/**************************************************************************
- *
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ *
+ * Copyright 2003 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
* distribute, sub license, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
- *
+ *
* The above copyright notice and this permission notice (including the
* next paragraph) shall be included in all copies or substantial portions
* of the Software.
- *
+ *
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
* ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
+ *
**************************************************************************/
/**
*/
-
-#include "main/imports.h"
#include "main/mtypes.h"
-#include "main/macros.h"
-#include "shader/program.h"
+#include "main/framebuffer.h"
+#include "main/state.h"
+#include "main/texobj.h"
+#include "main/texstate.h"
+#include "program/program.h"
#include "pipe/p_context.h"
#include "pipe/p_shader_tokens.h"
-
#include "util/u_simple_shaders.h"
-
#include "cso_cache/cso_context.h"
+#include "util/u_debug.h"
#include "st_context.h"
#include "st_atom.h"
#include "st_program.h"
-#include "st_atom_shader.h"
-#include "st_mesa_to_tgsi.h"
+#include "st_texture.h"
+#include "st_util.h"
-/**
- * This represents a vertex program, especially translated to match
- * the inputs of a particular fragment shader.
- */
-struct translated_vertex_program
+static unsigned
+get_texture_target(struct gl_context *ctx, const unsigned unit)
{
- struct st_vertex_program *master;
-
- /** The fragment shader "signature" this vertex shader is meant for: */
- GLbitfield frag_inputs;
-
- /** Compared against master vertex program's serialNo: */
- GLuint serialNo;
-
- /** Maps VERT_RESULT_x to slot */
- GLuint output_to_slot[VERT_RESULT_MAX];
- ubyte output_to_semantic_name[VERT_RESULT_MAX];
- ubyte output_to_semantic_index[VERT_RESULT_MAX];
-
- /** Pointer to the translated vertex program */
- struct st_vertex_program *vp;
-
- struct translated_vertex_program *next; /**< next in linked list */
-};
-
-
+ struct gl_texture_object *texObj = _mesa_get_tex_unit(ctx, unit)->_Current;
+ gl_texture_index index;
+
+ if (texObj) {
+ index = _mesa_tex_target_to_index(ctx, texObj->Target);
+ } else {
+ /* fallback for missing texture */
+ index = TEXTURE_2D_INDEX;
+ }
-/**
- * Given a vertex program output attribute, return the corresponding
- * fragment program input attribute.
- * \return -1 for vertex outputs that have no corresponding fragment input
- */
-static GLint
-vp_out_to_fp_in(GLuint vertResult)
-{
- if (vertResult >= VERT_RESULT_TEX0 &&
- vertResult < VERT_RESULT_TEX0 + MAX_TEXTURE_COORD_UNITS)
- return FRAG_ATTRIB_TEX0 + (vertResult - VERT_RESULT_TEX0);
-
- if (vertResult >= VERT_RESULT_VAR0 &&
- vertResult < VERT_RESULT_VAR0 + MAX_VARYING)
- return FRAG_ATTRIB_VAR0 + (vertResult - VERT_RESULT_VAR0);
-
- switch (vertResult) {
- case VERT_RESULT_HPOS:
- return FRAG_ATTRIB_WPOS;
- case VERT_RESULT_COL0:
- return FRAG_ATTRIB_COL0;
- case VERT_RESULT_COL1:
- return FRAG_ATTRIB_COL1;
- case VERT_RESULT_FOGC:
- return FRAG_ATTRIB_FOGC;
+ /* Map mesa texture target to TGSI texture target.
+ * Copied from st_mesa_to_tgsi.c, the shadow part is omitted */
+ switch(index) {
+ case TEXTURE_2D_MULTISAMPLE_INDEX: return TGSI_TEXTURE_2D_MSAA;
+ case TEXTURE_2D_MULTISAMPLE_ARRAY_INDEX: return TGSI_TEXTURE_2D_ARRAY_MSAA;
+ case TEXTURE_BUFFER_INDEX: return TGSI_TEXTURE_BUFFER;
+ case TEXTURE_1D_INDEX: return TGSI_TEXTURE_1D;
+ case TEXTURE_2D_INDEX: return TGSI_TEXTURE_2D;
+ case TEXTURE_3D_INDEX: return TGSI_TEXTURE_3D;
+ case TEXTURE_CUBE_INDEX: return TGSI_TEXTURE_CUBE;
+ case TEXTURE_CUBE_ARRAY_INDEX: return TGSI_TEXTURE_CUBE_ARRAY;
+ case TEXTURE_RECT_INDEX: return TGSI_TEXTURE_RECT;
+ case TEXTURE_1D_ARRAY_INDEX: return TGSI_TEXTURE_1D_ARRAY;
+ case TEXTURE_2D_ARRAY_INDEX: return TGSI_TEXTURE_2D_ARRAY;
+ case TEXTURE_EXTERNAL_INDEX: return TGSI_TEXTURE_2D;
default:
- /* Back-face colors, edge flags, etc */
- return -1;
+ debug_assert(0);
+ return TGSI_TEXTURE_1D;
}
}
/**
- * 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.
+ * Update fragment program state/atom. This involves translating the
+ * Mesa fragment program into a gallium fragment program and binding it.
*/
-static struct translated_vertex_program *
-find_translated_vp(struct st_context *st,
- struct st_vertex_program *stvp,
- struct st_fragment_program *stfp)
+void
+st_update_fp( struct st_context *st )
{
- static const GLuint UNUSED = ~0;
- struct translated_vertex_program *xvp;
- const GLbitfield fragInputsRead = stfp->Base.Base.InputsRead;
-
- /*
- * Translate fragment program if needed.
- */
- if (!stfp->state.tokens) {
- GLuint inAttr, numIn = 0;
+ struct st_program *stfp;
- for (inAttr = 0; inAttr < FRAG_ATTRIB_MAX; inAttr++) {
- if (fragInputsRead & (1 << inAttr)) {
- stfp->input_to_slot[inAttr] = numIn;
- numIn++;
- }
- else {
- stfp->input_to_slot[inAttr] = UNUSED;
+ assert(st->ctx->FragmentProgram._Current);
+ stfp = st_program(st->ctx->FragmentProgram._Current);
+ assert(stfp->Base.Target == GL_FRAGMENT_PROGRAM_ARB);
+
+ void *shader;
+
+ if (st->shader_has_one_variant[MESA_SHADER_FRAGMENT] &&
+ !stfp->ati_fs && /* ATI_fragment_shader always has multiple variants */
+ !stfp->Base.ExternalSamplersUsed && /* external samplers need variants */
+ stfp->variants &&
+ !st_fp_variant(stfp->variants)->key.drawpixels &&
+ !st_fp_variant(stfp->variants)->key.bitmap) {
+ shader = stfp->variants->driver_shader;
+ } else {
+ struct st_fp_variant_key key;
+
+ /* use memset, not an initializer to be sure all memory is zeroed */
+ memset(&key, 0, sizeof(key));
+
+ key.st = st->has_shareable_shaders ? NULL : st;
+
+ key.lower_flatshade = st->lower_flatshade &&
+ st->ctx->Light.ShadeModel == GL_FLAT;
+
+ /* _NEW_COLOR */
+ key.lower_alpha_func = COMPARE_FUNC_NEVER;
+ if (st->lower_alpha_test && _mesa_is_alpha_test_enabled(st->ctx))
+ key.lower_alpha_func = st->ctx->Color.AlphaFunc;
+
+ /* _NEW_LIGHT | _NEW_PROGRAM */
+ key.lower_two_sided_color = st->lower_two_sided_color &&
+ _mesa_vertex_program_two_side_enabled(st->ctx);
+
+ /* gl_driver_flags::NewFragClamp */
+ key.clamp_color = st->clamp_frag_color_in_shader &&
+ st->ctx->Color._ClampFragmentColor;
+
+ /* _NEW_MULTISAMPLE | _NEW_BUFFERS */
+ key.persample_shading =
+ st->force_persample_in_shader &&
+ _mesa_is_multisample_enabled(st->ctx) &&
+ st->ctx->Multisample.SampleShading &&
+ st->ctx->Multisample.MinSampleShadingValue *
+ _mesa_geometric_samples(st->ctx->DrawBuffer) > 1;
+
+ key.lower_depth_clamp =
+ st->clamp_frag_depth_in_shader &&
+ (st->ctx->Transform.DepthClampNear ||
+ st->ctx->Transform.DepthClampFar);
+
+ if (stfp->ati_fs) {
+ key.fog = st->ctx->Fog._PackedEnabledMode;
+
+ for (unsigned u = 0; u < MAX_NUM_FRAGMENT_REGISTERS_ATI; u++) {
+ key.texture_targets[u] = get_texture_target(st->ctx, u);
}
}
- stfp->num_input_slots = numIn;
+ key.external = st_get_external_sampler_key(st, &stfp->Base);
- assert(stfp->Base.Base.NumInstructions > 1);
-
- st_translate_fragment_program(st, stfp, stfp->input_to_slot);
+ shader = st_get_fp_variant(st, stfp, &key)->base.driver_shader;
}
+ st_reference_prog(st, &st->fp, stfp);
+
+ cso_set_fragment_shader_handle(st->cso_context, shader);
+}
+
+
+/**
+ * Update vertex program state/atom. This involves translating the
+ * Mesa vertex program into a gallium fragment program and binding it.
+ */
+void
+st_update_vp( struct st_context *st )
+{
+ struct st_program *stvp;
- /* See if we've got a translated vertex program whose outputs match
- * the fragment program's inputs.
- * XXX This could be a hash lookup, using InputsRead as the key.
+ /* find active shader and params -- Should be covered by
+ * ST_NEW_VERTEX_PROGRAM
*/
- for (xvp = stfp->vertex_programs; xvp; xvp = xvp->next) {
- if (xvp->master == stvp && xvp->frag_inputs == fragInputsRead) {
- break;
- }
+ assert(st->ctx->VertexProgram._Current);
+ stvp = st_program(st->ctx->VertexProgram._Current);
+ assert(stvp->Base.Target == GL_VERTEX_PROGRAM_ARB);
+
+ if (st->shader_has_one_variant[MESA_SHADER_VERTEX] &&
+ stvp->variants &&
+ st_common_variant(stvp->variants)->key.passthrough_edgeflags == st->vertdata_edgeflags &&
+ !st_common_variant(stvp->variants)->key.is_draw_shader) {
+ st->vp_variant = st_common_variant(stvp->variants);
+ } else {
+ struct st_common_variant_key key;
+
+ memset(&key, 0, sizeof(key));
+
+ key.st = st->has_shareable_shaders ? NULL : st;
+
+ /* 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.
+ */
+ key.passthrough_edgeflags = st->vertdata_edgeflags;
+
+ key.clamp_color = st->clamp_vert_color_in_shader &&
+ st->ctx->Light._ClampVertexColor &&
+ (stvp->Base.info.outputs_written &
+ (VARYING_SLOT_COL0 |
+ VARYING_SLOT_COL1 |
+ VARYING_SLOT_BFC0 |
+ VARYING_SLOT_BFC1));
+
+ key.lower_depth_clamp =
+ !st->gp && !st->tep &&
+ st->clamp_frag_depth_in_shader &&
+ (st->ctx->Transform.DepthClampNear ||
+ st->ctx->Transform.DepthClampFar);
+
+ if (key.lower_depth_clamp)
+ key.clip_negative_one_to_one =
+ st->ctx->Transform.ClipDepthMode == GL_NEGATIVE_ONE_TO_ONE;
+
+ /* _NEW_POINT */
+ key.lower_point_size = st->lower_point_size &&
+ !st_point_size_per_vertex(st->ctx);
+
+ /* _NEW_TRANSFORM */
+ if (st->lower_ucp && st_user_clip_planes_enabled(st->ctx))
+ key.lower_ucp = st->ctx->Transform.ClipPlanesEnabled;
+
+ st->vp_variant = st_get_vp_variant(st, stvp, &key);
}
- /* No? Allocate translated vp object now */
- if (!xvp) {
- xvp = ST_CALLOC_STRUCT(translated_vertex_program);
- xvp->frag_inputs = fragInputsRead;
- xvp->master = stvp;
+ st_reference_prog(st, &st->vp, stvp);
+
+ cso_set_vertex_shader_handle(st->cso_context,
+ st->vp_variant->base.driver_shader);
+}
+
+
+static void *
+st_update_common_program(struct st_context *st, struct gl_program *prog,
+ unsigned pipe_shader, struct st_program **dst)
+{
+ struct st_program *stp;
- xvp->next = stfp->vertex_programs;
- stfp->vertex_programs = xvp;
+ if (!prog) {
+ st_reference_prog(st, dst, NULL);
+ return NULL;
}
- /* See if we need to translate vertex program to TGSI form */
- if (xvp->serialNo != stvp->serialNo) {
- GLuint outAttr;
- const GLbitfield outputsWritten = stvp->Base.Base.OutputsWritten;
- GLuint numVpOuts = 0;
- GLboolean emitPntSize = GL_FALSE, emitBFC0 = GL_FALSE, emitBFC1 = GL_FALSE;
- GLbitfield usedGenerics = 0x0;
- GLbitfield usedOutputSlots = 0x0;
-
- /* Compute mapping of vertex program outputs to slots, which depends
- * on the fragment program's input->slot mapping.
- */
- for (outAttr = 0; outAttr < VERT_RESULT_MAX; outAttr++) {
- /* set defaults: */
- xvp->output_to_slot[outAttr] = UNUSED;
- xvp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_COUNT;
- xvp->output_to_semantic_index[outAttr] = 99;
-
- if (outAttr == VERT_RESULT_HPOS) {
- /* always put xformed position into slot zero */
- GLuint slot = 0;
- xvp->output_to_slot[VERT_RESULT_HPOS] = slot;
- xvp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_POSITION;
- xvp->output_to_semantic_index[outAttr] = 0;
- numVpOuts++;
- usedOutputSlots |= (1 << slot);
- }
- else if (outputsWritten & (1 << outAttr)) {
- /* see if the frag prog wants this vert output */
- GLint fpInAttrib = vp_out_to_fp_in(outAttr);
- if (fpInAttrib >= 0) {
- GLuint fpInSlot = stfp->input_to_slot[fpInAttrib];
- if (fpInSlot != ~0) {
- /* match this vp output to the fp input */
- GLuint vpOutSlot = stfp->input_map[fpInSlot];
- xvp->output_to_slot[outAttr] = vpOutSlot;
- xvp->output_to_semantic_name[outAttr] = stfp->input_semantic_name[fpInSlot];
- xvp->output_to_semantic_index[outAttr] = stfp->input_semantic_index[fpInSlot];
- numVpOuts++;
- usedOutputSlots |= (1 << vpOutSlot);
- }
- else {
-#if 0 /*debug*/
- printf("VP output %d not used by FP\n", outAttr);
-#endif
- }
- }
- else if (outAttr == VERT_RESULT_PSIZ)
- emitPntSize = GL_TRUE;
- else if (outAttr == VERT_RESULT_BFC0)
- emitBFC0 = GL_TRUE;
- else if (outAttr == VERT_RESULT_BFC1)
- emitBFC1 = GL_TRUE;
- }
-#if 0 /*debug*/
- printf("assign vp output_to_slot[%d] = %d\n", outAttr,
- xvp->output_to_slot[outAttr]);
-#endif
- }
+ stp = st_program(prog);
+ st_reference_prog(st, dst, stp);
- /* must do these last */
- if (emitPntSize) {
- GLuint slot = numVpOuts++;
- xvp->output_to_slot[VERT_RESULT_PSIZ] = slot;
- xvp->output_to_semantic_name[VERT_RESULT_PSIZ] = TGSI_SEMANTIC_PSIZE;
- xvp->output_to_semantic_index[VERT_RESULT_PSIZ] = 0;
- usedOutputSlots |= (1 << slot);
- }
- if (emitBFC0) {
- GLuint slot = numVpOuts++;
- xvp->output_to_slot[VERT_RESULT_BFC0] = slot;
- xvp->output_to_semantic_name[VERT_RESULT_BFC0] = TGSI_SEMANTIC_COLOR;
- xvp->output_to_semantic_index[VERT_RESULT_BFC0] = 0;
- usedOutputSlots |= (1 << slot);
- }
- if (emitBFC1) {
- GLuint slot = numVpOuts++;
- xvp->output_to_slot[VERT_RESULT_BFC1] = slot;
- xvp->output_to_semantic_name[VERT_RESULT_BFC1] = TGSI_SEMANTIC_COLOR;
- xvp->output_to_semantic_index[VERT_RESULT_BFC1] = 1;
- usedOutputSlots |= (1 << slot);
- }
+ if (st->shader_has_one_variant[prog->info.stage] && stp->variants)
+ return stp->variants->driver_shader;
- /* build usedGenerics mask */
- usedGenerics = 0x0;
- for (outAttr = 0; outAttr < VERT_RESULT_MAX; outAttr++) {
- if (xvp->output_to_semantic_name[outAttr] == TGSI_SEMANTIC_GENERIC) {
- usedGenerics |= (1 << xvp->output_to_semantic_index[outAttr]);
- }
- }
+ struct st_common_variant_key key;
- /* For each vertex program output that doesn't match up to a fragment
- * program input, map the vertex program output to a free slot and
- * free generic attribute.
- */
- for (outAttr = 0; outAttr < VERT_RESULT_MAX; outAttr++) {
- if (outputsWritten & (1 << outAttr)) {
- if (xvp->output_to_slot[outAttr] == UNUSED) {
- GLint freeGeneric = _mesa_ffs(~usedGenerics) - 1;
- GLint freeSlot = _mesa_ffs(~usedOutputSlots) - 1;
- usedGenerics |= (1 << freeGeneric);
- usedOutputSlots |= (1 << freeSlot);
- xvp->output_to_slot[outAttr] = freeSlot;
- xvp->output_to_semantic_name[outAttr] = TGSI_SEMANTIC_GENERIC;
- xvp->output_to_semantic_index[outAttr] = freeGeneric;
- }
- }
+ /* use memset, not an initializer to be sure all memory is zeroed */
+ memset(&key, 0, sizeof(key));
-#if 0 /*debug*/
- printf("vp output_to_slot[%d] = %d\n", outAttr,
- xvp->output_to_slot[outAttr]);
-#endif
- }
+ key.st = st->has_shareable_shaders ? NULL : st;
- assert(stvp->Base.Base.NumInstructions > 1);
+ if (pipe_shader == PIPE_SHADER_GEOMETRY ||
+ pipe_shader == PIPE_SHADER_TESS_EVAL) {
+ key.clamp_color = st->clamp_vert_color_in_shader &&
+ st->ctx->Light._ClampVertexColor &&
+ (stp->Base.info.outputs_written &
+ (VARYING_SLOT_COL0 |
+ VARYING_SLOT_COL1 |
+ VARYING_SLOT_BFC0 |
+ VARYING_SLOT_BFC1));
- st_translate_vertex_program(st, stvp, xvp->output_to_slot,
- xvp->output_to_semantic_name,
- xvp->output_to_semantic_index);
+ key.lower_depth_clamp =
+ (pipe_shader == PIPE_SHADER_GEOMETRY || !st->gp) &&
+ st->clamp_frag_depth_in_shader &&
+ (st->ctx->Transform.DepthClampNear ||
+ st->ctx->Transform.DepthClampFar);
- xvp->vp = stvp;
+ if (key.lower_depth_clamp)
+ key.clip_negative_one_to_one =
+ st->ctx->Transform.ClipDepthMode == GL_NEGATIVE_ONE_TO_ONE;
- /* translated VP is up to date now */
- xvp->serialNo = stvp->serialNo;
}
- return xvp;
+ return st_get_common_variant(st, stp, &key)->driver_shader;
}
void
-st_free_translated_vertex_programs(struct st_context *st,
- struct translated_vertex_program *xvp)
+st_update_gp(struct st_context *st)
{
- struct translated_vertex_program *next;
-
- while (xvp) {
- next = xvp->next;
- _mesa_free(xvp);
- xvp = next;
- }
+ void *shader = st_update_common_program(st,
+ st->ctx->GeometryProgram._Current,
+ PIPE_SHADER_GEOMETRY, &st->gp);
+ cso_set_geometry_shader_handle(st->cso_context, shader);
}
-static void *
-get_passthrough_fs(struct st_context *st)
+void
+st_update_tcp(struct st_context *st)
{
- if (!st->passthrough_fs) {
- st->passthrough_fs =
- util_make_fragment_passthrough_shader(st->pipe);
- }
-
- return st->passthrough_fs;
+ void *shader = st_update_common_program(st,
+ st->ctx->TessCtrlProgram._Current,
+ PIPE_SHADER_TESS_CTRL, &st->tcp);
+ cso_set_tessctrl_shader_handle(st->cso_context, shader);
}
-static void
-update_linkage( struct st_context *st )
+void
+st_update_tep(struct st_context *st)
{
- struct st_vertex_program *stvp;
- struct st_fragment_program *stfp;
- struct translated_vertex_program *xvp;
-
- /* find active shader and params -- Should be covered by
- * ST_NEW_VERTEX_PROGRAM
- */
- assert(st->ctx->VertexProgram._Current);
- stvp = st_vertex_program(st->ctx->VertexProgram._Current);
- assert(stvp->Base.Base.Target == GL_VERTEX_PROGRAM_ARB);
-
- assert(st->ctx->FragmentProgram._Current);
- stfp = st_fragment_program(st->ctx->FragmentProgram._Current);
- assert(stfp->Base.Base.Target == GL_FRAGMENT_PROGRAM_ARB);
-
- xvp = find_translated_vp(st, stvp, stfp);
-
- st_reference_vertprog(st, &st->vp, stvp);
- st_reference_fragprog(st, &st->fp, stfp);
-
- cso_set_vertex_shader_handle(st->cso_context, stvp->driver_shader);
-
- if (st->missing_textures) {
- /* use a pass-through frag shader that uses no textures */
- void *fs = get_passthrough_fs(st);
- cso_set_fragment_shader_handle(st->cso_context, fs);
- }
- else {
- cso_set_fragment_shader_handle(st->cso_context, stfp->driver_shader);
- }
-
- st->vertex_result_to_slot = xvp->output_to_slot;
+ void *shader = st_update_common_program(st,
+ st->ctx->TessEvalProgram._Current,
+ PIPE_SHADER_TESS_EVAL, &st->tep);
+ cso_set_tesseval_shader_handle(st->cso_context, shader);
}
-const struct st_tracked_state st_update_shader = {
- "st_update_shader", /* name */
- { /* dirty */
- 0, /* mesa */
- ST_NEW_VERTEX_PROGRAM | ST_NEW_FRAGMENT_PROGRAM /* st */
- },
- update_linkage /* update */
-};
+void
+st_update_cp(struct st_context *st)
+{
+ void *shader = st_update_common_program(st,
+ st->ctx->ComputeProgram._Current,
+ PIPE_SHADER_COMPUTE, &st->cp);
+ cso_set_compute_shader_handle(st->cso_context, shader);
+}
projects / mesa.git / blobdiff