summaryrefslogtreecommitdiffstats
path: root/src/mesa/state_tracker
diff options
context:
space:
mode:
Diffstat (limited to 'src/mesa/state_tracker')
-rw-r--r--src/mesa/state_tracker/st_atom_framebuffer.c5
-rw-r--r--src/mesa/state_tracker/st_atom_shader.c2
-rw-r--r--src/mesa/state_tracker/st_cb_bitmap.c8
-rw-r--r--src/mesa/state_tracker/st_cb_blit.c4
-rw-r--r--src/mesa/state_tracker/st_cb_clear.c6
-rw-r--r--src/mesa/state_tracker/st_cb_drawpixels.c8
-rw-r--r--src/mesa/state_tracker/st_cb_drawtex.c6
-rw-r--r--src/mesa/state_tracker/st_cb_fbo.c2
-rw-r--r--src/mesa/state_tracker/st_cb_flush.c35
-rw-r--r--src/mesa/state_tracker/st_cb_flush.h3
-rw-r--r--src/mesa/state_tracker/st_cb_program.c8
-rw-r--r--src/mesa/state_tracker/st_context.c13
-rw-r--r--src/mesa/state_tracker/st_context.h3
-rw-r--r--src/mesa/state_tracker/st_draw.c2
-rw-r--r--src/mesa/state_tracker/st_glsl_to_tgsi.cpp865
-rw-r--r--src/mesa/state_tracker/st_glsl_to_tgsi.h2
-rw-r--r--src/mesa/state_tracker/st_manager.c7
-rw-r--r--src/mesa/state_tracker/st_mesa_to_tgsi.c9
-rw-r--r--src/mesa/state_tracker/st_program.c30
-rw-r--r--src/mesa/state_tracker/st_program.h1
20 files changed, 662 insertions, 357 deletions
diff --git a/src/mesa/state_tracker/st_atom_framebuffer.c b/src/mesa/state_tracker/st_atom_framebuffer.c
index b195c55b347..ae883a2535e 100644
--- a/src/mesa/state_tracker/st_atom_framebuffer.c
+++ b/src/mesa/state_tracker/st_atom_framebuffer.c
@@ -134,7 +134,10 @@ update_framebuffer_state( struct st_context *st )
else {
strb = st_renderbuffer(fb->Attachment[BUFFER_STENCIL].Renderbuffer);
if (strb) {
- assert(strb->surface);
+ if (strb->is_rtt) {
+ /* rendering to a GL texture, may have to update surface */
+ st_update_renderbuffer_surface(st, strb);
+ }
pipe_surface_reference(&framebuffer->zsbuf, strb->surface);
update_framebuffer_size(framebuffer, strb->surface);
}
diff --git a/src/mesa/state_tracker/st_atom_shader.c b/src/mesa/state_tracker/st_atom_shader.c
index 629f54f25de..ad8d2624fc9 100644
--- a/src/mesa/state_tracker/st_atom_shader.c
+++ b/src/mesa/state_tracker/st_atom_shader.c
@@ -189,7 +189,7 @@ update_gp( struct st_context *st )
}
stgp = st_geometry_program(st->ctx->GeometryProgram._Current);
- assert(stgp->Base.Base.Target == MESA_GEOMETRY_PROGRAM);
+ assert(stgp->Base.Base.Target == GL_GEOMETRY_PROGRAM_NV);
memset(&key, 0, sizeof(key));
key.st = st;
diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c
index 2107ab16739..c881e194f70 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/src/mesa/state_tracker/st_cb_bitmap.c
@@ -452,6 +452,8 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z,
cso_save_fragment_shader(cso);
cso_save_stream_outputs(cso);
cso_save_vertex_shader(cso);
+ cso_save_tessctrl_shader(cso);
+ cso_save_tesseval_shader(cso);
cso_save_geometry_shader(cso);
cso_save_vertex_elements(cso);
cso_save_aux_vertex_buffer_slot(cso);
@@ -466,7 +468,9 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z,
/* vertex shader state: position + texcoord pass-through */
cso_set_vertex_shader_handle(cso, st->bitmap.vs);
- /* geometry shader state: disabled */
+ /* disable other shaders */
+ cso_set_tessctrl_shader_handle(cso, NULL);
+ cso_set_tesseval_shader_handle(cso, NULL);
cso_set_geometry_shader_handle(cso, NULL);
/* user samplers, plus our bitmap sampler */
@@ -536,6 +540,8 @@ draw_bitmap_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z,
cso_restore_viewport(cso);
cso_restore_fragment_shader(cso);
cso_restore_vertex_shader(cso);
+ cso_restore_tessctrl_shader(cso);
+ cso_restore_tesseval_shader(cso);
cso_restore_geometry_shader(cso);
cso_restore_vertex_elements(cso);
cso_restore_aux_vertex_buffer_slot(cso);
diff --git a/src/mesa/state_tracker/st_cb_blit.c b/src/mesa/state_tracker/st_cb_blit.c
index bbaedd108f6..6d9371852c5 100644
--- a/src/mesa/state_tracker/st_cb_blit.c
+++ b/src/mesa/state_tracker/st_cb_blit.c
@@ -36,6 +36,7 @@
#include "st_context.h"
#include "st_texture.h"
+#include "st_cb_bitmap.h"
#include "st_cb_blit.h"
#include "st_cb_fbo.h"
#include "st_atom.h"
@@ -93,6 +94,9 @@ st_BlitFramebuffer(struct gl_context *ctx,
st_validate_state(st);
+ /* Make sure bitmap rendering has landed in the framebuffers */
+ st_flush_bitmap_cache(st);
+
clip.srcX0 = srcX0;
clip.srcY0 = srcY0;
clip.srcX1 = srcX1;
diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c
index f10e9063ac7..137fac8a9a9 100644
--- a/src/mesa/state_tracker/st_cb_clear.c
+++ b/src/mesa/state_tracker/st_cb_clear.c
@@ -265,6 +265,8 @@ clear_with_quad(struct gl_context *ctx, unsigned clear_buffers)
cso_save_fragment_shader(st->cso_context);
cso_save_stream_outputs(st->cso_context);
cso_save_vertex_shader(st->cso_context);
+ cso_save_tessctrl_shader(st->cso_context);
+ cso_save_tesseval_shader(st->cso_context);
cso_save_geometry_shader(st->cso_context);
cso_save_vertex_elements(st->cso_context);
cso_save_aux_vertex_buffer_slot(st->cso_context);
@@ -347,6 +349,8 @@ clear_with_quad(struct gl_context *ctx, unsigned clear_buffers)
}
set_fragment_shader(st);
+ cso_set_tessctrl_shader_handle(st->cso_context, NULL);
+ cso_set_tesseval_shader_handle(st->cso_context, NULL);
if (num_layers > 1)
set_vertex_shader_layered(st);
@@ -371,6 +375,8 @@ clear_with_quad(struct gl_context *ctx, unsigned clear_buffers)
cso_restore_viewport(st->cso_context);
cso_restore_fragment_shader(st->cso_context);
cso_restore_vertex_shader(st->cso_context);
+ cso_restore_tessctrl_shader(st->cso_context);
+ cso_restore_tesseval_shader(st->cso_context);
cso_restore_geometry_shader(st->cso_context);
cso_restore_vertex_elements(st->cso_context);
cso_restore_aux_vertex_buffer_slot(st->cso_context);
diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c
index 3edf31bad52..a6a98c83aa6 100644
--- a/src/mesa/state_tracker/st_cb_drawpixels.c
+++ b/src/mesa/state_tracker/st_cb_drawpixels.c
@@ -693,6 +693,8 @@ draw_textured_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z,
cso_save_fragment_shader(cso);
cso_save_stream_outputs(cso);
cso_save_vertex_shader(cso);
+ cso_save_tessctrl_shader(cso);
+ cso_save_tesseval_shader(cso);
cso_save_geometry_shader(cso);
cso_save_vertex_elements(cso);
cso_save_aux_vertex_buffer_slot(cso);
@@ -746,7 +748,9 @@ draw_textured_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z,
/* vertex shader state: position + texcoord pass-through */
cso_set_vertex_shader_handle(cso, driver_vp);
- /* geometry shader state: disabled */
+ /* disable other shaders */
+ cso_set_tessctrl_shader_handle(cso, NULL);
+ cso_set_tesseval_shader_handle(cso, NULL);
cso_set_geometry_shader_handle(cso, NULL);
/* texture sampling state: */
@@ -816,6 +820,8 @@ draw_textured_quad(struct gl_context *ctx, GLint x, GLint y, GLfloat z,
cso_restore_sampler_views(cso, PIPE_SHADER_FRAGMENT);
cso_restore_fragment_shader(cso);
cso_restore_vertex_shader(cso);
+ cso_restore_tessctrl_shader(cso);
+ cso_restore_tesseval_shader(cso);
cso_restore_geometry_shader(cso);
cso_restore_vertex_elements(cso);
cso_restore_aux_vertex_buffer_slot(cso);
diff --git a/src/mesa/state_tracker/st_cb_drawtex.c b/src/mesa/state_tracker/st_cb_drawtex.c
index 1420b96e55a..2af4f6d4cf6 100644
--- a/src/mesa/state_tracker/st_cb_drawtex.c
+++ b/src/mesa/state_tracker/st_cb_drawtex.c
@@ -229,6 +229,8 @@ st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
cso_save_viewport(cso);
cso_save_stream_outputs(cso);
cso_save_vertex_shader(cso);
+ cso_save_tessctrl_shader(cso);
+ cso_save_tesseval_shader(cso);
cso_save_geometry_shader(cso);
cso_save_vertex_elements(cso);
cso_save_aux_vertex_buffer_slot(cso);
@@ -238,6 +240,8 @@ st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
semantic_names, semantic_indexes);
cso_set_vertex_shader_handle(cso, vs);
}
+ cso_set_tessctrl_shader_handle(cso, NULL);
+ cso_set_tesseval_shader_handle(cso, NULL);
cso_set_geometry_shader_handle(cso, NULL);
for (i = 0; i < numAttribs; i++) {
@@ -279,6 +283,8 @@ st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
/* restore state */
cso_restore_viewport(cso);
cso_restore_vertex_shader(cso);
+ cso_restore_tessctrl_shader(cso);
+ cso_restore_tesseval_shader(cso);
cso_restore_geometry_shader(cso);
cso_restore_vertex_elements(cso);
cso_restore_aux_vertex_buffer_slot(cso);
diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c
index 296ea1e0d29..0399eef7204 100644
--- a/src/mesa/state_tracker/st_cb_fbo.c
+++ b/src/mesa/state_tracker/st_cb_fbo.c
@@ -842,7 +842,7 @@ void st_init_fbo_functions(struct dd_function_table *functions)
functions->NewFramebuffer = st_new_framebuffer;
functions->NewRenderbuffer = st_new_renderbuffer;
functions->BindFramebuffer = st_bind_framebuffer;
- functions->FramebufferRenderbuffer = _mesa_framebuffer_renderbuffer;
+ functions->FramebufferRenderbuffer = _mesa_FramebufferRenderbuffer_sw;
functions->RenderTexture = st_render_texture;
functions->FinishRenderTexture = st_finish_render_texture;
functions->ValidateFramebuffer = st_validate_framebuffer;
diff --git a/src/mesa/state_tracker/st_cb_flush.c b/src/mesa/state_tracker/st_cb_flush.c
index ca51eeee366..82affd2de3e 100644
--- a/src/mesa/state_tracker/st_cb_flush.c
+++ b/src/mesa/state_tracker/st_cb_flush.c
@@ -141,11 +141,44 @@ static void st_glFinish(struct gl_context *ctx)
}
-void st_init_flush_functions(struct dd_function_table *functions)
+/**
+ * Query information about GPU resets observed by this context
+ *
+ * Called via \c dd_function_table::GetGraphicsResetStatus.
+ */
+static GLenum
+st_get_graphics_reset_status(struct gl_context *ctx)
+{
+ struct st_context *st = st_context(ctx);
+ enum pipe_reset_status status;
+
+ status = st->pipe->get_device_reset_status(st->pipe);
+
+ switch (status) {
+ case PIPE_NO_RESET:
+ return GL_NO_ERROR;
+ case PIPE_GUILTY_CONTEXT_RESET:
+ return GL_GUILTY_CONTEXT_RESET_ARB;
+ case PIPE_INNOCENT_CONTEXT_RESET:
+ return GL_INNOCENT_CONTEXT_RESET_ARB;
+ case PIPE_UNKNOWN_CONTEXT_RESET:
+ return GL_UNKNOWN_CONTEXT_RESET_ARB;
+ default:
+ assert(0);
+ return GL_NO_ERROR;
+ }
+}
+
+
+void st_init_flush_functions(struct pipe_screen *screen,
+ struct dd_function_table *functions)
{
functions->Flush = st_glFlush;
functions->Finish = st_glFinish;
+ if (screen->get_param(screen, PIPE_CAP_DEVICE_RESET_STATUS_QUERY))
+ functions->GetGraphicsResetStatus = st_get_graphics_reset_status;
+
/* Windows opengl32.dll calls glFinish prior to every swapbuffers.
* This is unnecessary and degrades performance. Luckily we have some
* scope to work around this, as the externally-visible behaviour of
diff --git a/src/mesa/state_tracker/st_cb_flush.h b/src/mesa/state_tracker/st_cb_flush.h
index 84ffc63ae13..f92dcd56b64 100644
--- a/src/mesa/state_tracker/st_cb_flush.h
+++ b/src/mesa/state_tracker/st_cb_flush.h
@@ -37,7 +37,8 @@ struct pipe_fence_handle;
struct st_context;
extern void
-st_init_flush_functions(struct dd_function_table *functions);
+st_init_flush_functions(struct pipe_screen *screen,
+ struct dd_function_table *functions);
extern void
st_flush(struct st_context *st,
diff --git a/src/mesa/state_tracker/st_cb_program.c b/src/mesa/state_tracker/st_cb_program.c
index c382d7d2ca3..6aa7d5796d9 100644
--- a/src/mesa/state_tracker/st_cb_program.c
+++ b/src/mesa/state_tracker/st_cb_program.c
@@ -65,7 +65,7 @@ st_bind_program(struct gl_context *ctx, GLenum target, struct gl_program *prog)
case GL_FRAGMENT_PROGRAM_ARB:
st->dirty.st |= ST_NEW_FRAGMENT_PROGRAM;
break;
- case MESA_GEOMETRY_PROGRAM:
+ case GL_GEOMETRY_PROGRAM_NV:
st->dirty.st |= ST_NEW_GEOMETRY_PROGRAM;
break;
}
@@ -105,7 +105,7 @@ st_new_program(struct gl_context *ctx, GLenum target, GLuint id)
return _mesa_init_fragment_program(ctx, &prog->Base, target, id);
}
- case MESA_GEOMETRY_PROGRAM: {
+ case GL_GEOMETRY_PROGRAM_NV: {
struct st_geometry_program *prog = ST_CALLOC_STRUCT(st_geometry_program);
return _mesa_init_geometry_program(ctx, &prog->Base, target, id);
}
@@ -135,7 +135,7 @@ st_delete_program(struct gl_context *ctx, struct gl_program *prog)
free_glsl_to_tgsi_visitor(stvp->glsl_to_tgsi);
}
break;
- case MESA_GEOMETRY_PROGRAM:
+ case GL_GEOMETRY_PROGRAM_NV:
{
struct st_geometry_program *stgp =
(struct st_geometry_program *) prog;
@@ -198,7 +198,7 @@ st_program_string_notify( struct gl_context *ctx,
if (st->fp == stfp)
st->dirty.st |= ST_NEW_FRAGMENT_PROGRAM;
}
- else if (target == MESA_GEOMETRY_PROGRAM) {
+ else if (target == GL_GEOMETRY_PROGRAM_NV) {
struct st_geometry_program *stgp = (struct st_geometry_program *) prog;
st_release_gp_variants(st, stgp);
diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c
index bfb9c8406bd..ed9ed0f1b6c 100644
--- a/src/mesa/state_tracker/st_context.c
+++ b/src/mesa/state_tracker/st_context.c
@@ -321,7 +321,7 @@ struct st_context *st_create_context(gl_api api, struct pipe_context *pipe,
struct st_context *st;
memset(&funcs, 0, sizeof(funcs));
- st_init_driver_functions(&funcs);
+ st_init_driver_functions(pipe->screen, &funcs);
ctx = _mesa_create_context(api, visual, shareCtx, &funcs);
if (!ctx) {
@@ -376,12 +376,6 @@ void st_destroy_context( struct st_context *st )
}
pipe_surface_reference(&st->state.framebuffer.zsbuf, NULL);
- pipe->set_index_buffer(pipe, NULL);
-
- for (i = 0; i < PIPE_SHADER_TYPES; i++) {
- pipe->set_constant_buffer(pipe, i, 0, NULL);
- }
-
_mesa_delete_program_cache(st->ctx, st->pixel_xfer.cache);
_vbo_DestroyContext(st->ctx);
@@ -401,7 +395,8 @@ void st_destroy_context( struct st_context *st )
}
-void st_init_driver_functions(struct dd_function_table *functions)
+void st_init_driver_functions(struct pipe_screen *screen,
+ struct dd_function_table *functions)
{
_mesa_init_shader_object_functions(functions);
_mesa_init_sampler_object_functions(functions);
@@ -429,7 +424,7 @@ void st_init_driver_functions(struct dd_function_table *functions)
st_init_readpixels_functions(functions);
st_init_texture_functions(functions);
st_init_texture_barrier_functions(functions);
- st_init_flush_functions(functions);
+ st_init_flush_functions(screen, functions);
st_init_string_functions(functions);
st_init_viewport_functions(functions);
diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h
index 8a9504bb7c1..dac5a4b9006 100644
--- a/src/mesa/state_tracker/st_context.h
+++ b/src/mesa/state_tracker/st_context.h
@@ -237,7 +237,8 @@ struct st_framebuffer
};
-extern void st_init_driver_functions(struct dd_function_table *functions);
+extern void st_init_driver_functions(struct pipe_screen *screen,
+ struct dd_function_table *functions);
void st_invalidate_state(struct gl_context * ctx, GLuint new_state);
diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c
index 488f6ead201..8b43582c14b 100644
--- a/src/mesa/state_tracker/st_draw.c
+++ b/src/mesa/state_tracker/st_draw.c
@@ -141,7 +141,7 @@ check_uniforms(struct gl_context *ctx)
if (shProg[j] == NULL || !shProg[j]->LinkStatus)
continue;
- for (i = 0; i < shProg[j]->NumUserUniformStorage; i++) {
+ for (i = 0; i < shProg[j]->NumUniformStorage; i++) {
const struct gl_uniform_storage *u = &shProg[j]->UniformStorage[i];
if (!u->initialized) {
_mesa_warning(ctx,
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp b/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
index 1fea8600a75..25e30c7deb2 100644
--- a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -57,11 +57,6 @@
(1 << PROGRAM_CONSTANT) | \
(1 << PROGRAM_UNIFORM))
-/**
- * Maximum number of arrays
- */
-#define MAX_ARRAYS 256
-
#define MAX_GLSL_TEXTURE_OFFSET 4
class st_src_reg;
@@ -89,6 +84,7 @@ public:
this->reladdr2 = NULL;
this->has_index2 = false;
this->double_reg2 = false;
+ this->array_id = 0;
}
st_src_reg(gl_register_file file, int index, int type)
@@ -103,6 +99,7 @@ public:
this->reladdr2 = NULL;
this->has_index2 = false;
this->double_reg2 = false;
+ this->array_id = 0;
}
st_src_reg(gl_register_file file, int index, int type, int index2D)
@@ -117,6 +114,7 @@ public:
this->reladdr2 = NULL;
this->has_index2 = false;
this->double_reg2 = false;
+ this->array_id = 0;
}
st_src_reg()
@@ -131,6 +129,7 @@ public:
this->reladdr2 = NULL;
this->has_index2 = false;
this->double_reg2 = false;
+ this->array_id = 0;
}
explicit st_src_reg(st_dst_reg reg);
@@ -150,6 +149,7 @@ public:
* currently used for input mapping only.
*/
bool double_reg2;
+ unsigned array_id;
};
class st_dst_reg {
@@ -162,6 +162,7 @@ public:
this->cond_mask = COND_TR;
this->reladdr = NULL;
this->type = type;
+ this->array_id = 0;
}
st_dst_reg(gl_register_file file, int writemask, int type)
@@ -172,6 +173,7 @@ public:
this->cond_mask = COND_TR;
this->reladdr = NULL;
this->type = type;
+ this->array_id = 0;
}
st_dst_reg()
@@ -182,6 +184,7 @@ public:
this->writemask = 0;
this->cond_mask = COND_TR;
this->reladdr = NULL;
+ this->array_id = 0;
}
explicit st_dst_reg(st_src_reg reg);
@@ -193,6 +196,7 @@ public:
int type; /** GLSL_TYPE_* from GLSL IR (enum glsl_base_type) */
/** Register index should be offset by the integer in this reg. */
st_src_reg *reladdr;
+ unsigned array_id;
};
st_src_reg::st_src_reg(st_dst_reg reg)
@@ -207,6 +211,7 @@ st_src_reg::st_src_reg(st_dst_reg reg)
this->reladdr2 = NULL;
this->has_index2 = false;
this->double_reg2 = false;
+ this->array_id = reg.array_id;
}
st_dst_reg::st_dst_reg(st_src_reg reg)
@@ -217,6 +222,7 @@ st_dst_reg::st_dst_reg(st_src_reg reg)
this->writemask = WRITEMASK_XYZW;
this->cond_mask = COND_TR;
this->reladdr = reg.reladdr;
+ this->array_id = reg.array_id;
}
class glsl_to_tgsi_instruction : public exec_node {
@@ -233,6 +239,7 @@ public:
st_src_reg sampler; /**< sampler register */
int sampler_array_size; /**< 1-based size of sampler array, 1 if not array */
int tex_target; /**< One of TEXTURE_*_INDEX */
+ glsl_base_type tex_type;
GLboolean tex_shadow;
st_src_reg tex_offsets[MAX_GLSL_TEXTURE_OFFSET];
@@ -244,8 +251,9 @@ public:
class variable_storage : public exec_node {
public:
- variable_storage(ir_variable *var, gl_register_file file, int index)
- : file(file), index(index), var(var)
+ variable_storage(ir_variable *var, gl_register_file file, int index,
+ unsigned array_id = 0)
+ : file(file), index(index), var(var), array_id(array_id)
{
/* empty */
}
@@ -253,6 +261,7 @@ public:
gl_register_file file;
int index;
ir_variable *var; /* variable that maps to this, if any */
+ unsigned array_id;
};
class immediate_storage : public exec_node {
@@ -302,6 +311,15 @@ public:
st_src_reg return_reg;
};
+static st_src_reg undef_src = st_src_reg(PROGRAM_UNDEFINED, 0, GLSL_TYPE_ERROR);
+static st_dst_reg undef_dst = st_dst_reg(PROGRAM_UNDEFINED, SWIZZLE_NOOP, GLSL_TYPE_ERROR);
+
+struct array_decl {
+ unsigned mesa_index;
+ unsigned array_id;
+ unsigned array_size;
+};
+
struct glsl_to_tgsi_visitor : public ir_visitor {
public:
glsl_to_tgsi_visitor();
@@ -317,11 +335,19 @@ public:
int next_temp;
- unsigned array_sizes[MAX_ARRAYS];
+ unsigned *array_sizes;
+ unsigned max_num_arrays;
unsigned next_array;
+ struct array_decl input_arrays[PIPE_MAX_SHADER_INPUTS];
+ unsigned num_input_arrays;
+ struct array_decl output_arrays[PIPE_MAX_SHADER_OUTPUTS];
+ unsigned num_output_arrays;
+
int num_address_regs;
int samplers_used;
+ glsl_base_type sampler_types[PIPE_MAX_SAMPLERS];
+ int sampler_targets[PIPE_MAX_SAMPLERS]; /**< One of TGSI_TEXTURE_* */
bool indirect_addr_consts;
int wpos_transform_const;
@@ -372,6 +398,7 @@ public:
virtual void visit(ir_if *);
virtual void visit(ir_emit_vertex *);
virtual void visit(ir_end_primitive *);
+ virtual void visit(ir_barrier *);
/*@}*/
st_src_reg result;
@@ -390,31 +417,19 @@ public:
/** List of glsl_to_tgsi_instruction */
exec_list instructions;
- glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op);
-
- glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_src_reg src0);
-
- glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_dst_reg dst1,
- st_src_reg src0);
+ glsl_to_tgsi_instruction *emit_asm(ir_instruction *ir, unsigned op,
+ st_dst_reg dst = undef_dst,
+ st_src_reg src0 = undef_src,
+ st_src_reg src1 = undef_src,
+ st_src_reg src2 = undef_src,
+ st_src_reg src3 = undef_src);
- glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_src_reg src0, st_src_reg src1);
-
- glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst,
- st_src_reg src0, st_src_reg src1, st_src_reg src2);
-
- glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst,
- st_src_reg src0, st_src_reg src1,
- st_src_reg src2, st_src_reg src3);
-
- glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_dst_reg dst1,
- st_src_reg src0, st_src_reg src1,
- st_src_reg src2, st_src_reg src3);
+ glsl_to_tgsi_instruction *emit_asm(ir_instruction *ir, unsigned op,
+ st_dst_reg dst, st_dst_reg dst1,
+ st_src_reg src0 = undef_src,
+ st_src_reg src1 = undef_src,
+ st_src_reg src2 = undef_src,
+ st_src_reg src3 = undef_src);
unsigned get_opcode(ir_instruction *ir, unsigned op,
st_dst_reg dst,
@@ -468,10 +483,6 @@ public:
void *mem_ctx;
};
-static st_src_reg undef_src = st_src_reg(PROGRAM_UNDEFINED, 0, GLSL_TYPE_ERROR);
-
-static st_dst_reg undef_dst = st_dst_reg(PROGRAM_UNDEFINED, SWIZZLE_NOOP, GLSL_TYPE_ERROR);
-
static st_dst_reg address_reg = st_dst_reg(PROGRAM_ADDRESS, WRITEMASK_X, GLSL_TYPE_FLOAT, 0);
static st_dst_reg address_reg2 = st_dst_reg(PROGRAM_ADDRESS, WRITEMASK_X, GLSL_TYPE_FLOAT, 1);
static st_dst_reg sampler_reladdr = st_dst_reg(PROGRAM_ADDRESS, WRITEMASK_X, GLSL_TYPE_FLOAT, 2);
@@ -526,10 +537,10 @@ num_inst_src_regs(unsigned opcode)
}
glsl_to_tgsi_instruction *
-glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_dst_reg dst1,
- st_src_reg src0, st_src_reg src1,
- st_src_reg src2, st_src_reg src3)
+glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
+ st_dst_reg dst, st_dst_reg dst1,
+ st_src_reg src0, st_src_reg src1,
+ st_src_reg src2, st_src_reg src3)
{
glsl_to_tgsi_instruction *inst = new(mem_ctx) glsl_to_tgsi_instruction();
int num_reladdr = 0, i, j;
@@ -571,6 +582,10 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
inst->src[3] = src3;
inst->ir = ir;
inst->dead_mask = 0;
+ /* default to float, for paths where this is not initialized
+ * (since 0==UINT which is likely wrong):
+ */
+ inst->tex_type = GLSL_TYPE_FLOAT;
inst->function = NULL;
@@ -716,48 +731,12 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
}
glsl_to_tgsi_instruction *
-glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst,
- st_src_reg src0, st_src_reg src1,
- st_src_reg src2, st_src_reg src3)
+glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
+ st_dst_reg dst,
+ st_src_reg src0, st_src_reg src1,
+ st_src_reg src2, st_src_reg src3)
{
- return emit(ir, op, dst, undef_dst, src0, src1, src2, src3);
-}
-
-glsl_to_tgsi_instruction *
-glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_src_reg src0,
- st_src_reg src1, st_src_reg src2)
-{
- return emit(ir, op, dst, undef_dst, src0, src1, src2, undef_src);
-}
-
-glsl_to_tgsi_instruction *
-glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_src_reg src0, st_src_reg src1)
-{
- return emit(ir, op, dst, undef_dst, src0, src1, undef_src, undef_src);
-}
-
-glsl_to_tgsi_instruction *
-glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_src_reg src0)
-{
- assert(dst.writemask != 0);
- return emit(ir, op, dst, undef_dst, src0, undef_src, undef_src, undef_src);
-}
-
-glsl_to_tgsi_instruction *
-glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
- st_dst_reg dst, st_dst_reg dst1, st_src_reg src0)
-{
- return emit(ir, op, dst, dst1, src0, undef_src, undef_src, undef_src);
-}
-
-glsl_to_tgsi_instruction *
-glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op)
-{
- return emit(ir, op, undef_dst, undef_dst, undef_src, undef_src, undef_src, undef_src);
+ return emit_asm(ir, op, dst, undef_dst, src0, src1, src2, src3);
}
/**
@@ -879,7 +858,7 @@ glsl_to_tgsi_visitor::emit_dp(ir_instruction *ir,
TGSI_OPCODE_DP2, TGSI_OPCODE_DP3, TGSI_OPCODE_DP4
};
- return emit(ir, dot_opcodes[elements - 2], dst, src0, src1);
+ return emit_asm(ir, dot_opcodes[elements - 2], dst, src0, src1);
}
/**
@@ -929,7 +908,7 @@ glsl_to_tgsi_visitor::emit_scalar(ir_instruction *ir, unsigned op,
src1_swiz, src1_swiz);
dst.writemask = this_mask;
- emit(ir, op, dst, src0, src1);
+ emit_asm(ir, op, dst, src0, src1);
done_mask |= this_mask;
}
}
@@ -958,7 +937,7 @@ glsl_to_tgsi_visitor::emit_arl(ir_instruction *ir,
if (dst.index >= this->num_address_regs)
this->num_address_regs = dst.index + 1;
- emit(NULL, op, dst, src0);
+ emit_asm(NULL, op, dst, src0);
}
int
@@ -1142,6 +1121,12 @@ glsl_to_tgsi_visitor::get_temp(const glsl_type *type)
if (!options->EmitNoIndirectTemp &&
(type->is_array() || type->is_matrix())) {
+ if (next_array >= max_num_arrays) {
+ max_num_arrays += 32;
+ array_sizes = (unsigned*)
+ realloc(array_sizes, sizeof(array_sizes[0]) * max_num_arrays);
+ }
+
src.file = PROGRAM_ARRAY;
src.index = next_array << 16 | 0x8000;
array_sizes[next_array] = type_size(type);
@@ -1242,7 +1227,7 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
*/
st_src_reg src(PROGRAM_STATE_VAR, index, GLSL_TYPE_FLOAT);
src.swizzle = slots[i].swizzle;
- emit(ir, TGSI_OPCODE_MOV, dst, src);
+ emit_asm(ir, TGSI_OPCODE_MOV, dst, src);
/* even a float takes up a whole vec4 reg in a struct/array. */
dst.index++;
}
@@ -1261,11 +1246,11 @@ glsl_to_tgsi_visitor::visit(ir_variable *ir)
void
glsl_to_tgsi_visitor::visit(ir_loop *ir)
{
- emit(NULL, TGSI_OPCODE_BGNLOOP);
+ emit_asm(NULL, TGSI_OPCODE_BGNLOOP);
visit_exec_list(&ir->body_instructions, this);
- emit(NULL, TGSI_OPCODE_ENDLOOP);
+ emit_asm(NULL, TGSI_OPCODE_ENDLOOP);
}
void
@@ -1273,10 +1258,10 @@ glsl_to_tgsi_visitor::visit(ir_loop_jump *ir)
{
switch (ir->mode) {
case ir_loop_jump::jump_break:
- emit(NULL, TGSI_OPCODE_BRK);
+ emit_asm(NULL, TGSI_OPCODE_BRK);
break;
case ir_loop_jump::jump_continue:
- emit(NULL, TGSI_OPCODE_CONT);
+ emit_asm(NULL, TGSI_OPCODE_CONT);
break;
}
}
@@ -1330,7 +1315,7 @@ glsl_to_tgsi_visitor::try_emit_mad(ir_expression *ir, int mul_operand)
this->result = get_temp(ir->type);
result_dst = st_dst_reg(this->result);
result_dst.writemask = (1 << ir->type->vector_elements) - 1;
- emit(ir, TGSI_OPCODE_MAD, result_dst, a, b, c);
+ emit_asm(ir, TGSI_OPCODE_MAD, result_dst, a, b, c);
return true;
}
@@ -1370,7 +1355,7 @@ glsl_to_tgsi_visitor::try_emit_mad_for_and_not(ir_expression *ir, int try_operan
b.negate = ~b.negate;
this->result = get_temp(ir->type);
- emit(ir, TGSI_OPCODE_MAD, st_dst_reg(this->result), a, b, a);
+ emit_asm(ir, TGSI_OPCODE_MAD, st_dst_reg(this->result), a, b, a);
return true;
}
@@ -1388,7 +1373,7 @@ glsl_to_tgsi_visitor::reladdr_to_temp(ir_instruction *ir,
if (*num_reladdr != 1) {
st_src_reg temp = get_temp(glsl_type::vec4_type);
- emit(ir, TGSI_OPCODE_MOV, st_dst_reg(temp), *reg);
+ emit_asm(ir, TGSI_OPCODE_MOV, st_dst_reg(temp), *reg);
*reg = temp;
}
@@ -1464,7 +1449,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
switch (ir->operation) {
case ir_unop_logic_not:
if (result_dst.type != GLSL_TYPE_FLOAT)
- emit(ir, TGSI_OPCODE_NOT, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_NOT, result_dst, op[0]);
else {
/* Previously 'SEQ dst, src, 0.0' was used for this. However, many
* older GPUs implement SEQ using multiple instructions (i915 uses two
@@ -1472,24 +1457,24 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
* 0.0 and 1.0, 1-x also implements !x.
*/
op[0].negate = ~op[0].negate;
- emit(ir, TGSI_OPCODE_ADD, result_dst, op[0], st_src_reg_for_float(1.0));
+ emit_asm(ir, TGSI_OPCODE_ADD, result_dst, op[0], st_src_reg_for_float(1.0));
}
break;
case ir_unop_neg:
if (result_dst.type == GLSL_TYPE_INT || result_dst.type == GLSL_TYPE_UINT)
- emit(ir, TGSI_OPCODE_INEG, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_INEG, result_dst, op[0]);
else if (result_dst.type == GLSL_TYPE_DOUBLE)
- emit(ir, TGSI_OPCODE_DNEG, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_DNEG, result_dst, op[0]);
else {
op[0].negate = ~op[0].negate;
result_src = op[0];
}
break;
case ir_unop_abs:
- emit(ir, TGSI_OPCODE_ABS, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_ABS, result_dst, op[0]);
break;
case ir_unop_sign:
- emit(ir, TGSI_OPCODE_SSG, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_SSG, result_dst, op[0]);
break;
case ir_unop_rcp:
emit_scalar(ir, TGSI_OPCODE_RCP, result_dst, op[0]);
@@ -1513,17 +1498,17 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
break;
case ir_unop_saturate: {
glsl_to_tgsi_instruction *inst;
- inst = emit(ir, TGSI_OPCODE_MOV, result_dst, op[0]);
+ inst = emit_asm(ir, TGSI_OPCODE_MOV, result_dst, op[0]);
inst->saturate = true;
break;
}
case ir_unop_dFdx:
case ir_unop_dFdx_coarse:
- emit(ir, TGSI_OPCODE_DDX, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_DDX, result_dst, op[0]);
break;
case ir_unop_dFdx_fine:
- emit(ir, TGSI_OPCODE_DDX_FINE, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_DDX_FINE, result_dst, op[0]);
break;
case ir_unop_dFdy:
case ir_unop_dFdy_coarse:
@@ -1547,18 +1532,18 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
st_src_reg temp = get_temp(glsl_type::vec4_type);
- emit(ir, TGSI_OPCODE_MUL, st_dst_reg(temp), transform_y, op[0]);
- emit(ir, ir->operation == ir_unop_dFdy_fine ?
+ emit_asm(ir, TGSI_OPCODE_MUL, st_dst_reg(temp), transform_y, op[0]);
+ emit_asm(ir, ir->operation == ir_unop_dFdy_fine ?
TGSI_OPCODE_DDY_FINE : TGSI_OPCODE_DDY, result_dst, temp);
break;
}
case ir_unop_frexp_sig:
- emit(ir, TGSI_OPCODE_DFRACEXP, result_dst, undef_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_DFRACEXP, result_dst, undef_dst, op[0]);
break;
case ir_unop_frexp_exp:
- emit(ir, TGSI_OPCODE_DFRACEXP, undef_dst, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_DFRACEXP, undef_dst, result_dst, op[0]);
break;
case ir_unop_noise: {
@@ -1568,50 +1553,50 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
* place to do this is in the GL state tracker, not the poor
* driver.
*/
- emit(ir, TGSI_OPCODE_MOV, result_dst, st_src_reg_for_float(0.5));
+ emit_asm(ir, TGSI_OPCODE_MOV, result_dst, st_src_reg_for_float(0.5));
break;
}
case ir_binop_add:
- emit(ir, TGSI_OPCODE_ADD, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_ADD, result_dst, op[0], op[1]);
break;
case ir_binop_sub:
- emit(ir, TGSI_OPCODE_SUB, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SUB, result_dst, op[0], op[1]);
break;
case ir_binop_mul:
- emit(ir, TGSI_OPCODE_MUL, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_MUL, result_dst, op[0], op[1]);
break;
case ir_binop_div:
if (result_dst.type == GLSL_TYPE_FLOAT || result_dst.type == GLSL_TYPE_DOUBLE)
assert(!"not reached: should be handled by ir_div_to_mul_rcp");
else
- emit(ir, TGSI_OPCODE_DIV, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_DIV, result_dst, op[0], op[1]);
break;
case ir_binop_mod:
if (result_dst.type == GLSL_TYPE_FLOAT)
assert(!"ir_binop_mod should have been converted to b * fract(a/b)");
else
- emit(ir, TGSI_OPCODE_MOD, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_MOD, result_dst, op[0], op[1]);
break;
case ir_binop_less:
- emit(ir, TGSI_OPCODE_SLT, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SLT, result_dst, op[0], op[1]);
break;
case ir_binop_greater:
- emit(ir, TGSI_OPCODE_SLT, result_dst, op[1], op[0]);
+ emit_asm(ir, TGSI_OPCODE_SLT, result_dst, op[1], op[0]);
break;
case ir_binop_lequal:
- emit(ir, TGSI_OPCODE_SGE, result_dst, op[1], op[0]);
+ emit_asm(ir, TGSI_OPCODE_SGE, result_dst, op[1], op[0]);
break;
case ir_binop_gequal:
- emit(ir, TGSI_OPCODE_SGE, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SGE, result_dst, op[0], op[1]);
break;
case ir_binop_equal:
- emit(ir, TGSI_OPCODE_SEQ, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SEQ, result_dst, op[0], op[1]);
break;
case ir_binop_nequal:
- emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
break;
case ir_binop_all_equal:
/* "==" operator producing a scalar boolean. */
@@ -1625,7 +1610,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
st_dst_reg temp_dst = st_dst_reg(temp);
st_src_reg temp1 = st_src_reg(temp), temp2 = st_src_reg(temp);
- emit(ir, TGSI_OPCODE_SEQ, st_dst_reg(temp), op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SEQ, st_dst_reg(temp), op[0], op[1]);
/* Emit 1-3 AND operations to combine the SEQ results. */
switch (ir->operands[0]->type->vector_elements) {
@@ -1635,24 +1620,24 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
temp_dst.writemask = WRITEMASK_Y;
temp1.swizzle = SWIZZLE_YYYY;
temp2.swizzle = SWIZZLE_ZZZZ;
- emit(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
+ emit_asm(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
break;
case 4:
temp_dst.writemask = WRITEMASK_X;
temp1.swizzle = SWIZZLE_XXXX;
temp2.swizzle = SWIZZLE_YYYY;
- emit(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
+ emit_asm(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
temp_dst.writemask = WRITEMASK_Y;
temp1.swizzle = SWIZZLE_ZZZZ;
temp2.swizzle = SWIZZLE_WWWW;
- emit(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
+ emit_asm(ir, TGSI_OPCODE_AND, temp_dst, temp1, temp2);
}
temp1.swizzle = SWIZZLE_XXXX;
temp2.swizzle = SWIZZLE_YYYY;
- emit(ir, TGSI_OPCODE_AND, result_dst, temp1, temp2);
+ emit_asm(ir, TGSI_OPCODE_AND, result_dst, temp1, temp2);
} else {
- emit(ir, TGSI_OPCODE_SNE, st_dst_reg(temp), op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SNE, st_dst_reg(temp), op[0], op[1]);
/* After the dot-product, the value will be an integer on the
* range [0,4]. Zero becomes 1.0, and positive values become zero.
@@ -1665,10 +1650,10 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
*/
st_src_reg sge_src = result_src;
sge_src.negate = ~sge_src.negate;
- emit(ir, TGSI_OPCODE_SGE, result_dst, sge_src, st_src_reg_for_float(0.0));
+ emit_asm(ir, TGSI_OPCODE_SGE, result_dst, sge_src, st_src_reg_for_float(0.0));
}
} else {
- emit(ir, TGSI_OPCODE_SEQ, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SEQ, result_dst, op[0], op[1]);
}
break;
case ir_binop_any_nequal:
@@ -1678,7 +1663,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
st_src_reg temp = get_temp(native_integers ?
glsl_type::uvec4_type :
glsl_type::vec4_type);
- emit(ir, TGSI_OPCODE_SNE, st_dst_reg(temp), op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SNE, st_dst_reg(temp), op[0], op[1]);
if (native_integers) {
st_dst_reg temp_dst = st_dst_reg(temp);
@@ -1692,22 +1677,22 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
temp_dst.writemask = WRITEMASK_Y;
temp1.swizzle = SWIZZLE_YYYY;
temp2.swizzle = SWIZZLE_ZZZZ;
- emit(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
+ emit_asm(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
break;
case 4:
temp_dst.writemask = WRITEMASK_X;
temp1.swizzle = SWIZZLE_XXXX;
temp2.swizzle = SWIZZLE_YYYY;
- emit(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
+ emit_asm(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
temp_dst.writemask = WRITEMASK_Y;
temp1.swizzle = SWIZZLE_ZZZZ;
temp2.swizzle = SWIZZLE_WWWW;
- emit(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
+ emit_asm(ir, TGSI_OPCODE_OR, temp_dst, temp1, temp2);
}
temp1.swizzle = SWIZZLE_XXXX;
temp2.swizzle = SWIZZLE_YYYY;
- emit(ir, TGSI_OPCODE_OR, result_dst, temp1, temp2);
+ emit_asm(ir, TGSI_OPCODE_OR, result_dst, temp1, temp2);
} else {
/* After the dot-product, the value will be an integer on the
* range [0,4]. Zero stays zero, and positive values become 1.0.
@@ -1726,11 +1711,11 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
*/
st_src_reg slt_src = result_src;
slt_src.negate = ~slt_src.negate;
- emit(ir, TGSI_OPCODE_SLT, result_dst, slt_src, st_src_reg_for_float(0.0));
+ emit_asm(ir, TGSI_OPCODE_SLT, result_dst, slt_src, st_src_reg_for_float(0.0));
}
}
} else {
- emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
}
break;
@@ -1763,7 +1748,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
GET_SWZ(op0_swizzle, 3),
GET_SWZ(op0_swizzle, 3),
GET_SWZ(op0_swizzle, 3));
- emit(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
+ emit_asm(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
accum = st_src_reg(result_dst);
accum.swizzle = dst_swizzle;
/* fallthrough */
@@ -1772,7 +1757,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
GET_SWZ(op0_swizzle, 2),
GET_SWZ(op0_swizzle, 2),
GET_SWZ(op0_swizzle, 2));
- emit(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
+ emit_asm(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
accum = st_src_reg(result_dst);
accum.swizzle = dst_swizzle;
/* fallthrough */
@@ -1781,7 +1766,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
GET_SWZ(op0_swizzle, 1),
GET_SWZ(op0_swizzle, 1),
GET_SWZ(op0_swizzle, 1));
- emit(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
+ emit_asm(ir, TGSI_OPCODE_OR, result_dst, accum, op[0]);
break;
default:
assert(!"Unexpected vector size");
@@ -1807,11 +1792,11 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
*/
st_src_reg slt_src = result_src;
slt_src.negate = ~slt_src.negate;
- emit(ir, TGSI_OPCODE_SLT, result_dst, slt_src, st_src_reg_for_float(0.0));
+ emit_asm(ir, TGSI_OPCODE_SLT, result_dst, slt_src, st_src_reg_for_float(0.0));
}
else {
/* Use SNE 0 if integers are being used as boolean values. */
- emit(ir, TGSI_OPCODE_SNE, result_dst, result_src, st_src_reg_for_int(0));
+ emit_asm(ir, TGSI_OPCODE_SNE, result_dst, result_src, st_src_reg_for_int(0));
}
}
break;
@@ -1819,9 +1804,9 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
case ir_binop_logic_xor:
if (native_integers)
- emit(ir, TGSI_OPCODE_XOR, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_XOR, result_dst, op[0], op[1]);
else
- emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SNE, result_dst, op[0], op[1]);
break;
case ir_binop_logic_or: {
@@ -1830,13 +1815,13 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
* instruction.
*/
assert(native_integers);
- emit(ir, TGSI_OPCODE_OR, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_OR, result_dst, op[0], op[1]);
} else {
/* After the addition, the value will be an integer on the
* range [0,2]. Zero stays zero, and positive values become 1.0.
*/
glsl_to_tgsi_instruction *add =
- emit(ir, TGSI_OPCODE_ADD, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_ADD, result_dst, op[0], op[1]);
if (this->prog->Target == GL_FRAGMENT_PROGRAM_ARB) {
/* The clamping to [0,1] can be done for free in the fragment
* shader with a saturate if floats are being used as boolean values.
@@ -1849,7 +1834,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
*/
st_src_reg slt_src = result_src;
slt_src.negate = ~slt_src.negate;
- emit(ir, TGSI_OPCODE_SLT, result_dst, slt_src, st_src_reg_for_float(0.0));
+ emit_asm(ir, TGSI_OPCODE_SLT, result_dst, slt_src, st_src_reg_for_float(0.0));
}
}
break;
@@ -1861,9 +1846,9 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
* actual AND opcode.
*/
if (native_integers)
- emit(ir, TGSI_OPCODE_AND, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_AND, result_dst, op[0], op[1]);
else
- emit(ir, TGSI_OPCODE_MUL, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_MUL, result_dst, op[0], op[1]);
break;
case ir_binop_dot:
@@ -1879,10 +1864,10 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
} else {
/* sqrt(x) = x * rsq(x). */
emit_scalar(ir, TGSI_OPCODE_RSQ, result_dst, op[0]);
- emit(ir, TGSI_OPCODE_MUL, result_dst, result_src, op[0]);
+ emit_asm(ir, TGSI_OPCODE_MUL, result_dst, result_src, op[0]);
/* For incoming channels <= 0, set the result to 0. */
op[0].negate = ~op[0].negate;
- emit(ir, TGSI_OPCODE_CMP, result_dst,
+ emit_asm(ir, TGSI_OPCODE_CMP, result_dst,
op[0], result_src, st_src_reg_for_float(0.0));
}
break;
@@ -1891,13 +1876,13 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
break;
case ir_unop_i2f:
if (native_integers) {
- emit(ir, TGSI_OPCODE_I2F, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_I2F, result_dst, op[0]);
break;
}
/* fallthrough to next case otherwise */
case ir_unop_b2f:
if (native_integers) {
- emit(ir, TGSI_OPCODE_AND, result_dst, op[0], st_src_reg_for_float(1.0));
+ emit_asm(ir, TGSI_OPCODE_AND, result_dst, op[0], st_src_reg_for_float(1.0));
break;
}
/* fallthrough to next case otherwise */
@@ -1912,7 +1897,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
* GLSL requires that int(bool) return 1 for true and 0 for false.
* This conversion is done with AND, but it could be done with NEG.
*/
- emit(ir, TGSI_OPCODE_AND, result_dst, op[0], st_src_reg_for_int(1));
+ emit_asm(ir, TGSI_OPCODE_AND, result_dst, op[0], st_src_reg_for_int(1));
} else {
/* Booleans and integers are both stored as floats when native
* integers are disabled.
@@ -1922,15 +1907,15 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
break;
case ir_unop_f2i:
if (native_integers)
- emit(ir, TGSI_OPCODE_F2I, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_F2I, result_dst, op[0]);
else
- emit(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
break;
case ir_unop_f2u:
if (native_integers)
- emit(ir, TGSI_OPCODE_F2U, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_F2U, result_dst, op[0]);
else
- emit(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
break;
case ir_unop_bitcast_f2i:
result_src = op[0];
@@ -1946,38 +1931,38 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
result_src.type = GLSL_TYPE_FLOAT;
break;
case ir_unop_f2b:
- emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_float(0.0));
+ emit_asm(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_float(0.0));
break;
case ir_unop_d2b:
- emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_double(0.0));
+ emit_asm(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_double(0.0));
break;
case ir_unop_i2b:
if (native_integers)
- emit(ir, TGSI_OPCODE_INEG, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_USNE, result_dst, op[0], st_src_reg_for_int(0));
else
- emit(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_float(0.0));
+ emit_asm(ir, TGSI_OPCODE_SNE, result_dst, op[0], st_src_reg_for_float(0.0));
break;
case ir_unop_trunc:
- emit(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_TRUNC, result_dst, op[0]);
break;
case ir_unop_ceil:
- emit(ir, TGSI_OPCODE_CEIL, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_CEIL, result_dst, op[0]);
break;
case ir_unop_floor:
- emit(ir, TGSI_OPCODE_FLR, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_FLR, result_dst, op[0]);
break;
case ir_unop_round_even:
- emit(ir, TGSI_OPCODE_ROUND, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_ROUND, result_dst, op[0]);
break;
case ir_unop_fract:
- emit(ir, TGSI_OPCODE_FRC, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_FRC, result_dst, op[0]);
break;
case ir_binop_min:
- emit(ir, TGSI_OPCODE_MIN, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_MIN, result_dst, op[0], op[1]);
break;
case ir_binop_max:
- emit(ir, TGSI_OPCODE_MAX, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_MAX, result_dst, op[0], op[1]);
break;
case ir_binop_pow:
emit_scalar(ir, TGSI_OPCODE_POW, result_dst, op[0], op[1]);
@@ -1985,37 +1970,37 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
case ir_unop_bit_not:
if (native_integers) {
- emit(ir, TGSI_OPCODE_NOT, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_NOT, result_dst, op[0]);
break;
}
case ir_unop_u2f:
if (native_integers) {
- emit(ir, TGSI_OPCODE_U2F, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_U2F, result_dst, op[0]);
break;
}
case ir_binop_lshift:
if (native_integers) {
- emit(ir, TGSI_OPCODE_SHL, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_SHL, result_dst, op[0], op[1]);
break;
}
case ir_binop_rshift:
if (native_integers) {
- emit(ir, TGSI_OPCODE_ISHR, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_ISHR, result_dst, op[0], op[1]);
break;
}
case ir_binop_bit_and:
if (native_integers) {
- emit(ir, TGSI_OPCODE_AND, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_AND, result_dst, op[0], op[1]);
break;
}
case ir_binop_bit_xor:
if (native_integers) {
- emit(ir, TGSI_OPCODE_XOR, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_XOR, result_dst, op[0], op[1]);
break;
}
case ir_binop_bit_or:
if (native_integers) {
- emit(ir, TGSI_OPCODE_OR, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_OR, result_dst, op[0], op[1]);
break;
}
@@ -2045,7 +2030,7 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
}
else {
/* Relative/variable index into constant buffer */
- emit(ir, TGSI_OPCODE_USHR, st_dst_reg(index_reg), op[1],
+ emit_asm(ir, TGSI_OPCODE_USHR, st_dst_reg(index_reg), op[1],
st_src_reg_for_int(4));
cbuf.reladdr = ralloc(mem_ctx, st_src_reg);
memcpy(cbuf.reladdr, &index_reg, sizeof(index_reg));
@@ -2078,88 +2063,88 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
const_offset % 16 / 4);
if (ir->type->base_type == GLSL_TYPE_BOOL) {
- emit(ir, TGSI_OPCODE_USNE, result_dst, cbuf, st_src_reg_for_int(0));
+ emit_asm(ir, TGSI_OPCODE_USNE, result_dst, cbuf, st_src_reg_for_int(0));
} else {
- emit(ir, TGSI_OPCODE_MOV, result_dst, cbuf);
+ emit_asm(ir, TGSI_OPCODE_MOV, result_dst, cbuf);
}
break;
}
case ir_triop_lrp:
/* note: we have to reorder the three args here */
- emit(ir, TGSI_OPCODE_LRP, result_dst, op[2], op[1], op[0]);
+ emit_asm(ir, TGSI_OPCODE_LRP, result_dst, op[2], op[1], op[0]);
break;
case ir_triop_csel:
if (this->ctx->Const.NativeIntegers)
- emit(ir, TGSI_OPCODE_UCMP, result_dst, op[0], op[1], op[2]);
+ emit_asm(ir, TGSI_OPCODE_UCMP, result_dst, op[0], op[1], op[2]);
else {
op[0].negate = ~op[0].negate;
- emit(ir, TGSI_OPCODE_CMP, result_dst, op[0], op[1], op[2]);
+ emit_asm(ir, TGSI_OPCODE_CMP, result_dst, op[0], op[1], op[2]);
}
break;
case ir_triop_bitfield_extract:
- emit(ir, TGSI_OPCODE_IBFE, result_dst, op[0], op[1], op[2]);
+ emit_asm(ir, TGSI_OPCODE_IBFE, result_dst, op[0], op[1], op[2]);
break;
case ir_quadop_bitfield_insert:
- emit(ir, TGSI_OPCODE_BFI, result_dst, op[0], op[1], op[2], op[3]);
+ emit_asm(ir, TGSI_OPCODE_BFI, result_dst, op[0], op[1], op[2], op[3]);
break;
case ir_unop_bitfield_reverse:
- emit(ir, TGSI_OPCODE_BREV, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_BREV, result_dst, op[0]);
break;
case ir_unop_bit_count:
- emit(ir, TGSI_OPCODE_POPC, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_POPC, result_dst, op[0]);
break;
case ir_unop_find_msb:
- emit(ir, TGSI_OPCODE_IMSB, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_IMSB, result_dst, op[0]);
break;
case ir_unop_find_lsb:
- emit(ir, TGSI_OPCODE_LSB, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_LSB, result_dst, op[0]);
break;
case ir_binop_imul_high:
- emit(ir, TGSI_OPCODE_IMUL_HI, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_IMUL_HI, result_dst, op[0], op[1]);
break;
case ir_triop_fma:
/* In theory, MAD is incorrect here. */
if (have_fma)
- emit(ir, TGSI_OPCODE_FMA, result_dst, op[0], op[1], op[2]);
+ emit_asm(ir, TGSI_OPCODE_FMA, result_dst, op[0], op[1], op[2]);
else
- emit(ir, TGSI_OPCODE_MAD, result_dst, op[0], op[1], op[2]);
+ emit_asm(ir, TGSI_OPCODE_MAD, result_dst, op[0], op[1], op[2]);
break;
case ir_unop_interpolate_at_centroid:
- emit(ir, TGSI_OPCODE_INTERP_CENTROID, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_INTERP_CENTROID, result_dst, op[0]);
break;
case ir_binop_interpolate_at_offset:
- emit(ir, TGSI_OPCODE_INTERP_OFFSET, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_INTERP_OFFSET, result_dst, op[0], op[1]);
break;
case ir_binop_interpolate_at_sample:
- emit(ir, TGSI_OPCODE_INTERP_SAMPLE, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_INTERP_SAMPLE, result_dst, op[0], op[1]);
break;
case ir_unop_d2f:
- emit(ir, TGSI_OPCODE_D2F, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_D2F, result_dst, op[0]);
break;
case ir_unop_f2d:
- emit(ir, TGSI_OPCODE_F2D, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_F2D, result_dst, op[0]);
break;
case ir_unop_d2i:
- emit(ir, TGSI_OPCODE_D2I, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_D2I, result_dst, op[0]);
break;
case ir_unop_i2d:
- emit(ir, TGSI_OPCODE_I2D, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_I2D, result_dst, op[0]);
break;
case ir_unop_d2u:
- emit(ir, TGSI_OPCODE_D2U, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_D2U, result_dst, op[0]);
break;
case ir_unop_u2d:
- emit(ir, TGSI_OPCODE_U2D, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_U2D, result_dst, op[0]);
break;
case ir_unop_unpack_double_2x32:
case ir_unop_pack_double_2x32:
- emit(ir, TGSI_OPCODE_MOV, result_dst, op[0]);
+ emit_asm(ir, TGSI_OPCODE_MOV, result_dst, op[0]);
break;
case ir_binop_ldexp:
if (ir->operands[0]->type->base_type == GLSL_TYPE_DOUBLE) {
- emit(ir, TGSI_OPCODE_DLDEXP, result_dst, op[0], op[1]);
+ emit_asm(ir, TGSI_OPCODE_DLDEXP, result_dst, op[0], op[1]);
} else {
assert(!"Invalid ldexp for non-double opcode in glsl_to_tgsi_visitor::visit()");
}
@@ -2243,11 +2228,38 @@ glsl_to_tgsi_visitor::visit(ir_swizzle *ir)
this->result = src;
}
+/* Test if the variable is an array. Note that geometry and
+ * tessellation shader inputs are outputs are always arrays (except
+ * for patch inputs), so only the array element type is considered.
+ */
+static bool
+is_inout_array(unsigned stage, ir_variable *var, bool *is_2d)
+{
+ const glsl_type *type = var->type;
+
+ if ((stage == MESA_SHADER_VERTEX && var->data.mode == ir_var_shader_in) ||
+ (stage == MESA_SHADER_FRAGMENT && var->data.mode == ir_var_shader_out))
+ return false;
+
+ *is_2d = false;
+
+ if (stage == MESA_SHADER_GEOMETRY && var->data.mode == ir_var_shader_in) {
+ if (!var->type->is_array())
+ return false; /* a system value probably */
+
+ type = var->type->fields.array;
+ *is_2d = true;
+ }
+
+ return type->is_array() || type->is_matrix();
+}
+
void
glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
{
variable_storage *entry = find_variable_storage(ir->var);
ir_variable *var = ir->var;
+ bool is_2d;
if (!entry) {
switch (var->data.mode) {
@@ -2263,16 +2275,56 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
* user-defined varyings.
*/
assert(var->data.location != -1);
- entry = new(mem_ctx) variable_storage(var,
- PROGRAM_INPUT,
- var->data.location);
+
+ if (is_inout_array(shader->Stage, var, &is_2d)) {
+ struct array_decl *decl = &input_arrays[num_input_arrays];
+
+ decl->mesa_index = var->data.location;
+ decl->array_id = num_input_arrays + 1;
+ if (is_2d)
+ decl->array_size = type_size(var->type->fields.array);
+ else
+ decl->array_size = type_size(var->type);
+ num_input_arrays++;
+
+ entry = new(mem_ctx) variable_storage(var,
+ PROGRAM_INPUT,
+ var->data.location,
+ decl->array_id);
+ }
+ else {
+ entry = new(mem_ctx) variable_storage(var,
+ PROGRAM_INPUT,
+ var->data.location);
+ }
+ this->variables.push_tail(entry);
break;
case ir_var_shader_out:
assert(var->data.location != -1);
- entry = new(mem_ctx) variable_storage(var,
- PROGRAM_OUTPUT,
- var->data.location
- + var->data.index);
+
+ if (is_inout_array(shader->Stage, var, &is_2d)) {
+ struct array_decl *decl = &output_arrays[num_output_arrays];
+
+ decl->mesa_index = var->data.location;
+ decl->array_id = num_output_arrays + 1;
+ if (is_2d)
+ decl->array_size = type_size(var->type->fields.array);
+ else
+ decl->array_size = type_size(var->type);
+ num_output_arrays++;
+
+ entry = new(mem_ctx) variable_storage(var,
+ PROGRAM_OUTPUT,
+ var->data.location,
+ decl->array_id);
+ }
+ else {
+ entry = new(mem_ctx) variable_storage(var,
+ PROGRAM_OUTPUT,
+ var->data.location
+ + var->data.index);
+ }
+ this->variables.push_tail(entry);
break;
case ir_var_system_value:
entry = new(mem_ctx) variable_storage(var,
@@ -2296,10 +2348,43 @@ glsl_to_tgsi_visitor::visit(ir_dereference_variable *ir)
}
this->result = st_src_reg(entry->file, entry->index, var->type);
+ this->result.array_id = entry->array_id;
if (!native_integers)
this->result.type = GLSL_TYPE_FLOAT;
}
+static void
+shrink_array_declarations(struct array_decl *arrays, unsigned count,
+ GLbitfield64 usage_mask)
+{
+ unsigned i, j;
+
+ /* Fix array declarations by removing unused array elements at both ends
+ * of the arrays. For example, mat4[3] where only mat[1] is used.
+ */
+ for (i = 0; i < count; i++) {
+ struct array_decl *decl = &arrays[i];
+
+ /* Shrink the beginning. */
+ for (j = 0; j < decl->array_size; j++) {
+ if (usage_mask & BITFIELD64_BIT(decl->mesa_index+j))
+ break;
+
+ decl->mesa_index++;
+ decl->array_size--;
+ j--;
+ }
+
+ /* Shrink the end. */
+ for (j = decl->array_size-1; j >= 0; j--) {
+ if (usage_mask & BITFIELD64_BIT(decl->mesa_index+j))
+ break;
+
+ decl->array_size--;
+ }
+ }
+}
+
void
glsl_to_tgsi_visitor::visit(ir_dereference_array *ir)
{
@@ -2341,7 +2426,7 @@ glsl_to_tgsi_visitor::visit(ir_dereference_array *ir)
index_reg = get_temp(native_integers ?
glsl_type::int_type : glsl_type::float_type);
- emit(ir, TGSI_OPCODE_MUL, st_dst_reg(index_reg),
+ emit_asm(ir, TGSI_OPCODE_MUL, st_dst_reg(index_reg),
this->result, st_src_reg_for_type(index_reg.type, element_size));
}
@@ -2352,7 +2437,7 @@ glsl_to_tgsi_visitor::visit(ir_dereference_array *ir)
st_src_reg accum_reg = get_temp(native_integers ?
glsl_type::int_type : glsl_type::float_type);
- emit(ir, TGSI_OPCODE_ADD, st_dst_reg(accum_reg),
+ emit_asm(ir, TGSI_OPCODE_ADD, st_dst_reg(accum_reg),
index_reg, *src.reladdr);
index_reg = accum_reg;
@@ -2589,16 +2674,16 @@ glsl_to_tgsi_visitor::emit_block_mov(ir_assignment *ir, const struct glsl_type *
l_src.swizzle = swizzle_for_size(type->vector_elements);
if (native_integers) {
- emit(ir, TGSI_OPCODE_UCMP, *l, *cond,
+ emit_asm(ir, TGSI_OPCODE_UCMP, *l, *cond,
cond_swap ? l_src : *r,
cond_swap ? *r : l_src);
} else {
- emit(ir, TGSI_OPCODE_CMP, *l, *cond,
+ emit_asm(ir, TGSI_OPCODE_CMP, *l, *cond,
cond_swap ? l_src : *r,
cond_swap ? *r : l_src);
}
} else {
- emit(ir, TGSI_OPCODE_MOV, *l, *r);
+ emit_asm(ir, TGSI_OPCODE_MOV, *l, *r);
}
l->index++;
r->index++;
@@ -2679,7 +2764,7 @@ glsl_to_tgsi_visitor::visit(ir_assignment *ir)
*/
glsl_to_tgsi_instruction *inst, *new_inst;
inst = (glsl_to_tgsi_instruction *)this->instructions.get_tail();
- new_inst = emit(ir, inst->op, l, inst->src[0], inst->src[1], inst->src[2]);
+ new_inst = emit_asm(ir, inst->op, l, inst->src[0], inst->src[1], inst->src[2]);
new_inst->saturate = inst->saturate;
inst->dead_mask = inst->dst[0].writemask;
} else {
@@ -2717,7 +2802,7 @@ glsl_to_tgsi_visitor::visit(ir_constant *ir)
src = this->result;
for (i = 0; i < (unsigned int)size; i++) {
- emit(ir, TGSI_OPCODE_MOV, temp, src);
+ emit_asm(ir, TGSI_OPCODE_MOV, temp, src);
src.index++;
temp.index++;
@@ -2739,7 +2824,7 @@ glsl_to_tgsi_visitor::visit(ir_constant *ir)
ir->array_elements[i]->accept(this);
src = this->result;
for (int j = 0; j < size; j++) {
- emit(ir, TGSI_OPCODE_MOV, temp, src);
+ emit_asm(ir, TGSI_OPCODE_MOV, temp, src);
src.index++;
temp.index++;
@@ -2764,7 +2849,7 @@ glsl_to_tgsi_visitor::visit(ir_constant *ir)
ir->type->vector_elements,
GL_FLOAT,
&src.swizzle);
- emit(ir, TGSI_OPCODE_MOV, mat_column, src);
+ emit_asm(ir, TGSI_OPCODE_MOV, mat_column, src);
mat_column.index++;
}
@@ -2889,7 +2974,7 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
l.cond_mask = COND_TR;
for (i = 0; i < type_size(param->type); i++) {
- emit(ir, TGSI_OPCODE_MOV, l, r);
+ emit_asm(ir, TGSI_OPCODE_MOV, l, r);
l.index++;
r.index++;
}
@@ -2897,7 +2982,7 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
}
/* Emit call instruction */
- call_inst = emit(ir, TGSI_OPCODE_CAL);
+ call_inst = emit_asm(ir, TGSI_OPCODE_CAL);
call_inst->function = entry;
/* Process out parameters. */
@@ -2922,7 +3007,7 @@ glsl_to_tgsi_visitor::visit(ir_call *ir)
st_dst_reg l = st_dst_reg(this->result);
for (i = 0; i < type_size(param->type); i++) {
- emit(ir, TGSI_OPCODE_MOV, l, r);
+ emit_asm(ir, TGSI_OPCODE_MOV, l, r);
l.index++;
r.index++;
}
@@ -2965,7 +3050,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
coord = get_temp(glsl_type::vec4_type);
coord_dst = st_dst_reg(coord);
coord_dst.writemask = (1 << ir->coordinate->type->vector_elements) - 1;
- emit(ir, TGSI_OPCODE_MOV, coord_dst, this->result);
+ emit_asm(ir, TGSI_OPCODE_MOV, coord_dst, this->result);
}
if (ir->projector) {
@@ -3074,7 +3159,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
if (opcode == TGSI_OPCODE_TEX) {
/* Slot the projector in as the last component of the coord. */
coord_dst.writemask = WRITEMASK_W;
- emit(ir, TGSI_OPCODE_MOV, coord_dst, projector);
+ emit_asm(ir, TGSI_OPCODE_MOV, coord_dst, projector);
coord_dst.writemask = WRITEMASK_XYZW;
opcode = TGSI_OPCODE_TXP;
} else {
@@ -3086,7 +3171,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
* projective divide now.
*/
coord_dst.writemask = WRITEMASK_W;
- emit(ir, TGSI_OPCODE_RCP, coord_dst, projector);
+ emit_asm(ir, TGSI_OPCODE_RCP, coord_dst, projector);
/* In the case where we have to project the coordinates "by hand,"
* the shadow comparator value must also be projected.
@@ -3105,14 +3190,14 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
assert(!sampler_type->sampler_array);
tmp_dst.writemask = WRITEMASK_Z;
- emit(ir, TGSI_OPCODE_MOV, tmp_dst, this->result);
+ emit_asm(ir, TGSI_OPCODE_MOV, tmp_dst, this->result);
tmp_dst.writemask = WRITEMASK_XY;
- emit(ir, TGSI_OPCODE_MOV, tmp_dst, coord);
+ emit_asm(ir, TGSI_OPCODE_MOV, tmp_dst, coord);
}
coord_dst.writemask = WRITEMASK_XYZ;
- emit(ir, TGSI_OPCODE_MUL, coord_dst, tmp_src, coord_w);
+ emit_asm(ir, TGSI_OPCODE_MUL, coord_dst, tmp_src, coord_w);
coord_dst.writemask = WRITEMASK_XYZW;
coord.swizzle = SWIZZLE_XYZW;
@@ -3133,7 +3218,7 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
cube_sc = get_temp(glsl_type::float_type);
cube_sc_dst = st_dst_reg(cube_sc);
cube_sc_dst.writemask = WRITEMASK_X;
- emit(ir, TGSI_OPCODE_MOV, cube_sc_dst, this->result);
+ emit_asm(ir, TGSI_OPCODE_MOV, cube_sc_dst, this->result);
cube_sc_dst.writemask = WRITEMASK_X;
}
else {
@@ -3144,20 +3229,20 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
} else {
coord_dst.writemask = WRITEMASK_Z;
}
- emit(ir, TGSI_OPCODE_MOV, coord_dst, this->result);
+ emit_asm(ir, TGSI_OPCODE_MOV, coord_dst, this->result);
coord_dst.writemask = WRITEMASK_XYZW;
}
}
if (ir->op == ir_txf_ms) {
coord_dst.writemask = WRITEMASK_W;
- emit(ir, TGSI_OPCODE_MOV, coord_dst, sample_index);
+ emit_asm(ir, TGSI_OPCODE_MOV, coord_dst, sample_index);
coord_dst.writemask = WRITEMASK_XYZW;
} else if (opcode == TGSI_OPCODE_TXL || opcode == TGSI_OPCODE_TXB ||
opcode == TGSI_OPCODE_TXF) {
/* TGSI stores LOD or LOD bias in the last channel of the coords. */
coord_dst.writemask = WRITEMASK_W;
- emit(ir, TGSI_OPCODE_MOV, coord_dst, lod_info);
+ emit_asm(ir, TGSI_OPCODE_MOV, coord_dst, lod_info);
coord_dst.writemask = WRITEMASK_XYZW;
}
@@ -3167,30 +3252,30 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
}
if (opcode == TGSI_OPCODE_TXD)
- inst = emit(ir, opcode, result_dst, coord, dx, dy);
+ inst = emit_asm(ir, opcode, result_dst, coord, dx, dy);
else if (opcode == TGSI_OPCODE_TXQ) {
if (ir->op == ir_query_levels) {
/* the level is stored in W */
- inst = emit(ir, opcode, st_dst_reg(levels_src), lod_info);
+ inst = emit_asm(ir, opcode, st_dst_reg(levels_src), lod_info);
result_dst.writemask = WRITEMASK_X;
levels_src.swizzle = SWIZZLE_WWWW;
- emit(ir, TGSI_OPCODE_MOV, result_dst, levels_src);
+ emit_asm(ir, TGSI_OPCODE_MOV, result_dst, levels_src);
} else
- inst = emit(ir, opcode, result_dst, lod_info);
+ inst = emit_asm(ir, opcode, result_dst, lod_info);
} else if (opcode == TGSI_OPCODE_TXF) {
- inst = emit(ir, opcode, result_dst, coord);
+ inst = emit_asm(ir, opcode, result_dst, coord);
} else if (opcode == TGSI_OPCODE_TXL2 || opcode == TGSI_OPCODE_TXB2) {
- inst = emit(ir, opcode, result_dst, coord, lod_info);
+ inst = emit_asm(ir, opcode, result_dst, coord, lod_info);
} else if (opcode == TGSI_OPCODE_TEX2) {
- inst = emit(ir, opcode, result_dst, coord, cube_sc);
+ inst = emit_asm(ir, opcode, result_dst, coord, cube_sc);
} else if (opcode == TGSI_OPCODE_TG4) {
if (is_cube_array && ir->shadow_comparitor) {
- inst = emit(ir, opcode, result_dst, coord, cube_sc);
+ inst = emit_asm(ir, opcode, result_dst, coord, cube_sc);
} else {
- inst = emit(ir, opcode, result_dst, coord, component);
+ inst = emit_asm(ir, opcode, result_dst, coord, component);
}
} else
- inst = emit(ir, opcode, result_dst, coord);
+ inst = emit_asm(ir, opcode, result_dst, coord);
if (ir->shadow_comparitor)
inst->tex_shadow = GL_TRUE;
@@ -3246,6 +3331,8 @@ glsl_to_tgsi_visitor::visit(ir_texture *ir)
assert(!"Should not get here.");
}
+ inst->tex_type = ir->type->base_type;
+
this->result = result_src;
}
@@ -3264,13 +3351,13 @@ glsl_to_tgsi_visitor::visit(ir_return *ir)
l = st_dst_reg(current_function->return_reg);
for (i = 0; i < type_size(current_function->sig->return_type); i++) {
- emit(ir, TGSI_OPCODE_MOV, l, r);
+ emit_asm(ir, TGSI_OPCODE_MOV, l, r);
l.index++;
r.index++;
}
}
- emit(ir, TGSI_OPCODE_RET);
+ emit_asm(ir, TGSI_OPCODE_RET);
}
void
@@ -3283,16 +3370,16 @@ glsl_to_tgsi_visitor::visit(ir_discard *ir)
/* Convert the bool condition to a float so we can negate. */
if (native_integers) {
st_src_reg temp = get_temp(ir->condition->type);
- emit(ir, TGSI_OPCODE_AND, st_dst_reg(temp),
+ emit_asm(ir, TGSI_OPCODE_AND, st_dst_reg(temp),
condition, st_src_reg_for_float(1.0));
condition = temp;
}
condition.negate = ~condition.negate;
- emit(ir, TGSI_OPCODE_KILL_IF, undef_dst, condition);
+ emit_asm(ir, TGSI_OPCODE_KILL_IF, undef_dst, condition);
} else {
/* unconditional kil */
- emit(ir, TGSI_OPCODE_KILL);
+ emit_asm(ir, TGSI_OPCODE_KILL);
}
}
@@ -3307,18 +3394,18 @@ glsl_to_tgsi_visitor::visit(ir_if *ir)
if_opcode = native_integers ? TGSI_OPCODE_UIF : TGSI_OPCODE_IF;
- if_inst = emit(ir->condition, if_opcode, undef_dst, this->result);
+ if_inst = emit_asm(ir->condition, if_opcode, undef_dst, this->result);
this->instructions.push_tail(if_inst);
visit_exec_list(&ir->then_instructions, this);
if (!ir->else_instructions.is_empty()) {
- emit(ir->condition, TGSI_OPCODE_ELSE);
+ emit_asm(ir->condition, TGSI_OPCODE_ELSE);
visit_exec_list(&ir->else_instructions, this);
}
- if_inst = emit(ir->condition, TGSI_OPCODE_ENDIF);
+ if_inst = emit_asm(ir->condition, TGSI_OPCODE_ENDIF);
}
@@ -3328,7 +3415,7 @@ glsl_to_tgsi_visitor::visit(ir_emit_vertex *ir)
assert(this->prog->Target == GL_GEOMETRY_PROGRAM_NV);
ir->stream->accept(this);
- emit(ir, TGSI_OPCODE_EMIT, undef_dst, this->result);
+ emit_asm(ir, TGSI_OPCODE_EMIT, undef_dst, this->result);
}
void
@@ -3337,14 +3424,24 @@ glsl_to_tgsi_visitor::visit(ir_end_primitive *ir)
assert(this->prog->Target == GL_GEOMETRY_PROGRAM_NV);
ir->stream->accept(this);
- emit(ir, TGSI_OPCODE_ENDPRIM, undef_dst, this->result);
+ emit_asm(ir, TGSI_OPCODE_ENDPRIM, undef_dst, this->result);
+}
+
+void
+glsl_to_tgsi_visitor::visit(ir_barrier *ir)
+{
+ unreachable("Not implemented!");
}
glsl_to_tgsi_visitor::glsl_to_tgsi_visitor()
{
result.file = PROGRAM_UNDEFINED;
next_temp = 1;
+ array_sizes = NULL;
+ max_num_arrays = 0;
next_array = 0;
+ num_input_arrays = 0;
+ num_output_arrays = 0;
next_signature_id = 1;
num_immediates = 0;
current_function = NULL;
@@ -3366,6 +3463,7 @@ glsl_to_tgsi_visitor::glsl_to_tgsi_visitor()
glsl_to_tgsi_visitor::~glsl_to_tgsi_visitor()
{
+ free(array_sizes);
ralloc_free(mem_ctx);
}
@@ -3387,7 +3485,13 @@ count_resources(glsl_to_tgsi_visitor *v, gl_program *prog)
foreach_in_list(glsl_to_tgsi_instruction, inst, &v->instructions) {
if (is_tex_instruction(inst->op)) {
for (int i = 0; i < inst->sampler_array_size; i++) {
- v->samplers_used |= 1 << (inst->sampler.index + i);
+ unsigned idx = inst->sampler.index + i;
+ v->samplers_used |= 1 << idx;
+
+ debug_assert(idx < (int)ARRAY_SIZE(v->sampler_types));
+ v->sampler_types[idx] = inst->tex_type;
+ v->sampler_targets[idx] =
+ st_translate_texture_target(inst->tex_target, inst->tex_shadow);
if (inst->tex_shadow) {
prog->ShadowSamplers |= 1 << (inst->sampler.index + i);
@@ -3734,6 +3838,7 @@ glsl_to_tgsi_visitor::copy_propagate(void)
inst->src[r].index2D = first->src[0].index2D;
inst->src[r].has_index2 = first->src[0].has_index2;
inst->src[r].double_reg2 = first->src[0].double_reg2;
+ inst->src[r].array_id = first->src[0].array_id;
int swizzle = 0;
for (int i = 0; i < 4; i++) {
@@ -4177,7 +4282,7 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
coord = st_src_reg(PROGRAM_INPUT, VARYING_SLOT_TEX0, glsl_type::vec2_type);
src0 = v->get_temp(glsl_type::vec4_type);
dst0 = st_dst_reg(src0);
- inst = v->emit(NULL, TGSI_OPCODE_TEX, dst0, coord);
+ inst = v->emit_asm(NULL, TGSI_OPCODE_TEX, dst0, coord);
inst->sampler_array_size = 1;
inst->tex_target = TEXTURE_2D_INDEX;
@@ -4201,7 +4306,7 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
/* MAD colorTemp, colorTemp, scale, bias; */
scale = st_src_reg(PROGRAM_STATE_VAR, scale_p, GLSL_TYPE_FLOAT);
bias = st_src_reg(PROGRAM_STATE_VAR, bias_p, GLSL_TYPE_FLOAT);
- inst = v->emit(NULL, TGSI_OPCODE_MAD, dst0, src0, scale, bias);
+ inst = v->emit_asm(NULL, TGSI_OPCODE_MAD, dst0, src0, scale, bias);
}
if (pixel_maps) {
@@ -4209,6 +4314,7 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
st_dst_reg temp_dst = st_dst_reg(temp);
assert(st->pixel_xfer.pixelmap_texture);
+ (void) st;
/* With a little effort, we can do four pixel map look-ups with
* two TEX instructions:
@@ -4216,7 +4322,7 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
/* TEX temp.rg, colorTemp.rgba, texture[1], 2D; */
temp_dst.writemask = WRITEMASK_XY; /* write R,G */
- inst = v->emit(NULL, TGSI_OPCODE_TEX, temp_dst, src0);
+ inst = v->emit_asm(NULL, TGSI_OPCODE_TEX, temp_dst, src0);
inst->sampler.index = 1;
inst->sampler_array_size = 1;
inst->tex_target = TEXTURE_2D_INDEX;
@@ -4224,7 +4330,7 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
/* TEX temp.ba, colorTemp.baba, texture[1], 2D; */
src0.swizzle = MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_W, SWIZZLE_Z, SWIZZLE_W);
temp_dst.writemask = WRITEMASK_ZW; /* write B,A */
- inst = v->emit(NULL, TGSI_OPCODE_TEX, temp_dst, src0);
+ inst = v->emit_asm(NULL, TGSI_OPCODE_TEX, temp_dst, src0);
inst->sampler.index = 1;
inst->sampler_array_size = 1;
inst->tex_target = TEXTURE_2D_INDEX;
@@ -4233,7 +4339,7 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
v->samplers_used |= (1 << 1);
/* MOV colorTemp, temp; */
- inst = v->emit(NULL, TGSI_OPCODE_MOV, dst0, temp);
+ inst = v->emit_asm(NULL, TGSI_OPCODE_MOV, dst0, temp);
}
/* Now copy the instructions from the original glsl_to_tgsi_visitor into the
@@ -4256,7 +4362,7 @@ get_pixel_transfer_visitor(struct st_fragment_program *fp,
prog->InputsRead |= BITFIELD64_BIT(src_regs[i].index);
}
- newinst = v->emit(NULL, inst->op, inst->dst[0], src_regs[0], src_regs[1], src_regs[2]);
+ newinst = v->emit_asm(NULL, inst->op, inst->dst[0], src_regs[0], src_regs[1], src_regs[2]);
newinst->tex_target = inst->tex_target;
newinst->sampler_array_size = inst->sampler_array_size;
}
@@ -4306,7 +4412,7 @@ get_bitmap_visitor(struct st_fragment_program *fp,
coord = st_src_reg(PROGRAM_INPUT, VARYING_SLOT_TEX0, glsl_type::vec2_type);
src0 = v->get_temp(glsl_type::vec4_type);
dst0 = st_dst_reg(src0);
- inst = v->emit(NULL, TGSI_OPCODE_TEX, dst0, coord);
+ inst = v->emit_asm(NULL, TGSI_OPCODE_TEX, dst0, coord);
inst->sampler.index = samplerIndex;
inst->sampler_array_size = 1;
inst->tex_target = TEXTURE_2D_INDEX;
@@ -4319,7 +4425,7 @@ get_bitmap_visitor(struct st_fragment_program *fp,
src0.negate = NEGATE_XYZW;
if (st->bitmap.tex_format == PIPE_FORMAT_L8_UNORM)
src0.swizzle = SWIZZLE_XXXX;
- inst = v->emit(NULL, TGSI_OPCODE_KILL_IF, undef_dst, src0);
+ inst = v->emit_asm(NULL, TGSI_OPCODE_KILL_IF, undef_dst, src0);
/* Now copy the instructions from the original glsl_to_tgsi_visitor into the
* new visitor. */
@@ -4336,7 +4442,7 @@ get_bitmap_visitor(struct st_fragment_program *fp,
prog->InputsRead |= BITFIELD64_BIT(src_regs[i].index);
}
- newinst = v->emit(NULL, inst->op, inst->dst[0], src_regs[0], src_regs[1], src_regs[2]);
+ newinst = v->emit_asm(NULL, inst->op, inst->dst[0], src_regs[0], src_regs[1], src_regs[2]);
newinst->tex_target = inst->tex_target;
newinst->sampler_array_size = inst->sampler_array_size;
}
@@ -4362,7 +4468,8 @@ struct st_translate {
unsigned temps_size;
struct ureg_dst *temps;
- struct ureg_dst arrays[MAX_ARRAYS];
+ struct ureg_dst *arrays;
+ unsigned num_temp_arrays;
struct ureg_src *constants;
int num_constants;
struct ureg_src *immediates;
@@ -4373,7 +4480,9 @@ struct st_translate {
struct ureg_src samplers[PIPE_MAX_SAMPLERS];
struct ureg_src systemValues[SYSTEM_VALUE_MAX];
struct tgsi_texture_offset tex_offsets[MAX_GLSL_TEXTURE_OFFSET];
- unsigned array_sizes[MAX_ARRAYS];
+ unsigned *array_sizes;
+ struct array_decl *input_arrays;
+ struct array_decl *output_arrays;
const GLuint *inputMapping;
const GLuint *outputMapping;
@@ -4497,9 +4606,8 @@ emit_immediate(struct st_translate *t,
* Map a glsl_to_tgsi dst register to a TGSI ureg_dst register.
*/
static struct ureg_dst
-dst_register(struct st_translate *t,
- gl_register_file file,
- GLuint index)
+dst_register(struct st_translate *t, gl_register_file file, unsigned index,
+ unsigned array_id)
{
unsigned array;
@@ -4530,7 +4638,7 @@ dst_register(struct st_translate *t,
case PROGRAM_ARRAY:
array = index >> 16;
- assert(array < ARRAY_SIZE(t->arrays));
+ assert(array < t->num_temp_arrays);
if (ureg_dst_is_undef(t->arrays[array]))
t->arrays[array] = ureg_DECL_array_temporary(
@@ -4540,16 +4648,25 @@ dst_register(struct st_translate *t,
(int)(index & 0xFFFF) - 0x8000);
case PROGRAM_OUTPUT:
- if (t->procType == TGSI_PROCESSOR_VERTEX)
- assert(index < VARYING_SLOT_MAX);
- else if (t->procType == TGSI_PROCESSOR_FRAGMENT)
- assert(index < FRAG_RESULT_MAX);
- else
- assert(index < VARYING_SLOT_MAX);
+ if (!array_id) {
+ if (t->procType == TGSI_PROCESSOR_FRAGMENT)
+ assert(index < FRAG_RESULT_MAX);
+ else
+ assert(index < VARYING_SLOT_MAX);
- assert(t->outputMapping[index] < ARRAY_SIZE(t->outputs));
+ assert(t->outputMapping[index] < ARRAY_SIZE(t->outputs));
+ assert(t->outputs[t->outputMapping[index]].File != TGSI_FILE_NULL);
+ return t->outputs[t->outputMapping[index]];
+ }
+ else {
+ struct array_decl *decl = &t->output_arrays[array_id-1];
+ unsigned mesa_index = decl->mesa_index;
+ int slot = t->outputMapping[mesa_index];
- return t->outputs[t->outputMapping[index]];
+ assert(slot != -1 && t->outputs[slot].File == TGSI_FILE_OUTPUT);
+ assert(t->outputs[slot].ArrayID == array_id);
+ return ureg_dst_array_offset(t->outputs[slot], index - mesa_index);
+ }
case PROGRAM_ADDRESS:
return t->address[index];
@@ -4575,7 +4692,8 @@ src_register(struct st_translate *t, const st_src_reg *reg)
case PROGRAM_TEMPORARY:
case PROGRAM_ARRAY:
- return ureg_src(dst_register(t, reg->file, reg->index));
+ case PROGRAM_OUTPUT:
+ return ureg_src(dst_register(t, reg->file, reg->index, reg->array_id));
case PROGRAM_UNIFORM:
assert(reg->index >= 0);
@@ -4598,12 +4716,20 @@ src_register(struct st_translate *t, const st_src_reg *reg)
* map back to the original index and add the offset after
* mapping. */
index -= double_reg2;
- assert(t->inputMapping[index] < ARRAY_SIZE(t->inputs));
- return t->inputs[t->inputMapping[index] + double_reg2];
+ if (!reg->array_id) {
+ assert(t->inputMapping[index] < ARRAY_SIZE(t->inputs));
+ assert(t->inputs[t->inputMapping[index]].File != TGSI_FILE_NULL);
+ return t->inputs[t->inputMapping[index]];
+ }
+ else {
+ struct array_decl *decl = &t->input_arrays[reg->array_id-1];
+ unsigned mesa_index = decl->mesa_index;
+ int slot = t->inputMapping[mesa_index];
- case PROGRAM_OUTPUT:
- assert(t->outputMapping[reg->index] < ARRAY_SIZE(t->outputs));
- return ureg_src(t->outputs[t->outputMapping[reg->index]]); /* not needed? */
+ assert(slot != -1 && t->inputs[slot].File == TGSI_FILE_INPUT);
+ assert(t->inputs[slot].ArrayID == reg->array_id);
+ return ureg_src_array_offset(t->inputs[slot], index - mesa_index);
+ }
case PROGRAM_ADDRESS:
return ureg_src(t->address[reg->index]);
@@ -4626,9 +4752,8 @@ translate_dst(struct st_translate *t,
const st_dst_reg *dst_reg,
bool saturate, bool clamp_color)
{
- struct ureg_dst dst = dst_register(t,
- dst_reg->file,
- dst_reg->index);
+ struct ureg_dst dst = dst_register(t, dst_reg->file, dst_reg->index,
+ dst_reg->array_id);
if (dst.File == TGSI_FILE_NULL)
return dst;
@@ -4738,7 +4863,7 @@ translate_tex_offset(struct st_translate *t,
array = in_offset->index >> 16;
assert(array >= 0);
- assert(array < (int) ARRAY_SIZE(t->arrays));
+ assert(array < (int)t->num_temp_arrays);
dst = t->arrays[array];
offset.File = dst.File;
@@ -5060,6 +5185,25 @@ emit_edgeflags(struct st_translate *t)
ureg_MOV(ureg, edge_dst, edge_src);
}
+static bool
+find_array(unsigned attr, struct array_decl *arrays, unsigned count,
+ unsigned *array_id, unsigned *array_size)
+{
+ unsigned i;
+
+ for (i = 0; i < count; i++) {
+ struct array_decl *decl = &arrays[i];
+
+ if (attr == decl->mesa_index) {
+ *array_id = decl->array_id;
+ *array_size = decl->array_size;
+ assert(*array_size);
+ return true;
+ }
+ }
+ return false;
+}
+
/**
* Translate intermediate IR (glsl_to_tgsi_instruction) to TGSI format.
* \param program the program to translate
@@ -5089,12 +5233,14 @@ st_translate_program(
const struct gl_program *proginfo,
GLuint numInputs,
const GLuint inputMapping[],
+ const GLuint inputSlotToAttr[],
const ubyte inputSemanticName[],
const ubyte inputSemanticIndex[],
const GLuint interpMode[],
const GLuint interpLocation[],
GLuint numOutputs,
const GLuint outputMapping[],
+ const GLuint outputSlotToAttr[],
const ubyte outputSemanticName[],
const ubyte outputSemanticIndex[],
boolean passthrough_edgeflags,
@@ -5132,25 +5278,101 @@ st_translate_program(
goto out;
}
- memset(t, 0, sizeof *t);
-
t->procType = procType;
t->inputMapping = inputMapping;
t->outputMapping = outputMapping;
t->ureg = ureg;
+ t->num_temp_arrays = program->next_array;
+ if (t->num_temp_arrays)
+ t->arrays = (struct ureg_dst*)
+ calloc(1, sizeof(t->arrays[0]) * t->num_temp_arrays);
/*
* Declare input attributes.
*/
- if (procType == TGSI_PROCESSOR_FRAGMENT) {
+ switch (procType) {
+ case TGSI_PROCESSOR_FRAGMENT:
for (i = 0; i < numInputs; i++) {
- t->inputs[i] = ureg_DECL_fs_input_cyl_centroid(ureg,
- inputSemanticName[i],
- inputSemanticIndex[i],
- interpMode[i], 0,
- interpLocation[i]);
+ unsigned array_id = 0;
+ unsigned array_size;
+
+ if (find_array(inputSlotToAttr[i], program->input_arrays,
+ program->num_input_arrays, &array_id, &array_size)) {
+ /* We've found an array. Declare it so. */
+ t->inputs[i] = ureg_DECL_fs_input_cyl_centroid(ureg,
+ inputSemanticName[i], inputSemanticIndex[i],
+ interpMode[i], 0, interpLocation[i],
+ array_id, array_size);
+ i += array_size - 1;
+ }
+ else {
+ t->inputs[i] = ureg_DECL_fs_input_cyl_centroid(ureg,
+ inputSemanticName[i], inputSemanticIndex[i],
+ interpMode[i], 0, interpLocation[i], 0, 1);
+ }
+ }
+ break;
+ case TGSI_PROCESSOR_GEOMETRY:
+ for (i = 0; i < numInputs; i++) {
+ unsigned array_id = 0;
+ unsigned array_size;
+
+ if (find_array(inputSlotToAttr[i], program->input_arrays,
+ program->num_input_arrays, &array_id, &array_size)) {
+ /* We've found an array. Declare it so. */
+ t->inputs[i] = ureg_DECL_input(ureg, inputSemanticName[i],
+ inputSemanticIndex[i],
+ array_id, array_size);
+ i += array_size - 1;
+ }
+ else {
+ t->inputs[i] = ureg_DECL_input(ureg, inputSemanticName[i],
+ inputSemanticIndex[i], 0, 1);
+ }
}
+ break;
+ case TGSI_PROCESSOR_VERTEX:
+ for (i = 0; i < numInputs; i++) {
+ t->inputs[i] = ureg_DECL_vs_input(ureg, i);
+ }
+ break;
+ default:
+ assert(0);
+ }
+ /*
+ * Declare output attributes.
+ */
+ switch (procType) {
+ case TGSI_PROCESSOR_FRAGMENT:
+ break;
+ case TGSI_PROCESSOR_GEOMETRY:
+ case TGSI_PROCESSOR_VERTEX:
+ for (i = 0; i < numOutputs; i++) {
+ unsigned array_id = 0;
+ unsigned array_size;
+
+ if (find_array(outputSlotToAttr[i], program->output_arrays,
+ program->num_output_arrays, &array_id, &array_size)) {
+ /* We've found an array. Declare it so. */
+ t->outputs[i] = ureg_DECL_output_array(ureg,
+ outputSemanticName[i],
+ outputSemanticIndex[i],
+ array_id, array_size);
+ i += array_size - 1;
+ }
+ else {
+ t->outputs[i] = ureg_DECL_output(ureg,
+ outputSemanticName[i],
+ outputSemanticIndex[i]);
+ }
+ }
+ break;
+ default:
+ assert(0);
+ }
+
+ if (procType == TGSI_PROCESSOR_FRAGMENT) {
if (proginfo->InputsRead & VARYING_BIT_POS) {
/* Must do this after setting up t->inputs. */
emit_wpos(st_context(ctx), t, proginfo, ureg,
@@ -5160,9 +5382,6 @@ st_translate_program(
if (proginfo->InputsRead & VARYING_BIT_FACE)
emit_face_var(ctx, t);
- /*
- * Declare output attributes.
- */
for (i = 0; i < numOutputs; i++) {
switch (outputSemanticName[i]) {
case TGSI_SEMANTIC_POSITION:
@@ -5198,31 +5417,8 @@ st_translate_program(
}
}
}
- else if (procType == TGSI_PROCESSOR_GEOMETRY) {
- for (i = 0; i < numInputs; i++) {
- t->inputs[i] = ureg_DECL_gs_input(ureg,
- i,
- inputSemanticName[i],
- inputSemanticIndex[i]);
- }
-
+ else if (procType == TGSI_PROCESSOR_VERTEX) {
for (i = 0; i < numOutputs; i++) {
- t->outputs[i] = ureg_DECL_output(ureg,
- outputSemanticName[i],
- outputSemanticIndex[i]);
- }
- }
- else {
- assert(procType == TGSI_PROCESSOR_VERTEX);
-
- for (i = 0; i < numInputs; i++) {
- t->inputs[i] = ureg_DECL_vs_input(ureg, i);
- }
-
- for (i = 0; i < numOutputs; i++) {
- t->outputs[i] = ureg_DECL_output(ureg,
- outputSemanticName[i],
- outputSemanticIndex[i]);
if (outputSemanticName[i] == TGSI_SEMANTIC_FOG) {
/* force register to contain a fog coordinate in the form (F, 0, 0, 1). */
ureg_MOV(ureg,
@@ -5277,9 +5473,9 @@ st_translate_program(
}
}
- /* Copy over array sizes
- */
- memcpy(t->array_sizes, program->array_sizes, sizeof(unsigned) * program->next_array);
+ t->array_sizes = program->array_sizes;
+ t->input_arrays = program->input_arrays;
+ t->output_arrays = program->output_arrays;
/* Emit constants and uniforms. TGSI uses a single index space for these,
* so we put all the translated regs in t->constants.
@@ -5355,7 +5551,26 @@ st_translate_program(
/* texture samplers */
for (i = 0; i < ctx->Const.Program[MESA_SHADER_FRAGMENT].MaxTextureImageUnits; i++) {
if (program->samplers_used & (1 << i)) {
+ unsigned type;
+
t->samplers[i] = ureg_DECL_sampler(ureg, i);
+
+ switch (program->sampler_types[i]) {
+ case GLSL_TYPE_INT:
+ type = TGSI_RETURN_TYPE_SINT;
+ break;
+ case GLSL_TYPE_UINT:
+ type = TGSI_RETURN_TYPE_UINT;
+ break;
+ case GLSL_TYPE_FLOAT:
+ type = TGSI_RETURN_TYPE_FLOAT;
+ break;
+ default:
+ unreachable("not reached");
+ }
+
+ ureg_DECL_sampler_view( ureg, i, program->sampler_targets[i],
+ type, type, type, type );
}
}
@@ -5375,6 +5590,7 @@ st_translate_program(
out:
if (t) {
+ free(t->arrays);
free(t->temps);
free(t->insn);
free(t->labels);
@@ -5470,7 +5686,7 @@ get_mesa_program(struct gl_context *ctx,
if (!entry->bgn_inst) {
v->current_function = entry;
- entry->bgn_inst = v->emit(NULL, TGSI_OPCODE_BGNSUB);
+ entry->bgn_inst = v->emit_asm(NULL, TGSI_OPCODE_BGNSUB);
entry->bgn_inst->function = entry;
visit_exec_list(&entry->sig->body, v);
@@ -5478,10 +5694,10 @@ get_mesa_program(struct gl_context *ctx,
glsl_to_tgsi_instruction *last;
last = (glsl_to_tgsi_instruction *)v->instructions.get_tail();
if (last->op != TGSI_OPCODE_RET)
- v->emit(NULL, TGSI_OPCODE_RET);
+ v->emit_asm(NULL, TGSI_OPCODE_RET);
glsl_to_tgsi_instruction *end;
- end = v->emit(NULL, TGSI_OPCODE_ENDSUB);
+ end = v->emit_asm(NULL, TGSI_OPCODE_ENDSUB);
end->function = entry;
progress = GL_TRUE;
@@ -5513,7 +5729,7 @@ get_mesa_program(struct gl_context *ctx,
v->renumber_registers();
/* Write the END instruction. */
- v->emit(NULL, TGSI_OPCODE_END);
+ v->emit_asm(NULL, TGSI_OPCODE_END);
if (ctx->_Shader->Flags & GLSL_DUMP) {
_mesa_log("\n");
@@ -5528,6 +5744,10 @@ get_mesa_program(struct gl_context *ctx,
prog->NumInstructions = 0;
do_set_program_inouts(shader->ir, prog, shader->Stage);
+ shrink_array_declarations(v->input_arrays, v->num_input_arrays,
+ prog->InputsRead);
+ shrink_array_declarations(v->output_arrays, v->num_output_arrays,
+ prog->OutputsWritten);
count_resources(v, prog);
/* This must be done before the uniform storage is associated. */
@@ -5549,6 +5769,7 @@ get_mesa_program(struct gl_context *ctx,
*/
_mesa_associate_uniform_storage(ctx, shader_program, prog->Parameters);
if (!shader_program->LinkStatus) {
+ free_glsl_to_tgsi_visitor(v);
return NULL;
}
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi.h b/src/mesa/state_tracker/st_glsl_to_tgsi.h
index 2cb80bcf961..4af747fa9de 100644
--- a/src/mesa/state_tracker/st_glsl_to_tgsi.h
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi.h
@@ -43,12 +43,14 @@ enum pipe_error st_translate_program(
const struct gl_program *proginfo,
GLuint numInputs,
const GLuint inputMapping[],
+ const GLuint inputSlotToAttr[],
const ubyte inputSemanticName[],
const ubyte inputSemanticIndex[],
const GLuint interpMode[],
const GLuint interpLocation[],
GLuint numOutputs,
const GLuint outputMapping[],
+ const GLuint outputSlotToAttr[],
const ubyte outputSemanticName[],
const ubyte outputSemanticIndex[],
boolean passthrough_edgeflags,
diff --git a/src/mesa/state_tracker/st_manager.c b/src/mesa/state_tracker/st_manager.c
index 840f76a1307..a2dee6298fa 100644
--- a/src/mesa/state_tracker/st_manager.c
+++ b/src/mesa/state_tracker/st_manager.c
@@ -680,6 +680,10 @@ st_api_create_context(struct st_api *stapi, struct st_manager *smapi,
if (attribs->flags & ST_CONTEXT_FLAG_FORWARD_COMPATIBLE)
st->ctx->Const.ContextFlags |= GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT;
+ if (attribs->flags & ST_CONTEXT_FLAG_ROBUST_ACCESS)
+ st->ctx->Const.ContextFlags |= GL_CONTEXT_FLAG_ROBUST_ACCESS_BIT_ARB;
+ if (attribs->flags & ST_CONTEXT_FLAG_RESET_NOTIFICATION_ENABLED)
+ st->ctx->Const.ResetStrategy = GL_LOSE_CONTEXT_ON_RESET_ARB;
/* need to perform version check */
if (attribs->major > 1 || attribs->minor > 0) {
@@ -920,8 +924,7 @@ static unsigned get_version(struct pipe_screen *screen,
struct gl_extensions extensions = {0};
GLuint version;
- if ((api == API_OPENGL_COMPAT || api == API_OPENGL_CORE) &&
- _mesa_override_gl_version_contextless(&consts, &api, &version)) {
+ if (_mesa_override_gl_version_contextless(&consts, &api, &version)) {
return version;
}
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index 98d525c86c2..896e239ee68 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -665,7 +665,7 @@ compile_instruction(
if (num_dst)
dst[0] = translate_dst( t,
&inst->DstReg,
- inst->SaturateMode,
+ inst->Saturate,
clamp_dst_color_output);
for (i = 0; i < num_src; i++)
@@ -1095,10 +1095,9 @@ st_translate_mesa_program(
}
else if (procType == TGSI_PROCESSOR_GEOMETRY) {
for (i = 0; i < numInputs; i++) {
- t->inputs[i] = ureg_DECL_gs_input(ureg,
- i,
- inputSemanticName[i],
- inputSemanticIndex[i]);
+ t->inputs[i] = ureg_DECL_input(ureg,
+ inputSemanticName[i],
+ inputSemanticIndex[i], 0, 1);
}
for (i = 0; i < numOutputs; i++) {
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index a9110d3c674..fa792bc349b 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -215,6 +215,7 @@ st_prepare_vertex_program(struct gl_context *ctx,
unsigned slot = stvp->num_outputs++;
stvp->result_to_output[attr] = slot;
+ stvp->output_slot_to_attr[slot] = attr;
switch (attr) {
case VARYING_SLOT_POS:
@@ -285,7 +286,8 @@ st_prepare_vertex_program(struct gl_context *ctx,
/* fall through */
case VARYING_SLOT_VAR0:
default:
- assert(attr < VARYING_SLOT_MAX);
+ assert(attr >= VARYING_SLOT_VAR0 ||
+ (attr >= VARYING_SLOT_TEX0 && attr <= VARYING_SLOT_TEX7));
stvp->output_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
stvp->output_semantic_index[slot] =
st_get_generic_varying_index(st, attr);
@@ -321,7 +323,7 @@ st_translate_vertex_program(struct st_context *st,
_mesa_remove_output_reads(&stvp->Base.Base, PROGRAM_OUTPUT);
}
- ureg = ureg_create( TGSI_PROCESSOR_VERTEX );
+ ureg = ureg_create_with_screen(TGSI_PROCESSOR_VERTEX, st->pipe->screen);
if (ureg == NULL) {
free(vpv);
return NULL;
@@ -351,6 +353,7 @@ st_translate_vertex_program(struct st_context *st,
/* inputs */
vpv->num_inputs,
stvp->input_to_index,
+ NULL, /* inputSlotToAttr */
NULL, /* input semantic name */
NULL, /* input semantic index */
NULL, /* interp mode */
@@ -358,6 +361,7 @@ st_translate_vertex_program(struct st_context *st,
/* outputs */
num_outputs,
stvp->result_to_output,
+ stvp->output_slot_to_attr,
stvp->output_semantic_name,
stvp->output_semantic_index,
key->passthrough_edgeflags,
@@ -482,6 +486,7 @@ st_translate_fragment_program(struct st_context *st,
GLuint outputMapping[FRAG_RESULT_MAX];
GLuint inputMapping[VARYING_SLOT_MAX];
+ GLuint inputSlotToAttr[VARYING_SLOT_MAX];
GLuint interpMode[PIPE_MAX_SHADER_INPUTS]; /* XXX size? */
GLuint interpLocation[PIPE_MAX_SHADER_INPUTS];
GLuint attr;
@@ -502,6 +507,7 @@ st_translate_fragment_program(struct st_context *st,
return NULL;
assert(!(key->bitmap && key->drawpixels));
+ memset(inputSlotToAttr, ~0, sizeof(inputSlotToAttr));
if (key->bitmap) {
/* glBitmap drawing */
@@ -543,6 +549,7 @@ st_translate_fragment_program(struct st_context *st,
const GLuint slot = fs_num_inputs++;
inputMapping[attr] = slot;
+ inputSlotToAttr[slot] = attr;
if (stfp->Base.IsCentroid & BITFIELD64_BIT(attr))
interpLocation[slot] = TGSI_INTERPOLATE_LOC_CENTROID;
else if (stfp->Base.IsSample & BITFIELD64_BIT(attr))
@@ -657,7 +664,8 @@ st_translate_fragment_program(struct st_context *st,
* consumed for the TEXi varyings, and we can base the locations of
* the user varyings on VAR0. Otherwise, we use TEX0 as base index.
*/
- assert(attr >= VARYING_SLOT_TEX0);
+ assert(attr >= VARYING_SLOT_VAR0 || attr == VARYING_SLOT_PNTC ||
+ (attr >= VARYING_SLOT_TEX0 && attr <= VARYING_SLOT_TEX7));
input_semantic_name[slot] = TGSI_SEMANTIC_GENERIC;
input_semantic_index[slot] = st_get_generic_varying_index(st, attr);
if (attr == VARYING_SLOT_PNTC)
@@ -732,7 +740,7 @@ st_translate_fragment_program(struct st_context *st,
}
}
- ureg = ureg_create( TGSI_PROCESSOR_FRAGMENT );
+ ureg = ureg_create_with_screen(TGSI_PROCESSOR_FRAGMENT, st->pipe->screen);
if (ureg == NULL) {
free(variant);
return NULL;
@@ -778,6 +786,7 @@ st_translate_fragment_program(struct st_context *st,
/* inputs */
fs_num_inputs,
inputMapping,
+ inputSlotToAttr,
input_semantic_name,
input_semantic_index,
interpMode,
@@ -785,6 +794,7 @@ st_translate_fragment_program(struct st_context *st,
/* outputs */
fs_num_outputs,
outputMapping,
+ NULL,
fs_output_semantic_name,
fs_output_semantic_index, FALSE,
key->clamp_color );
@@ -867,7 +877,9 @@ st_translate_geometry_program(struct st_context *st,
struct st_geometry_program *stgp,
const struct st_gp_variant_key *key)
{
+ GLuint inputSlotToAttr[VARYING_SLOT_MAX];
GLuint inputMapping[VARYING_SLOT_MAX];
+ GLuint outputSlotToAttr[VARYING_SLOT_MAX];
GLuint outputMapping[VARYING_SLOT_MAX];
struct pipe_context *pipe = st->pipe;
GLuint attr;
@@ -890,13 +902,15 @@ st_translate_geometry_program(struct st_context *st,
if (!gpv)
return NULL;
- ureg = ureg_create(TGSI_PROCESSOR_GEOMETRY);
+ ureg = ureg_create_with_screen(TGSI_PROCESSOR_GEOMETRY, st->pipe->screen);
if (ureg == NULL) {
free(gpv);
return NULL;
}
+ memset(inputSlotToAttr, 0, sizeof(inputSlotToAttr));
memset(inputMapping, 0, sizeof(inputMapping));
+ memset(outputSlotToAttr, 0, sizeof(outputSlotToAttr));
memset(outputMapping, 0, sizeof(outputMapping));
/*
@@ -907,6 +921,7 @@ st_translate_geometry_program(struct st_context *st,
const GLuint slot = gs_num_inputs++;
inputMapping[attr] = slot;
+ inputSlotToAttr[slot] = attr;
switch (attr) {
case VARYING_SLOT_PRIMITIVE_ID:
@@ -985,6 +1000,7 @@ st_translate_geometry_program(struct st_context *st,
GLuint slot = gs_num_outputs++;
outputMapping[attr] = slot;
+ outputSlotToAttr[slot] = attr;
switch (attr) {
case VARYING_SLOT_POS:
@@ -1080,6 +1096,7 @@ st_translate_geometry_program(struct st_context *st,
/* inputs */
gs_num_inputs,
inputMapping,
+ inputSlotToAttr,
input_semantic_name,
input_semantic_index,
NULL,
@@ -1087,6 +1104,7 @@ st_translate_geometry_program(struct st_context *st,
/* outputs */
gs_num_outputs,
outputMapping,
+ outputSlotToAttr,
gs_output_semantic_name,
gs_output_semantic_index,
FALSE,
@@ -1201,7 +1219,7 @@ destroy_program_variants(struct st_context *st, struct gl_program *program)
}
}
break;
- case MESA_GEOMETRY_PROGRAM:
+ case GL_GEOMETRY_PROGRAM_NV:
{
struct st_geometry_program *stgp =
(struct st_geometry_program *) program;
diff --git a/src/mesa/state_tracker/st_program.h b/src/mesa/state_tracker/st_program.h
index a2c56062d6e..bb77eb6ed65 100644
--- a/src/mesa/state_tracker/st_program.h
+++ b/src/mesa/state_tracker/st_program.h
@@ -163,6 +163,7 @@ struct st_vertex_program
/** Maps VARYING_SLOT_x to slot */
GLuint result_to_output[VARYING_SLOT_MAX];
+ GLuint output_slot_to_attr[VARYING_SLOT_MAX];
ubyte output_semantic_name[VARYING_SLOT_MAX];
ubyte output_semantic_index[VARYING_SLOT_MAX];
GLuint num_outputs;
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-26 11:54:45 +0000