diff options
70 files changed, 1435 insertions, 595 deletions
diff --git a/docs/GL3.txt b/docs/GL3.txt index 889edefbce..a485878af2 100644 --- a/docs/GL3.txt +++ b/docs/GL3.txt @@ -16,6 +16,7 @@ GLSL changes (GL_EXT_gpu_shader4, etc) not started Conditional rendering (GL_NV_conditional_render) DONE (swrast & softpipe) Map buffer subranges (GL_APPLE_flush_buffer_range) not started Float textures, renderbuffers some infrastructure done + (incl. GL_EXT_packed_float, GL_EXT_shared_exponent) Framebuffer objects (GL_EXT_framebuffer_object) DONE Half-float some infrastructure done Multisample blit DONE diff --git a/progs/tests/stencil_twoside.c b/progs/tests/stencil_twoside.c index 7d871e5877..1010139a20 100644 --- a/progs/tests/stencil_twoside.c +++ b/progs/tests/stencil_twoside.c @@ -26,7 +26,7 @@ * \file stencil_twoside.c * * Simple test of GL_ATI_separate_stencil (or the OGL 2.0 equivalent) functionality. - * Four squares are drawn + * Five squares (or six if stencil wrap is available) are drawn * with different stencil modes, but all should be rendered with the same * final color. */ @@ -37,7 +37,7 @@ #include <GL/glut.h> static int use20syntax = 1; -static int Width = 550; +static int Width = 650; static int Height = 200; static const GLfloat Near = 5.0, Far = 25.0; @@ -70,7 +70,7 @@ static void Display( void ) */ glDisable(GL_STENCIL_TEST); - glTranslatef(-6.0, 0, 0); + glTranslatef(-7.0, 0, 0); glBegin(GL_QUADS); glColor3f( 0.5, 0.5, 0.5 ); glVertex2f(-1, -1); @@ -85,6 +85,9 @@ static void Display( void ) /* Draw the first two squares using incr for the affected face */ + /************************************************************************* + * 2nd square + */ if (use20syntax) { stencil_func_separate(GL_FRONT, GL_ALWAYS, 0, ~0); stencil_func_separate(GL_BACK, GL_ALWAYS, 0, ~0); @@ -98,8 +101,8 @@ static void Display( void ) glTranslatef(3.0, 0, 0); glBegin(GL_QUADS); glColor3f( 0.9, 0.9, 0.9 ); - /* this should be front facing */ for ( i = 0 ; i < (max_stencil + 5) ; i++ ) { + /* this should be front facing */ glVertex2f(-1, -1); glVertex2f( 1, -1); glVertex2f( 1, 1); @@ -107,6 +110,7 @@ static void Display( void ) } glEnd(); + /* stencil vals should be equal to max_stencil */ glStencilFunc(GL_EQUAL, max_stencil, ~0); glBegin(GL_QUADS); glColor3f( 0.5, 0.5, 0.5 ); @@ -116,6 +120,9 @@ static void Display( void ) glVertex2f(-1, 1); glEnd(); + /************************************************************************* + * 3rd square + */ if (use20syntax) { stencil_func_separate(GL_FRONT, GL_ALWAYS, 0, ~0); stencil_func_separate(GL_BACK, GL_ALWAYS, 0, ~0); @@ -129,9 +136,8 @@ static void Display( void ) glTranslatef(3.0, 0, 0); glBegin(GL_QUADS); glColor3f( 0.9, 0.9, 0.9 ); - - /* this should be back facing */ for ( i = 0 ; i < (max_stencil + 5) ; i++ ) { + /* this should be back facing */ glVertex2f(-1, -1); glVertex2f(-1, 1); glVertex2f( 1, 1); @@ -139,6 +145,7 @@ static void Display( void ) } glEnd(); + /* stencil vals should be equal to max_stencil */ glStencilFunc(GL_EQUAL, max_stencil, ~0); glBegin(GL_QUADS); glColor3f( 0.5, 0.5, 0.5 ); @@ -148,6 +155,9 @@ static void Display( void ) glVertex2f(-1, 1); glEnd(); + /************************************************************************* + * 4th square + */ if (use20syntax) { stencil_func_separate(GL_FRONT, GL_NEVER, 0, ~0); stencil_func_separate(GL_BACK, GL_ALWAYS, 0, ~0); @@ -161,15 +171,13 @@ static void Display( void ) glTranslatef(3.0, 0, 0); glBegin(GL_QUADS); glColor3f( 0.9, 0.9, 0.9 ); - - /* this should be back facing */ for ( i = 0 ; i < (max_stencil + 5) ; i++ ) { - /* this should be back facing */ + /* this should be back facing */ glVertex2f(-1, -1); glVertex2f(-1, 1); glVertex2f( 1, 1); glVertex2f( 1, -1); - /* this should be front facing */ + /* this should be front facing */ glVertex2f(-1, -1); glVertex2f( 1, -1); glVertex2f( 1, 1); @@ -177,6 +185,7 @@ static void Display( void ) } glEnd(); + /* stencil vals should be equal to max_stencil */ glStencilFunc(GL_EQUAL, max_stencil, ~0); glBegin(GL_QUADS); glColor3f( 0.5, 0.5, 0.5 ); @@ -186,6 +195,9 @@ static void Display( void ) glVertex2f(-1, 1); glEnd(); + /************************************************************************* + * 5th square + */ if (use20syntax) { stencil_func_separate(GL_FRONT, GL_ALWAYS, 0, ~0); stencil_func_separate(GL_BACK, GL_ALWAYS, 0, ~0); @@ -193,21 +205,19 @@ static void Display( void ) else { stencil_func_separate_ati(GL_ALWAYS, GL_ALWAYS, 0, ~0); } - stencil_op_separate(GL_FRONT, GL_KEEP, GL_KEEP, GL_DECR); - stencil_op_separate(GL_BACK, GL_KEEP, GL_KEEP, GL_INCR); + stencil_op_separate(GL_FRONT, GL_KEEP, GL_KEEP, GL_INCR); + stencil_op_separate(GL_BACK, GL_KEEP, GL_KEEP, GL_DECR); glTranslatef(3.0, 0, 0); glBegin(GL_QUADS); glColor3f( 0.9, 0.9, 0.9 ); - - /* this should be back facing */ for ( i = 0 ; i < (max_stencil + 5) ; i++ ) { - /* this should be back facing */ + /* this should be back facing */ glVertex2f(-1, -1); glVertex2f(-1, 1); glVertex2f( 1, 1); glVertex2f( 1, -1); - /* this should be front facing */ + /* this should be front facing */ glVertex2f(-1, -1); glVertex2f( 1, -1); glVertex2f( 1, 1); @@ -224,6 +234,47 @@ static void Display( void ) glVertex2f(-1, 1); glEnd(); + /************************************************************************* + * 6th square + */ + if (glutExtensionSupported("GL_EXT_stencil_wrap")) { + if (use20syntax) { + stencil_func_separate(GL_FRONT, GL_ALWAYS, 0, ~0); + stencil_func_separate(GL_BACK, GL_ALWAYS, 0, ~0); + } + else { + stencil_func_separate_ati(GL_ALWAYS, GL_ALWAYS, 0, ~0); + } + stencil_op_separate(GL_FRONT, GL_KEEP, GL_KEEP, GL_KEEP); + stencil_op_separate(GL_BACK, GL_KEEP, GL_KEEP, GL_INCR_WRAP); + + glTranslatef(3.0, 0, 0); + glBegin(GL_QUADS); + glColor3f( 0.9, 0.9, 0.9 ); + for ( i = 0 ; i < (max_stencil + 5) ; i++ ) { + /* this should be back facing */ + glVertex2f(-1, -1); + glVertex2f(-1, 1); + glVertex2f( 1, 1); + glVertex2f( 1, -1); + /* this should be front facing */ + glVertex2f(-1, -1); + glVertex2f( 1, -1); + glVertex2f( 1, 1); + glVertex2f(-1, 1); + } + glEnd(); + + glStencilFunc(GL_EQUAL, 260 - 255, ~0); + glBegin(GL_QUADS); + glColor3f( 0.5, 0.5, 0.5 ); + glVertex2f(-1, -1); + glVertex2f( 1, -1); + glVertex2f( 1, 1); + glVertex2f(-1, 1); + glEnd(); + } + glPopMatrix(); glutSwapBuffers(); @@ -278,7 +329,7 @@ static void Init( void ) stencil_func_separate_ati = (PFNGLSTENCILFUNCSEPARATEATIPROC) glutGetProcAddress( "glStencilFuncSeparateATI" ); stencil_op_separate = (PFNGLSTENCILOPSEPARATEPROC) glutGetProcAddress( "glStencilOpSeparate" ); - printf("\nAll 5 squares should be the same color.\n"); + printf("\nAll 5 (or 6) squares should be the same color.\n"); } diff --git a/src/gallium/auxiliary/cso_cache/cso_context.c b/src/gallium/auxiliary/cso_cache/cso_context.c index 4ed9e09c52..d6f8dd34bf 100644 --- a/src/gallium/auxiliary/cso_cache/cso_context.c +++ b/src/gallium/auxiliary/cso_cache/cso_context.c @@ -618,74 +618,6 @@ cso_restore_vertex_samplers(struct cso_context *ctx) } -enum pipe_error cso_set_sampler_textures( struct cso_context *ctx, - uint count, - struct pipe_texture **textures ) -{ - uint i; - - ctx->nr_fragment_sampler_views = count; - - for (i = 0; i < count; i++) { - struct pipe_sampler_view templ, *view; - - u_sampler_view_default_template(&templ, - textures[i], - textures[i]->format); - - view = ctx->pipe->create_sampler_view(ctx->pipe, - textures[i], - &templ); - - pipe_sampler_view_reference(&ctx->fragment_sampler_views[i], view); - } - for ( ; i < PIPE_MAX_SAMPLERS; i++) { - pipe_sampler_view_reference(&ctx->fragment_sampler_views[i], NULL); - } - - ctx->pipe->set_fragment_sampler_views(ctx->pipe, - count, - ctx->fragment_sampler_views); - - return PIPE_OK; -} - -void cso_save_sampler_textures( struct cso_context *ctx ) -{ - uint i; - - ctx->nr_fragment_sampler_views_saved = ctx->nr_fragment_sampler_views; - for (i = 0; i < ctx->nr_fragment_sampler_views; i++) { - assert(!ctx->fragment_sampler_views_saved[i]); - - pipe_sampler_view_reference(&ctx->fragment_sampler_views_saved[i], - ctx->fragment_sampler_views[i]); - } -} - -void cso_restore_sampler_textures( struct cso_context *ctx ) -{ - uint i; - - ctx->nr_fragment_sampler_views = ctx->nr_fragment_sampler_views_saved; - - for (i = 0; i < ctx->nr_fragment_sampler_views; i++) { - pipe_sampler_view_reference(&ctx->fragment_sampler_views[i], NULL); - ctx->fragment_sampler_views[i] = ctx->fragment_sampler_views_saved[i]; - ctx->fragment_sampler_views_saved[i] = NULL; - } - for ( ; i < PIPE_MAX_SAMPLERS; i++) { - pipe_sampler_view_reference(&ctx->fragment_sampler_views[i], NULL); - } - - ctx->pipe->set_fragment_sampler_views(ctx->pipe, - ctx->nr_fragment_sampler_views, - ctx->fragment_sampler_views); - - ctx->nr_fragment_sampler_views_saved = 0; -} - - enum pipe_error cso_set_depth_stencil_alpha(struct cso_context *ctx, const struct pipe_depth_stencil_alpha_state *templ) { diff --git a/src/gallium/auxiliary/cso_cache/cso_context.h b/src/gallium/auxiliary/cso_cache/cso_context.h index a24077e009..d6bcb1fe8f 100644 --- a/src/gallium/auxiliary/cso_cache/cso_context.h +++ b/src/gallium/auxiliary/cso_cache/cso_context.h @@ -103,14 +103,6 @@ void cso_single_vertex_sampler_done(struct cso_context *cso); - -enum pipe_error cso_set_sampler_textures( struct cso_context *cso, - uint count, - struct pipe_texture **textures ); -void cso_save_sampler_textures( struct cso_context *cso ); -void cso_restore_sampler_textures( struct cso_context *cso ); - - enum pipe_error cso_set_vertex_elements(struct cso_context *ctx, unsigned count, const struct pipe_vertex_element *states); diff --git a/src/gallium/auxiliary/gallivm/lp_bld_depth.c b/src/gallium/auxiliary/gallivm/lp_bld_depth.c index e4500e5aef..4ce1a27a06 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_depth.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_depth.c @@ -52,7 +52,14 @@ * Z31 Z32 Z41 Z42 Z33 Z34 Z43 Z44 ... * ... ... ... ... ... ... ... ... ... * - * FIXME: Code generate stencil test + * + * Stencil test: + * Two-sided stencil test is supported but probably not as efficient as + * it could be. Currently, we use if/then/else constructs to do the + * operations for front vs. back-facing polygons. We could probably do + * both the front and back arithmetic then use a Select() instruction to + * choose the result depending on polyon orientation. We'd have to + * measure performance both ways and see which is better. * * @author Jose Fonseca <jfonseca@vmware.com> */ @@ -70,23 +77,35 @@ #include "lp_bld_swizzle.h" +/** Used to select fields from pipe_stencil_state */ +enum stencil_op { + S_FAIL_OP, + Z_FAIL_OP, + Z_PASS_OP +}; + + /** - * Do the stencil test comparison (compare fb Z values against ref value. - * \param stencilVals vector of stencil values from framebuffer + * Do the stencil test comparison (compare FB stencil values against ref value). + * This will be used twice when generating two-sided stencil code. + * \param stencil the front/back stencil state * \param stencilRef the stencil reference value, replicated as a vector - * \return mask of pass/fail values + * \param stencilVals vector of stencil values from framebuffer + * \return vector mask of pass/fail values (~0 or 0) */ static LLVMValueRef -lp_build_stencil_test(struct lp_build_context *bld, - const struct pipe_stencil_state *stencil, - LLVMValueRef stencilRef, - LLVMValueRef stencilVals) +lp_build_stencil_test_single(struct lp_build_context *bld, + const struct pipe_stencil_state *stencil, + LLVMValueRef stencilRef, + LLVMValueRef stencilVals) { const unsigned stencilMax = 255; /* XXX fix */ struct lp_type type = bld->type; LLVMValueRef res; + assert(type.sign); + assert(stencil->enabled); if (stencil->valuemask != stencilMax) { @@ -104,23 +123,101 @@ lp_build_stencil_test(struct lp_build_context *bld, /** + * Do the one or two-sided stencil test comparison. + * \sa lp_build_stencil_test_single + * \param face an integer indicating front (+) or back (-) facing polygon. + * If NULL, assume front-facing. + */ +static LLVMValueRef +lp_build_stencil_test(struct lp_build_context *bld, + const struct pipe_stencil_state stencil[2], + LLVMValueRef stencilRefs[2], + LLVMValueRef stencilVals, + LLVMValueRef face) +{ + LLVMValueRef res; + + assert(stencil[0].enabled); + + if (stencil[1].enabled && face) { + /* do two-sided test */ + struct lp_build_flow_context *flow_ctx; + struct lp_build_if_state if_ctx; + LLVMValueRef front_facing; + LLVMValueRef zero = LLVMConstReal(LLVMFloatType(), 0.0); + LLVMValueRef result = bld->undef; + + flow_ctx = lp_build_flow_create(bld->builder); + lp_build_flow_scope_begin(flow_ctx); + + lp_build_flow_scope_declare(flow_ctx, &result); + + /* front_facing = face > 0.0 */ + front_facing = LLVMBuildFCmp(bld->builder, LLVMRealUGT, face, zero, ""); + + lp_build_if(&if_ctx, flow_ctx, bld->builder, front_facing); + { + result = lp_build_stencil_test_single(bld, &stencil[0], + stencilRefs[0], stencilVals); + } + lp_build_else(&if_ctx); + { + result = lp_build_stencil_test_single(bld, &stencil[1], + stencilRefs[1], stencilVals); + } + lp_build_endif(&if_ctx); + + lp_build_flow_scope_end(flow_ctx); + lp_build_flow_destroy(flow_ctx); + + res = result; + } + else { + /* do single-side test */ + res = lp_build_stencil_test_single(bld, &stencil[0], + stencilRefs[0], stencilVals); + } + + return res; +} + + +/** * Apply the stencil operator (add/sub/keep/etc) to the given vector * of stencil values. * \return new stencil values vector */ static LLVMValueRef -lp_build_stencil_op(struct lp_build_context *bld, - const struct pipe_stencil_state *stencil, - unsigned stencil_op, - LLVMValueRef stencilRef, - LLVMValueRef stencilVals, - LLVMValueRef mask) +lp_build_stencil_op_single(struct lp_build_context *bld, + const struct pipe_stencil_state *stencil, + enum stencil_op op, + LLVMValueRef stencilRef, + LLVMValueRef stencilVals, + LLVMValueRef mask) { const unsigned stencilMax = 255; /* XXX fix */ struct lp_type type = bld->type; LLVMValueRef res; LLVMValueRef max = lp_build_const_int_vec(type, stencilMax); + unsigned stencil_op; + + assert(type.sign); + + switch (op) { + case S_FAIL_OP: + stencil_op = stencil->fail_op; + break; + case Z_FAIL_OP: + stencil_op = stencil->zfail_op; + break; + case Z_PASS_OP: + stencil_op = stencil->zpass_op; + break; + default: + assert(0 && "Invalid stencil_op mode"); + stencil_op = PIPE_STENCIL_OP_KEEP; + } switch (stencil_op) { case PIPE_STENCIL_OP_KEEP: @@ -151,6 +248,7 @@ lp_build_stencil_op(struct lp_build_context *bld, break; case PIPE_STENCIL_OP_INVERT: res = LLVMBuildNot(bld->builder, stencilVals, ""); + res = LLVMBuildAnd(bld->builder, res, max, ""); break; default: assert(0 && "bad stencil op mode"); @@ -174,6 +272,63 @@ lp_build_stencil_op(struct lp_build_context *bld, /** + * Do the one or two-sided stencil test op/update. + */ +static LLVMValueRef +lp_build_stencil_op(struct lp_build_context *bld, + const struct pipe_stencil_state stencil[2], + enum stencil_op op, + LLVMValueRef stencilRefs[2], + LLVMValueRef stencilVals, + LLVMValueRef mask, + LLVMValueRef face) + +{ + assert(stencil[0].enabled); + + if (stencil[1].enabled && face) { + /* do two-sided op */ + struct lp_build_flow_context *flow_ctx; + struct lp_build_if_state if_ctx; + LLVMValueRef front_facing; + LLVMValueRef zero = LLVMConstReal(LLVMFloatType(), 0.0); + LLVMValueRef result = bld->undef; + + flow_ctx = lp_build_flow_create(bld->builder); + lp_build_flow_scope_begin(flow_ctx); + + lp_build_flow_scope_declare(flow_ctx, &result); + + /* front_facing = face > 0.0 */ + front_facing = LLVMBuildFCmp(bld->builder, LLVMRealUGT, face, zero, ""); + + lp_build_if(&if_ctx, flow_ctx, bld->builder, front_facing); + { + result = lp_build_stencil_op_single(bld, &stencil[0], op, + stencilRefs[0], stencilVals, mask); + } + lp_build_else(&if_ctx); + { + result = lp_build_stencil_op_single(bld, &stencil[1], op, + stencilRefs[1], stencilVals, mask); + } + lp_build_endif(&if_ctx); + + lp_build_flow_scope_end(flow_ctx); + lp_build_flow_destroy(flow_ctx); + + return result; + } + else { + /* do single-sided op */ + return lp_build_stencil_op_single(bld, &stencil[0], op, + stencilRefs[0], stencilVals, mask); + } +} + + + +/** * Return a type appropriate for depth/stencil testing. */ struct lp_type @@ -213,36 +368,96 @@ lp_depth_type(const struct util_format_description *format_desc, } -static LLVMValueRef -lp_build_get_stencil_ref(struct lp_build_context *bld, - struct lp_type type, LLVMValueRef stencil_refs_ptr) +/** + * Compute bitmask and bit shift to apply to the incoming fragment Z values + * and the Z buffer values needed before doing the Z comparison. + * + * Note that we leave the Z bits in the position that we find them + * in the Z buffer (typically 0xffffff00 or 0x00ffffff). That lets us + * get by with fewer bit twiddling steps. + */ +static boolean +get_z_shift_and_mask(const struct util_format_description *format_desc, + unsigned *shift, unsigned *mask) +{ + const unsigned total_bits = format_desc->block.bits; + unsigned z_swizzle; + int chan; + unsigned padding_left, padding_right; + + assert(format_desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS); + assert(format_desc->block.width == 1); + assert(format_desc->block.height == 1); + + z_swizzle = format_desc->swizzle[0]; + + if (z_swizzle == UTIL_FORMAT_SWIZZLE_NONE) + return FALSE; + + padding_right = 0; + for (chan = 0; chan < z_swizzle; ++chan) + padding_right += format_desc->channel[chan].size; + + padding_left = + total_bits - (padding_right + format_desc->channel[z_swizzle].size); + + if (padding_left || padding_right) { + unsigned long long mask_left = (1ULL << (total_bits - padding_left)) - 1; + unsigned long long mask_right = (1ULL << (padding_right)) - 1; + *mask = mask_left ^ mask_right; + } + else { + *mask = 0xffffffff; + } + + *shift = padding_left; + + return TRUE; +} + + +/** + * Compute bitmask and bit shift to apply to the framebuffer pixel values + * to put the stencil bits in the least significant position. + * (i.e. 0x000000ff) + */ +static boolean +get_s_shift_and_mask(const struct util_format_description *format_desc, + unsigned *shift, unsigned *mask) { - LLVMValueRef indexes[2], ptr, ref, ref_vec; + unsigned s_swizzle; + int chan, sz; + + s_swizzle = format_desc->swizzle[1]; - /* load 0th element of the array */ - indexes[0] = indexes[1] = LLVMConstInt(LLVMInt32Type(), 0, 0); - ptr = LLVMBuildGEP(bld->builder, stencil_refs_ptr, indexes, 2, ""); - ref = LLVMBuildLoad(bld->builder, ptr, ""); + if (s_swizzle == UTIL_FORMAT_SWIZZLE_NONE) + return FALSE; - /* convert int8 value to i32 */ - ref = LLVMBuildZExt(bld->builder, ref, LLVMIntType(type.width), ""); + *shift = 0; + for (chan = 0; chan < s_swizzle; chan++) + *shift += format_desc->channel[chan].size; - /* make scalar into vector */ - ref_vec = lp_build_broadcast_scalar(bld, ref); + sz = format_desc->channel[s_swizzle].size; + *mask = (1U << sz) - 1U; - return ref_vec; + return TRUE; } + /** * Generate code for performing depth and/or stencil tests. * We operate on a vector of values (typically a 2x2 quad). * + * \param depth the depth test state + * \param stencil the front/back stencil state * \param type the data type of the fragment depth/stencil values * \param format_desc description of the depth/stencil surface - * \param mask the alive/dead pixel mask for the quad - * \param src the incoming depth/stencil values (a 2x2 quad) - * \param dst_ptr the outgoing/updated depth/stencil values + * \param mask the alive/dead pixel mask for the quad (vector) + * \param stencil_refs the front/back stencil ref values (scalar) + * \param z_src the incoming depth/stencil values (a 2x2 quad) + * \param zs_dst_ptr pointer to depth/stencil values in framebuffer + * \param facing contains float value indicating front/back facing polygon */ void lp_build_depth_stencil_test(LLVMBuilderRef builder, @@ -251,119 +466,131 @@ lp_build_depth_stencil_test(LLVMBuilderRef builder, struct lp_type type, const struct util_format_description *format_desc, struct lp_build_mask_context *mask, - LLVMValueRef stencil_refs, + LLVMValueRef stencil_refs[2], LLVMValueRef z_src, - LLVMValueRef zs_dst_ptr) + LLVMValueRef zs_dst_ptr, + LLVMValueRef face) { struct lp_build_context bld; - unsigned z_swizzle, s_swizzle; + struct lp_build_context sbld; + struct lp_type s_type; LLVMValueRef zs_dst, z_dst = NULL; LLVMValueRef stencil_vals = NULL; - LLVMValueRef z_bitmask = NULL, s_bitmask = NULL; + LLVMValueRef z_bitmask = NULL, stencil_shift = NULL; LLVMValueRef z_pass = NULL, s_pass_mask = NULL; LLVMValueRef orig_mask = mask->value; - assert(depth->enabled || stencil[0].enabled); + /* Sanity checking */ + { + const unsigned z_swizzle = format_desc->swizzle[0]; + const unsigned s_swizzle = format_desc->swizzle[1]; - assert(format_desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS); - assert(format_desc->block.width == 1); - assert(format_desc->block.height == 1); + assert(z_swizzle != UTIL_FORMAT_SWIZZLE_NONE || + s_swizzle != UTIL_FORMAT_SWIZZLE_NONE); - z_swizzle = format_desc->swizzle[0]; - s_swizzle = format_desc->swizzle[1]; + assert(depth->enabled || stencil[0].enabled); - assert(z_swizzle != UTIL_FORMAT_SWIZZLE_NONE || - s_swizzle != UTIL_FORMAT_SWIZZLE_NONE); + assert(format_desc->colorspace == UTIL_FORMAT_COLORSPACE_ZS); + assert(format_desc->block.width == 1); + assert(format_desc->block.height == 1); - /* Sanity checking */ - assert(z_swizzle < 4); - assert(format_desc->block.bits == type.width); - if(type.floating) { - assert(z_swizzle == 0); - assert(format_desc->channel[z_swizzle].type == UTIL_FORMAT_TYPE_FLOAT); - assert(format_desc->channel[z_swizzle].size == format_desc->block.bits); - } - else { - assert(format_desc->channel[z_swizzle].type == UTIL_FORMAT_TYPE_UNSIGNED); - assert(format_desc->channel[z_swizzle].normalized); - assert(!type.fixed); - assert(!type.sign); - assert(type.norm); + if (stencil[0].enabled) { + assert(format_desc->format == PIPE_FORMAT_Z24S8_UNORM || + format_desc->format == PIPE_FORMAT_S8Z24_UNORM); + } + + assert(z_swizzle < 4); + assert(format_desc->block.bits == type.width); + if (type.floating) { + assert(z_swizzle == 0); + assert(format_desc->channel[z_swizzle].type == + UTIL_FORMAT_TYPE_FLOAT); + assert(format_desc->channel[z_swizzle].size == + format_desc->block.bits); + } + else { + assert(format_desc->channel[z_swizzle].type == + UTIL_FORMAT_TYPE_UNSIGNED); + assert(format_desc->channel[z_swizzle].normalized); + assert(!type.fixed); + assert(!type.sign); + assert(type.norm); + } } - /* Setup build context */ + + /* Setup build context for Z vals */ lp_build_context_init(&bld, builder, type); + /* Setup build context for stencil vals */ + s_type = lp_type_int_vec(type.width); + lp_build_context_init(&sbld, builder, s_type); + /* Load current z/stencil value from z/stencil buffer */ zs_dst = LLVMBuildLoad(builder, zs_dst_ptr, ""); lp_build_name(zs_dst, "zsbufval"); - /* Align the source depth bits with the destination's, and mask out any - * stencil or padding bits from both */ - if(format_desc->channel[z_swizzle].size == format_desc->block.bits) { - assert(z_swizzle == 0); - z_dst = zs_dst; - } - else { - /* shift/mask bits to right-justify the Z bits */ - unsigned padding_left; - unsigned padding_right; - unsigned chan; - - assert(format_desc->layout == UTIL_FORMAT_LAYOUT_PLAIN); - assert(format_desc->channel[z_swizzle].type == UTIL_FORMAT_TYPE_UNSIGNED); - assert(format_desc->channel[z_swizzle].size <= format_desc->block.bits); - assert(format_desc->channel[z_swizzle].normalized); - - padding_right = 0; - for(chan = 0; chan < z_swizzle; ++chan) - padding_right += format_desc->channel[chan].size; - padding_left = format_desc->block.bits - - (padding_right + format_desc->channel[z_swizzle].size); - - if(padding_left || padding_right) { - const unsigned long long mask_left = (1ULL << (format_desc->block.bits - padding_left)) - 1; - const unsigned long long mask_right = (1ULL << (padding_right)) - 1; - z_bitmask = lp_build_const_int_vec(type, mask_left ^ mask_right); - } - - s_bitmask = LLVMBuildNot(builder, z_bitmask, ""); - stencil_vals = LLVMBuildAnd(builder, zs_dst, s_bitmask, ""); + /* Compute and apply the Z/stencil bitmasks and shifts. + */ + { + unsigned z_shift, z_mask; + unsigned s_shift, s_mask; + + if (get_z_shift_and_mask(format_desc, &z_shift, &z_mask)) { + if (z_shift) { + LLVMValueRef shift = lp_build_const_int_vec(type, z_shift); + z_src = LLVMBuildLShr(builder, z_src, shift, ""); + } + + if (z_mask != 0xffffffff) { + LLVMValueRef mask = lp_build_const_int_vec(type, z_mask); + z_src = LLVMBuildAnd(builder, z_src, mask, ""); + z_dst = LLVMBuildAnd(builder, zs_dst, mask, ""); + z_bitmask = mask; /* used below */ + } + else { + z_dst = zs_dst; + } + + lp_build_name(z_dst, "zsbuf.z"); + } - if(padding_left) - z_src = LLVMBuildLShr(builder, z_src, - lp_build_const_int_vec(type, padding_left), ""); - if(padding_right) - z_src = LLVMBuildAnd(builder, z_src, z_bitmask, ""); - if(padding_left || padding_right) - z_dst = LLVMBuildAnd(builder, zs_dst, z_bitmask, ""); - else - z_dst = zs_dst; + if (get_s_shift_and_mask(format_desc, &s_shift, &s_mask)) { + if (s_shift) { + LLVMValueRef shift = lp_build_const_int_vec(type, s_shift); + stencil_vals = LLVMBuildLShr(builder, zs_dst, shift, ""); + stencil_shift = shift; /* used below */ + } + else { + stencil_vals = zs_dst; + } + + if (s_mask != 0xffffffff) { + LLVMValueRef mask = lp_build_const_int_vec(type, s_mask); + stencil_vals = LLVMBuildAnd(builder, stencil_vals, mask, ""); + } + + lp_build_name(stencil_vals, "stencil"); + } } - lp_build_name(z_dst, "zsbuf.z"); - - /* - printf("build depth %d stencil %d\n", - depth->enabled, - stencil[0].enabled); - */ if (stencil[0].enabled) { - /* Incoming stencil_refs is ptr to int8[2]. Get/convert to int32[4]. */ - stencil_refs = lp_build_get_stencil_ref(&bld, type, stencil_refs); + /* convert scalar stencil refs into vectors */ + stencil_refs[0] = lp_build_broadcast_scalar(&bld, stencil_refs[0]); + stencil_refs[1] = lp_build_broadcast_scalar(&bld, stencil_refs[1]); - s_pass_mask = lp_build_stencil_test(&bld, stencil, - stencil_refs, stencil_vals); + s_pass_mask = lp_build_stencil_test(&sbld, stencil, + stencil_refs, stencil_vals, face); /* apply stencil-fail operator */ { LLVMValueRef s_fail_mask = lp_build_andc(&bld, orig_mask, s_pass_mask); - stencil_vals = lp_build_stencil_op(&bld, stencil, stencil[0].fail_op, + stencil_vals = lp_build_stencil_op(&sbld, stencil, S_FAIL_OP, stencil_refs, stencil_vals, - s_fail_mask); + s_fail_mask, face); } } @@ -394,15 +621,15 @@ lp_build_depth_stencil_test(LLVMBuilderRef builder, /* apply Z-fail operator */ z_fail_mask = lp_build_andc(&bld, orig_mask, z_pass); - stencil_vals = lp_build_stencil_op(&bld, stencil, stencil[0].zfail_op, + stencil_vals = lp_build_stencil_op(&sbld, stencil, Z_FAIL_OP, stencil_refs, stencil_vals, - z_fail_mask); + z_fail_mask, face); /* apply Z-pass operator */ z_pass_mask = LLVMBuildAnd(bld.builder, orig_mask, z_pass, ""); - stencil_vals = lp_build_stencil_op(&bld, stencil, stencil[0].zpass_op, + stencil_vals = lp_build_stencil_op(&sbld, stencil, Z_PASS_OP, stencil_refs, stencil_vals, - z_pass_mask); + z_pass_mask, face); } } else { @@ -410,10 +637,18 @@ lp_build_depth_stencil_test(LLVMBuilderRef builder, * passed the stencil test. */ s_pass_mask = LLVMBuildAnd(bld.builder, orig_mask, s_pass_mask, ""); - stencil_vals = lp_build_stencil_op(&bld, stencil, stencil[0].zpass_op, - stencil_refs, stencil_vals, s_pass_mask); + stencil_vals = lp_build_stencil_op(&sbld, stencil, Z_PASS_OP, + stencil_refs, stencil_vals, + s_pass_mask, face); } + /* The Z bits are already in the right place but we may need to shift the + * stencil bits before ORing Z with Stencil to make the final pixel value. + */ + if (stencil_vals && stencil_shift) + stencil_vals = LLVMBuildShl(bld.builder, stencil_vals, + stencil_shift, ""); + /* Finally, merge/store the z/stencil values */ if ((depth->enabled && depth->writemask) || (stencil[0].enabled && stencil[0].writemask)) { diff --git a/src/gallium/auxiliary/gallivm/lp_bld_depth.h b/src/gallium/auxiliary/gallivm/lp_bld_depth.h index eedc1e419b..27dd46b625 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_depth.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_depth.h @@ -57,9 +57,10 @@ lp_build_depth_stencil_test(LLVMBuilderRef builder, struct lp_type type, const struct util_format_description *format_desc, struct lp_build_mask_context *mask, - LLVMValueRef stencil_refs, + LLVMValueRef stencil_refs[2], LLVMValueRef zs_src, - LLVMValueRef zs_dst_ptr); + LLVMValueRef zs_dst_ptr, + LLVMValueRef facing); #endif /* !LP_BLD_DEPTH_H */ diff --git a/src/gallium/auxiliary/util/u_format.csv b/src/gallium/auxiliary/util/u_format.csv index 96a0fa6550..11243e7349 100644 --- a/src/gallium/auxiliary/util/u_format.csv +++ b/src/gallium/auxiliary/util/u_format.csv @@ -63,6 +63,7 @@ PIPE_FORMAT_A8R8G8B8_UNORM , plain, 1, 1, un8 , un8 , un8 , un8 , yzwx, r PIPE_FORMAT_X8R8G8B8_UNORM , plain, 1, 1, un8 , un8 , un8 , un8 , yzw1, rgb PIPE_FORMAT_A8B8G8R8_UNORM , plain, 1, 1, un8 , un8 , un8 , un8 , wzyx, rgb PIPE_FORMAT_X8B8G8R8_UNORM , plain, 1, 1, un8 , un8 , un8 , un8 , wzy1, rgb +PIPE_FORMAT_B5G5R5X1_UNORM , plain, 1, 1, un5 , un5 , un5 , un1 , zyx1, rgb PIPE_FORMAT_B5G5R5A1_UNORM , plain, 1, 1, un5 , un5 , un5 , un1 , zyxw, rgb PIPE_FORMAT_B4G4R4A4_UNORM , plain, 1, 1, un4 , un4 , un4 , un4 , zyxw, rgb PIPE_FORMAT_B5G6R5_UNORM , plain, 1, 1, un5 , un6 , un5 , , zyx1, rgb diff --git a/src/gallium/auxiliary/util/u_format_tests.c b/src/gallium/auxiliary/util/u_format_tests.c index 182a474044..9d6debcd8c 100644 --- a/src/gallium/auxiliary/util/u_format_tests.c +++ b/src/gallium/auxiliary/util/u_format_tests.c @@ -120,6 +120,13 @@ util_format_test_cases[] = * 16-bit rendertarget formats */ + {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x0000), {0.0, 0.0, 0.0, 0.0}}, + {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x001f), {0.0, 0.0, 1.0, 0.0}}, + {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x03e0), {0.0, 1.0, 0.0, 0.0}}, + {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x7c00), {1.0, 0.0, 0.0, 0.0}}, + {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0x8000), {0.0, 0.0, 0.0, 1.0}}, + {PIPE_FORMAT_B5G5R5X1_UNORM, PACKED_1x16(0x7fff), PACKED_1x16(0xffff), {1.0, 1.0, 1.0, 1.0}}, + {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x0000), {0.0, 0.0, 0.0, 0.0}}, {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x001f), {0.0, 0.0, 1.0, 0.0}}, {PIPE_FORMAT_B5G5R5A1_UNORM, PACKED_1x16(0xffff), PACKED_1x16(0x03e0), {0.0, 1.0, 0.0, 0.0}}, diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c index 61d64cff6d..509d38754f 100644 --- a/src/gallium/auxiliary/util/u_gen_mipmap.c +++ b/src/gallium/auxiliary/util/u_gen_mipmap.c @@ -938,6 +938,7 @@ format_to_type_comps(enum pipe_format pformat, *datatype = DTYPE_UBYTE; *comps = 4; return; + case PIPE_FORMAT_B5G5R5X1_UNORM: case PIPE_FORMAT_B5G5R5A1_UNORM: *datatype = DTYPE_USHORT_1_5_5_5_REV; *comps = 4; diff --git a/src/gallium/auxiliary/util/u_pack_color.h b/src/gallium/auxiliary/util/u_pack_color.h index 50f1b1670b..c5fd7a6783 100644 --- a/src/gallium/auxiliary/util/u_pack_color.h +++ b/src/gallium/auxiliary/util/u_pack_color.h @@ -92,6 +92,11 @@ util_pack_color_ub(ubyte r, ubyte g, ubyte b, ubyte a, uc->us = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3); } return; + case PIPE_FORMAT_B5G5R5X1_UNORM: + { + uc->us = ((0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3); + } + return; case PIPE_FORMAT_B5G5R5A1_UNORM: { uc->us = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3); @@ -216,6 +221,15 @@ util_unpack_color_ub(enum pipe_format format, union util_color *uc, *a = (ubyte) 0xff; } return; + case PIPE_FORMAT_B5G5R5X1_UNORM: + { + ushort p = uc->us; + *r = (ubyte) (((p >> 7) & 0xf8) | ((p >> 12) & 0x7)); + *g = (ubyte) (((p >> 2) & 0xf8) | ((p >> 7) & 0x7)); + *b = (ubyte) (((p << 3) & 0xf8) | ((p >> 2) & 0x7)); + *a = (ubyte) 0xff; + } + return; case PIPE_FORMAT_B5G5R5A1_UNORM: { ushort p = uc->us; @@ -361,6 +375,11 @@ util_pack_color(const float rgba[4], enum pipe_format format, union util_color * uc->us = ((r & 0xf8) << 8) | ((g & 0xfc) << 3) | (b >> 3); } return; + case PIPE_FORMAT_B5G5R5X1_UNORM: + { + uc->us = ((0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3); + } + return; case PIPE_FORMAT_B5G5R5A1_UNORM: { uc->us = ((a & 0x80) << 8) | ((r & 0xf8) << 7) | ((g & 0xf8) << 2) | (b >> 3); diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c index 82e44192aa..09b2382733 100644 --- a/src/gallium/auxiliary/util/u_tile.c +++ b/src/gallium/auxiliary/util/u_tile.c @@ -295,6 +295,55 @@ r8g8b8a8_put_tile_rgba(unsigned *dst, } +/*** PIPE_FORMAT_B5G5R5X1_UNORM ***/ + +static void +x1r5g5b5_get_tile_rgba(const ushort *src, + unsigned w, unsigned h, + float *p, + unsigned dst_stride) +{ + unsigned i, j; + + for (i = 0; i < h; i++) { + float *pRow = p; + for (j = 0; j < w; j++, pRow += 4) { + const ushort pixel = *src++; + pRow[0] = ((pixel >> 10) & 0x1f) * (1.0f / 31.0f); + pRow[1] = ((pixel >> 5) & 0x1f) * (1.0f / 31.0f); + pRow[2] = ((pixel ) & 0x1f) * (1.0f / 31.0f); + pRow[3] = 1.0f; + } + p += dst_stride; + } +} + + +static void +x1r5g5b5_put_tile_rgba(ushort *dst, + unsigned w, unsigned h, + const float *p, + unsigned src_stride) +{ + unsigned i, j; + + for (i = 0; i < h; i++) { + const float *pRow = p; + for (j = 0; j < w; j++, pRow += 4) { + unsigned r, g, b; + r = float_to_ubyte(pRow[0]); + g = float_to_ubyte(pRow[1]); + b = float_to_ubyte(pRow[2]); + r = r >> 3; /* 5 bits */ + g = g >> 3; /* 5 bits */ + b = b >> 3; /* 5 bits */ + *dst++ = (1 << 15) | (r << 10) | (g << 5) | b; + } + p += src_stride; + } +} + + /*** PIPE_FORMAT_B5G5R5A1_UNORM ***/ static void @@ -1174,6 +1223,9 @@ pipe_tile_raw_to_rgba(enum pipe_format format, case PIPE_FORMAT_A8B8G8R8_UNORM: r8g8b8a8_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride); break; + case PIPE_FORMAT_B5G5R5X1_UNORM: + x1r5g5b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride); + break; case PIPE_FORMAT_B5G5R5A1_UNORM: a1r5g5b5_get_tile_rgba((ushort *) src, w, h, dst, dst_stride); break; @@ -1368,6 +1420,9 @@ pipe_put_tile_rgba(struct pipe_context *pipe, case PIPE_FORMAT_A8B8G8R8_UNORM: r8g8b8a8_put_tile_rgba((unsigned *) packed, w, h, p, src_stride); break; + case PIPE_FORMAT_B5G5R5X1_UNORM: + x1r5g5b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride); + break; case PIPE_FORMAT_B5G5R5A1_UNORM: a1r5g5b5_put_tile_rgba((ushort *) packed, w, h, p, src_stride); break; diff --git a/src/gallium/docs/source/context.rst b/src/gallium/docs/source/context.rst index 1f022570db..2b0941010b 100644 --- a/src/gallium/docs/source/context.rst +++ b/src/gallium/docs/source/context.rst @@ -50,6 +50,7 @@ Non-CSO State These pieces of state are too small, variable, and/or trivial to have CSO objects. They all follow simple, one-method binding calls, e.g. ``set_blend_color``. + * ``set_stencil_ref`` sets the stencil front and back reference values which are used as comparison values in stencil test. * ``set_blend_color`` @@ -73,11 +74,11 @@ is being cast to another format. Casting can be done only between compatible formats, that is formats that have matching component order and sizes. Swizzle fields specify they way in which fetched texel components are placed -in the result register. For example, swizzle_r specifies what is going to be -placed in destination register x (AKA r). +in the result register. For example, ``swizzle_r`` specifies what is going to be +placed in first component of result register. -first_level and last_level fields of sampler view template specify the LOD -range the texture is going to be constrained to. +The ``first_level`` and ``last_level`` fields of sampler view template specify +the LOD range the texture is going to be constrained to. * ``set_fragment_sampler_views`` binds an array of sampler views to fragment shader stage. Every binding point acquires a reference @@ -88,7 +89,7 @@ range the texture is going to be constrained to. shader stage. Every binding point acquires a reference to a respective sampler view and releases a reference to the previous sampler view. -* ``create_sampler_view`` creates a new sampler view. texture is associated +* ``create_sampler_view`` creates a new sampler view. ``texture`` is associated with the sampler view which results in sampler view holding a reference to the texture. Format specified in template must be compatible with texture format. diff --git a/src/gallium/docs/source/cso/velems.rst b/src/gallium/docs/source/cso/velems.rst index 8e758fae10..92cde014fb 100644 --- a/src/gallium/docs/source/cso/velems.rst +++ b/src/gallium/docs/source/cso/velems.rst @@ -1,4 +1,4 @@ -.. _vertex,elements +.. _vertexelements: Vertex Elements =============== diff --git a/src/gallium/drivers/llvmpipe/lp_jit.c b/src/gallium/drivers/llvmpipe/lp_jit.c index 1eee9212e6..927e472ff2 100644 --- a/src/gallium/drivers/llvmpipe/lp_jit.c +++ b/src/gallium/drivers/llvmpipe/lp_jit.c @@ -96,7 +96,8 @@ lp_jit_init_globals(struct llvmpipe_screen *screen) elem_types[LP_JIT_CTX_CONSTANTS] = LLVMPointerType(LLVMFloatType(), 0); elem_types[LP_JIT_CTX_ALPHA_REF] = LLVMFloatType(); - elem_types[LP_JIT_CTX_STENCIL_REF] = LLVMArrayType(LLVMInt8Type(), 2); + elem_types[LP_JIT_CTX_STENCIL_REF_FRONT] = LLVMInt32Type(); + elem_types[LP_JIT_CTX_STENCIL_REF_BACK] = LLVMInt32Type(); elem_types[LP_JIT_CTX_SCISSOR_XMIN] = LLVMFloatType(); elem_types[LP_JIT_CTX_SCISSOR_YMIN] = LLVMFloatType(); elem_types[LP_JIT_CTX_SCISSOR_XMAX] = LLVMFloatType(); @@ -113,9 +114,12 @@ lp_jit_init_globals(struct llvmpipe_screen *screen) LP_CHECK_MEMBER_OFFSET(struct lp_jit_context, alpha_ref_value, screen->target, context_type, LP_JIT_CTX_ALPHA_REF); - LP_CHECK_MEMBER_OFFSET(struct lp_jit_context, stencil_ref, + LP_CHECK_MEMBER_OFFSET(struct lp_jit_context, stencil_ref_front, screen->target, context_type, - LP_JIT_CTX_STENCIL_REF); + LP_JIT_CTX_STENCIL_REF_FRONT); + LP_CHECK_MEMBER_OFFSET(struct lp_jit_context, stencil_ref_back, + screen->target, context_type, + LP_JIT_CTX_STENCIL_REF_BACK); LP_CHECK_MEMBER_OFFSET(struct lp_jit_context, scissor_xmin, screen->target, context_type, LP_JIT_CTX_SCISSOR_XMIN); diff --git a/src/gallium/drivers/llvmpipe/lp_jit.h b/src/gallium/drivers/llvmpipe/lp_jit.h index 63e05c5d5e..4930ff02e6 100644 --- a/src/gallium/drivers/llvmpipe/lp_jit.h +++ b/src/gallium/drivers/llvmpipe/lp_jit.h @@ -84,7 +84,7 @@ struct lp_jit_context float alpha_ref_value; - uint8_t stencil_ref[2]; + uint32_t stencil_ref_front, stencil_ref_back; /** floats, not ints */ float scissor_xmin, scissor_ymin, scissor_xmax, scissor_ymax; @@ -103,7 +103,8 @@ struct lp_jit_context enum { LP_JIT_CTX_CONSTANTS = 0, LP_JIT_CTX_ALPHA_REF, - LP_JIT_CTX_STENCIL_REF, + LP_JIT_CTX_STENCIL_REF_FRONT, + LP_JIT_CTX_STENCIL_REF_BACK, LP_JIT_CTX_SCISSOR_XMIN, LP_JIT_CTX_SCISSOR_YMIN, LP_JIT_CTX_SCISSOR_XMAX, @@ -120,8 +121,11 @@ enum { #define lp_jit_context_alpha_ref_value(_builder, _ptr) \ lp_build_struct_get(_builder, _ptr, LP_JIT_CTX_ALPHA_REF, "alpha_ref_value") -#define lp_jit_context_stencil_ref_values(_builder, _ptr) \ - lp_build_struct_get_ptr(_builder, _ptr, LP_JIT_CTX_STENCIL_REF, "stencil_ref") +#define lp_jit_context_stencil_ref_front_value(_builder, _ptr) \ + lp_build_struct_get(_builder, _ptr, LP_JIT_CTX_STENCIL_REF_FRONT, "stencil_ref_front") + +#define lp_jit_context_stencil_ref_back_value(_builder, _ptr) \ + lp_build_struct_get(_builder, _ptr, LP_JIT_CTX_STENCIL_REF_BACK, "stencil_ref_back") #define lp_jit_context_scissor_xmin_value(_builder, _ptr) \ lp_build_struct_get(_builder, _ptr, LP_JIT_CTX_SCISSOR_XMIN, "scissor_xmin") @@ -142,10 +146,16 @@ enum { lp_build_struct_get_ptr(_builder, _ptr, LP_JIT_CONTEXT_TEXTURES, "textures") +/** Indexes into jit_function[] array */ +#define RAST_WHOLE 0 +#define RAST_EDGE_TEST 1 + + typedef void (*lp_jit_frag_func)(const struct lp_jit_context *context, uint32_t x, uint32_t y, + float facing, const void *a0, const void *dadx, const void *dady, diff --git a/src/gallium/drivers/llvmpipe/lp_rast.c b/src/gallium/drivers/llvmpipe/lp_rast.c index 81ea11a16b..3a51800c40 100644 --- a/src/gallium/drivers/llvmpipe/lp_rast.c +++ b/src/gallium/drivers/llvmpipe/lp_rast.c @@ -312,15 +312,16 @@ lp_rast_shade_tile(struct lp_rasterizer_task *task, depth = lp_rast_depth_pointer(rast, tile_x + x, tile_y + y); /* run shader */ - state->jit_function[0]( &state->jit_context, - tile_x + x, tile_y + y, - inputs->a0, - inputs->dadx, - inputs->dady, - color, - depth, - INT_MIN, INT_MIN, INT_MIN, - NULL, NULL, NULL ); + state->jit_function[RAST_WHOLE]( &state->jit_context, + tile_x + x, tile_y + y, + inputs->facing, + inputs->a0, + inputs->dadx, + inputs->dady, + color, + depth, + INT_MIN, INT_MIN, INT_MIN, + NULL, NULL, NULL ); } } } @@ -375,15 +376,18 @@ void lp_rast_shade_quads( struct lp_rasterizer_task *task, assert(lp_check_alignment(inputs->step[2], 16)); /* run shader */ - state->jit_function[1]( &state->jit_context, - x, y, - inputs->a0, - inputs->dadx, - inputs->dady, - color, - depth, - c1, c2, c3, - inputs->step[0], inputs->step[1], inputs->step[2]); + state->jit_function[RAST_EDGE_TEST]( &state->jit_context, + x, y, + inputs->facing, + inputs->a0, + inputs->dadx, + inputs->dady, + color, + depth, + c1, c2, c3, + inputs->step[0], + inputs->step[1], + inputs->step[2]); } diff --git a/src/gallium/drivers/llvmpipe/lp_rast.h b/src/gallium/drivers/llvmpipe/lp_rast.h index 303f6e3f7e..ae838f3fbe 100644 --- a/src/gallium/drivers/llvmpipe/lp_rast.h +++ b/src/gallium/drivers/llvmpipe/lp_rast.h @@ -82,6 +82,8 @@ struct lp_rast_state { * These pointers point into the bin data buffer. */ struct lp_rast_shader_inputs { + float facing; /** Positive for front-facing, negative for back-facing */ + float (*a0)[4]; float (*dadx)[4]; float (*dady)[4]; diff --git a/src/gallium/drivers/llvmpipe/lp_rast_priv.h b/src/gallium/drivers/llvmpipe/lp_rast_priv.h index 39bf2c2587..6ee9bcaae3 100644 --- a/src/gallium/drivers/llvmpipe/lp_rast_priv.h +++ b/src/gallium/drivers/llvmpipe/lp_rast_priv.h @@ -195,6 +195,7 @@ lp_rast_shade_quads_all( struct lp_rasterizer_task *task, /* run shader */ state->jit_function[0]( &state->jit_context, x, y, + inputs->facing, inputs->a0, inputs->dadx, inputs->dady, diff --git a/src/gallium/drivers/llvmpipe/lp_setup.c b/src/gallium/drivers/llvmpipe/lp_setup.c index bcc9d1fc1a..fbb0d6f8a6 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup.c +++ b/src/gallium/drivers/llvmpipe/lp_setup.c @@ -407,10 +407,10 @@ lp_setup_set_stencil_ref_values( struct lp_setup_context *setup, { LP_DBG(DEBUG_SETUP, "%s %d %d\n", __FUNCTION__, refs[0], refs[1]); - if (setup->fs.current.jit_context.stencil_ref[0] != refs[0] || - setup->fs.current.jit_context.stencil_ref[1] != refs[1]) { - setup->fs.current.jit_context.stencil_ref[0] = refs[0]; - setup->fs.current.jit_context.stencil_ref[1] = refs[1]; + if (setup->fs.current.jit_context.stencil_ref_front != refs[0] || + setup->fs.current.jit_context.stencil_ref_back != refs[1]) { + setup->fs.current.jit_context.stencil_ref_front = refs[0]; + setup->fs.current.jit_context.stencil_ref_back = refs[1]; setup->dirty |= LP_SETUP_NEW_FS; } } diff --git a/src/gallium/drivers/llvmpipe/lp_setup_tri.c b/src/gallium/drivers/llvmpipe/lp_setup_tri.c index ac6264dc73..ce689d3d56 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup_tri.c +++ b/src/gallium/drivers/llvmpipe/lp_setup_tri.c @@ -361,6 +361,8 @@ do_triangle_ccw(struct lp_setup_context *setup, */ setup_tri_coefficients( setup, tri, oneoverarea, v1, v2, v3, frontfacing ); + tri->inputs.facing = frontfacing ? 1.0F : -1.0F; + /* half-edge constants, will be interated over the whole render target. */ tri->c1 = tri->dy12 * x1 - tri->dx12 * y1; diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c index b38e0f393d..7bbf348e0b 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_fs.c +++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c @@ -145,13 +145,17 @@ generate_depth_stencil(LLVMBuilderRef builder, const struct lp_fragment_shader_variant_key *key, struct lp_type src_type, struct lp_build_mask_context *mask, - LLVMValueRef stencil_refs, + LLVMValueRef stencil_refs[2], LLVMValueRef src, - LLVMValueRef dst_ptr) + LLVMValueRef dst_ptr, + LLVMValueRef facing) { const struct util_format_description *format_desc; struct lp_type dst_type; + if (!key->depth.enabled && !key->stencil[0].enabled && !key->stencil[1].enabled) + return; + format_desc = util_format_description(key->zsbuf_format); assert(format_desc); @@ -190,7 +194,8 @@ generate_depth_stencil(LLVMBuilderRef builder, mask, stencil_refs, src, - dst_ptr); + dst_ptr, + facing); } @@ -390,6 +395,7 @@ generate_fs(struct llvmpipe_context *lp, LLVMValueRef *pmask, LLVMValueRef (*color)[4], LLVMValueRef depth_ptr, + LLVMValueRef facing, unsigned do_tri_test, LLVMValueRef c0, LLVMValueRef c1, @@ -405,7 +411,7 @@ generate_fs(struct llvmpipe_context *lp, LLVMValueRef consts_ptr; LLVMValueRef outputs[PIPE_MAX_SHADER_OUTPUTS][NUM_CHANNELS]; LLVMValueRef z = interp->pos[2]; - LLVMValueRef stencil_refs; + LLVMValueRef stencil_refs[2]; struct lp_build_flow_context *flow; struct lp_build_mask_context mask; boolean early_depth_stencil_test; @@ -415,7 +421,8 @@ generate_fs(struct llvmpipe_context *lp, assert(i < 4); - stencil_refs = lp_jit_context_stencil_ref_values(builder, context_ptr); + stencil_refs[0] = lp_jit_context_stencil_ref_front_value(builder, context_ptr); + stencil_refs[1] = lp_jit_context_stencil_ref_back_value(builder, context_ptr); elem_type = lp_build_elem_type(type); vec_type = lp_build_vec_type(type); @@ -465,7 +472,7 @@ generate_fs(struct llvmpipe_context *lp, if (early_depth_stencil_test) generate_depth_stencil(builder, key, type, &mask, - stencil_refs, z, depth_ptr); + stencil_refs, z, depth_ptr, facing); lp_build_tgsi_soa(builder, tokens, type, &mask, consts_ptr, interp->pos, interp->inputs, @@ -512,7 +519,7 @@ generate_fs(struct llvmpipe_context *lp, if (!early_depth_stencil_test) generate_depth_stencil(builder, key, type, &mask, - stencil_refs, z, depth_ptr); + stencil_refs, z, depth_ptr, facing); lp_build_mask_end(&mask); @@ -623,7 +630,7 @@ generate_fragment(struct llvmpipe_context *lp, LLVMTypeRef fs_int_vec_type; LLVMTypeRef blend_vec_type; LLVMTypeRef blend_int_vec_type; - LLVMTypeRef arg_types[14]; + LLVMTypeRef arg_types[15]; LLVMTypeRef func_type; LLVMTypeRef int32_vec4_type = lp_build_int32_vec4_type(); LLVMValueRef context_ptr; @@ -646,6 +653,7 @@ generate_fragment(struct llvmpipe_context *lp, LLVMValueRef blend_mask; LLVMValueRef blend_in_color[NUM_CHANNELS]; LLVMValueRef function; + LLVMValueRef facing; unsigned num_fs; unsigned i; unsigned chan; @@ -685,20 +693,21 @@ generate_fragment(struct llvmpipe_context *lp, arg_types[0] = screen->context_ptr_type; /* context */ arg_types[1] = LLVMInt32Type(); /* x */ arg_types[2] = LLVMInt32Type(); /* y */ - arg_types[3] = LLVMPointerType(fs_elem_type, 0); /* a0 */ - arg_types[4] = LLVMPointerType(fs_elem_type, 0); /* dadx */ - arg_types[5] = LLVMPointerType(fs_elem_type, 0); /* dady */ - arg_types[6] = LLVMPointerType(LLVMPointerType(blend_vec_type, 0), 0); /* color */ - arg_types[7] = LLVMPointerType(fs_int_vec_type, 0); /* depth */ - arg_types[8] = LLVMInt32Type(); /* c0 */ - arg_types[9] = LLVMInt32Type(); /* c1 */ - arg_types[10] = LLVMInt32Type(); /* c2 */ + arg_types[3] = LLVMFloatType(); /* facing */ + arg_types[4] = LLVMPointerType(fs_elem_type, 0); /* a0 */ + arg_types[5] = LLVMPointerType(fs_elem_type, 0); /* dadx */ + arg_types[6] = LLVMPointerType(fs_elem_type, 0); /* dady */ + arg_types[7] = LLVMPointerType(LLVMPointerType(blend_vec_type, 0), 0); /* color */ + arg_types[8] = LLVMPointerType(fs_int_vec_type, 0); /* depth */ + arg_types[9] = LLVMInt32Type(); /* c0 */ + arg_types[10] = LLVMInt32Type(); /* c1 */ + arg_types[11] = LLVMInt32Type(); /* c2 */ /* Note: the step arrays are built as int32[16] but we interpret * them here as int32_vec4[4]. */ - arg_types[11] = LLVMPointerType(int32_vec4_type, 0);/* step0 */ - arg_types[12] = LLVMPointerType(int32_vec4_type, 0);/* step1 */ - arg_types[13] = LLVMPointerType(int32_vec4_type, 0);/* step2 */ + arg_types[12] = LLVMPointerType(int32_vec4_type, 0);/* step0 */ + arg_types[13] = LLVMPointerType(int32_vec4_type, 0);/* step1 */ + arg_types[14] = LLVMPointerType(int32_vec4_type, 0);/* step2 */ func_type = LLVMFunctionType(LLVMVoidType(), arg_types, Elements(arg_types), 0); @@ -718,17 +727,18 @@ generate_fragment(struct llvmpipe_context *lp, context_ptr = LLVMGetParam(function, 0); x = LLVMGetParam(function, 1); y = LLVMGetParam(function, 2); - a0_ptr = LLVMGetParam(function, 3); - dadx_ptr = LLVMGetParam(function, 4); - dady_ptr = LLVMGetParam(function, 5); - color_ptr_ptr = LLVMGetParam(function, 6); - depth_ptr = LLVMGetParam(function, 7); - c0 = LLVMGetParam(function, 8); - c1 = LLVMGetParam(function, 9); - c2 = LLVMGetParam(function, 10); - step0_ptr = LLVMGetParam(function, 11); - step1_ptr = LLVMGetParam(function, 12); - step2_ptr = LLVMGetParam(function, 13); + facing = LLVMGetParam(function, 3); + a0_ptr = LLVMGetParam(function, 4); + dadx_ptr = LLVMGetParam(function, 5); + dady_ptr = LLVMGetParam(function, 6); + color_ptr_ptr = LLVMGetParam(function, 7); + depth_ptr = LLVMGetParam(function, 8); + c0 = LLVMGetParam(function, 9); + c1 = LLVMGetParam(function, 10); + c2 = LLVMGetParam(function, 11); + step0_ptr = LLVMGetParam(function, 12); + step1_ptr = LLVMGetParam(function, 13); + step2_ptr = LLVMGetParam(function, 14); lp_build_name(context_ptr, "context"); lp_build_name(x, "x"); @@ -787,6 +797,7 @@ generate_fragment(struct llvmpipe_context *lp, &fs_mask[i], /* output */ out_color, depth_ptr_i, + facing, do_tri_test, c0, c1, c2, step0_ptr, step1_ptr, step2_ptr); @@ -1166,6 +1177,7 @@ llvmpipe_update_fs(struct llvmpipe_context *lp) opaque = !key.blend.logicop_enable && !key.blend.rt[0].blend_enable && key.blend.rt[0].colormask == 0xf && + !key.stencil[0].enabled && !key.alpha.enabled && !key.depth.enabled && !key.scissor && @@ -1173,7 +1185,7 @@ llvmpipe_update_fs(struct llvmpipe_context *lp) ? TRUE : FALSE; lp_setup_set_fs_functions(lp->setup, - shader->current->jit_function[0], - shader->current->jit_function[1], + shader->current->jit_function[RAST_WHOLE], + shader->current->jit_function[RAST_EDGE_TEST], opaque); } diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c index d994a46ccf..a55d2db039 100644 --- a/src/gallium/drivers/r300/r300_context.c +++ b/src/gallium/drivers/r300/r300_context.c @@ -72,29 +72,23 @@ static void r300_destroy_context(struct pipe_context* context) } static unsigned int -r300_is_texture_referenced(struct pipe_context *pipe, +r300_is_texture_referenced(struct pipe_context *context, struct pipe_texture *texture, unsigned face, unsigned level) { - return 0; + struct r300_context* r300 = r300_context(context); + struct r300_texture* tex = (struct r300_texture*)texture; + + return r300->rws->is_buffer_referenced(r300->rws, tex->buffer); } static unsigned int -r300_is_buffer_referenced(struct pipe_context *pipe, +r300_is_buffer_referenced(struct pipe_context *context, struct pipe_buffer *buf) { - /* This only checks to see whether actual hardware buffers are - * referenced. Since we use managed BOs and transfers, it's actually not - * possible for pipe_buffers to ever reference the actual hardware, so - * buffers are never referenced. - */ - - /* XXX: that doesn't make sense given that - * r300_is_texture_referenced is implemented on top of this - * function and hardware can certainly refer to textures - * directly... - */ - return 0; + struct r300_context* r300 = r300_context(context); + + return r300_buffer_is_referenced(r300, buf); } static void r300_flush_cb(void *data) @@ -202,6 +196,8 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen, r300_setup_atoms(r300); + r300->sprite_coord_index = -1; + /* Open up the OQ BO. */ r300->oqbo = screen->buffer_create(screen, 4096, PIPE_BUFFER_USAGE_VERTEX, 4096); diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h index 0c8fb6860e..edebeab86f 100644 --- a/src/gallium/drivers/r300/r300_context.h +++ b/src/gallium/drivers/r300/r300_context.h @@ -98,6 +98,16 @@ struct r300_rs_state { uint32_t line_stipple_value; /* R300_GA_LINE_STIPPLE_VALUE: 0x4260 */ uint32_t color_control; /* R300_GA_COLOR_CONTROL: 0x4278 */ uint32_t polygon_mode; /* R300_GA_POLY_MODE: 0x4288 */ + + /* Specifies top of Raster pipe specific enable controls, + * i.e. texture coordinates stuffing for points, lines, triangles */ + uint32_t stuffing_enable; /* R300_GB_ENABLE: 0x4008 */ + + /* Point sprites texture coordinates, 0: lower left, 1: upper right */ + float point_texcoord_left; /* R300_GA_POINT_S0: 0x4200 */ + float point_texcoord_bottom; /* R300_GA_POINT_T0: 0x4204 */ + float point_texcoord_right; /* R300_GA_POINT_S1: 0x4208 */ + float point_texcoord_top; /* R300_GA_POINT_T1: 0x420c */ }; struct r300_rs_block { @@ -390,6 +400,9 @@ struct r300_context { uint32_t zbuffer_bpp; /* Whether scissor is enabled. */ boolean scissor_enabled; + /* Point sprites texcoord index, -1 = unused. */ + int sprite_coord_index; + /* upload managers */ struct u_upload_mgr *upload_vb; struct u_upload_mgr *upload_ib; diff --git a/src/gallium/drivers/r300/r300_emit.c b/src/gallium/drivers/r300/r300_emit.c index 3ad0e561bc..c897df628d 100644 --- a/src/gallium/drivers/r300/r300_emit.c +++ b/src/gallium/drivers/r300/r300_emit.c @@ -616,6 +616,12 @@ void r300_emit_rs_state(struct r300_context* r300, unsigned size, void* state) OUT_CS_REG(R300_GA_LINE_STIPPLE_CONFIG, rs->line_stipple_config); OUT_CS_REG(R300_GA_LINE_STIPPLE_VALUE, rs->line_stipple_value); OUT_CS_REG(R300_GA_POLY_MODE, rs->polygon_mode); + OUT_CS_REG(R300_GB_ENABLE, rs->stuffing_enable); + OUT_CS_REG_SEQ(R300_GA_POINT_S0, 4); + OUT_CS_32F(rs->point_texcoord_left); + OUT_CS_32F(rs->point_texcoord_bottom); + OUT_CS_32F(rs->point_texcoord_right); + OUT_CS_32F(rs->point_texcoord_top); END_CS; } @@ -1044,9 +1050,11 @@ validate: } /* ...textures... */ for (i = 0; i < texstate->count; i++) { - tex = (struct r300_texture*)texstate->fragment_sampler_views[i]->texture; - if (!tex || !texstate->sampler_states[i]) + if (!(texstate->tx_enable & (1 << i))) { continue; + } + + tex = (struct r300_texture*)texstate->fragment_sampler_views[i]->texture; if (!r300_add_texture(r300->rws, tex, RADEON_GEM_DOMAIN_GTT | RADEON_GEM_DOMAIN_VRAM, 0)) { r300->context.flush(&r300->context, 0, NULL); diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c index 47100c83b0..2fed263a7a 100644 --- a/src/gallium/drivers/r300/r300_render.c +++ b/src/gallium/drivers/r300/r300_render.c @@ -141,7 +141,7 @@ static boolean immd_is_good_idea(struct r300_context *r300, unsigned vertex_element_count = r300->velems->count; unsigned i, vbi; - if (count > 4) { + if (count > 10) { return FALSE; } @@ -155,8 +155,7 @@ static boolean immd_is_good_idea(struct r300_context *r300, if (!checked[vbi]) { vbuf = &r300->vertex_buffer[vbi]; - if (r300_buffer_is_referenced(r300, - vbuf->buffer)) { + if (r300_buffer_is_referenced(r300, vbuf->buffer)) { /* It's a very bad idea to map it... */ return FALSE; } @@ -304,10 +303,9 @@ static void r300_emit_draw_elements(struct r300_context *r300, #endif CS_LOCALS(r300); - assert((start * indexSize) % 4 == 0); assert(count < (1 << 24)); - maxIndex = MIN3(maxIndex, r300->vertex_buffer_max_index, count - minIndex); + maxIndex = MIN2(maxIndex, r300->vertex_buffer_max_index); DBG(r300, DBG_DRAW, "r300: Indexbuf of %u indices, min %u max %u\n", count, minIndex, maxIndex); @@ -354,6 +352,7 @@ static void r300_emit_draw_elements(struct r300_context *r300, static void r300_shorten_ubyte_elts(struct r300_context* r300, struct pipe_buffer** elts, + unsigned start, unsigned count) { struct pipe_screen* screen = r300->context.screen; @@ -371,6 +370,8 @@ static void r300_shorten_ubyte_elts(struct r300_context* r300, in_map = pipe_buffer_map(screen, *elts, PIPE_BUFFER_USAGE_CPU_READ); out_map = pipe_buffer_map(screen, new_elts, PIPE_BUFFER_USAGE_CPU_WRITE); + in_map += start; + for (i = 0; i < count; i++) { *out_map = (unsigned short)*in_map; in_map++; @@ -383,6 +384,32 @@ static void r300_shorten_ubyte_elts(struct r300_context* r300, *elts = new_elts; } +static void r300_align_ushort_elts(struct r300_context *r300, + struct pipe_buffer **elts, + unsigned start, unsigned count) +{ + struct pipe_screen* screen = r300->context.screen; + struct pipe_buffer* new_elts; + unsigned short *in_map; + unsigned short *out_map; + + new_elts = screen->buffer_create(screen, 32, + PIPE_BUFFER_USAGE_INDEX | + PIPE_BUFFER_USAGE_CPU_WRITE | + PIPE_BUFFER_USAGE_GPU_READ, + 2 * count); + + in_map = pipe_buffer_map(screen, *elts, PIPE_BUFFER_USAGE_CPU_READ); + out_map = pipe_buffer_map(screen, new_elts, PIPE_BUFFER_USAGE_CPU_WRITE); + + memcpy(out_map, in_map+start, 2 * count); + + pipe_buffer_unmap(screen, *elts); + pipe_buffer_unmap(screen, new_elts); + + *elts = new_elts; +} + /* This is the fast-path drawing & emission for HW TCL. */ void r300_draw_range_elements(struct pipe_context* pipe, struct pipe_buffer* indexBuffer, @@ -408,8 +435,12 @@ void r300_draw_range_elements(struct pipe_context* pipe, } if (indexSize == 1) { - r300_shorten_ubyte_elts(r300, &indexBuffer, count); + r300_shorten_ubyte_elts(r300, &indexBuffer, start, count); indexSize = 2; + start = 0; + } else if (indexSize == 2 && start % 2 != 0) { + r300_align_ushort_elts(r300, &indexBuffer, start, count); + start = 0; } r300_update_derived_state(r300); diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c index d7d654dc31..f396b42e95 100644 --- a/src/gallium/drivers/r300/r300_state.c +++ b/src/gallium/drivers/r300/r300_state.c @@ -715,6 +715,7 @@ static void* r300_create_rs_state(struct pipe_context* pipe, { struct r300_screen* r300screen = r300_screen(pipe->screen); struct r300_rs_state* rs = CALLOC_STRUCT(r300_rs_state); + unsigned coord_index; /* Copy rasterizer state for Draw. */ rs->rs = *state; @@ -807,6 +808,32 @@ static void* r300_create_rs_state(struct pipe_context* pipe, rs->color_control = R300_SHADE_MODEL_SMOOTH; } + /* Point sprites */ + if (state->sprite_coord_enable) { + coord_index = ffs(state->sprite_coord_enable)-1; + + SCREEN_DBG(r300screen, DBG_DRAW, + "r300: point sprite: shader coord=%d\n", coord_index); + + rs->stuffing_enable = + R300_GB_POINT_STUFF_ENABLE | + R300_GB_TEX_ST << (R300_GB_TEX0_SOURCE_SHIFT + (coord_index*2)); + + rs->point_texcoord_left = 0.0f; + rs->point_texcoord_right = 1.0f; + + switch (state->sprite_coord_mode) { + case PIPE_SPRITE_COORD_UPPER_LEFT: + rs->point_texcoord_top = 0.0f; + rs->point_texcoord_bottom = 1.0f; + break; + case PIPE_SPRITE_COORD_LOWER_LEFT: + rs->point_texcoord_top = 1.0f; + rs->point_texcoord_bottom = 0.0f; + break; + } + } + return (void*)rs; } @@ -816,6 +843,7 @@ static void r300_bind_rs_state(struct pipe_context* pipe, void* state) struct r300_context* r300 = r300_context(pipe); struct r300_rs_state* rs = (struct r300_rs_state*)state; boolean scissor_was_enabled = r300->scissor_enabled; + int last_sprite_coord_index = r300->sprite_coord_index; if (r300->draw) { draw_flush(r300->draw); @@ -825,17 +853,22 @@ static void r300_bind_rs_state(struct pipe_context* pipe, void* state) if (rs) { r300->polygon_offset_enabled = rs->rs.offset_cw || rs->rs.offset_ccw; r300->scissor_enabled = rs->rs.scissor; + r300->sprite_coord_index = ffs(rs->rs.sprite_coord_enable)-1; } else { r300->polygon_offset_enabled = FALSE; r300->scissor_enabled = FALSE; + r300->sprite_coord_index = -1; } UPDATE_STATE(state, r300->rs_state); - r300->rs_state.size = 17 + (r300->polygon_offset_enabled ? 5 : 0); + r300->rs_state.size = 24 + (r300->polygon_offset_enabled ? 5 : 0); if (scissor_was_enabled != r300->scissor_enabled) { r300->scissor_state.dirty = TRUE; } + if (last_sprite_coord_index != r300->sprite_coord_index) { + r300->rs_block_state.dirty = TRUE; + } } /* Free rasterizer state. */ @@ -940,6 +973,7 @@ static void r300_set_fragment_sampler_views(struct pipe_context* pipe, struct r300_context* r300 = r300_context(pipe); struct r300_textures_state* state = (struct r300_textures_state*)r300->textures_state.state; + struct r300_texture *texture; unsigned i; boolean is_r500 = r300_screen(r300->context.screen)->caps->is_r500; boolean dirty_tex = FALSE; @@ -951,15 +985,18 @@ static void r300_set_fragment_sampler_views(struct pipe_context* pipe, for (i = 0; i < count; i++) { if (state->fragment_sampler_views[i] != views[i]) { - struct r300_texture *texture; - pipe_sampler_view_reference(&state->fragment_sampler_views[i], views[i]); - dirty_tex = TRUE; - texture = (struct r300_texture *)views[i]->texture; + if (!views[i]) { + continue; + } + + /* A new sampler view (= texture)... */ + dirty_tex = TRUE; /* R300-specific - set the texrect factor in the fragment shader */ + texture = (struct r300_texture *)views[i]->texture; if (!is_r500 && texture->is_npot) { /* XXX It would be nice to re-emit just 1 constant, * XXX not all of them */ @@ -1002,7 +1039,6 @@ r300_create_sampler_view(struct pipe_context *pipe, return view; } - static void r300_sampler_view_destroy(struct pipe_context *pipe, struct pipe_sampler_view *view) @@ -1072,26 +1108,62 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe, const struct pipe_vertex_buffer* buffers) { struct r300_context* r300 = r300_context(pipe); - int i; - unsigned max_index = (1 << 24) - 1; - boolean any_user_buffer = false; + struct pipe_vertex_buffer *vbo; + unsigned i, max_index = (1 << 24) - 1; + boolean any_user_buffer = FALSE; if (count == r300->vertex_buffer_count && - memcmp(r300->vertex_buffer, buffers, count * sizeof(buffers[0])) == 0) + memcmp(r300->vertex_buffer, buffers, + sizeof(struct pipe_vertex_buffer) * count) == 0) { return; + } + /* Check if the stride is aligned to the size of DWORD. */ for (i = 0; i < count; i++) { - pipe_buffer_reference(&r300->vertex_buffer[i].buffer, buffers[i].buffer); - if (r300_buffer_is_user_buffer(buffers[i].buffer)) - any_user_buffer = true; - max_index = MIN2(buffers[i].max_index, max_index); + if (buffers[i].buffer) { + if (buffers[i].stride % 4 != 0) { + // XXX Shouldn't we align the buffer? + fprintf(stderr, "r300_set_vertex_buffers: " + "Unaligned buffer stride %i isn't supported.\n", + buffers[i].stride); + assert(0); + abort(); + } + } } - for ( ; i < r300->vertex_buffer_count; i++) - pipe_buffer_reference(&r300->vertex_buffer[i].buffer, NULL); + for (i = 0; i < count; i++) { + /* Why, yes, I AM casting away constness. How did you know? */ + vbo = (struct pipe_vertex_buffer*)&buffers[i]; + + /* Reference our buffer. */ + pipe_buffer_reference(&r300->vertex_buffer[i].buffer, vbo->buffer); + + /* Skip NULL buffers */ + if (!buffers[i].buffer) { + continue; + } + + if (r300_buffer_is_user_buffer(vbo->buffer)) { + any_user_buffer = TRUE; + } + + if (vbo->max_index == ~0) { + /* Bogus value from broken state tracker; hax it. */ + vbo->max_index = + (vbo->buffer->size - vbo->buffer_offset) / vbo->stride; + } + + max_index = MIN2(vbo->max_index, max_index); + } + + for (; i < r300->vertex_buffer_count; i++) { + /* Dereference any old buffers. */ + pipe_buffer_reference(&r300->vertex_buffer[i].buffer, NULL); + } memcpy(r300->vertex_buffer, buffers, - sizeof(struct pipe_vertex_buffer) * count); + sizeof(struct pipe_vertex_buffer) * count); r300->vertex_buffer_count = count; r300->vertex_buffer_max_index = max_index; @@ -1103,22 +1175,6 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe, } } -static boolean r300_validate_aos(struct r300_context *r300) -{ - struct pipe_vertex_buffer *vbuf = r300->vertex_buffer; - struct pipe_vertex_element *velem = r300->velems->velem; - int i; - - /* Check if formats and strides are aligned to the size of DWORD. */ - for (i = 0; i < r300->velems->count; i++) { - if (vbuf[velem[i].vertex_buffer_index].stride % 4 != 0 || - util_format_get_blocksize(velem[i].src_format) % 4 != 0) { - return FALSE; - } - } - return TRUE; -} - static void r300_draw_emit_attrib(struct r300_context* r300, enum attrib_emit emit, enum interp_mode interp, @@ -1288,6 +1344,7 @@ static void* r300_create_vertex_elements_state(struct pipe_context* pipe, struct r300_context *r300 = r300_context(pipe); struct r300_screen* r300screen = r300_screen(pipe->screen); struct r300_vertex_element_state *velems; + unsigned i, size; assert(count <= PIPE_MAX_ATTRIBS); velems = CALLOC_STRUCT(r300_vertex_element_state); @@ -1296,6 +1353,20 @@ static void* r300_create_vertex_elements_state(struct pipe_context* pipe, memcpy(velems->velem, attribs, sizeof(struct pipe_vertex_element) * count); if (r300screen->caps->has_tcl) { + /* Check if the format is aligned to the size of DWORD. */ + for (i = 0; i < count; i++) { + size = util_format_get_blocksize(attribs[i].src_format); + + if (size % 4 != 0) { + /* XXX Shouldn't we align the format? */ + fprintf(stderr, "r300_create_vertex_elements_state: " + "Unaligned format %s:%i isn't supported\n", + util_format_name(attribs[i].src_format), size); + assert(0); + abort(); + } + } + r300_vertex_psc(velems); } else { memset(&r300->vertex_info, 0, sizeof(struct vertex_info)); @@ -1324,12 +1395,6 @@ static void r300_bind_vertex_elements_state(struct pipe_context *pipe, draw_set_vertex_elements(r300->draw, velems->count, velems->velem); } - if (!r300_validate_aos(r300)) { - /* XXX We should fallback using draw. */ - assert(0); - abort(); - } - UPDATE_STATE(&velems->vertex_stream, r300->vertex_stream_state); r300->vertex_stream_state.size = (1 + velems->vertex_stream.count) * 2; } diff --git a/src/gallium/drivers/r300/r300_state_derived.c b/src/gallium/drivers/r300/r300_state_derived.c index 7947ec6641..8178d55dc9 100644 --- a/src/gallium/drivers/r300/r300_state_derived.c +++ b/src/gallium/drivers/r300/r300_state_derived.c @@ -178,7 +178,8 @@ static void r300_update_rs_block(struct r300_context* r300, /* Rasterize texture coordinates. */ for (i = 0; i < ATTR_GENERIC_COUNT; i++) { - if (vs_outputs->generic[i] != ATTR_UNUSED) { + if (vs_outputs->generic[i] != ATTR_UNUSED || + r300->sprite_coord_index == i) { /* Always rasterize if it's written by the VS, * otherwise it locks up. */ rX00_rs_tex(&rs, tex_count, tex_count, FALSE); @@ -186,6 +187,8 @@ static void r300_update_rs_block(struct r300_context* r300, /* Write it to the FS input register if it's used by the FS. */ if (fs_inputs->generic[i] != ATTR_UNUSED) { rX00_rs_tex_write(&rs, tex_count, fp_offset); + if (r300->sprite_coord_index == i) + debug_printf("r300: SpriteCoord (generic index %i) is being written to reg %i\n", i, fp_offset); fp_offset++; } tex_count++; @@ -332,20 +335,25 @@ static void r300_merge_textures_and_samplers(struct r300_context* r300) (struct r300_textures_state*)r300->textures_state.state; struct r300_texture_sampler_state *texstate; struct r300_sampler_state *sampler; + struct pipe_sampler_view *view; struct r300_texture *tex; unsigned min_level, max_level, i, size; unsigned count = MIN2(state->texture_count, state->sampler_count); state->tx_enable = 0; + state->count = 0; size = 2; for (i = 0; i < count; i++) { if (state->fragment_sampler_views[i] && state->sampler_states[i]) { state->tx_enable |= 1 << i; - tex = (struct r300_texture *)state->fragment_sampler_views[i]->texture; + view = state->fragment_sampler_views[i]; + tex = (struct r300_texture *)view->texture; sampler = state->sampler_states[i]; + assert(view->format == tex->tex.format); + texstate = &state->regs[i]; memcpy(texstate->format, &tex->state, sizeof(uint32_t)*3); texstate->filter[0] = sampler->filter0; @@ -367,8 +375,10 @@ static void r300_merge_textures_and_samplers(struct r300_context* r300) } else { /* determine min/max levels */ /* the MAX_MIP level is the largest (finest) one */ - max_level = MIN2(sampler->max_lod, tex->tex.last_level); - min_level = MIN2(sampler->min_lod, max_level); + max_level = MIN3(sampler->max_lod + view->first_level, + tex->tex.last_level, view->last_level); + min_level = MIN2(sampler->min_lod + view->first_level, + max_level); texstate->format[0] |= R300_TX_NUM_LEVELS(max_level); texstate->filter[0] |= R300_TX_MAX_MIP_LEVEL(min_level); } diff --git a/src/gallium/drivers/r300/r300_state_invariant.c b/src/gallium/drivers/r300/r300_state_invariant.c index 4a2c68269b..2d9a63d29a 100644 --- a/src/gallium/drivers/r300/r300_state_invariant.c +++ b/src/gallium/drivers/r300/r300_state_invariant.c @@ -44,13 +44,9 @@ void r300_emit_invariant_state(struct r300_context* r300, struct r300_capabilities* caps = r300_screen(r300->context.screen)->caps; CS_LOCALS(r300); - BEGIN_CS(14 + (caps->has_tcl ? 2: 0)); + BEGIN_CS(12 + (caps->has_tcl ? 2: 0)); /*** Graphics Backend (GB) ***/ - /* Various GB enables */ - OUT_CS_REG(R300_GB_ENABLE, R300_GB_POINT_STUFF_ENABLE | - R300_GB_LINE_STUFF_ENABLE | - R300_GB_TRIANGLE_STUFF_ENABLE); /* Subpixel multisampling for AA * These are commented out because glisse's CS checker doesn't like them. * I presume these will be re-enabled later. @@ -78,7 +74,7 @@ void r300_emit_invariant_state(struct r300_context* r300, END_CS; /* XXX unsorted stuff from surface_fill */ - BEGIN_CS(44 + (caps->has_tcl ? 7 : 0) + + BEGIN_CS(40 + (caps->has_tcl ? 7 : 0) + (caps->family >= CHIP_FAMILY_RV350 ? 4 : 0)); if (caps->has_tcl) { @@ -90,11 +86,6 @@ void r300_emit_invariant_state(struct r300_context* r300, OUT_CS_32F(1.0); OUT_CS_32F(1.0); } - /* XXX point tex stuffing */ - OUT_CS_REG_SEQ(R300_GA_POINT_S0, 1); - OUT_CS_32F(0.0); - OUT_CS_REG_SEQ(R300_GA_POINT_S1, 1); - OUT_CS_32F(1.0); /* XXX line tex stuffing */ OUT_CS_REG_SEQ(R300_GA_LINE_S0, 1); OUT_CS_32F(0.0); diff --git a/src/gallium/drivers/r300/r300_texture.c b/src/gallium/drivers/r300/r300_texture.c index 7c7656068b..cb53619fe5 100644 --- a/src/gallium/drivers/r300/r300_texture.c +++ b/src/gallium/drivers/r300/r300_texture.c @@ -302,7 +302,6 @@ static uint32_t r300_translate_colorformat(enum pipe_format format) case PIPE_FORMAT_A8_UNORM: case PIPE_FORMAT_I8_UNORM: case PIPE_FORMAT_L8_UNORM: - case PIPE_FORMAT_L8_SRGB: case PIPE_FORMAT_R8_UNORM: case PIPE_FORMAT_R8_SNORM: return R300_COLOR_FORMAT_I8; @@ -311,24 +310,19 @@ static uint32_t r300_translate_colorformat(enum pipe_format format) case PIPE_FORMAT_B5G6R5_UNORM: return R300_COLOR_FORMAT_RGB565; case PIPE_FORMAT_B5G5R5A1_UNORM: + case PIPE_FORMAT_B5G5R5X1_UNORM: return R300_COLOR_FORMAT_ARGB1555; case PIPE_FORMAT_B4G4R4A4_UNORM: return R300_COLOR_FORMAT_ARGB4444; /* 32-bit buffers. */ case PIPE_FORMAT_B8G8R8A8_UNORM: - case PIPE_FORMAT_B8G8R8A8_SRGB: case PIPE_FORMAT_B8G8R8X8_UNORM: - case PIPE_FORMAT_B8G8R8X8_SRGB: case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_A8R8G8B8_SRGB: case PIPE_FORMAT_X8R8G8B8_UNORM: - case PIPE_FORMAT_X8R8G8B8_SRGB: case PIPE_FORMAT_A8B8G8R8_UNORM: case PIPE_FORMAT_R8G8B8A8_SNORM: - case PIPE_FORMAT_A8B8G8R8_SRGB: case PIPE_FORMAT_X8B8G8R8_UNORM: - case PIPE_FORMAT_X8B8G8R8_SRGB: case PIPE_FORMAT_R8SG8SB8UX8U_NORM: return R300_COLOR_FORMAT_ARGB8888; case PIPE_FORMAT_R10G10B10A2_UNORM: @@ -393,12 +387,7 @@ static uint32_t r300_translate_out_fmt(enum pipe_format format) desc = util_format_description(format); /* Specifies how the shader output is written to the fog unit. */ - if (desc->colorspace == UTIL_FORMAT_COLORSPACE_SRGB) { - /* The gamma correction causes precision loss so we need - * higher precision to maintain reasonable quality. - * It has nothing to do with the colorbuffer format. */ - modifier |= R300_US_OUT_FMT_C4_10_GAMMA; - } else if (desc->channel[0].type == UTIL_FORMAT_TYPE_FLOAT) { + if (desc->channel[0].type == UTIL_FORMAT_TYPE_FLOAT) { if (desc->channel[0].size == 32) { modifier |= R300_US_OUT_FMT_C4_32_FP; } else { @@ -428,46 +417,39 @@ static uint32_t r300_translate_out_fmt(enum pipe_format format) return modifier | R300_C2_SEL_A; case PIPE_FORMAT_I8_UNORM: case PIPE_FORMAT_L8_UNORM: - case PIPE_FORMAT_L8_SRGB: case PIPE_FORMAT_R8_UNORM: case PIPE_FORMAT_R8_SNORM: return modifier | R300_C2_SEL_R; - /* ARGB 32-bit outputs. */ + /* BGRA outputs. */ case PIPE_FORMAT_B5G6R5_UNORM: case PIPE_FORMAT_B5G5R5A1_UNORM: + case PIPE_FORMAT_B5G5R5X1_UNORM: case PIPE_FORMAT_B4G4R4A4_UNORM: case PIPE_FORMAT_B8G8R8A8_UNORM: - case PIPE_FORMAT_B8G8R8A8_SRGB: case PIPE_FORMAT_B8G8R8X8_UNORM: - case PIPE_FORMAT_B8G8R8X8_SRGB: return modifier | R300_C0_SEL_B | R300_C1_SEL_G | R300_C2_SEL_R | R300_C3_SEL_A; - /* BGRA 32-bit outputs. */ + /* ARGB outputs. */ case PIPE_FORMAT_A8R8G8B8_UNORM: - case PIPE_FORMAT_A8R8G8B8_SRGB: case PIPE_FORMAT_X8R8G8B8_UNORM: - case PIPE_FORMAT_X8R8G8B8_SRGB: return modifier | R300_C0_SEL_A | R300_C1_SEL_R | R300_C2_SEL_G | R300_C3_SEL_B; - /* RGBA 32-bit outputs. */ + /* ABGR outputs. */ case PIPE_FORMAT_A8B8G8R8_UNORM: - case PIPE_FORMAT_R8G8B8A8_SNORM: - case PIPE_FORMAT_A8B8G8R8_SRGB: case PIPE_FORMAT_X8B8G8R8_UNORM: - case PIPE_FORMAT_X8B8G8R8_SRGB: return modifier | R300_C0_SEL_A | R300_C1_SEL_B | R300_C2_SEL_G | R300_C3_SEL_R; - /* ABGR 32-bit outputs. */ + /* RGBA outputs. */ + case PIPE_FORMAT_R8G8B8A8_SNORM: case PIPE_FORMAT_R8SG8SB8UX8U_NORM: case PIPE_FORMAT_R10G10B10A2_UNORM: - /* RGBA high precision outputs (same swizzles as ABGR low precision) */ case PIPE_FORMAT_R16G16B16A16_UNORM: case PIPE_FORMAT_R16G16B16A16_SNORM: //case PIPE_FORMAT_R16G16B16A16_FLOAT: /* not in pipe_format */ diff --git a/src/gallium/include/pipe/p_format.h b/src/gallium/include/pipe/p_format.h index cbf3273ec8..ba2985f449 100644 --- a/src/gallium/include/pipe/p_format.h +++ b/src/gallium/include/pipe/p_format.h @@ -157,6 +157,7 @@ enum pipe_format { PIPE_FORMAT_DXT5_SRGBA = 109, PIPE_FORMAT_A8B8G8R8_UNORM = 110, + PIPE_FORMAT_B5G5R5X1_UNORM = 111, PIPE_FORMAT_COUNT }; diff --git a/src/gallium/state_trackers/es/Makefile b/src/gallium/state_trackers/es/Makefile index e33685d247..089d441167 100644 --- a/src/gallium/state_trackers/es/Makefile +++ b/src/gallium/state_trackers/es/Makefile @@ -38,6 +38,7 @@ SYS_LIBS = -lm -pthread INCLUDE_DIRS = \ + -I$(TOP)/include \ -I$(TOP)/src/mesa \ -I$(TOP)/src/gallium/include diff --git a/src/gallium/state_trackers/vega/api_filters.c b/src/gallium/state_trackers/vega/api_filters.c index 18e2cc1f25..a643f38624 100644 --- a/src/gallium/state_trackers/vega/api_filters.c +++ b/src/gallium/state_trackers/vega/api_filters.c @@ -40,6 +40,7 @@ #include "util/u_format.h" #include "util/u_memory.h" +#include "util/u_sampler.h" #include "asm_filters.h" @@ -53,7 +54,7 @@ struct filter_info { const void *const_buffer; VGint const_buffer_len; VGTilingMode tiling_mode; - struct pipe_texture *extra_texture; + struct pipe_sampler_view *extra_texture_view; }; static INLINE struct pipe_texture *create_texture_1d(struct vg_context *ctx, @@ -91,13 +92,35 @@ static INLINE struct pipe_texture *create_texture_1d(struct vg_context *ctx, return tex; } +static INLINE struct pipe_sampler_view *create_texture_1d_view(struct vg_context *ctx, + const VGuint *color_data, + const VGint color_data_len) +{ + struct pipe_context *pipe = ctx->pipe; + struct pipe_texture *texture; + struct pipe_sampler_view view_templ; + struct pipe_sampler_view *view; + + texture = create_texture_1d(ctx, color_data, color_data_len); + + if (!texture) + return NULL; + + u_sampler_view_default_template(&view_templ, texture, texture->format); + view = pipe->create_sampler_view(pipe, texture, &view_templ); + /* want the texture to go away if the view is freed */ + pipe_texture_reference(&texture, NULL); + + return view; +} + static INLINE struct pipe_surface * setup_framebuffer(struct vg_image *dst) { struct vg_context *ctx = vg_current_context(); struct pipe_context *pipe = ctx->pipe; struct pipe_framebuffer_state fb; struct pipe_surface *dst_surf = pipe->screen->get_tex_surface( - pipe->screen, dst->texture, 0, 0, 0, + pipe->screen, dst->sampler_view->texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); /* drawing dest */ @@ -168,7 +191,7 @@ static void setup_constant_buffer(struct vg_context *ctx, const void *buffer, static void setup_samplers(struct vg_context *ctx, struct filter_info *info) { struct pipe_sampler_state *samplers[PIPE_MAX_SAMPLERS]; - struct pipe_texture *textures[PIPE_MAX_SAMPLERS]; + struct pipe_sampler_view *sampler_views[PIPE_MAX_SAMPLERS]; struct pipe_sampler_state sampler[3]; int num_samplers = 0; int num_textures = 0; @@ -177,10 +200,10 @@ static void setup_samplers(struct vg_context *ctx, struct filter_info *info) samplers[1] = NULL; samplers[2] = NULL; samplers[3] = NULL; - textures[0] = NULL; - textures[1] = NULL; - textures[2] = NULL; - textures[3] = NULL; + sampler_views[0] = NULL; + sampler_views[1] = NULL; + sampler_views[2] = NULL; + sampler_views[3] = NULL; memset(&sampler[0], 0, sizeof(struct pipe_sampler_state)); sampler[0].wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; @@ -215,21 +238,21 @@ static void setup_samplers(struct vg_context *ctx, struct filter_info *info) } samplers[0] = &sampler[0]; - textures[0] = info->src->texture; + sampler_views[0] = info->src->sampler_view; ++num_samplers; ++num_textures; - if (info->extra_texture) { + if (info->extra_texture_view) { memcpy(&sampler[1], &sampler[0], sizeof(struct pipe_sampler_state)); samplers[1] = &sampler[1]; - textures[1] = info->extra_texture; + sampler_views[1] = info->extra_texture_view; ++num_samplers; ++num_textures; } cso_set_samplers(ctx->cso_context, num_samplers, (const struct pipe_sampler_state **)samplers); - cso_set_sampler_textures(ctx->cso_context, num_textures, textures); + cso_set_fragment_sampler_views(ctx->cso_context, num_textures, sampler_views); } static struct vg_shader * setup_color_matrix(struct vg_context *ctx, void *user_data) @@ -308,7 +331,7 @@ static void execute_filter(struct vg_context *ctx, cso_save_viewport(ctx->cso_context); cso_save_blend(ctx->cso_context); cso_save_samplers(ctx->cso_context); - cso_save_sampler_textures(ctx->cso_context); + cso_save_fragment_sampler_views(ctx->cso_context); dst_surf = setup_framebuffer(info->dst); setup_viewport(info->dst); @@ -318,7 +341,7 @@ static void execute_filter(struct vg_context *ctx, setup_samplers(ctx, info); renderer_draw_texture(ctx->renderer, - info->src->texture, + info->src->sampler_view->texture, info->dst->x, info->dst->y, info->dst->x + info->dst->width, info->dst->y + info->dst->height, @@ -331,7 +354,7 @@ static void execute_filter(struct vg_context *ctx, cso_restore_viewport(ctx->cso_context); cso_restore_blend(ctx->cso_context); cso_restore_samplers(ctx->cso_context); - cso_restore_sampler_textures(ctx->cso_context); + cso_restore_fragment_sampler_views(ctx->cso_context); vg_shader_destroy(ctx, shader); @@ -369,7 +392,7 @@ void vgColorMatrix(VGImage dst, VGImage src, info.const_buffer = matrix; info.const_buffer_len = 20 * sizeof(VGfloat); info.tiling_mode = VG_TILE_PAD; - info.extra_texture = 0; + info.extra_texture_view = NULL; execute_filter(ctx, &info); } @@ -479,7 +502,7 @@ void vgConvolve(VGImage dst, VGImage src, info.const_buffer = buffer; info.const_buffer_len = buffer_len * sizeof(VGfloat); info.tiling_mode = tilingMode; - info.extra_texture = 0; + info.extra_texture_view = NULL; execute_filter(ctx, &info); free(buffer); @@ -669,7 +692,7 @@ void vgGaussianBlur(VGImage dst, VGImage src, info.const_buffer = buffer; info.const_buffer_len = buffer_len * sizeof(VGfloat); info.tiling_mode = tilingMode; - info.extra_texture = 0; + info.extra_texture_view = NULL; execute_filter(ctx, &info); free(buffer); @@ -688,7 +711,7 @@ void vgLookup(VGImage dst, VGImage src, struct vg_image *d, *s; VGuint color_data[256]; VGint i; - struct pipe_texture *lut_texture; + struct pipe_sampler_view *lut_texture_view; VGfloat buffer[4]; struct filter_info info; @@ -714,7 +737,7 @@ void vgLookup(VGImage dst, VGImage src, color_data[i] = blueLUT[i] << 24 | greenLUT[i] << 16 | redLUT[i] << 8 | alphaLUT[i]; } - lut_texture = create_texture_1d(ctx, color_data, 255); + lut_texture_view = create_texture_1d_view(ctx, color_data, 255); buffer[0] = 0.f; buffer[1] = 0.f; @@ -728,11 +751,11 @@ void vgLookup(VGImage dst, VGImage src, info.const_buffer = buffer; info.const_buffer_len = 4 * sizeof(VGfloat); info.tiling_mode = VG_TILE_PAD; - info.extra_texture = lut_texture; + info.extra_texture_view = lut_texture_view; execute_filter(ctx, &info); - pipe_texture_reference(&lut_texture, NULL); + pipe_sampler_view_reference(&lut_texture_view, NULL); } void vgLookupSingle(VGImage dst, VGImage src, @@ -743,7 +766,7 @@ void vgLookupSingle(VGImage dst, VGImage src, { struct vg_context *ctx = vg_current_context(); struct vg_image *d, *s; - struct pipe_texture *lut_texture; + struct pipe_sampler_view *lut_texture_view; VGfloat buffer[4]; struct filter_info info; VGuint color_data[256]; @@ -783,7 +806,7 @@ void vgLookupSingle(VGImage dst, VGImage src, color_data[i] = blue << 24 | green << 16 | red << 8 | alpha; } - lut_texture = create_texture_1d(ctx, color_data, 256); + lut_texture_view = create_texture_1d_view(ctx, color_data, 256); buffer[0] = 0.f; buffer[1] = 0.f; @@ -797,9 +820,9 @@ void vgLookupSingle(VGImage dst, VGImage src, info.const_buffer = buffer; info.const_buffer_len = 4 * sizeof(VGfloat); info.tiling_mode = VG_TILE_PAD; - info.extra_texture = lut_texture; + info.extra_texture_view = lut_texture_view; execute_filter(ctx, &info); - pipe_texture_reference(&lut_texture, NULL); + pipe_sampler_view_reference(&lut_texture_view, NULL); } diff --git a/src/gallium/state_trackers/vega/image.c b/src/gallium/state_trackers/vega/image.c index a71579cd26..c3268a84a6 100644 --- a/src/gallium/state_trackers/vega/image.c +++ b/src/gallium/state_trackers/vega/image.c @@ -43,6 +43,7 @@ #include "util/u_tile.h" #include "util/u_memory.h" #include "util/u_math.h" +#include "util/u_sampler.h" static enum pipe_format vg_format_to_pipe(VGImageFormat format) { @@ -81,7 +82,7 @@ static INLINE void vg_sync_size(VGfloat *src_loc, VGfloat *dst_loc) static void vg_copy_texture(struct vg_context *ctx, struct pipe_texture *dst, VGint dx, VGint dy, - struct pipe_texture *src, VGint sx, VGint sy, + struct pipe_sampler_view *src, VGint sx, VGint sy, VGint width, VGint height) { VGfloat dst_loc[4], src_loc[4]; @@ -103,8 +104,8 @@ static void vg_copy_texture(struct vg_context *ctx, src_loc[3] = height; src_bounds[0] = 0.f; src_bounds[1] = 0.f; - src_bounds[2] = src->width0; - src_bounds[3] = src->height0; + src_bounds[2] = src->texture->width0; + src_bounds[3] = src->texture->height0; vg_bound_rect(src_loc, src_bounds, src_shift); vg_bound_rect(dst_loc, dst_bounds, dst_shift); @@ -218,7 +219,7 @@ void vg_copy_surface(struct vg_context *ctx, static struct pipe_texture *image_texture(struct vg_image *img) { - struct pipe_texture *tex = img->texture; + struct pipe_texture *tex = img->sampler_view->texture; return tex; } @@ -247,9 +248,12 @@ struct vg_image * image_create(VGImageFormat format, VGint width, VGint height) { struct vg_context *ctx = vg_current_context(); + struct pipe_context *pipe = ctx->pipe; struct vg_image *image = CALLOC_STRUCT(vg_image); enum pipe_format pformat = vg_format_to_pipe(format); struct pipe_texture pt, *newtex; + struct pipe_sampler_view view_templ; + struct pipe_sampler_view *view; struct pipe_screen *screen = ctx->pipe->screen; vg_init_object(&image->base, ctx, VG_OBJECT_IMAGE); @@ -281,7 +285,12 @@ struct vg_image * image_create(VGImageFormat format, debug_assert(newtex); - image->texture = newtex; + u_sampler_view_default_template(&view_templ, newtex, newtex->format); + view = pipe->create_sampler_view(pipe, newtex, &view_templ); + /* want the texture to go away if the view is freed */ + pipe_texture_reference(&newtex, NULL); + + image->sampler_view = view; vg_context_add_object(ctx, VG_OBJECT_IMAGE, image); @@ -345,7 +354,7 @@ void image_destroy(struct vg_image *img) array_destroy(img->children_array); } - pipe_texture_reference(&img->texture, NULL); + pipe_sampler_view_reference(&img->sampler_view, NULL); free(img); } @@ -444,7 +453,7 @@ void image_get_sub_data(struct vg_image * image, { struct pipe_transfer *transfer = pipe->get_tex_transfer(pipe, - image->texture, 0, 0, 0, + image->sampler_view->texture, 0, 0, 0, PIPE_TRANSFER_READ, 0, 0, image->x + image->width, @@ -478,9 +487,9 @@ struct vg_image * image_child_image(struct vg_image *parent, image->width = width; image->height = height; image->parent = parent; - image->texture = 0; - pipe_texture_reference(&image->texture, - parent->texture); + image->sampler_view = NULL; + pipe_sampler_view_reference(&image->sampler_view, + parent->sampler_view); image->sampler.wrap_s = PIPE_TEX_WRAP_CLAMP_TO_EDGE; image->sampler.wrap_t = PIPE_TEX_WRAP_CLAMP_TO_EDGE; @@ -514,8 +523,8 @@ void image_copy(struct vg_image *dst, VGint dx, VGint dy, } /* make sure rendering has completed */ ctx->pipe->flush(ctx->pipe, PIPE_FLUSH_RENDER_CACHE, NULL); - vg_copy_texture(ctx, dst->texture, dst->x + dx, dst->y + dy, - src->texture, src->x + sx, src->y + sy, width, height); + vg_copy_texture(ctx, dst->sampler_view->texture, dst->x + dx, dst->y + dy, + src->sampler_view, src->x + sx, src->y + sy, width, height); } void image_draw(struct vg_image *img) @@ -624,12 +633,12 @@ VGboolean vg_image_overlaps(struct vg_image *dst, } VGint image_bind_samplers(struct vg_image *img, struct pipe_sampler_state **samplers, - struct pipe_texture **textures) + struct pipe_sampler_view **sampler_views) { img->sampler.min_img_filter = image_sampler_filter(img->base.ctx); img->sampler.mag_img_filter = image_sampler_filter(img->base.ctx); samplers[3] = &img->sampler; - textures[3] = img->texture; + sampler_views[3] = img->sampler_view; return 1; } diff --git a/src/gallium/state_trackers/vega/image.h b/src/gallium/state_trackers/vega/image.h index 78e17cffa6..805b35fab9 100644 --- a/src/gallium/state_trackers/vega/image.h +++ b/src/gallium/state_trackers/vega/image.h @@ -43,7 +43,7 @@ struct vg_image { struct vg_image *parent; - struct pipe_texture *texture; + struct pipe_sampler_view *sampler_view; struct pipe_sampler_state sampler; struct array *children_array; @@ -89,7 +89,7 @@ void image_get_pixels(struct vg_image *dst, VGint dx, VGint dy, VGint width, VGint height); VGint image_bind_samplers(struct vg_image *dst, struct pipe_sampler_state **samplers, - struct pipe_texture **textures); + struct pipe_sampler_view **sampler_views); VGboolean vg_image_overlaps(struct vg_image *dst, struct vg_image *src); diff --git a/src/gallium/state_trackers/vega/mask.c b/src/gallium/state_trackers/vega/mask.c index 839dc19a3b..316ea7a9c9 100644 --- a/src/gallium/state_trackers/vega/mask.c +++ b/src/gallium/state_trackers/vega/mask.c @@ -45,7 +45,7 @@ struct vg_mask_layer { VGint width; VGint height; - struct pipe_texture *texture; + struct pipe_sampler_view *sampler_view; }; static INLINE struct pipe_surface * @@ -54,7 +54,7 @@ alpha_mask_surface(struct vg_context *ctx, int usage) struct pipe_screen *screen = ctx->pipe->screen; struct st_framebuffer *stfb = ctx->draw_buffer; return screen->get_tex_surface(screen, - stfb->alpha_mask, + stfb->alpha_mask_view->texture, 0, 0, 0, usage); } @@ -284,35 +284,33 @@ static void setup_mask_operation(VGMaskOperation operation) cso_set_fragment_shader_handle(ctx->cso_context, shader); } -static void setup_mask_samplers(struct pipe_texture *umask) +static void setup_mask_samplers(struct pipe_sampler_view *umask) { struct vg_context *ctx = vg_current_context(); struct pipe_sampler_state *samplers[PIPE_MAX_SAMPLERS]; - struct pipe_texture *textures[PIPE_MAX_SAMPLERS]; + struct pipe_sampler_view *sampler_views[PIPE_MAX_SAMPLERS]; struct st_framebuffer *fb_buffers = ctx->draw_buffer; - struct pipe_texture *uprev = NULL; + struct pipe_sampler_view *uprev = NULL; struct pipe_sampler_state sampler; - uprev = fb_buffers->blend_texture; + uprev = fb_buffers->blend_texture_view; sampler = ctx->mask.sampler; sampler.normalized_coords = 1; samplers[0] = NULL; samplers[1] = NULL; - samplers[2] = NULL; - textures[0] = NULL; - textures[1] = NULL; - textures[2] = NULL; + sampler_views[0] = NULL; + sampler_views[1] = NULL; samplers[0] = &sampler; samplers[1] = &ctx->mask.sampler; - textures[0] = umask; - textures[1] = uprev; + sampler_views[0] = umask; + sampler_views[1] = uprev; cso_set_samplers(ctx->cso_context, 2, (const struct pipe_sampler_state **)samplers); - cso_set_sampler_textures(ctx->cso_context, 2, textures); + cso_set_fragment_sampler_views(ctx->cso_context, 2, sampler_views); } @@ -411,12 +409,13 @@ static void surface_fill(struct pipe_surface *surf, } -static void mask_using_texture(struct pipe_texture *texture, +static void mask_using_texture(struct pipe_sampler_view *sampler_view, VGMaskOperation operation, VGint x, VGint y, VGint width, VGint height) { struct vg_context *ctx = vg_current_context(); + struct pipe_texture *texture = sampler_view->texture; struct pipe_surface *surface = alpha_mask_surface(ctx, PIPE_BUFFER_USAGE_GPU_WRITE); VGint offsets[4], loc[4]; @@ -439,13 +438,13 @@ static void mask_using_texture(struct pipe_texture *texture, vg_prepare_blend_surface_from_mask(ctx); cso_save_samplers(ctx->cso_context); - cso_save_sampler_textures(ctx->cso_context); + cso_save_fragment_sampler_views(ctx->cso_context); cso_save_framebuffer(ctx->cso_context); cso_save_blend(ctx->cso_context); cso_save_fragment_shader(ctx->cso_context); cso_save_viewport(ctx->cso_context); - setup_mask_samplers(texture); + setup_mask_samplers(sampler_view); setup_mask_blend(); setup_mask_operation(operation); setup_mask_framebuffer(surface, surface->width, surface->height); @@ -463,7 +462,7 @@ static void mask_using_texture(struct pipe_texture *texture, cso_restore_framebuffer(ctx->cso_context); cso_restore_fragment_shader(ctx->cso_context); cso_restore_samplers(ctx->cso_context); - cso_restore_sampler_textures(ctx->cso_context); + cso_restore_fragment_sampler_views(ctx->cso_context); cso_restore_viewport(ctx->cso_context); pipe_surface_reference(&surface, NULL); @@ -484,7 +483,11 @@ struct vg_mask_layer * mask_layer_create(VGint width, VGint height) { struct pipe_texture pt; + struct pipe_context *pipe = ctx->pipe; struct pipe_screen *screen = ctx->pipe->screen; + struct pipe_sampler_view view_templ; + struct pipe_sampler_view *view = NULL; + struct pipe_texture *texture; memset(&pt, 0, sizeof(pt)); pt.target = PIPE_TEXTURE_2D; @@ -496,7 +499,14 @@ struct vg_mask_layer * mask_layer_create(VGint width, VGint height) pt.tex_usage = PIPE_TEXTURE_USAGE_SAMPLER; pt.compressed = 0; - mask->texture = screen->texture_create(screen, &pt); + texture = screen->texture_create(screen, &pt); + + if (texture) { + u_sampler_view_default_template(&view_templ, texture, texture->format); + view = pipe->create_sampler_view(pipe, texture, &view_templ); + } + pipe_texture_reference(&texture, NULL); + mask->sampler_view = view; } vg_context_add_object(ctx, VG_OBJECT_MASK, mask); @@ -525,7 +535,7 @@ void mask_layer_fill(struct vg_mask_layer *layer, alpha_color[3] = value; surface = ctx->pipe->screen->get_tex_surface( - ctx->pipe->screen, layer->texture, + ctx->pipe->screen, layer->sampler_view->texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); @@ -545,10 +555,10 @@ void mask_copy(struct vg_mask_layer *layer, struct st_framebuffer *fb_buffers = ctx->draw_buffer; renderer_copy_texture(ctx->renderer, - layer->texture, + layer->sampler_view, sx, sy, sx + width, sy + height, - fb_buffers->alpha_mask, + fb_buffers->alpha_mask_view->texture, dx, dy, dx + width, dy + height); } @@ -562,7 +572,7 @@ static void mask_layer_render_to(struct vg_mask_layer *layer, struct pipe_screen *screen = ctx->pipe->screen; struct pipe_surface *surface; - surface = screen->get_tex_surface(screen, layer->texture, 0, 0, 0, + surface = screen->get_tex_surface(screen, layer->sampler_view->texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); cso_save_framebuffer(ctx->cso_context); @@ -604,8 +614,8 @@ void mask_render_to(struct path *path, struct vg_mask_layer *temp_layer; VGint width, height; - width = fb_buffers->alpha_mask->width0; - height = fb_buffers->alpha_mask->width0; + width = fb_buffers->alpha_mask_view->texture->width0; + height = fb_buffers->alpha_mask_view->texture->width0; temp_layer = mask_layer_create(width, height); @@ -622,7 +632,7 @@ void mask_using_layer(struct vg_mask_layer *layer, VGint x, VGint y, VGint width, VGint height) { - mask_using_texture(layer->texture, operation, + mask_using_texture(layer->sampler_view, operation, x, y, width, height); } @@ -644,7 +654,7 @@ void mask_using_image(struct vg_image *image, VGint x, VGint y, VGint width, VGint height) { - mask_using_texture(image->texture, operation, + mask_using_texture(image->sampler_view, operation, x, y, width, height); } @@ -672,7 +682,7 @@ void mask_fill(VGint x, VGint y, VGint width, VGint height, } VGint mask_bind_samplers(struct pipe_sampler_state **samplers, - struct pipe_texture **textures) + struct pipe_sampler_view **sampler_views) { struct vg_context *ctx = vg_current_context(); @@ -680,7 +690,7 @@ VGint mask_bind_samplers(struct pipe_sampler_state **samplers, struct st_framebuffer *fb_buffers = ctx->draw_buffer; samplers[1] = &ctx->mask.sampler; - textures[1] = fb_buffers->alpha_mask; + sampler_views[1] = fb_buffers->alpha_mask_view; return 1; } else return 0; diff --git a/src/gallium/state_trackers/vega/mask.h b/src/gallium/state_trackers/vega/mask.h index 5eaaede0e3..4feacbefda 100644 --- a/src/gallium/state_trackers/vega/mask.h +++ b/src/gallium/state_trackers/vega/mask.h @@ -63,6 +63,6 @@ void mask_fill(VGint x, VGint y, VGfloat value); VGint mask_bind_samplers(struct pipe_sampler_state **samplers, - struct pipe_texture **textures); + struct pipe_sampler_view **sampler_views); #endif diff --git a/src/gallium/state_trackers/vega/paint.c b/src/gallium/state_trackers/vega/paint.c index dc56b8c5f3..508e1863a5 100644 --- a/src/gallium/state_trackers/vega/paint.c +++ b/src/gallium/state_trackers/vega/paint.c @@ -37,6 +37,7 @@ #include "util/u_format.h" #include "util/u_memory.h" #include "util/u_math.h" +#include "util/u_sampler.h" #include "cso_cache/cso_context.h" @@ -61,7 +62,7 @@ struct vg_paint { VGfloat vals[5]; VGint valsi[5]; } radial; - struct pipe_texture *texture; + struct pipe_sampler_view *sampler_view; struct pipe_sampler_state sampler; VGfloat *ramp_stops; @@ -72,7 +73,7 @@ struct vg_paint { } gradient; struct { - struct pipe_texture *texture; + struct pipe_sampler_view *sampler_view; VGTilingMode tiling_mode; struct pipe_sampler_state sampler; } pattern; @@ -173,6 +174,26 @@ static INLINE struct pipe_texture *create_gradient_texture(struct vg_paint *p) return tex; } +static INLINE struct pipe_sampler_view *create_gradient_sampler_view(struct vg_paint *p) +{ + struct pipe_context *pipe = p->base.ctx->pipe; + struct pipe_texture *texture; + struct pipe_sampler_view view_templ; + struct pipe_sampler_view *view; + + texture = create_gradient_texture(p); + + if (!texture) + return NULL; + + u_sampler_view_default_template(&view_templ, texture, texture->format); + view = pipe->create_sampler_view(pipe, texture, &view_templ); + /* want the texture to go away if the view is freed */ + pipe_texture_reference(&texture, NULL); + + return view; +} + struct vg_paint * paint_create(struct vg_context *ctx) { struct vg_paint *paint = CALLOC_STRUCT(vg_paint); @@ -207,8 +228,9 @@ struct vg_paint * paint_create(struct vg_context *ctx) void paint_destroy(struct vg_paint *paint) { struct vg_context *ctx = paint->base.ctx; - if (paint->pattern.texture) - pipe_texture_reference(&paint->pattern.texture, NULL); + pipe_sampler_view_reference(&paint->gradient.sampler_view, NULL); + if (paint->pattern.sampler_view) + pipe_sampler_view_reference(&paint->pattern.sampler_view, NULL); if (ctx) vg_context_remove_object(ctx, VG_OBJECT_PAINT, paint); @@ -329,8 +351,8 @@ static INLINE void paint_pattern_buffer(struct vg_paint *paint, void *buffer) map[4] = 0.f; map[5] = 1.f; - map[6] = paint->pattern.texture->width0; - map[7] = paint->pattern.texture->height0; + map[6] = paint->pattern.sampler_view->texture->width0; + map[7] = paint->pattern.sampler_view->texture->height0; { struct matrix mat; memcpy(&mat, &ctx->state.vg.fill_paint_to_user_matrix, @@ -394,12 +416,12 @@ void paint_set_ramp_stops(struct vg_paint *paint, const VGfloat *stops, create_gradient_data(stops, num / 5, paint->gradient.color_data, 1024); - if (paint->gradient.texture) { - pipe_texture_reference(&paint->gradient.texture, NULL); - paint->gradient.texture = 0; + if (paint->gradient.sampler_view) { + pipe_sampler_view_reference(&paint->gradient.sampler_view, NULL); + paint->gradient.sampler_view = NULL; } - paint->gradient.texture = create_gradient_texture(paint); + paint->gradient.sampler_view = create_gradient_sampler_view(paint); } void paint_set_colori(struct vg_paint *p, @@ -459,12 +481,12 @@ void paint_set_radial_gradient(struct vg_paint *paint, void paint_set_pattern(struct vg_paint *paint, struct vg_image *img) { - if (paint->pattern.texture) - pipe_texture_reference(&paint->pattern.texture, NULL); + if (paint->pattern.sampler_view) + pipe_sampler_view_reference(&paint->pattern.sampler_view, NULL); - paint->pattern.texture = 0; - pipe_texture_reference(&paint->pattern.texture, - img->texture); + paint->pattern.sampler_view = NULL; + pipe_sampler_view_reference(&paint->pattern.sampler_view, + img->sampler_view); } void paint_set_pattern_tiling(struct vg_paint *paint, @@ -611,18 +633,18 @@ VGTilingMode paint_pattern_tiling(struct vg_paint *paint) } VGint paint_bind_samplers(struct vg_paint *paint, struct pipe_sampler_state **samplers, - struct pipe_texture **textures) + struct pipe_sampler_view **sampler_views) { struct vg_context *ctx = vg_current_context(); switch(paint->type) { case VG_PAINT_TYPE_LINEAR_GRADIENT: case VG_PAINT_TYPE_RADIAL_GRADIENT: { - if (paint->gradient.texture) { + if (paint->gradient.sampler_view) { paint->gradient.sampler.min_img_filter = image_sampler_filter(ctx); paint->gradient.sampler.mag_img_filter = image_sampler_filter(ctx); samplers[0] = &paint->gradient.sampler; - textures[0] = paint->gradient.texture; + sampler_views[0] = paint->gradient.sampler_view; return 1; } } @@ -634,7 +656,7 @@ VGint paint_bind_samplers(struct vg_paint *paint, struct pipe_sampler_state **sa paint->pattern.sampler.min_img_filter = image_sampler_filter(ctx); paint->pattern.sampler.mag_img_filter = image_sampler_filter(ctx); samplers[0] = &paint->pattern.sampler; - textures[0] = paint->pattern.texture; + sampler_views[0] = paint->pattern.sampler_view; return 1; } break; @@ -647,7 +669,7 @@ VGint paint_bind_samplers(struct vg_paint *paint, struct pipe_sampler_state **sa void paint_resolve_type(struct vg_paint *paint) { if (paint->type == VG_PAINT_TYPE_PATTERN && - !paint->pattern.texture) { + !paint->pattern.sampler_view) { paint->type = VG_PAINT_TYPE_COLOR; } } diff --git a/src/gallium/state_trackers/vega/paint.h b/src/gallium/state_trackers/vega/paint.h index 999b5c167c..9ea67c4b1e 100644 --- a/src/gallium/state_trackers/vega/paint.h +++ b/src/gallium/state_trackers/vega/paint.h @@ -108,7 +108,7 @@ VGboolean paint_color_ramp_premultiplied(struct vg_paint *paint); VGint paint_bind_samplers(struct vg_paint *paint, struct pipe_sampler_state **samplers, - struct pipe_texture **textures); + struct pipe_sampler_view **sampler_views); VGint paint_constant_buffer_size(struct vg_paint *paint); void paint_fill_constant_buffer(struct vg_paint *paint, diff --git a/src/gallium/state_trackers/vega/renderer.c b/src/gallium/state_trackers/vega/renderer.c index 47e8b470a1..2bb4c8bc75 100644 --- a/src/gallium/state_trackers/vega/renderer.c +++ b/src/gallium/state_trackers/vega/renderer.c @@ -39,6 +39,7 @@ #include "util/u_simple_shaders.h" #include "util/u_memory.h" #include "util/u_rect.h" +#include "util/u_sampler.h" #include "cso_cache/cso_context.h" @@ -263,7 +264,7 @@ void renderer_draw_texture(struct renderer *r, } void renderer_copy_texture(struct renderer *ctx, - struct pipe_texture *src, + struct pipe_sampler_view *src, VGfloat sx1, VGfloat sy1, VGfloat sx2, VGfloat sy2, struct pipe_texture *dst, @@ -272,6 +273,7 @@ void renderer_copy_texture(struct renderer *ctx, { struct pipe_context *pipe = ctx->pipe; struct pipe_screen *screen = pipe->screen; + struct pipe_texture *tex = src->texture; struct pipe_buffer *buf; struct pipe_surface *dst_surf = screen->get_tex_surface( screen, dst, 0, 0, 0, @@ -279,8 +281,8 @@ void renderer_copy_texture(struct renderer *ctx, struct pipe_framebuffer_state fb; float s0, t0, s1, t1; - assert(src->width0 != 0); - assert(src->height0 != 0); + assert(tex->width0 != 0); + assert(tex->height0 != 0); assert(dst->width0 != 0); assert(dst->height0 != 0); @@ -290,10 +292,10 @@ void renderer_copy_texture(struct renderer *ctx, #endif #if 1 - s0 = sx1 / src->width0; - s1 = sx2 / src->width0; - t0 = sy1 / src->height0; - t1 = sy2 / src->height0; + s0 = sx1 / tex->width0; + s1 = sx2 / tex->width0; + t0 = sy1 / tex->height0; + t1 = sy2 / tex->height0; #else s0 = 0; s1 = 1; @@ -307,7 +309,7 @@ void renderer_copy_texture(struct renderer *ctx, /* save state (restored below) */ cso_save_blend(ctx->cso); cso_save_samplers(ctx->cso); - cso_save_sampler_textures(ctx->cso); + cso_save_fragment_sampler_views(ctx->cso); cso_save_framebuffer(ctx->cso); cso_save_fragment_shader(ctx->cso); cso_save_vertex_shader(ctx->cso); @@ -345,7 +347,7 @@ void renderer_copy_texture(struct renderer *ctx, vg_set_viewport(ctx->owner, VEGA_Y0_TOP); /* texture */ - cso_set_sampler_textures(ctx->cso, 1, &src); + cso_set_fragment_sampler_views(ctx->cso, 1, &src); /* shaders */ cso_set_vertex_shader_handle(ctx->cso, vg_texture_vs(ctx->owner)); @@ -385,7 +387,7 @@ void renderer_copy_texture(struct renderer *ctx, /* restore state we changed */ cso_restore_blend(ctx->cso); cso_restore_samplers(ctx->cso); - cso_restore_sampler_textures(ctx->cso); + cso_restore_fragment_sampler_views(ctx->cso); cso_restore_framebuffer(ctx->cso); cso_restore_vertex_shader(ctx->cso); cso_restore_fragment_shader(ctx->cso); @@ -406,6 +408,8 @@ void renderer_copy_surface(struct renderer *ctx, struct pipe_context *pipe = ctx->pipe; struct pipe_screen *screen = pipe->screen; struct pipe_buffer *buf; + struct pipe_sampler_view view_templ; + struct pipe_sampler_view *view; struct pipe_texture texTemp, *tex; struct pipe_surface *texSurf; struct pipe_framebuffer_state fb; @@ -457,6 +461,12 @@ void renderer_copy_surface(struct renderer *ctx, if (!tex) return; + u_sampler_view_default_template(&view_templ, tex, tex->format); + view = pipe->create_sampler_view(pipe, tex, &view_templ); + + if (!view) + return; + texSurf = screen->get_tex_surface(screen, tex, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); @@ -479,7 +489,7 @@ void renderer_copy_surface(struct renderer *ctx, /* save state (restored below) */ cso_save_blend(ctx->cso); cso_save_samplers(ctx->cso); - cso_save_sampler_textures(ctx->cso); + cso_save_fragment_sampler_views(ctx->cso); cso_save_framebuffer(ctx->cso); cso_save_fragment_shader(ctx->cso); cso_save_vertex_shader(ctx->cso); @@ -515,7 +525,7 @@ void renderer_copy_surface(struct renderer *ctx, } /* texture */ - cso_set_sampler_textures(ctx->cso, 1, &tex); + cso_set_fragment_sampler_views(ctx->cso, 1, &view); /* shaders */ cso_set_fragment_shader_handle(ctx->cso, ctx->fs); @@ -552,13 +562,14 @@ void renderer_copy_surface(struct renderer *ctx, /* restore state we changed */ cso_restore_blend(ctx->cso); cso_restore_samplers(ctx->cso); - cso_restore_sampler_textures(ctx->cso); + cso_restore_fragment_sampler_views(ctx->cso); cso_restore_framebuffer(ctx->cso); cso_restore_fragment_shader(ctx->cso); cso_restore_vertex_shader(ctx->cso); cso_restore_viewport(ctx->cso); pipe_texture_reference(&tex, NULL); + pipe_sampler_view_reference(&view, NULL); } void renderer_texture_quad(struct renderer *r, diff --git a/src/gallium/state_trackers/vega/renderer.h b/src/gallium/state_trackers/vega/renderer.h index 990cd32c31..03366f1361 100644 --- a/src/gallium/state_trackers/vega/renderer.h +++ b/src/gallium/state_trackers/vega/renderer.h @@ -33,6 +33,7 @@ struct renderer; struct vg_context; struct pipe_texture; +struct pipe_sampler_view; struct pipe_surface; struct renderer *renderer_create(struct vg_context *owner); @@ -57,7 +58,7 @@ void renderer_texture_quad(struct renderer *, VGfloat x3, VGfloat y3, VGfloat x4, VGfloat y4); void renderer_copy_texture(struct renderer *r, - struct pipe_texture *src, + struct pipe_sampler_view *src, VGfloat sx1, VGfloat sy1, VGfloat sx2, VGfloat sy2, struct pipe_texture *dst, diff --git a/src/gallium/state_trackers/vega/shader.c b/src/gallium/state_trackers/vega/shader.c index 0e71a507bf..f2ec24c57f 100644 --- a/src/gallium/state_trackers/vega/shader.c +++ b/src/gallium/state_trackers/vega/shader.c @@ -119,7 +119,7 @@ static void setup_constant_buffer(struct shader *shader) static VGint blend_bind_samplers(struct vg_context *ctx, struct pipe_sampler_state **samplers, - struct pipe_texture **textures) + struct pipe_sampler_view **sampler_views) { VGBlendMode bmode = ctx->state.vg.blend_mode; @@ -132,15 +132,15 @@ static VGint blend_bind_samplers(struct vg_context *ctx, vg_prepare_blend_surface(ctx); samplers[2] = &ctx->blend_sampler; - textures[2] = stfb->blend_texture; + sampler_views[2] = stfb->blend_texture_view; - if (!samplers[0] || !textures[0]) { + if (!samplers[0] || !sampler_views[0]) { samplers[0] = samplers[2]; - textures[0] = textures[2]; + sampler_views[0] = sampler_views[2]; } - if (!samplers[1] || !textures[1]) { + if (!samplers[1] || !sampler_views[1]) { samplers[1] = samplers[0]; - textures[1] = textures[0]; + sampler_views[1] = sampler_views[0]; } return 1; @@ -151,7 +151,7 @@ static VGint blend_bind_samplers(struct vg_context *ctx, static void setup_samplers(struct shader *shader) { struct pipe_sampler_state *samplers[PIPE_MAX_SAMPLERS]; - struct pipe_texture *textures[PIPE_MAX_SAMPLERS]; + struct pipe_sampler_view *sampler_views[PIPE_MAX_SAMPLERS]; struct vg_context *ctx = shader->context; /* a little wonky: we use the num as a boolean that just says * whether any sampler/textures have been set. the actual numbering @@ -167,20 +167,20 @@ static void setup_samplers(struct shader *shader) samplers[1] = NULL; samplers[2] = NULL; samplers[3] = NULL; - textures[0] = NULL; - textures[1] = NULL; - textures[2] = NULL; - textures[3] = NULL; - - num += paint_bind_samplers(shader->paint, samplers, textures); - num += mask_bind_samplers(samplers, textures); - num += blend_bind_samplers(ctx, samplers, textures); + sampler_views[0] = NULL; + sampler_views[1] = NULL; + sampler_views[2] = NULL; + sampler_views[3] = NULL; + + num += paint_bind_samplers(shader->paint, samplers, sampler_views); + num += mask_bind_samplers(samplers, sampler_views); + num += blend_bind_samplers(ctx, samplers, sampler_views); if (shader->drawing_image && shader->image) - num += image_bind_samplers(shader->image, samplers, textures); + num += image_bind_samplers(shader->image, samplers, sampler_views); if (num) { cso_set_samplers(ctx->cso_context, 4, (const struct pipe_sampler_state **)samplers); - cso_set_sampler_textures(ctx->cso_context, 4, textures); + cso_set_fragment_sampler_views(ctx->cso_context, 4, sampler_views); } } diff --git a/src/gallium/state_trackers/vega/vg_context.c b/src/gallium/state_trackers/vega/vg_context.c index 7769112a31..11ebbbe544 100644 --- a/src/gallium/state_trackers/vega/vg_context.c +++ b/src/gallium/state_trackers/vega/vg_context.c @@ -43,6 +43,7 @@ #include "util/u_simple_shaders.h" #include "util/u_memory.h" #include "util/u_blit.h" +#include "util/u_sampler.h" struct vg_context *_vg_context = 0; @@ -436,19 +437,24 @@ void vg_prepare_blend_surface(struct vg_context *ctx) { struct pipe_surface *dest_surface = NULL; struct pipe_context *pipe = ctx->pipe; + struct pipe_sampler_view *view; + struct pipe_sampler_view view_templ; struct st_framebuffer *stfb = ctx->draw_buffer; struct st_renderbuffer *strb = stfb->strb; /* first finish all pending rendering */ vgFinish(); + u_sampler_view_default_template(&view_templ, strb->texture, strb->texture->format); + view = pipe->create_sampler_view(pipe, strb->texture, &view_templ); + dest_surface = pipe->screen->get_tex_surface(pipe->screen, - stfb->blend_texture, + stfb->blend_texture_view->texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); /* flip it, because we want to use it as a sampler */ util_blit_pixels_tex(ctx->blit, - strb->texture, + view, 0, strb->height, strb->width, 0, dest_surface, @@ -461,6 +467,8 @@ void vg_prepare_blend_surface(struct vg_context *ctx) /* make sure it's complete */ vgFinish(); + + pipe_sampler_view_reference(&view, NULL); } @@ -477,13 +485,13 @@ void vg_prepare_blend_surface_from_mask(struct vg_context *ctx) vgFinish(); dest_surface = pipe->screen->get_tex_surface(pipe->screen, - stfb->blend_texture, + stfb->blend_texture_view->texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); /* flip it, because we want to use it as a sampler */ util_blit_pixels_tex(ctx->blit, - stfb->alpha_mask, + stfb->alpha_mask_view, 0, strb->height, strb->width, 0, dest_surface, diff --git a/src/gallium/state_trackers/vega/vg_context.h b/src/gallium/state_trackers/vega/vg_context.h index 17c7d2ad1c..c9e36d7d76 100644 --- a/src/gallium/state_trackers/vega/vg_context.h +++ b/src/gallium/state_trackers/vega/vg_context.h @@ -55,9 +55,10 @@ struct st_framebuffer { struct st_renderbuffer *strb; struct st_renderbuffer *dsrb; - struct pipe_texture *alpha_mask; + struct pipe_sampler_view *alpha_mask_view; + + struct pipe_sampler_view *blend_texture_view; - struct pipe_texture *blend_texture; struct st_framebuffer_iface *iface; enum st_attachment_type strb_att; diff --git a/src/gallium/state_trackers/vega/vg_tracker.c b/src/gallium/state_trackers/vega/vg_tracker.c index ea5c2ce41f..f438e34087 100644 --- a/src/gallium/state_trackers/vega/vg_tracker.c +++ b/src/gallium/state_trackers/vega/vg_tracker.c @@ -32,6 +32,7 @@ #include "util/u_inlines.h" #include "pipe/p_screen.h" #include "util/u_format.h" +#include "util/u_sampler.h" #include "util/u_memory.h" #include "util/u_math.h" #include "util/u_rect.h" @@ -41,7 +42,7 @@ PUBLIC const int st_api_OpenVG = 1; static struct pipe_texture * create_texture(struct pipe_context *pipe, enum pipe_format format, - VGint width, VGint height) + VGint width, VGint height) { struct pipe_texture templ; @@ -71,6 +72,27 @@ create_texture(struct pipe_context *pipe, enum pipe_format format, return pipe->screen->texture_create(pipe->screen, &templ); } +static struct pipe_sampler_view * +create_tex_and_view(struct pipe_context *pipe, enum pipe_format format, + VGint width, VGint height) +{ + struct pipe_texture *texture; + struct pipe_sampler_view view_templ; + struct pipe_sampler_view *view; + + texture = create_texture(pipe, format, width, height); + + if (!texture) + return NULL; + + u_sampler_view_default_template(&view_templ, texture, texture->format); + view = pipe->create_sampler_view(pipe, texture, &view_templ); + /* want the texture to go away if the view is freed */ + pipe_texture_reference(&texture, NULL); + + return view; +} + /** * Allocate a renderbuffer for a an on-screen window (not a user-created * renderbuffer). The window system code determines the format. @@ -115,8 +137,8 @@ st_renderbuffer_alloc_storage(struct vg_context * ctx, surface_usage = (PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE); - strb->texture = create_texture(pipe, strb->format, - width, height); + strb->texture = create_texture(pipe, strb->format, width, height); + if (!strb->texture) return FALSE; @@ -191,7 +213,7 @@ struct st_framebuffer * st_create_framebuffer(const void *visual, /*### currently we always allocate it but it's possible it's not necessary if EGL_ALPHA_MASK_SIZE was 0 */ - stfb->alpha_mask = 0; + stfb->alpha_mask_view = NULL; stfb->width = width; stfb->height = height; @@ -206,19 +228,19 @@ static void setup_new_alpha_mask(struct vg_context *ctx, uint width, uint height) { struct pipe_context *pipe = ctx->pipe; - struct pipe_texture *old_texture = stfb->alpha_mask; + struct pipe_sampler_view *old_sampler_view = stfb->alpha_mask_view; /* we use PIPE_FORMAT_B8G8R8A8_UNORM because we want to render to this texture and use it as a sampler, so while this wastes some space it makes both of those a lot simpler */ - stfb->alpha_mask = - create_texture(pipe, PIPE_FORMAT_B8G8R8A8_UNORM, width, height); + stfb->alpha_mask_view = + create_tex_and_view(pipe, PIPE_FORMAT_B8G8R8A8_UNORM, width, height); - if (!stfb->alpha_mask) { - if (old_texture) - pipe_texture_reference(&old_texture, NULL); + if (!stfb->alpha_mask_view) { + if (old_sampler_view) + pipe_sampler_view_reference(&old_sampler_view, NULL); return; } @@ -228,15 +250,15 @@ static void setup_new_alpha_mask(struct vg_context *ctx, mask_fill(0, 0, width, height, 1.f); /* if we had an old surface copy it over */ - if (old_texture) { + if (old_sampler_view) { struct pipe_surface *surface = pipe->screen->get_tex_surface( pipe->screen, - stfb->alpha_mask, + stfb->alpha_mask_view->texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_WRITE); struct pipe_surface *old_surface = pipe->screen->get_tex_surface( pipe->screen, - old_texture, + old_sampler_view->texture, 0, 0, 0, PIPE_BUFFER_USAGE_GPU_READ); if (pipe->surface_copy) { @@ -264,8 +286,8 @@ static void setup_new_alpha_mask(struct vg_context *ctx, /* Free the old texture */ - if (old_texture) - pipe_texture_reference(&old_texture, NULL); + if (old_sampler_view) + pipe_sampler_view_reference(&old_sampler_view, NULL); } void st_resize_framebuffer(struct st_framebuffer *stfb, @@ -326,9 +348,9 @@ void st_resize_framebuffer(struct st_framebuffer *stfb, setup_new_alpha_mask(ctx, stfb, width, height); - pipe_texture_reference( &stfb->blend_texture, NULL ); - stfb->blend_texture = create_texture(ctx->pipe, PIPE_FORMAT_B8G8R8A8_UNORM, - width, height); + pipe_sampler_view_reference( &stfb->blend_texture_view, NULL ); + stfb->blend_texture_view = create_tex_and_view(ctx->pipe, PIPE_FORMAT_B8G8R8A8_UNORM, + width, height); } void st_set_framebuffer_surface(struct st_framebuffer *stfb, diff --git a/src/gallium/winsys/drm/radeon/core/radeon_drm_buffer.c b/src/gallium/winsys/drm/radeon/core/radeon_drm_buffer.c index cc56a2bb8f..2472b6bf95 100644 --- a/src/gallium/winsys/drm/radeon/core/radeon_drm_buffer.c +++ b/src/gallium/winsys/drm/radeon/core/radeon_drm_buffer.c @@ -23,7 +23,6 @@ struct radeon_drm_buffer { boolean flinked; uint32_t flink; - boolean mapped; struct radeon_drm_buffer *next, *prev; }; @@ -56,10 +55,10 @@ radeon_drm_buffer_destroy(struct pb_buffer *_buf) { struct radeon_drm_buffer *buf = radeon_drm_buffer(_buf); - if (buf->mapped) { + if (buf->bo->ptr != NULL) { remove_from_list(buf); radeon_bo_unmap(buf->bo); - buf->mapped = false; + buf->bo->ptr = NULL; } radeon_bo_unref(buf->bo); @@ -73,7 +72,7 @@ radeon_drm_buffer_map(struct pb_buffer *_buf, struct radeon_drm_buffer *buf = radeon_drm_buffer(_buf); int write; - if (buf->mapped) + if (buf->bo->ptr != NULL) return buf->bo->ptr; if (flags & PIPE_BUFFER_USAGE_DONTBLOCK) { @@ -95,7 +94,6 @@ radeon_drm_buffer_map(struct pb_buffer *_buf, if (radeon_bo_map(buf->bo, write)) { return NULL; } - buf->mapped = true; insert_at_tail(&buf->mgr->buffer_map_list, buf); return buf->bo->ptr; } @@ -289,7 +287,7 @@ boolean radeon_drm_bufmgr_get_handle(struct pb_buffer *_buf, retval = ioctl(fd, DRM_IOCTL_GEM_FLINK, &flink); if (retval) { - return false; + return FALSE; } buf->flinked = TRUE; @@ -306,13 +304,21 @@ boolean radeon_drm_bufmgr_get_handle(struct pb_buffer *_buf, void radeon_drm_bufmgr_set_tiling(struct pb_buffer *_buf, boolean microtiled, boolean macrotiled, uint32_t pitch) { struct radeon_drm_buffer *buf = get_drm_buffer(_buf); - uint32_t flags = 0; - + uint32_t flags = 0, old_flags, old_pitch; if (microtiled) flags |= RADEON_BO_FLAGS_MICRO_TILE; if (macrotiled) flags |= RADEON_BO_FLAGS_MACRO_TILE; + radeon_bo_get_tiling(buf->bo, &old_flags, &old_pitch); + + if (flags != old_flags || pitch != old_pitch) { + /* Tiling determines how DRM treats the buffer data. + * We must flush CS when changing it if the buffer is referenced. */ + if (radeon_bo_is_referenced_by_cs(buf->bo, buf->mgr->rws->cs)) { + buf->mgr->rws->flush_cb(buf->mgr->rws->flush_data); + } + } radeon_bo_set_tiling(buf->bo, flags, pitch); } @@ -323,7 +329,7 @@ boolean radeon_drm_bufmgr_add_buffer(struct pb_buffer *_buf, struct radeon_drm_buffer *buf = get_drm_buffer(_buf); radeon_cs_space_add_persistent_bo(buf->mgr->rws->cs, buf->bo, rd, wd); - return true; + return TRUE; } void radeon_drm_bufmgr_write_reloc(struct pb_buffer *_buf, @@ -357,12 +363,10 @@ void radeon_drm_bufmgr_flush_maps(struct pb_manager *_mgr) struct radeon_drm_buffer *rpb, *t_rpb; foreach_s(rpb, t_rpb, &mgr->buffer_map_list) { - rpb->mapped = 0; radeon_bo_unmap(rpb->bo); + rpb->bo->ptr = NULL; remove_from_list(rpb); } make_empty_list(&mgr->buffer_map_list); - - } diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c index d6a1ba6952..c86dd1d0d9 100644 --- a/src/mesa/drivers/dri/intel/intel_context.c +++ b/src/mesa/drivers/dri/intel/intel_context.c @@ -880,12 +880,12 @@ intelMakeCurrent(__DRIcontext * driContextPriv, struct gl_framebuffer *fb = driDrawPriv->driverPrivate; struct gl_framebuffer *readFb = driReadPriv->driverPrivate; - _mesa_make_current(&intel->ctx, fb, readFb); intel->driReadDrawable = driReadPriv; intel->driDrawable = driDrawPriv; driContextPriv->dri2.draw_stamp = driDrawPriv->dri2.stamp - 1; driContextPriv->dri2.read_stamp = driReadPriv->dri2.stamp - 1; intel_prepare_render(intel); + _mesa_make_current(&intel->ctx, fb, readFb); } else { _mesa_make_current(NULL, NULL, NULL); diff --git a/src/mesa/drivers/dri/nouveau/nouveau_gldefs.h b/src/mesa/drivers/dri/nouveau/nouveau_gldefs.h index 00007a9a35..fbeed3baea 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_gldefs.h +++ b/src/mesa/drivers/dri/nouveau/nouveau_gldefs.h @@ -260,4 +260,23 @@ nvgl_filter_mode(unsigned filter) } } +static inline unsigned +nvgl_texgen_mode(unsigned mode) +{ + switch (mode) { + case GL_EYE_LINEAR: + return 0x2400; + case GL_OBJECT_LINEAR: + return 0x2401; + case GL_SPHERE_MAP: + return 0x2402; + case GL_NORMAL_MAP: + return 0x8511; + case GL_REFLECTION_MAP: + return 0x8512; + default: + assert(0); + } +} + #endif diff --git a/src/mesa/drivers/dri/nouveau/nouveau_render_t.c b/src/mesa/drivers/dri/nouveau/nouveau_render_t.c index c0505781cf..7ccd7e6416 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_render_t.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_render_t.c @@ -254,7 +254,7 @@ get_scratch_vbo(GLcontext *ctx, unsigned size, struct nouveau_bo **bo, */ static inline unsigned get_max_vertices(GLcontext *ctx, const struct _mesa_index_buffer *ib, - unsigned n) + int n) { struct nouveau_render_state *render = to_render_state(ctx); diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.c b/src/mesa/drivers/dri/nouveau/nouveau_state.c index bc610451b4..ef2cc787de 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_state.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_state.c @@ -234,6 +234,13 @@ nouveau_enable(GLcontext *ctx, GLenum cap, GLboolean state) context_dirty_i(ctx, TEX_ENV, ctx->Texture.CurrentUnit); context_dirty_i(ctx, TEX_OBJ, ctx->Texture.CurrentUnit); break; + case GL_TEXTURE_GEN_S: + case GL_TEXTURE_GEN_T: + case GL_TEXTURE_GEN_R: + case GL_TEXTURE_GEN_Q: + context_dirty_i(ctx, TEX_GEN, ctx->Texture.CurrentUnit); + context_dirty(ctx, MODELVIEW); + break; } } @@ -368,7 +375,15 @@ static void nouveau_tex_gen(GLcontext *ctx, GLenum coord, GLenum pname, const GLfloat *params) { - context_dirty_i(ctx, TEX_GEN, ctx->Texture.CurrentUnit); + switch (pname) { + case GL_TEXTURE_GEN_MODE: + context_dirty_i(ctx, TEX_GEN, ctx->Texture.CurrentUnit); + context_dirty(ctx, MODELVIEW); + break; + default: + context_dirty_i(ctx, TEX_GEN, ctx->Texture.CurrentUnit); + break; + } } static void @@ -454,12 +469,19 @@ nouveau_state_emit(GLcontext *ctx) static void nouveau_update_state(GLcontext *ctx, GLbitfield new_state) { + int i; + if (new_state & (_NEW_PROJECTION | _NEW_MODELVIEW)) context_dirty(ctx, PROJECTION); if (new_state & _NEW_MODELVIEW) context_dirty(ctx, MODELVIEW); + if (new_state & _NEW_TEXTURE_MATRIX) { + for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) + context_dirty_i(ctx, TEX_MAT, i); + } + if (new_state & _NEW_CURRENT_ATTRIB && new_state & _NEW_LIGHT) { context_dirty(ctx, MATERIAL_FRONT_AMBIENT); diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.h b/src/mesa/drivers/dri/nouveau/nouveau_state.h index d01d962c9f..38ac9753c8 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_state.h +++ b/src/mesa/drivers/dri/nouveau/nouveau_state.h @@ -89,6 +89,10 @@ enum { NOUVEAU_STATE_TEX_GEN1, NOUVEAU_STATE_TEX_GEN2, NOUVEAU_STATE_TEX_GEN3, + NOUVEAU_STATE_TEX_MAT0, + NOUVEAU_STATE_TEX_MAT1, + NOUVEAU_STATE_TEX_MAT2, + NOUVEAU_STATE_TEX_MAT3, NOUVEAU_STATE_TEX_OBJ0, NOUVEAU_STATE_TEX_OBJ1, NOUVEAU_STATE_TEX_OBJ2, diff --git a/src/mesa/drivers/dri/nouveau/nouveau_texture.c b/src/mesa/drivers/dri/nouveau/nouveau_texture.c index bf365bfca3..e89018653b 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_texture.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_texture.c @@ -177,15 +177,15 @@ nouveau_choose_tex_format(GLcontext *ctx, GLint internalFormat, } static GLboolean -teximage_fits(struct gl_texture_object *t, int level, - struct gl_texture_image *ti) +teximage_fits(struct gl_texture_object *t, int level) { struct nouveau_surface *s = &to_nouveau_texture(t)->surfaces[level]; + struct gl_texture_image *ti = t->Image[0][level]; - return t->Target == GL_TEXTURE_RECTANGLE || - (s->bo && s->width == ti->Width && - s->height == ti->Height && - s->format == ti->TexFormat); + return ti && (t->Target == GL_TEXTURE_RECTANGLE || + (s->bo && s->width == ti->Width && + s->height == ti->Height && + s->format == ti->TexFormat)); } static GLboolean @@ -195,7 +195,7 @@ validate_teximage(GLcontext *ctx, struct gl_texture_object *t, { struct gl_texture_image *ti = t->Image[0][level]; - if (ti && teximage_fits(t, level, ti)) { + if (teximage_fits(t, level)) { struct nouveau_surface *ss = to_nouveau_texture(t)->surfaces; struct nouveau_surface *s = &to_nouveau_teximage(ti)->surface; @@ -296,6 +296,8 @@ nouveau_texture_validate(GLcontext *ctx, struct gl_texture_object *t) validate_teximage(ctx, t, i, 0, 0, 0, s->width, s->height, 1); } + + FIRE_RING(context_chan(ctx)); } return GL_TRUE; @@ -304,9 +306,12 @@ nouveau_texture_validate(GLcontext *ctx, struct gl_texture_object *t) void nouveau_texture_reallocate(GLcontext *ctx, struct gl_texture_object *t) { - texture_dirty(t); - relayout_texture(ctx, t); - nouveau_texture_validate(ctx, t); + if (!teximage_fits(t, t->BaseLevel) || + !teximage_fits(t, get_last_level(t))) { + texture_dirty(t); + relayout_texture(ctx, t); + nouveau_texture_validate(ctx, t); + } } static unsigned @@ -364,7 +369,7 @@ nouveau_teximage(GLcontext *ctx, GLint dims, GLenum target, GLint level, } if (level == t->BaseLevel) { - if (!teximage_fits(t, level, ti)) + if (!teximage_fits(t, level)) relayout_texture(ctx, t); nouveau_texture_validate(ctx, t); } diff --git a/src/mesa/drivers/dri/nouveau/nouveau_util.h b/src/mesa/drivers/dri/nouveau/nouveau_util.h index d6007aba2b..584cb80ef6 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_util.h +++ b/src/mesa/drivers/dri/nouveau/nouveau_util.h @@ -191,4 +191,22 @@ is_texture_source(int s) return s == GL_TEXTURE || (s >= GL_TEXTURE0 && s <= GL_TEXTURE31); } +static inline struct gl_texgen * +get_texgen_coord(struct gl_texture_unit *u, int i) +{ + return ((struct gl_texgen *[]) + { &u->GenS, &u->GenT, &u->GenR, &u->GenQ }) [i]; +} + +static inline float * +get_texgen_coeff(struct gl_texgen *c) +{ + if (c->Mode == GL_OBJECT_LINEAR) + return c->ObjectPlane; + else if (c->Mode == GL_EYE_LINEAR) + return c->EyePlane; + else + return NULL; +} + #endif diff --git a/src/mesa/drivers/dri/nouveau/nouveau_vbo_t.c b/src/mesa/drivers/dri/nouveau/nouveau_vbo_t.c index a365b977f2..0c29eec8ee 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_vbo_t.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_vbo_t.c @@ -224,9 +224,11 @@ vbo_choose_attrs(GLcontext *ctx, const struct gl_client_array **arrays) if (ctx->Fog.Enabled && ctx->Fog.FogCoordinateSource == GL_FOG_COORD) vbo_emit_attr(ctx, arrays, VERT_ATTRIB_FOG); - if (ctx->Light.Enabled) { + if (ctx->Light.Enabled || + (ctx->Texture._GenFlags & TEXGEN_NEED_NORMALS)) vbo_emit_attr(ctx, arrays, VERT_ATTRIB_NORMAL); + if (ctx->Light.Enabled) { vbo_emit_attr(ctx, arrays, MAT(FRONT_AMBIENT)); vbo_emit_attr(ctx, arrays, MAT(FRONT_DIFFUSE)); vbo_emit_attr(ctx, arrays, MAT(FRONT_SPECULAR)); @@ -244,17 +246,20 @@ vbo_choose_attrs(GLcontext *ctx, const struct gl_client_array **arrays) } static unsigned -get_max_client_stride(GLcontext *ctx) +get_max_client_stride(GLcontext *ctx, const struct gl_client_array **arrays) { struct nouveau_render_state *render = to_render_state(ctx); int i, s = 0; for (i = 0; i < render->attr_count; i++) { int attr = render->map[i]; - struct nouveau_array_state *a = &render->attrs[attr]; - if (attr >= 0 && !a->bo) - s = MAX2(a->stride, s); + if (attr >= 0) { + const struct gl_client_array *a = arrays[attr]; + + if (!_mesa_is_bufferobj(a->BufferObj)) + s = MAX2(a->StrideB, s); + } } return s; @@ -275,14 +280,15 @@ vbo_maybe_split(GLcontext *ctx, const struct gl_client_array **arrays, { struct nouveau_context *nctx = to_nouveau_context(ctx); struct nouveau_render_state *render = to_render_state(ctx); - unsigned pushbuf_avail = PUSHBUF_DWORDS - 2 * nctx->bo.count, + unsigned pushbuf_avail = PUSHBUF_DWORDS - 2 * (nctx->bo.count + + render->attr_count), vert_avail = get_max_vertices(ctx, NULL, pushbuf_avail), idx_avail = get_max_vertices(ctx, ib, pushbuf_avail); int stride; /* Try to keep client buffers smaller than the scratch BOs. */ if (render->mode == VBO && - (stride = get_max_client_stride(ctx))) + (stride = get_max_client_stride(ctx, arrays))) vert_avail = MIN2(vert_avail, RENDER_SCRATCH_SIZE / stride); @@ -371,8 +377,6 @@ vbo_draw_vbo(GLcontext *ctx, const struct gl_client_array **arrays, dispatch(ctx, start, delta, count); BATCH_END(); } - - FIRE_RING(chan); } /* Immediate rendering path. */ @@ -416,8 +420,6 @@ vbo_draw_imm(GLcontext *ctx, const struct gl_client_array **arrays, BATCH_END(); } - - FIRE_RING(chan); } /* draw_prims entry point when we're doing hw-tnl. */ diff --git a/src/mesa/drivers/dri/nouveau/nv04_context.c b/src/mesa/drivers/dri/nouveau/nv04_context.c index 3624b3af92..6834f7cd3d 100644 --- a/src/mesa/drivers/dri/nouveau/nv04_context.c +++ b/src/mesa/drivers/dri/nouveau/nv04_context.c @@ -276,6 +276,10 @@ const struct nouveau_driver nv04_driver = { nouveau_emit_nothing, nouveau_emit_nothing, nouveau_emit_nothing, + nouveau_emit_nothing, + nouveau_emit_nothing, + nouveau_emit_nothing, + nouveau_emit_nothing, nv04_emit_tex_obj, nv04_emit_tex_obj, nouveau_emit_nothing, diff --git a/src/mesa/drivers/dri/nouveau/nv10_context.c b/src/mesa/drivers/dri/nouveau/nv10_context.c index 860d0aeb8f..b6d10361de 100644 --- a/src/mesa/drivers/dri/nouveau/nv10_context.c +++ b/src/mesa/drivers/dri/nouveau/nv10_context.c @@ -212,7 +212,7 @@ nv10_hwctx_init(GLcontext *ctx) OUT_RING(chan, 0); BEGIN_RING(chan, celsius, NV10TCL_CULL_FACE_ENABLE, 1); OUT_RING(chan, 0); - BEGIN_RING(chan, celsius, NV10TCL_TX_GEN_S(0), 8); + BEGIN_RING(chan, celsius, NV10TCL_TX_GEN_MODE_S(0), 8); for (i = 0; i < 8; i++) OUT_RING(chan, 0); @@ -412,6 +412,10 @@ const struct nouveau_driver nv10_driver = { nv10_emit_tex_gen, nouveau_emit_nothing, nouveau_emit_nothing, + nv10_emit_tex_mat, + nv10_emit_tex_mat, + nouveau_emit_nothing, + nouveau_emit_nothing, nv10_emit_tex_obj, nv10_emit_tex_obj, nouveau_emit_nothing, diff --git a/src/mesa/drivers/dri/nouveau/nv10_driver.h b/src/mesa/drivers/dri/nouveau/nv10_driver.h index d662712533..cefd6c6fba 100644 --- a/src/mesa/drivers/dri/nouveau/nv10_driver.h +++ b/src/mesa/drivers/dri/nouveau/nv10_driver.h @@ -134,6 +134,9 @@ void nv10_emit_tex_gen(GLcontext *ctx, int emit); void +nv10_emit_tex_mat(GLcontext *ctx, int emit); + +void nv10_emit_tex_obj(GLcontext *ctx, int emit); /* nv10_state_tnl.c */ diff --git a/src/mesa/drivers/dri/nouveau/nv10_state_fb.c b/src/mesa/drivers/dri/nouveau/nv10_state_fb.c index 6bd383ebcd..f7c3d36e1c 100644 --- a/src/mesa/drivers/dri/nouveau/nv10_state_fb.c +++ b/src/mesa/drivers/dri/nouveau/nv10_state_fb.c @@ -71,6 +71,7 @@ setup_lma_buffer(GLcontext *ctx) nouveau_bo_markl(bctx, celsius, NV17TCL_LMA_DEPTH_BUFFER_OFFSET, nfb->lma_bo, 0, NOUVEAU_BO_VRAM | NOUVEAU_BO_RDWR); + WAIT_RING(chan, 9); BEGIN_RING(chan, celsius, NV17TCL_LMA_DEPTH_WINDOW_X, 4); OUT_RINGf(chan, - 1792); OUT_RINGf(chan, - 2304 + fb->Height); diff --git a/src/mesa/drivers/dri/nouveau/nv10_state_tex.c b/src/mesa/drivers/dri/nouveau/nv10_state_tex.c index 02a5ca797a..35f41d7295 100644 --- a/src/mesa/drivers/dri/nouveau/nv10_state_tex.c +++ b/src/mesa/drivers/dri/nouveau/nv10_state_tex.c @@ -32,9 +32,67 @@ #include "nouveau_util.h" #include "nv10_driver.h" +#define TX_GEN_MODE(i, j) (NV10TCL_TX_GEN_MODE_S(i) + 4 * (j)) +#define TX_GEN_COEFF(i, j) (NV10TCL_TX_GEN_COEFF_S_A(i) + 16 * (j)) +#define TX_MATRIX(i) (NV10TCL_TX0_MATRIX(0) + 64 * (i)) + void nv10_emit_tex_gen(GLcontext *ctx, int emit) { + const int i = emit - NOUVEAU_STATE_TEX_GEN0; + struct nouveau_context *nctx = to_nouveau_context(ctx); + struct nouveau_channel *chan = context_chan(ctx); + struct nouveau_grobj *celsius = context_eng3d(ctx); + struct gl_texture_unit *unit = &ctx->Texture.Unit[i]; + int j; + + for (j = 0; j < 4; j++) { + if (nctx->fallback == HWTNL && (unit->TexGenEnabled & 1 << j)) { + struct gl_texgen *coord = get_texgen_coord(unit, j); + float *k = get_texgen_coeff(coord); + + if (k) { + BEGIN_RING(chan, celsius, + TX_GEN_COEFF(i, j), 4); + OUT_RINGf(chan, k[0]); + OUT_RINGf(chan, k[1]); + OUT_RINGf(chan, k[2]); + OUT_RINGf(chan, k[3]); + } + + BEGIN_RING(chan, celsius, TX_GEN_MODE(i, j), 1); + OUT_RING(chan, nvgl_texgen_mode(coord->Mode)); + + } else { + BEGIN_RING(chan, celsius, TX_GEN_MODE(i, j), 1); + OUT_RING(chan, 0); + } + } + + context_dirty_i(ctx, TEX_MAT, i); +} + +void +nv10_emit_tex_mat(GLcontext *ctx, int emit) +{ + const int i = emit - NOUVEAU_STATE_TEX_MAT0; + struct nouveau_context *nctx = to_nouveau_context(ctx); + struct nouveau_channel *chan = context_chan(ctx); + struct nouveau_grobj *celsius = context_eng3d(ctx); + + if (nctx->fallback == HWTNL && + ((ctx->Texture._TexMatEnabled & 1 << i) || + ctx->Texture.Unit[i]._GenFlags)) { + BEGIN_RING(chan, celsius, NV10TCL_TX_MATRIX_ENABLE(i), 1); + OUT_RING(chan, 1); + + BEGIN_RING(chan, celsius, TX_MATRIX(i), 16); + OUT_RINGm(chan, ctx->TextureMatrixStack[i].Top->m); + + } else { + BEGIN_RING(chan, celsius, NV10TCL_TX_MATRIX_ENABLE(i), 1); + OUT_RING(chan, 0); + } } static uint32_t diff --git a/src/mesa/drivers/dri/nouveau/nv10_state_tnl.c b/src/mesa/drivers/dri/nouveau/nv10_state_tnl.c index 406e24c455..2624c9bf30 100644 --- a/src/mesa/drivers/dri/nouveau/nv10_state_tnl.c +++ b/src/mesa/drivers/dri/nouveau/nv10_state_tnl.c @@ -474,12 +474,14 @@ nv10_emit_modelview(GLcontext *ctx, int emit) if (nctx->fallback != HWTNL) return; - if (ctx->Light._NeedEyeCoords || ctx->Fog.Enabled) { + if (ctx->Light._NeedEyeCoords || ctx->Fog.Enabled || + (ctx->Texture._GenFlags & TEXGEN_NEED_EYE_COORD)) { BEGIN_RING(chan, celsius, NV10TCL_MODELVIEW0_MATRIX(0), 16); OUT_RINGm(chan, m->m); } - if (ctx->Light.Enabled) { + if (ctx->Light.Enabled || + (ctx->Texture._GenFlags & TEXGEN_NEED_EYE_COORD)) { int i, j; BEGIN_RING(chan, celsius, diff --git a/src/mesa/drivers/dri/nouveau/nv20_context.c b/src/mesa/drivers/dri/nouveau/nv20_context.c index db39ef7075..789dcaa6b4 100644 --- a/src/mesa/drivers/dri/nouveau/nv20_context.c +++ b/src/mesa/drivers/dri/nouveau/nv20_context.c @@ -297,9 +297,9 @@ nv20_hwctx_init(GLcontext *ctx) BEGIN_RING(chan, kelvin, NV20TCL_POLYGON_STIPPLE_ENABLE, 1); OUT_RING (chan, 0); - BEGIN_RING(chan, kelvin, NV20TCL_TX_GEN_S(0), - 4 * NV20TCL_TX_GEN_S__SIZE); - for (i=0; i < 4 * NV20TCL_TX_GEN_S__SIZE; i++) + BEGIN_RING(chan, kelvin, NV20TCL_TX_GEN_MODE_S(0), + 4 * NV20TCL_TX_GEN_MODE_S__SIZE); + for (i=0; i < 4 * NV20TCL_TX_GEN_MODE_S__SIZE; i++) OUT_RING(chan, 0); BEGIN_RING(chan, kelvin, NV20TCL_FOG_EQUATION_CONSTANT, 3); @@ -497,10 +497,14 @@ const struct nouveau_driver nv20_driver = { nv20_emit_tex_env, nv20_emit_tex_env, nv20_emit_tex_env, - nv10_emit_tex_gen, - nv10_emit_tex_gen, - nv10_emit_tex_gen, - nv10_emit_tex_gen, + nv20_emit_tex_gen, + nv20_emit_tex_gen, + nv20_emit_tex_gen, + nv20_emit_tex_gen, + nv20_emit_tex_mat, + nv20_emit_tex_mat, + nv20_emit_tex_mat, + nv20_emit_tex_mat, nv20_emit_tex_obj, nv20_emit_tex_obj, nv20_emit_tex_obj, diff --git a/src/mesa/drivers/dri/nouveau/nv20_driver.h b/src/mesa/drivers/dri/nouveau/nv20_driver.h index 18574e9be6..8adecef2c4 100644 --- a/src/mesa/drivers/dri/nouveau/nv20_driver.h +++ b/src/mesa/drivers/dri/nouveau/nv20_driver.h @@ -68,6 +68,12 @@ nv20_emit_frag(GLcontext *ctx, int emit); /* nv20_state_tex.c */ void +nv20_emit_tex_gen(GLcontext *ctx, int emit); + +void +nv20_emit_tex_mat(GLcontext *ctx, int emit); + +void nv20_emit_tex_obj(GLcontext *ctx, int emit); void diff --git a/src/mesa/drivers/dri/nouveau/nv20_state_tex.c b/src/mesa/drivers/dri/nouveau/nv20_state_tex.c index 92870105f9..bb8a79c2c9 100644 --- a/src/mesa/drivers/dri/nouveau/nv20_state_tex.c +++ b/src/mesa/drivers/dri/nouveau/nv20_state_tex.c @@ -32,6 +32,65 @@ #include "nouveau_util.h" #include "nv20_driver.h" +#define TX_GEN_MODE(i, j) (NV20TCL_TX_GEN_MODE_S(i) + 4 * (j)) +#define TX_GEN_COEFF(i, j) (NV20TCL_TX_GEN_COEFF_S_A(i) + 16 * (j)) +#define TX_MATRIX(i) (NV20TCL_TX0_MATRIX(0) + 64 * (i)) + +void +nv20_emit_tex_gen(GLcontext *ctx, int emit) +{ + const int i = emit - NOUVEAU_STATE_TEX_GEN0; + struct nouveau_context *nctx = to_nouveau_context(ctx); + struct nouveau_channel *chan = context_chan(ctx); + struct nouveau_grobj *kelvin = context_eng3d(ctx); + struct gl_texture_unit *unit = &ctx->Texture.Unit[i]; + int j; + + for (j = 0; j < 4; j++) { + if (nctx->fallback == HWTNL && (unit->TexGenEnabled & 1 << j)) { + struct gl_texgen *coord = get_texgen_coord(unit, j); + float *k = get_texgen_coeff(coord); + + if (k) { + BEGIN_RING(chan, kelvin, TX_GEN_COEFF(i, j), 4); + OUT_RINGf(chan, k[0]); + OUT_RINGf(chan, k[1]); + OUT_RINGf(chan, k[2]); + OUT_RINGf(chan, k[3]); + } + + BEGIN_RING(chan, kelvin, TX_GEN_MODE(i, j), 1); + OUT_RING(chan, nvgl_texgen_mode(coord->Mode)); + + } else { + BEGIN_RING(chan, kelvin, TX_GEN_MODE(i, j), 1); + OUT_RING(chan, 0); + } + } +} + +void +nv20_emit_tex_mat(GLcontext *ctx, int emit) +{ + const int i = emit - NOUVEAU_STATE_TEX_MAT0; + struct nouveau_context *nctx = to_nouveau_context(ctx); + struct nouveau_channel *chan = context_chan(ctx); + struct nouveau_grobj *kelvin = context_eng3d(ctx); + + if (nctx->fallback == HWTNL && + (ctx->Texture._TexMatEnabled & 1 << i)) { + BEGIN_RING(chan, kelvin, NV20TCL_TX_MATRIX_ENABLE(i), 1); + OUT_RING(chan, 1); + + BEGIN_RING(chan, kelvin, TX_MATRIX(i), 16); + OUT_RINGm(chan, ctx->TextureMatrixStack[i].Top->m); + + } else { + BEGIN_RING(chan, kelvin, NV20TCL_TX_MATRIX_ENABLE(i), 1); + OUT_RING(chan, 0); + } +} + static uint32_t get_tex_format_pot(struct gl_texture_image *ti) { diff --git a/src/mesa/drivers/dri/nouveau/nv20_state_tnl.c b/src/mesa/drivers/dri/nouveau/nv20_state_tnl.c index 43f8c72312..df22adf272 100644 --- a/src/mesa/drivers/dri/nouveau/nv20_state_tnl.c +++ b/src/mesa/drivers/dri/nouveau/nv20_state_tnl.c @@ -359,12 +359,14 @@ nv20_emit_modelview(GLcontext *ctx, int emit) if (nctx->fallback != HWTNL) return; - if (ctx->Light._NeedEyeCoords || ctx->Fog.Enabled) { + if (ctx->Light._NeedEyeCoords || ctx->Fog.Enabled || + (ctx->Texture._GenFlags & TEXGEN_NEED_EYE_COORD)) { BEGIN_RING(chan, kelvin, NV20TCL_MODELVIEW0_MATRIX(0), 16); OUT_RINGm(chan, m->m); } - if (ctx->Light.Enabled) { + if (ctx->Light.Enabled || + (ctx->Texture._GenFlags & TEXGEN_NEED_EYE_COORD)) { int i, j; BEGIN_RING(chan, kelvin, diff --git a/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c b/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c index 710cae727a..4e84eefd65 100644 --- a/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c +++ b/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c @@ -469,9 +469,8 @@ void r500BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compi if (compiler->Base.Error) return; - assert(code->inst_end >= 0); - - if ((code->inst[code->inst_end].inst0 & R500_INST_TYPE_MASK) != R500_INST_TYPE_OUT) { + if (code->inst_end == -1 || + (code->inst[code->inst_end].inst0 & R500_INST_TYPE_MASK) != R500_INST_TYPE_OUT) { /* This may happen when dead-code elimination is disabled or * when most of the fragment program logic is leading to a KIL */ if (code->inst_end >= 511) { diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c index 71d1514fe4..dbaba2975e 100644 --- a/src/mesa/main/bufferobj.c +++ b/src/mesa/main/bufferobj.c @@ -32,6 +32,7 @@ #include "glheader.h" +#include "enums.h" #include "hash.h" #include "imports.h" #include "image.h" @@ -1376,31 +1377,49 @@ _mesa_GetBufferParameterivARB(GLenum target, GLenum pname, GLint *params) bufObj = get_buffer(ctx, target); if (!bufObj) { - _mesa_error(ctx, GL_INVALID_ENUM, "GetBufferParameterivARB(target)" ); + _mesa_error(ctx, GL_INVALID_ENUM, "glGetBufferParameterivARB(target)" ); return; } if (!_mesa_is_bufferobj(bufObj)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "GetBufferParameterivARB" ); + _mesa_error(ctx, GL_INVALID_OPERATION, "glGetBufferParameterivARB" ); return; } switch (pname) { case GL_BUFFER_SIZE_ARB: *params = (GLint) bufObj->Size; - break; + return; case GL_BUFFER_USAGE_ARB: *params = bufObj->Usage; - break; + return; case GL_BUFFER_ACCESS_ARB: *params = simplified_access_mode(bufObj->AccessFlags); - break; + return; case GL_BUFFER_MAPPED_ARB: *params = _mesa_bufferobj_mapped(bufObj); - break; - default: - _mesa_error(ctx, GL_INVALID_ENUM, "glGetBufferParameterivARB(pname)"); return; + case GL_BUFFER_ACCESS_FLAGS: + if (ctx->VersionMajor < 3) + goto invalid_pname; + *params = bufObj->AccessFlags; + return; + case GL_BUFFER_MAP_OFFSET: + if (ctx->VersionMajor < 3) + goto invalid_pname; + *params = (GLint) bufObj->Offset; + return; + case GL_BUFFER_MAP_LENGTH: + if (ctx->VersionMajor < 3) + goto invalid_pname; + *params = (GLint) bufObj->Length; + return; + default: + ; /* fall-through */ } + +invalid_pname: + _mesa_error(ctx, GL_INVALID_ENUM, "glGetBufferParameterivARB(pname=%s)", + _mesa_lookup_enum_by_nr(pname)); } @@ -1418,31 +1437,50 @@ _mesa_GetBufferParameteri64v(GLenum target, GLenum pname, GLint64 *params) bufObj = get_buffer(ctx, target); if (!bufObj) { - _mesa_error(ctx, GL_INVALID_ENUM, "GetBufferParameteri64v(target)" ); + _mesa_error(ctx, GL_INVALID_ENUM, "glGetBufferParameteri64v(target)" ); return; } if (!_mesa_is_bufferobj(bufObj)) { - _mesa_error(ctx, GL_INVALID_OPERATION, "GetBufferParameteri64v" ); + _mesa_error(ctx, GL_INVALID_OPERATION, "glGetBufferParameteri64v" ); return; } switch (pname) { case GL_BUFFER_SIZE_ARB: *params = bufObj->Size; - break; + return; case GL_BUFFER_USAGE_ARB: *params = bufObj->Usage; - break; + return; case GL_BUFFER_ACCESS_ARB: *params = simplified_access_mode(bufObj->AccessFlags); - break; + return; + case GL_BUFFER_ACCESS_FLAGS: + if (ctx->VersionMajor < 3) + goto invalid_pname; + *params = bufObj->AccessFlags; + return; case GL_BUFFER_MAPPED_ARB: *params = _mesa_bufferobj_mapped(bufObj); - break; + return; + case GL_BUFFER_MAP_OFFSET: + if (ctx->VersionMajor < 3) + goto invalid_pname; + *params = bufObj->Offset; + return; + case GL_BUFFER_MAP_LENGTH: + if (ctx->VersionMajor < 3) + goto invalid_pname; + *params = bufObj->Length; + return; default: - _mesa_error(ctx, GL_INVALID_ENUM, "glGetBufferParameteri64v(pname)"); + ; /* fall-through */ return; } + +invalid_pname: + _mesa_error(ctx, GL_INVALID_ENUM, "glGetBufferParameteri64v(pname=%s)", + _mesa_lookup_enum_by_nr(pname)); } diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c index 7c442e390c..0782734869 100644 --- a/src/mesa/main/fbobject.c +++ b/src/mesa/main/fbobject.c @@ -180,8 +180,12 @@ _mesa_get_attachment(GLcontext *ctx, struct gl_framebuffer *fb, return &fb->Attachment[BUFFER_COLOR0 + i]; case GL_DEPTH_STENCIL_ATTACHMENT: /* fall-through */ + case GL_DEPTH_BUFFER: + /* fall-through / new in GL 3.0 */ case GL_DEPTH_ATTACHMENT_EXT: return &fb->Attachment[BUFFER_DEPTH]; + case GL_STENCIL_BUFFER: + /* fall-through / new in GL 3.0 */ case GL_STENCIL_ATTACHMENT_EXT: return &fb->Attachment[BUFFER_STENCIL]; default: diff --git a/src/mesa/main/get.c b/src/mesa/main/get.c index edc4400912..523dc2e4f7 100644 --- a/src/mesa/main/get.c +++ b/src/mesa/main/get.c @@ -917,6 +917,10 @@ _mesa_GetBooleanv( GLenum pname, GLboolean *params ) CHECK_EXT1(MESA_texture_array, "GetBooleanv"); params[0] = INT_TO_BOOLEAN(ctx->Texture.Unit[ctx->Texture.CurrentUnit].CurrentTex[TEXTURE_2D_ARRAY_INDEX]->Name); break; + case GL_MAX_ARRAY_TEXTURE_LAYERS_EXT: + CHECK_EXT1(MESA_texture_array, "GetBooleanv"); + params[0] = INT_TO_BOOLEAN(ctx->Const.MaxArrayTextureLayers); + break; case GL_TEXTURE_GEN_S: params[0] = ((ctx->Texture.Unit[ctx->Texture.CurrentUnit].TexGenEnabled & S_BIT) ? 1 : 0); break; @@ -1932,6 +1936,9 @@ _mesa_GetBooleanv( GLenum pname, GLboolean *params ) case GL_MINOR_VERSION: params[0] = INT_TO_BOOLEAN(ctx->VersionMinor); break; + case GL_CONTEXT_FLAGS: + params[0] = INT_TO_BOOLEAN(ctx->Const.ContextFlags); + break; default: _mesa_error(ctx, GL_INVALID_ENUM, "glGetBooleanv(pname=0x%x)", pname); } @@ -2786,6 +2793,10 @@ _mesa_GetFloatv( GLenum pname, GLfloat *params ) CHECK_EXT1(MESA_texture_array, "GetFloatv"); params[0] = (GLfloat)(ctx->Texture.Unit[ctx->Texture.CurrentUnit].CurrentTex[TEXTURE_2D_ARRAY_INDEX]->Name); break; + case GL_MAX_ARRAY_TEXTURE_LAYERS_EXT: + CHECK_EXT1(MESA_texture_array, "GetFloatv"); + params[0] = (GLfloat)(ctx->Const.MaxArrayTextureLayers); + break; case GL_TEXTURE_GEN_S: params[0] = BOOLEAN_TO_FLOAT(((ctx->Texture.Unit[ctx->Texture.CurrentUnit].TexGenEnabled & S_BIT) ? 1 : 0)); break; @@ -3801,6 +3812,9 @@ _mesa_GetFloatv( GLenum pname, GLfloat *params ) case GL_MINOR_VERSION: params[0] = (GLfloat)(ctx->VersionMinor); break; + case GL_CONTEXT_FLAGS: + params[0] = (GLfloat)(ctx->Const.ContextFlags); + break; default: _mesa_error(ctx, GL_INVALID_ENUM, "glGetFloatv(pname=0x%x)", pname); } @@ -4655,6 +4669,10 @@ _mesa_GetIntegerv( GLenum pname, GLint *params ) CHECK_EXT1(MESA_texture_array, "GetIntegerv"); params[0] = ctx->Texture.Unit[ctx->Texture.CurrentUnit].CurrentTex[TEXTURE_2D_ARRAY_INDEX]->Name; break; + case GL_MAX_ARRAY_TEXTURE_LAYERS_EXT: + CHECK_EXT1(MESA_texture_array, "GetIntegerv"); + params[0] = ctx->Const.MaxArrayTextureLayers; + break; case GL_TEXTURE_GEN_S: params[0] = BOOLEAN_TO_INT(((ctx->Texture.Unit[ctx->Texture.CurrentUnit].TexGenEnabled & S_BIT) ? 1 : 0)); break; @@ -5670,6 +5688,9 @@ _mesa_GetIntegerv( GLenum pname, GLint *params ) case GL_MINOR_VERSION: params[0] = ctx->VersionMinor; break; + case GL_CONTEXT_FLAGS: + params[0] = ctx->Const.ContextFlags; + break; default: _mesa_error(ctx, GL_INVALID_ENUM, "glGetIntegerv(pname=0x%x)", pname); } @@ -6525,6 +6546,10 @@ _mesa_GetInteger64v( GLenum pname, GLint64 *params ) CHECK_EXT1(MESA_texture_array, "GetInteger64v"); params[0] = (GLint64)(ctx->Texture.Unit[ctx->Texture.CurrentUnit].CurrentTex[TEXTURE_2D_ARRAY_INDEX]->Name); break; + case GL_MAX_ARRAY_TEXTURE_LAYERS_EXT: + CHECK_EXT1(MESA_texture_array, "GetInteger64v"); + params[0] = (GLint64)(ctx->Const.MaxArrayTextureLayers); + break; case GL_TEXTURE_GEN_S: params[0] = BOOLEAN_TO_INT64(((ctx->Texture.Unit[ctx->Texture.CurrentUnit].TexGenEnabled & S_BIT) ? 1 : 0)); break; @@ -7540,6 +7565,9 @@ _mesa_GetInteger64v( GLenum pname, GLint64 *params ) case GL_MINOR_VERSION: params[0] = (GLint64)(ctx->VersionMinor); break; + case GL_CONTEXT_FLAGS: + params[0] = (GLint64)(ctx->Const.ContextFlags); + break; default: _mesa_error(ctx, GL_INVALID_ENUM, "glGetInteger64v(pname=0x%x)", pname); } diff --git a/src/mesa/main/get_gen.py b/src/mesa/main/get_gen.py index 9d5a51d58c..0ef9d8fe94 100644 --- a/src/mesa/main/get_gen.py +++ b/src/mesa/main/get_gen.py @@ -457,6 +457,8 @@ StateVars = [ ["ctx->Texture.Unit[ctx->Texture.CurrentUnit].CurrentTex[TEXTURE_1D_ARRAY_INDEX]->Name"], "", ["MESA_texture_array"] ), ( "GL_TEXTURE_BINDING_2D_ARRAY_EXT", GLint, ["ctx->Texture.Unit[ctx->Texture.CurrentUnit].CurrentTex[TEXTURE_2D_ARRAY_INDEX]->Name"], "", ["MESA_texture_array"] ), + ( "GL_MAX_ARRAY_TEXTURE_LAYERS_EXT", GLint, + ["ctx->Const.MaxArrayTextureLayers"], "", ["MESA_texture_array"] ), ( "GL_TEXTURE_GEN_S", GLboolean, ["((ctx->Texture.Unit[ctx->Texture.CurrentUnit].TexGenEnabled & S_BIT) ? 1 : 0)"], "", None ), ( "GL_TEXTURE_GEN_T", GLboolean, @@ -1062,7 +1064,8 @@ StateVars = [ # GL3 ( "GL_NUM_EXTENSIONS", GLint, ["_mesa_get_extension_count(ctx)"], "", None ), ( "GL_MAJOR_VERSION", GLint, ["ctx->VersionMajor"], "", None ), - ( "GL_MINOR_VERSION", GLint, ["ctx->VersionMinor"], "", None ) + ( "GL_MINOR_VERSION", GLint, ["ctx->VersionMinor"], "", None ), + ( "GL_CONTEXT_FLAGS", GLint, ["ctx->Const.ContextFlags"], "", None ) ] diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h index 9d9b475dd1..82e004f348 100644 --- a/src/mesa/main/mtypes.h +++ b/src/mesa/main/mtypes.h @@ -2385,6 +2385,9 @@ struct gl_constants /**< GL_EXT_provoking_vertex */ GLboolean QuadsFollowProvokingVertexConvention; + + /**< OpenGL version 3.x */ + GLbitfield ContextFlags; /**< Ex: GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT */ }; diff --git a/src/mesa/main/texparam.c b/src/mesa/main/texparam.c index 0fde89b507..714c4cfd52 100644 --- a/src/mesa/main/texparam.c +++ b/src/mesa/main/texparam.c @@ -940,6 +940,18 @@ _mesa_GetTexLevelParameteriv( GLenum target, GLint level, "glGetTexLevelParameter[if]v(pname)"); } break; + case GL_TEXTURE_SHARED_SIZE: + if (ctx->VersionMajor >= 3) { + /* XXX return number of exponent bits for shared exponent texture + * formats, like GL_RGB9_E5. + */ + *params = 0; + } + else { + _mesa_error(ctx, GL_INVALID_ENUM, + "glGetTexLevelParameter[if]v(pname)"); + } + break; /* GL_ARB_texture_compression */ case GL_TEXTURE_COMPRESSED_IMAGE_SIZE: |