *variant = CALLOC_STRUCT(st_fp_variant); - ubyte input_semantic_name[PIPE_MAX_SHADER_INPUTS]; - ubyte input_semantic_index[PIPE_MAX_SHADER_INPUTS]; - uint fs_num_inputs = 0; + if (!variant) + return NULL; - ubyte fs_output_semantic_name[PIPE_MAX_SHADER_OUTPUTS]; - ubyte fs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS]; - uint fs_num_outputs = 0; + assert(!(key->bitmap && key->drawpixels)); - _mesa_remove_output_reads(&stfp->Base.Base, PROGRAM_OUTPUT); + if (key->bitmap) { + /* glBitmap drawing */ + struct gl_fragment_program *fp; - /* - * Convert Mesa program inputs to TGSI input register semantics. - */ - for (attr = 0; attr < FRAG_ATTRIB_MAX; attr++) { - if (inputsRead & (1 << attr)) { - const GLuint slot = fs_num_inputs++; + st_make_bitmap_fragment_program(st, &stfp->Base, + &fp, &variant->bitmap_sampler); - inputMapping[attr] = slot; + variant->parameters = _mesa_clone_parameter_list(fp->Base.Parameters); + stfp = st_fragment_program(fp); + } + else if (key->drawpixels) { + /* glDrawPixels drawing */ + struct gl_fragment_program *fp; - switch (attr) { - case FRAG_ATTRIB_WPOS: - input_semantic_name[slot] = TGSI_SEMANTIC_POSITION; - input_semantic_index[slot] = 0; - interpMode[slot] = TGSI_INTERPOLATE_LINEAR; - break; - case FRAG_ATTRIB_COL0: - input_semantic_name[slot] = TGSI_SEMANTIC_COLOR; - input_semantic_index[slot] = 0; - interpMode[slot] = TGSI_INTERPOLATE_LINEAR; - break; - case FRAG_ATTRIB_COL1: - input_semantic_name[slot] = TGSI_SEMANTIC_COLOR; - input_semantic_index[slot] = 1; - interpMode[slot] = TGSI_INTERPOLATE_LINEAR; - break; - case FRAG_ATTRIB_FOGC: - input_semantic_name[slot] = TGSI_SEMANTIC_FOG; - input_semantic_index[slot] = 0; - interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE; - break; - case FRAG_ATTRIB_FACE: - input_semantic_name[slot] = TGSI_SEMANTIC_FACE; - input_semantic_index[slot] = 0; - interpMode[slot] = TGSI_INTERPOLATE_CONSTANT; - break; - /* In most cases, there is nothing special about these - * inputs, so adopt a convention to use the generic - * semantic name and the mesa FRAG_ATTRIB_ number as the - * index. - * - * All that is required is that the vertex shader labels - * its own outputs similarly, and that the vertex shader - * generates at least every output required by the - * fragment shader plus fixed-function hardware (such as - * BFC). - * - * There is no requirement that semantic indexes start at - * zero or be restricted to a particular range -- nobody - * should be building tables based on semantic index. - */ - case FRAG_ATTRIB_PNTC: - 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: - case FRAG_ATTRIB_VAR0: - default: - /* Actually, let's try and zero-base this just for - * readability of the generated TGSI. - */ - assert(attr >= FRAG_ATTRIB_TEX0); - input_semantic_index[slot] = (attr - FRAG_ATTRIB_TEX0); - input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC; - if (attr == FRAG_ATTRIB_PNTC) - interpMode[slot] = TGSI_INTERPOLATE_LINEAR; - else - interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE; - break; - } + if (key->drawpixels_z || key->drawpixels_stencil) { + fp = st_make_drawpix_z_stencil_program(st, key->drawpixels_z, + key->drawpixels_stencil); } else { - inputMapping[attr] = -1; + /* RGBA */ + st_make_drawpix_fragment_program(st, &stfp->Base, &fp); + variant->parameters = _mesa_clone_parameter_list(fp->Base.Parameters); } + stfp = st_fragment_program(fp); } - /* - * Semantics and mapping for outputs - */ - { - uint numColors = 0; - GLbitfield64 outputsWritten = stfp->Base.Base.OutputsWritten; - - /* if z is written, emit that first */ - if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) { - fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_POSITION; - fs_output_semantic_index[fs_num_outputs] = 0; - outputMapping[FRAG_RESULT_DEPTH] = fs_num_outputs; - fs_num_outputs++; - outputsWritten &= ~(1 << FRAG_RESULT_DEPTH); - } + if (!stfp->tgsi.tokens) { + /* need to translate Mesa instructions to TGSI now */ + GLuint outputMapping[FRAG_RESULT_MAX]; + GLuint inputMapping[FRAG_ATTRIB_MAX]; + GLuint interpMode[PIPE_MAX_SHADER_INPUTS]; /* XXX size? */ + GLuint attr; + enum pipe_error error; + const GLbitfield inputsRead = stfp->Base.Base.InputsRead; + struct ureg_program *ureg; - if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_STENCIL)) { - fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_STENCIL; - fs_output_semantic_index[fs_num_outputs] = 0; - outputMapping[FRAG_RESULT_STENCIL] = fs_num_outputs; - fs_num_outputs++; - outputsWritten &= ~(1 << FRAG_RESULT_STENCIL); - } + ubyte input_semantic_name[PIPE_MAX_SHADER_INPUTS]; + ubyte input_semantic_index[PIPE_MAX_SHADER_INPUTS]; + uint fs_num_inputs = 0; + + ubyte fs_output_semantic_name[PIPE_MAX_SHADER_OUTPUTS]; + ubyte fs_output_semantic_index[PIPE_MAX_SHADER_OUTPUTS]; + uint fs_num_outputs = 0; + + + _mesa_remove_output_reads(&stfp->Base.Base, PROGRAM_OUTPUT); + + /* + * Convert Mesa program inputs to TGSI input register semantics. + */ + for (attr = 0; attr < FRAG_ATTRIB_MAX; attr++) { + if (inputsRead & (1 << attr)) { + const GLuint slot = fs_num_inputs++; + + inputMapping[attr] = slot; - /* handle remaning outputs (color) */ - for (attr = 0; attr < FRAG_RESULT_MAX; attr++) { - if (outputsWritten & BITFIELD64_BIT(attr)) { switch (attr) { - case FRAG_RESULT_DEPTH: - case FRAG_RESULT_STENCIL: - /* handled above */ - assert(0); + case FRAG_ATTRIB_WPOS: + input_semantic_name[slot] = TGSI_SEMANTIC_POSITION; + input_semantic_index[slot] = 0; + interpMode[slot] = TGSI_INTERPOLATE_LINEAR; + break; + case FRAG_ATTRIB_COL0: + input_semantic_name[slot] = TGSI_SEMANTIC_COLOR; + input_semantic_index[slot] = 0; + interpMode[slot] = TGSI_INTERPOLATE_LINEAR; break; + case FRAG_ATTRIB_COL1: + input_semantic_name[slot] = TGSI_SEMANTIC_COLOR; + input_semantic_index[slot] = 1; + interpMode[slot] = TGSI_INTERPOLATE_LINEAR; + break; + case FRAG_ATTRIB_FOGC: + input_semantic_name[slot] = TGSI_SEMANTIC_FOG; + input_semantic_index[slot] = 0; + interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE; + break; + case FRAG_ATTRIB_FACE: + input_semantic_name[slot] = TGSI_SEMANTIC_FACE; + input_semantic_index[slot] = 0; + interpMode[slot] = TGSI_INTERPOLATE_CONSTANT; + break; + /* In most cases, there is nothing special about these + * inputs, so adopt a convention to use the generic + * semantic name and the mesa FRAG_ATTRIB_ number as the + * index. + * + * All that is required is that the vertex shader labels + * its own outputs similarly, and that the vertex shader + * generates at least every output required by the + * fragment shader plus fixed-function hardware (such as + * BFC). + * + * There is no requirement that semantic indexes start at + * zero or be restricted to a particular range -- nobody + * should be building tables based on semantic index. + */ + case FRAG_ATTRIB_PNTC: + 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: + case FRAG_ATTRIB_VAR0: default: - assert(attr == FRAG_RESULT_COLOR || - (FRAG_RESULT_DATA0 <= attr && attr < FRAG_RESULT_MAX)); - fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_COLOR; - fs_output_semantic_index[fs_num_outputs] = numColors; - outputMapping[attr] = fs_num_outputs; - numColors++; + /* Actually, let's try and zero-base this just for + * readability of the generated TGSI. + */ + assert(attr >= FRAG_ATTRIB_TEX0); + input_semantic_index[slot] = (attr - FRAG_ATTRIB_TEX0); + input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC; + if (attr == FRAG_ATTRIB_PNTC) + interpMode[slot] = TGSI_INTERPOLATE_LINEAR; + else + interpMode[slot] = TGSI_INTERPOLATE_PERSPECTIVE; break; } + } + else { + inputMapping[attr] = -1; + } + } + /* + * Semantics and mapping for outputs + */ + { + uint numColors = 0; + GLbitfield64 outputsWritten = stfp->Base.Base.OutputsWritten; + + /* if z is written, emit that first */ + if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) { + fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_POSITION; + fs_output_semantic_index[fs_num_outputs] = 0; + outputMapping[FRAG_RESULT_DEPTH] = fs_num_outputs; fs_num_outputs++; + outputsWritten &= ~(1 << FRAG_RESULT_DEPTH); + } + + if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_STENCIL)) { + fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_STENCIL; + fs_output_semantic_index[fs_num_outputs] = 0; + outputMapping[FRAG_RESULT_STENCIL] = fs_num_outputs; + fs_num_outputs++; + outputsWritten &= ~(1 << FRAG_RESULT_STENCIL); + } + + /* handle remaning outputs (color) */ + for (attr = 0; attr < FRAG_RESULT_MAX; attr++) { + if (outputsWritten & BITFIELD64_BIT(attr)) { + switch (attr) { + case FRAG_RESULT_DEPTH: + case FRAG_RESULT_STENCIL: + /* handled above */ + assert(0); + break; + default: + assert(attr == FRAG_RESULT_COLOR || + (FRAG_RESULT_DATA0 <= attr && attr < FRAG_RESULT_MAX)); + fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_COLOR; + fs_output_semantic_index[fs_num_outputs] = numColors; + outputMapping[attr] = fs_num_outputs; + numColors++; + break; + } + + fs_num_outputs++; + } } } - } - ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT ); - if (ureg == NULL) - return; + ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT ); + if (ureg == NULL) + return NULL; - if (ST_DEBUG & DEBUG_MESA) { - _mesa_print_program(&stfp->Base.Base); - _mesa_print_program_parameters(st->ctx, &stfp->Base.Base); - debug_printf("\n"); + if (ST_DEBUG & DEBUG_MESA) { + _mesa_print_program(&stfp->Base.Base); + _mesa_print_program_parameters(st->ctx, &stfp->Base.Base); + debug_printf("\n"); + } + + error = st_translate_mesa_program(st->ctx, + TGSI_PROCESSOR_FRAGMENT, + ureg, + &stfp->Base.Base, + /* inputs */ + fs_num_inputs, + inputMapping, + input_semantic_name, + input_semantic_index, + interpMode, + /* outputs */ + fs_num_outputs, + outputMapping, + fs_output_semantic_name, + fs_output_semantic_index, FALSE ); + + stfp->tgsi.tokens = ureg_get_tokens( ureg, NULL ); + ureg_destroy( ureg ); } - error = - st_translate_mesa_program(st->ctx, - TGSI_PROCESSOR_FRAGMENT, - ureg, - &stfp->Base.Base, - /* inputs */ - fs_num_inputs, - inputMapping, - input_semantic_name, - input_semantic_index, - interpMode, - /* outputs */ - fs_num_outputs, - outputMapping, - fs_output_semantic_name, - fs_output_semantic_index, FALSE ); - - stfp->tgsi.tokens = ureg_get_tokens( ureg, NULL ); - ureg_destroy( ureg ); - stfp->driver_shader = pipe->create_fs_state(pipe, &stfp->tgsi); + /* fill in variant */ + variant->driver_shader = pipe->create_fs_state(pipe, &stfp->tgsi); + variant->key = *key; if (ST_DEBUG & DEBUG_TGSI) { tgsi_dump( stfp->tgsi.tokens, 0/*TGSI_DUMP_VERBOSE*/ ); debug_printf("\n"); } + + return variant; } -void + +/** + * Translate fragment program if needed. + */ +struct st_fp_variant * +st_get_fp_variant(struct st_context *st, + struct st_fragment_program *stfp, + const struct st_fp_variant_key *key) +{ + struct st_fp_variant *fpv; + + /* Search for existing variant */ + for (fpv = stfp->variants; fpv; fpv = fpv->next) { + if (memcmp(&fpv->key, key, sizeof(*key)) == 0) { + break; + } + } + + if (!fpv) { + /* create new */ + fpv = st_translate_fragment_program(st, stfp, key); + if (fpv) { + /* insert into list */ + fpv->next = stfp->variants; + stfp->variants = fpv; + } + } + + return fpv; +} + + +/** + * Translate a geometry program to create a new variant. + */ +static struct st_gp_variant * st_translate_geometry_program(struct st_context *st, - struct st_geometry_program *stgp) + struct st_geometry_program *stgp, + const struct st_gp_variant_key *key) { GLuint inputMapping[GEOM_ATTRIB_MAX]; GLuint outputMapping[GEOM_RESULT_MAX]; @@ -494,12 +686,19 @@ st_translate_geometry_program(struct st_context *st, GLuint maxSlot = 0; struct ureg_program *ureg; + struct st_gp_variant *gpv; + + gpv = CALLOC_STRUCT(st_gp_variant); + if (!gpv) + return NULL; + _mesa_remove_output_reads(&stgp->Base.Base, PROGRAM_OUTPUT); _mesa_remove_output_reads(&stgp->Base.Base, PROGRAM_VARYING); ureg = ureg_create( TGSI_PROCESSOR_GEOMETRY ); if (ureg == NULL) { - return; + FREE(gpv); + return NULL; } /* which vertex output goes to the first geometry input */ @@ -529,7 +728,7 @@ st_translate_geometry_program(struct st_context *st, } else ++gs_builtin_inputs; -#if 1 +#if 0 debug_printf("input map at %d = %d\n", slot + gs_array_offset, stgp->input_map[slot + gs_array_offset]); #endif @@ -671,37 +870,35 @@ st_translate_geometry_program(struct st_context *st, st_free_tokens(stgp->tgsi.tokens); stgp->tgsi.tokens = NULL; } - if (stgp->driver_shader) { - cso_delete_geometry_shader(st->cso_context, stgp->driver_shader); - stgp->driver_shader = NULL; - } ureg_property_gs_input_prim(ureg, stgp->Base.InputType); ureg_property_gs_output_prim(ureg, stgp->Base.OutputType); ureg_property_gs_max_vertices(ureg, stgp->Base.VerticesOut); - error = st_translate_mesa_program(st->ctx, - TGSI_PROCESSOR_GEOMETRY, - ureg, - &stgp->Base.Base, - /* inputs */ - gs_num_inputs, - inputMapping, - stgp->input_semantic_name, - stgp->input_semantic_index, - NULL, - /* outputs */ - gs_num_outputs, - outputMapping, - gs_output_semantic_name, - gs_output_semantic_index, - FALSE); - + error = st_translate_mesa_program(st->ctx, + TGSI_PROCESSOR_GEOMETRY, + ureg, + &stgp->Base.Base, + /* inputs */ + gs_num_inputs, + inputMapping, + stgp->input_semantic_name, + stgp->input_semantic_index, + NULL, + /* outputs */ + gs_num_outputs, + outputMapping, + gs_output_semantic_name, + gs_output_semantic_index, + FALSE); stgp->num_inputs = gs_num_inputs; stgp->tgsi.tokens = ureg_get_tokens( ureg, NULL ); ureg_destroy( ureg ); - stgp->driver_shader = pipe->create_gs_state(pipe, &stgp->tgsi); + + /* fill in new variant */ + gpv->driver_shader = pipe->create_gs_state(pipe, &stgp->tgsi); + gpv->key = *key; if ((ST_DEBUG & DEBUG_TGSI) && (ST_DEBUG & DEBUG_MESA)) { _mesa_print_program(&stgp->Base.Base); @@ -712,8 +909,44 @@ st_translate_geometry_program(struct st_context *st, tgsi_dump(stgp->tgsi.tokens, 0); debug_printf("\n"); } + + return gpv; +} + + +/** + * Get/create geometry program variant. + */ +struct st_gp_variant * +st_get_gp_variant(struct st_context *st, + struct st_geometry_program *stgp, + const struct st_gp_variant_key *key) +{ + struct st_gp_variant *gpv; + + /* Search for existing variant */ + for (gpv = stgp->variants; gpv; gpv = gpv->next) { + if (memcmp(&gpv->key, key, sizeof(*key)) == 0) { + break; + } + } + + if (!gpv) { + /* create new */ + gpv = st_translate_geometry_program(st, stgp, key); + if (gpv) { + /* insert into list */ + gpv->next = stgp->variants; + stgp->variants = gpv; + } + } + + return gpv; } + + + /** * Debug- print current shader text */ @@ -759,3 +992,155 @@ st_print_shaders(struct gl_context *ctx) } } } + + +/** + * Vert/Geom/Frag programs have per-context variants. Free all the + * variants attached to the given program which match the given context. + */ +static void +destroy_program_variants(struct st_context *st, struct gl_program *program) +{ + if (!program) + return; + + switch (program->Target) { + case GL_VERTEX_PROGRAM_ARB: + { + struct st_vertex_program *stvp = (struct st_vertex_program *) program; + struct st_vp_variant *vpv, **prevPtr = &stvp->variants; + + for (vpv = stvp->variants; vpv; ) { + struct st_vp_variant *next = vpv->next; + if (vpv->key.st == st) { + /* unlink from list */ + *prevPtr = next; + /* destroy this variant */ + delete_vp_variant(st, vpv); + } + else { + prevPtr = &vpv->next; + } + vpv = next; + } + } + break; + case GL_FRAGMENT_PROGRAM_ARB: + { + struct st_fragment_program *stfp = + (struct st_fragment_program *) program; + struct st_fp_variant *fpv, **prevPtr = &stfp->variants; + + for (fpv = stfp->variants; fpv; ) { + struct st_fp_variant *next = fpv->next; + if (fpv->key.st == st) { + /* unlink from list */ + *prevPtr = next; + /* destroy this variant */ + delete_fp_variant(st, fpv); + } + else { + prevPtr = &fpv->next; + } + fpv = next; + } + } + break; + case MESA_GEOMETRY_PROGRAM: + { + struct st_geometry_program *stgp = + (struct st_geometry_program *) program; + struct st_gp_variant *gpv, **prevPtr = &stgp->variants; + + for (gpv = stgp->variants; gpv; ) { + struct st_gp_variant *next = gpv->next; + if (gpv->key.st == st) { + /* unlink from list */ + *prevPtr = next; + /* destroy this variant */ + delete_gp_variant(st, gpv); + } + else { + prevPtr = &gpv->next; + } + gpv = next; + } + } + break; + default: + _mesa_problem(NULL, "Unexpected program target in " + "destroy_program_variants_cb()"); + } +} + + +/** + * Callback for _mesa_HashWalk. Free all the shader's program variants + * which match the given context. + */ +static void +destroy_shader_program_variants_cb(GLuint key, void *data, void *userData) +{ + struct st_context *st = (struct st_context *) userData; + struct gl_shader *shader = (struct gl_shader *) data; + + switch (shader->Type) { + case GL_SHADER_PROGRAM_MESA: + { + struct gl_shader_program *shProg = (struct gl_shader_program *) data; + GLuint i; + + for (i = 0; i < shProg->NumShaders; i++) { + destroy_program_variants(st, shProg->Shaders[i]->Program); + } + + destroy_program_variants(st, (struct gl_program *) + shProg->VertexProgram); + destroy_program_variants(st, (struct gl_program *) + shProg->FragmentProgram); + destroy_program_variants(st, (struct gl_program *) + shProg->GeometryProgram); + } + break; + case GL_VERTEX_SHADER: + case GL_FRAGMENT_SHADER: + case GL_GEOMETRY_SHADER: + { + destroy_program_variants(st, shader->Program); + } + break; + default: + assert(0); + } +} + + +/** + * Callback for _mesa_HashWalk. Free all the program variants which match + * the given context. + */ +static void +destroy_program_variants_cb(GLuint key, void *data, void *userData) +{ + struct st_context *st = (struct st_context *) userData; + struct gl_program *program = (struct gl_program *) data; + destroy_program_variants(st, program); +} + + +/** + * Walk over all shaders and programs to delete any variants which + * belong to the given context. + * This is called during context tear-down. + */ +void +st_destroy_program_variants(struct st_context *st) +{ + /* ARB vert/frag program */ + _mesa_HashWalk(st->ctx->Shared->Programs, + destroy_program_variants_cb, st); + + /* GLSL vert/frag/geom shaders */ + _mesa_HashWalk(st->ctx->Shared->ShaderObjects, + destroy_shader_program_variants_cb, st); +} diff --git a/src/mesa/state_tracker/st_program.h b/src/mesa/state_tracker/st_program.h index 72dbc715fe1..c4244df939e 100644 --- a/src/mesa/state_tracker/st_program.h +++ b/src/mesa/state_tracker/st_program.h @@ -40,26 +40,61 @@ #include "st_context.h" +/** Fragment program variant key */ +struct st_fp_variant_key +{ + struct st_context *st; /**< variants are per-context */ + + /** for glBitmap */ + GLuint bitmap:1; /**< glBitmap variant? */ + + /** for glDrawPixels */ + GLuint drawpixels:1; /**< glDrawPixels variant */ + GLuint scaleAndBias:1; /**< glDrawPixels w/ scale and/or bias? */ + GLuint pixelMaps:1; /**< glDrawPixels w/ pixel lookup map? */ + GLuint drawpixels_z:1; /**< glDrawPixels(GL_DEPTH) */ + GLuint drawpixels_stencil:1; /**< glDrawPixels(GL_STENCIL) */ +}; + + +/** + * Variant of a fragment program. + */ +struct st_fp_variant +{ + /** Parameters which generated this version of fragment program */ + struct st_fp_variant_key key; + + /** Driver's compiled shader */ + void *driver_shader; + + /** For glBitmap variants */ + struct gl_program_parameter_list *parameters; + uint bitmap_sampler; + + /** next in linked list */ + struct st_fp_variant *next; +}; + + /** * Derived from Mesa gl_fragment_program: */ struct st_fragment_program { struct gl_fragment_program Base; - GLuint serialNo; struct pipe_shader_state tgsi; - void *driver_shader; - /** Program prefixed with glBitmap prologue */ - struct st_fragment_program *bitmap_program; - uint bitmap_sampler; + struct st_fp_variant *variants; }; -struct st_vp_varient_key +/** Vertex program variant key */ +struct st_vp_variant_key { + struct st_context *st; /**< variants are per-context */ boolean passthrough_edgeflags; }; @@ -68,12 +103,12 @@ struct st_vp_varient_key * This represents a vertex program, especially translated to match * the inputs of a particular fragment shader. */ -struct st_vp_varient +struct st_vp_variant { /* Parameters which generated this translated version of a vertex * shader: */ - struct st_vp_varient_key key; + struct st_vp_variant_key key; /** * TGSI tokens (to later generate a 'draw' module shader for @@ -88,9 +123,9 @@ struct st_vp_varient struct draw_vertex_shader *draw_shader; /** Next in linked list */ - struct st_vp_varient *next; + struct st_vp_variant *next; - /** similar to that in st_vertex_program, but with information about edgeflags too */ + /** similar to that in st_vertex_program, but with edgeflags info too */ GLuint num_inputs; }; @@ -101,7 +136,6 @@ struct st_vp_varient struct st_vertex_program { struct gl_vertex_program Base; /**< The Mesa vertex program */ - GLuint serialNo, lastSerialNo; /** maps a Mesa VERT_ATTRIB_x to a packed TGSI input index */ GLuint input_to_index[VERT_ATTRIB_MAX]; @@ -115,18 +149,41 @@ struct st_vertex_program ubyte output_semantic_index[VERT_RESULT_MAX]; GLuint num_outputs; - /** List of translated varients of this vertex program. + /** List of translated variants of this vertex program. */ - struct st_vp_varient *varients; + struct st_vp_variant *variants; +}; + + + +/** Geometry program variant key */ +struct st_gp_variant_key +{ + struct st_context *st; /**< variants are per-context */ + /* no other fields yet */ +}; + + +/** + * Geometry program variant. + */ +struct st_gp_variant +{ + /* Parameters which generated this translated version of a vertex */ + struct st_gp_variant_key key; + + void *driver_shader; + + struct st_gp_variant *next; }; + /** * Derived from Mesa gl_geometry_program: */ struct st_geometry_program { struct gl_geometry_program Base; /**< The Mesa geometry program */ - GLuint serialNo; /** map GP input back to VP output */ GLuint input_map[PIPE_MAX_SHADER_INPUTS]; @@ -145,9 +202,12 @@ struct st_geometry_program ubyte input_semantic_index[PIPE_MAX_SHADER_INPUTS]; struct pipe_shader_state tgsi; - void *driver_shader; + + struct st_gp_variant *variants; }; + + static INLINE struct st_fragment_program * st_fragment_program( struct gl_fragment_program *fp ) { @@ -162,9 +222,9 @@ st_vertex_program( struct gl_vertex_program *vp ) } static INLINE struct st_geometry_program * -st_geometry_program( struct gl_geometry_program *vp ) +st_geometry_program( struct gl_geometry_program *gp ) { - return (struct st_geometry_program *)vp; + return (struct st_geometry_program *)gp; } static INLINE void @@ -198,32 +258,43 @@ st_reference_fragprog(struct st_context *st, } -extern void -st_translate_fragment_program(struct st_context *st, - struct st_fragment_program *fp); +extern struct st_vp_variant * +st_get_vp_variant(struct st_context *st, + struct st_vertex_program *stvp, + const struct st_vp_variant_key *key); + + +extern struct st_fp_variant * +st_get_fp_variant(struct st_context *st, + struct st_fragment_program *stfp, + const struct st_fp_variant_key *key); + + +extern struct st_gp_variant * +st_get_gp_variant(struct st_context *st, + struct st_geometry_program *stgp, + const struct st_gp_variant_key *key); + + extern void -st_translate_geometry_program(struct st_context *st, - struct st_geometry_program *stgp); +st_release_vp_variants( struct st_context *st, + struct st_vertex_program *stvp ); -/* Called after program string change, discard all previous - * compilation results. - */ extern void -st_prepare_vertex_program(struct st_context *st, - struct st_vertex_program *stvp); +st_release_fp_variants( struct st_context *st, + struct st_fragment_program *stfp ); -extern struct st_vp_varient * -st_translate_vertex_program(struct st_context *st, - struct st_vertex_program *stvp, - const struct st_vp_varient_key *key); +extern void +st_release_gp_variants(struct st_context *st, + struct st_geometry_program *stgp); -void -st_vp_release_varients( struct st_context *st, - struct st_vertex_program *stvp ); extern void st_print_shaders(struct gl_context *ctx); +extern void +st_destroy_program_variants(struct st_context *st); + #endif diff --git a/src/mesa/tnl/t_vb_program.c b/src/mesa/tnl/t_vb_program.c index 76f8fde3f52..94372bbafbc 100644 --- a/src/mesa/tnl/t_vb_program.c +++ b/src/mesa/tnl/t_vb_program.c @@ -67,6 +67,8 @@ struct vp_stage_data { GLvector4f ndcCoords; /**< normalized device coords */ GLubyte *clipmask; /**< clip flags */ GLubyte ormask, andmask; /**< for clipping */ + + struct gl_program_machine machine; }; @@ -311,7 +313,7 @@ run_vp( struct gl_context *ctx, struct tnl_pipeline_stage *stage ) struct vp_stage_data *store = VP_STAGE_DATA(stage); struct vertex_buffer *VB = &tnl->vb; struct gl_vertex_program *program = ctx->VertexProgram._Current; - struct gl_program_machine machine; + struct gl_program_machine *machine = &store->machine; GLuint outputs[VERT_RESULT_MAX], numOutputs; GLuint i, j; @@ -339,7 +341,7 @@ run_vp( struct gl_context *ctx, struct tnl_pipeline_stage *stage ) for (i = 0; i < VB->Count; i++) { GLuint attr; - init_machine(ctx, &machine); + init_machine(ctx, machine); #if 0 printf("Input %d: %f, %f, %f, %f\n", i, @@ -372,23 +374,23 @@ run_vp( struct gl_context *ctx, struct tnl_pipeline_stage *stage ) check_float(data[2]); check_float(data[3]); #endif - COPY_CLEAN_4V(machine.VertAttribs[attr], size, data); + COPY_CLEAN_4V(machine->VertAttribs[attr], size, data); } } /* execute the program */ - _mesa_execute_program(ctx, &program->Base, &machine); + _mesa_execute_program(ctx, &program->Base, machine); /* copy the output registers into the VB->attribs arrays */ for (j = 0; j < numOutputs; j++) { const GLuint attr = outputs[j]; #ifdef NAN_CHECK - check_float(machine.Outputs[attr][0]); - check_float(machine.Outputs[attr][1]); - check_float(machine.Outputs[attr][2]); - check_float(machine.Outputs[attr][3]); + check_float(machine->Outputs[attr][0]); + check_float(machine->Outputs[attr][1]); + check_float(machine->Outputs[attr][2]); + check_float(machine->Outputs[attr][3]); #endif - COPY_4V(store->results[attr].data[i], machine.Outputs[attr]); + COPY_4V(store->results[attr].data[i], machine->Outputs[attr]); } /* FOGC is a special case. Fragment shader expects (f,0,0,1) */ @@ -398,14 +400,14 @@ run_vp( struct gl_context *ctx, struct tnl_pipeline_stage *stage ) store->results[VERT_RESULT_FOGC].data[i][3] = 1.0; } #ifdef NAN_CHECK - ASSERT(machine.Outputs[0][3] != 0.0F); + ASSERT(machine->Outputs[0][3] != 0.0F); #endif #if 0 printf("HPOS: %f %f %f %f\n", - machine.Outputs[0][0], - machine.Outputs[0][1], - machine.Outputs[0][2], - machine.Outputs[0][3]); + machine->Outputs[0][0], + machine->Outputs[0][1], + machine->Outputs[0][2], + machine->Outputs[0][3]); #endif } @@ -501,7 +503,7 @@ init_vp(struct gl_context *ctx, struct tnl_pipeline_stage *stage) const GLuint size = VB->Size; GLuint i; - stage->privatePtr = MALLOC(sizeof(*store)); + stage->privatePtr = CALLOC(sizeof(*store)); store = VP_STAGE_DATA(stage); if (!store) return GL_FALSE; |