diff options
Diffstat (limited to 'src')
233 files changed, 3256 insertions, 2327 deletions
diff --git a/src/gallium/auxiliary/draw/draw_pipe_aaline.c b/src/gallium/auxiliary/draw/draw_pipe_aaline.c index 70d7dbdfc75..f4615064e65 100644 --- a/src/gallium/auxiliary/draw/draw_pipe_aaline.c +++ b/src/gallium/auxiliary/draw/draw_pipe_aaline.c @@ -436,9 +436,9 @@ aaline_create_texture(struct aaline_stage *aaline) /* This texture is new, no need to flush. */ - transfer = screen->get_tex_transfer(screen, aaline->texture, 0, level, 0, + transfer = pipe->get_tex_transfer(pipe, aaline->texture, 0, level, 0, PIPE_TRANSFER_WRITE, 0, 0, size, size); - data = screen->transfer_map(screen, transfer); + data = pipe->transfer_map(pipe, transfer); if (data == NULL) return FALSE; @@ -462,8 +462,8 @@ aaline_create_texture(struct aaline_stage *aaline) } /* unmap */ - screen->transfer_unmap(screen, transfer); - screen->tex_transfer_destroy(transfer); + pipe->transfer_unmap(pipe, transfer); + pipe->tex_transfer_destroy(pipe, transfer); } return TRUE; } diff --git a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c index e03081d65c5..794fd81d70f 100644 --- a/src/gallium/auxiliary/draw/draw_pipe_pstipple.c +++ b/src/gallium/auxiliary/draw/draw_pipe_pstipple.c @@ -377,19 +377,21 @@ pstip_update_texture(struct pstip_stage *pstip) { static const uint bit31 = 1 << 31; struct pipe_context *pipe = pstip->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *transfer; const uint *stipple = pstip->state.stipple->stipple; uint i, j; ubyte *data; /* XXX: want to avoid flushing just because we use stipple: + * + * Flush should no longer be necessary if driver is properly + * interleaving drawing and transfers on a given context: */ pipe->flush( pipe, PIPE_FLUSH_TEXTURE_CACHE, NULL ); - transfer = screen->get_tex_transfer(screen, pstip->texture, 0, 0, 0, - PIPE_TRANSFER_WRITE, 0, 0, 32, 32); - data = screen->transfer_map(screen, transfer); + transfer = pipe->get_tex_transfer(pipe, pstip->texture, 0, 0, 0, + PIPE_TRANSFER_WRITE, 0, 0, 32, 32); + data = pipe->transfer_map(pipe, transfer); /* * Load alpha texture. @@ -411,8 +413,8 @@ pstip_update_texture(struct pstip_stage *pstip) } /* unmap */ - screen->transfer_unmap(screen, transfer); - screen->tex_transfer_destroy(transfer); + pipe->transfer_unmap(pipe, transfer); + pipe->tex_transfer_destroy(pipe, transfer); } diff --git a/src/gallium/auxiliary/gallivm/lp_bld_alpha.h b/src/gallium/auxiliary/gallivm/lp_bld_alpha.h index 634575670db..fe3cedcc48c 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_alpha.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_alpha.h @@ -35,7 +35,7 @@ #define LP_BLD_ALPHA_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct pipe_alpha_state; struct lp_type; diff --git a/src/gallium/auxiliary/gallivm/lp_bld_arit.c b/src/gallium/auxiliary/gallivm/lp_bld_arit.c index e2c67883972..aa47338b323 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_arit.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_arit.c @@ -232,6 +232,37 @@ lp_build_add(struct lp_build_context *bld, } +/** Return the sum of the elements of a */ +LLVMValueRef +lp_build_sum_vector(struct lp_build_context *bld, + LLVMValueRef a) +{ + const struct lp_type type = bld->type; + LLVMValueRef index, res; + int i; + + if (a == bld->zero) + return bld->zero; + if (a == bld->undef) + return bld->undef; + assert(type.length > 1); + + assert(!bld->type.norm); + + index = LLVMConstInt(LLVMInt32Type(), 0, 0); + res = LLVMBuildExtractElement(bld->builder, a, index, ""); + + for (i = 1; i < type.length; i++) { + index = LLVMConstInt(LLVMInt32Type(), i, 0); + res = LLVMBuildAdd(bld->builder, res, + LLVMBuildExtractElement(bld->builder, a, index, ""), + ""); + } + + return res; +} + + /** * Generate a - b */ @@ -689,12 +720,12 @@ lp_build_negate(struct lp_build_context *bld, } +/** Return -1, 0 or +1 depending on the sign of a */ LLVMValueRef lp_build_sgn(struct lp_build_context *bld, LLVMValueRef a) { const struct lp_type type = bld->type; - LLVMTypeRef vec_type = lp_build_vec_type(type); LLVMValueRef cond; LLVMValueRef res; @@ -704,14 +735,29 @@ lp_build_sgn(struct lp_build_context *bld, res = bld->one; } else if(type.floating) { - /* Take the sign bit and add it to 1 constant */ - LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); - LLVMValueRef mask = lp_build_int_const_scalar(type, (unsigned long long)1 << (type.width - 1)); + LLVMTypeRef vec_type; + LLVMTypeRef int_type; + LLVMValueRef mask; LLVMValueRef sign; LLVMValueRef one; - sign = LLVMBuildBitCast(bld->builder, a, int_vec_type, ""); + unsigned long long maskBit = (unsigned long long)1 << (type.width - 1); + + if (type.length == 1) { + int_type = lp_build_int_elem_type(type); + vec_type = lp_build_elem_type(type); + mask = LLVMConstInt(int_type, maskBit, 0); + } + else { + /* vector */ + int_type = lp_build_int_vec_type(type); + vec_type = lp_build_vec_type(type); + mask = lp_build_int_const_scalar(type, maskBit); + } + + /* Take the sign bit and add it to 1 constant */ + sign = LLVMBuildBitCast(bld->builder, a, int_type, ""); sign = LLVMBuildAnd(bld->builder, sign, mask, ""); - one = LLVMConstBitCast(bld->one, int_vec_type); + one = LLVMConstBitCast(bld->one, int_type); res = LLVMBuildOr(bld->builder, sign, one, ""); res = LLVMBuildBitCast(bld->builder, res, vec_type, ""); } @@ -883,6 +929,10 @@ lp_build_floor(struct lp_build_context *bld, assert(type.floating); + if (type.length == 1) { + return LLVMBuildFPTrunc(bld->builder, a, LLVMFloatType(), ""); + } + if(util_cpu_caps.has_sse4_1) return lp_build_round_sse41(bld, a, LP_BUILD_ROUND_SSE41_FLOOR); else { @@ -953,6 +1003,9 @@ lp_build_itrunc(struct lp_build_context *bld, } +/** + * Convert float[] to int[] with round(). + */ LLVMValueRef lp_build_iround(struct lp_build_context *bld, LLVMValueRef a) @@ -1013,6 +1066,14 @@ lp_build_ifloor(struct lp_build_context *bld, LLVMValueRef res; assert(type.floating); + + if (type.length == 1) { + /* scalar float to int */ + LLVMTypeRef int_type = LLVMIntType(type.width); + res = LLVMBuildFPToSI(bld->builder, a, int_type, ""); + return res; + } + assert(lp_check_value(type, a)); if(util_cpu_caps.has_sse4_1) { diff --git a/src/gallium/auxiliary/gallivm/lp_bld_arit.h b/src/gallium/auxiliary/gallivm/lp_bld_arit.h index 55385e3a66a..7a10fe12209 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_arit.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_arit.h @@ -37,7 +37,7 @@ #define LP_BLD_ARIT_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct lp_type; @@ -57,6 +57,10 @@ lp_build_add(struct lp_build_context *bld, LLVMValueRef b); LLVMValueRef +lp_build_sum_vector(struct lp_build_context *bld, + LLVMValueRef a); + +LLVMValueRef lp_build_sub(struct lp_build_context *bld, LLVMValueRef a, LLVMValueRef b); diff --git a/src/gallium/auxiliary/gallivm/lp_bld_blend.h b/src/gallium/auxiliary/gallivm/lp_bld_blend.h index da272e549f3..5a9e1c1fb2f 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_blend.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_blend.h @@ -40,7 +40,7 @@ * for a standalone example. */ -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include "pipe/p_format.h" diff --git a/src/gallium/auxiliary/gallivm/lp_bld_const.c b/src/gallium/auxiliary/gallivm/lp_bld_const.c index 53447757e8e..8a275fa72f3 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_const.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_const.c @@ -221,8 +221,16 @@ lp_build_undef(struct lp_type type) LLVMValueRef lp_build_zero(struct lp_type type) { - LLVMTypeRef vec_type = lp_build_vec_type(type); - return LLVMConstNull(vec_type); + if (type.length == 1) { + if (type.floating) + return LLVMConstReal(LLVMFloatType(), 0.0); + else + return LLVMConstInt(LLVMIntType(type.width), 0, 0); + } + else { + LLVMTypeRef vec_type = lp_build_vec_type(type); + return LLVMConstNull(vec_type); + } } diff --git a/src/gallium/auxiliary/gallivm/lp_bld_const.h b/src/gallium/auxiliary/gallivm/lp_bld_const.h index cb8e1c7b006..40786361031 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_const.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_const.h @@ -37,7 +37,7 @@ #define LP_BLD_CONST_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include <pipe/p_compiler.h> diff --git a/src/gallium/auxiliary/gallivm/lp_bld_conv.h b/src/gallium/auxiliary/gallivm/lp_bld_conv.h index 948e68fae4f..78e8155ff73 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_conv.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_conv.h @@ -37,7 +37,7 @@ #define LP_BLD_CONV_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct lp_type; diff --git a/src/gallium/auxiliary/gallivm/lp_bld_debug.h b/src/gallium/auxiliary/gallivm/lp_bld_debug.h index 583e6132b4b..441ad94786f 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_debug.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_debug.h @@ -30,7 +30,7 @@ #define LP_BLD_DEBUG_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include "pipe/p_compiler.h" #include "util/u_string.h" diff --git a/src/gallium/auxiliary/gallivm/lp_bld_depth.h b/src/gallium/auxiliary/gallivm/lp_bld_depth.h index 79d6981bb51..8be80024ae8 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_depth.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_depth.h @@ -36,7 +36,7 @@ #define LP_BLD_DEPTH_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct pipe_depth_state; diff --git a/src/gallium/auxiliary/gallivm/lp_bld_flow.c b/src/gallium/auxiliary/gallivm/lp_bld_flow.c index bc831389085..106fc03e46f 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_flow.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_flow.c @@ -308,7 +308,7 @@ lp_build_flow_scope_end(struct lp_build_flow_context *flow) * Note: this function has no dependencies on the flow code and could * be used elsewhere. */ -static LLVMBasicBlockRef +LLVMBasicBlockRef lp_build_insert_new_block(LLVMBuilderRef builder, const char *name) { LLVMBasicBlockRef current_block; @@ -648,7 +648,9 @@ lp_build_if(struct lp_build_if_state *ctx, ifthen->phi[i] = LLVMBuildPhi(builder, LLVMTypeOf(*flow->variables[i]), ""); /* add add the initial value of the var from the entry block */ - LLVMAddIncoming(ifthen->phi[i], flow->variables[i], &ifthen->entry_block, 1); + if (!LLVMIsUndef(*flow->variables[i])) + LLVMAddIncoming(ifthen->phi[i], flow->variables[i], + &ifthen->entry_block, 1); } /* create/insert true_block before merge_block */ @@ -695,18 +697,21 @@ lp_build_endif(struct lp_build_if_state *ctx) { struct lp_build_flow_context *flow = ctx->flow; struct lp_build_flow_if *ifthen; + LLVMBasicBlockRef curBlock = LLVMGetInsertBlock(ctx->builder); unsigned i; ifthen = &lp_build_flow_pop(flow, LP_BUILD_FLOW_IF)->ifthen; assert(ifthen); + /* Insert branch to the merge block from current block */ + LLVMBuildBr(ctx->builder, ifthen->merge_block); + if (ifthen->false_block) { LLVMPositionBuilderAtEnd(ctx->builder, ifthen->merge_block); /* for each variable, update the Phi node with a (variable, block) pair */ for (i = 0; i < flow->num_variables; i++) { assert(*flow->variables[i]); - LLVMAddIncoming(ifthen->phi[i], flow->variables[i], &ifthen->false_block, 1); - + LLVMAddIncoming(ifthen->phi[i], flow->variables[i], &curBlock, 1); /* replace the variable ref with the phi function */ *flow->variables[i] = ifthen->phi[i]; } @@ -742,15 +747,18 @@ lp_build_endif(struct lp_build_if_state *ctx) ifthen->true_block, ifthen->merge_block); } - /* Append an unconditional Br(anch) instruction on the true_block */ - LLVMPositionBuilderAtEnd(ctx->builder, ifthen->true_block); - LLVMBuildBr(ctx->builder, ifthen->merge_block); + /* Insert branch from end of true_block to merge_block */ if (ifthen->false_block) { - /* Append an unconditional Br(anch) instruction on the false_block */ - LLVMPositionBuilderAtEnd(ctx->builder, ifthen->false_block); + /* Append an unconditional Br(anch) instruction on the true_block */ + LLVMPositionBuilderAtEnd(ctx->builder, ifthen->true_block); LLVMBuildBr(ctx->builder, ifthen->merge_block); } - + else { + /* No else clause. + * Note that we've already inserted the branch at the end of + * true_block. See the very first LLVMBuildBr() call in this function. + */ + } /* Resume building code at end of the ifthen->merge_block */ LLVMPositionBuilderAtEnd(ctx->builder, ifthen->merge_block); diff --git a/src/gallium/auxiliary/gallivm/lp_bld_flow.h b/src/gallium/auxiliary/gallivm/lp_bld_flow.h index 4c225a0d4f9..e1588365491 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_flow.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_flow.h @@ -35,7 +35,7 @@ #define LP_BLD_FLOW_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct lp_type; @@ -145,7 +145,9 @@ lp_build_else(struct lp_build_if_state *ctx); void lp_build_endif(struct lp_build_if_state *ctx); - + +LLVMBasicBlockRef +lp_build_insert_new_block(LLVMBuilderRef builder, const char *name); #endif /* !LP_BLD_FLOW_H */ diff --git a/src/gallium/auxiliary/gallivm/lp_bld_format.h b/src/gallium/auxiliary/gallivm/lp_bld_format.h index 970bee379f5..8972c0dc178 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_format.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_format.h @@ -34,7 +34,7 @@ * Pixel format helpers. */ -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include "pipe/p_format.h" diff --git a/src/gallium/auxiliary/gallivm/lp_bld_interp.h b/src/gallium/auxiliary/gallivm/lp_bld_interp.h index ca958cdf343..177b5e943ee 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_interp.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_interp.h @@ -41,7 +41,7 @@ #define LP_BLD_INTERP_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include "tgsi/tgsi_exec.h" diff --git a/src/gallium/auxiliary/gallivm/lp_bld_intr.h b/src/gallium/auxiliary/gallivm/lp_bld_intr.h index f813f27074b..7d5506c7338 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_intr.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_intr.h @@ -37,7 +37,7 @@ #define LP_BLD_INTR_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" /** diff --git a/src/gallium/auxiliary/gallivm/lp_bld_logic.c b/src/gallium/auxiliary/gallivm/lp_bld_logic.c index 7c585fda788..f3df3dd1388 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_logic.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_logic.c @@ -42,6 +42,26 @@ #include "lp_bld_logic.h" +/* + * XXX + * + * Selection with vector conditional like + * + * select <4 x i1> %C, %A, %B + * + * is valid IR (e.g. llvm/test/Assembler/vector-select.ll), but it is not + * supported on any backend. + * + * Expanding the boolean vector to full SIMD register width, as in + * + * sext <4 x i1> %C to <4 x i32> + * + * is valid and supported (e.g., llvm/test/CodeGen/X86/vec_compare.ll), but + * it causes assertion failures in LLVM 2.6. It appears to work correctly on + * LLVM 2.7. + */ + + /** * Build code to compare two values 'a' and 'b' of 'type' using the given func. * \param func one of PIPE_FUNC_x @@ -54,13 +74,11 @@ lp_build_compare(LLVMBuilderRef builder, LLVMValueRef a, LLVMValueRef b) { - LLVMTypeRef vec_type = lp_build_vec_type(type); LLVMTypeRef int_vec_type = lp_build_int_vec_type(type); LLVMValueRef zeros = LLVMConstNull(int_vec_type); LLVMValueRef ones = LLVMConstAllOnes(int_vec_type); LLVMValueRef cond; LLVMValueRef res; - unsigned i; assert(func >= PIPE_FUNC_NEVER); assert(func <= PIPE_FUNC_ALWAYS); @@ -74,10 +92,12 @@ lp_build_compare(LLVMBuilderRef builder, /* XXX: It is not clear if we should use the ordered or unordered operators */ +#if HAVE_LLVM < 0x0207 #if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64) if(type.width * type.length == 128) { if(type.floating && util_cpu_caps.has_sse) { /* float[4] comparison */ + LLVMTypeRef vec_type = lp_build_vec_type(type); LLVMValueRef args[3]; unsigned cc; boolean swap; @@ -147,6 +167,7 @@ lp_build_compare(LLVMBuilderRef builder, const char *pcmpgt; LLVMValueRef args[2]; LLVMValueRef res; + LLVMTypeRef vec_type = lp_build_vec_type(type); switch (type.width) { case 8: @@ -200,6 +221,7 @@ lp_build_compare(LLVMBuilderRef builder, } } /* if (type.width * type.length == 128) */ #endif +#endif /* HAVE_LLVM < 0x0207 */ if(type.floating) { LLVMRealPredicate op; @@ -233,16 +255,19 @@ lp_build_compare(LLVMBuilderRef builder, return lp_build_undef(type); } -#if 0 - /* XXX: Although valid IR, no LLVM target currently support this */ +#if HAVE_LLVM >= 0x0207 cond = LLVMBuildFCmp(builder, op, a, b, ""); - res = LLVMBuildSelect(builder, cond, ones, zeros, ""); + res = LLVMBuildSExt(builder, cond, int_vec_type, ""); #else - res = LLVMGetUndef(int_vec_type); if (type.length == 1) { - res = LLVMBuildFCmp(builder, op, a, b, ""); + cond = LLVMBuildFCmp(builder, op, a, b, ""); + res = LLVMBuildSExt(builder, cond, int_vec_type, ""); } else { + unsigned i; + + res = LLVMGetUndef(int_vec_type); + debug_printf("%s: warning: using slow element-wise float" " vector comparison\n", __FUNCTION__); for (i = 0; i < type.length; ++i) { @@ -286,16 +311,19 @@ lp_build_compare(LLVMBuilderRef builder, return lp_build_undef(type); } -#if 0 - /* XXX: Although valid IR, no LLVM target currently support this */ +#if HAVE_LLVM >= 0x0207 cond = LLVMBuildICmp(builder, op, a, b, ""); - res = LLVMBuildSelect(builder, cond, ones, zeros, ""); + res = LLVMBuildSExt(builder, cond, int_vec_type, ""); #else - res = LLVMGetUndef(int_vec_type); if (type.length == 1) { - res = LLVMBuildICmp(builder, op, a, b, ""); + cond = LLVMBuildICmp(builder, op, a, b, ""); + res = LLVMBuildSExt(builder, cond, int_vec_type, ""); } else { + unsigned i; + + res = LLVMGetUndef(int_vec_type); + debug_printf("%s: warning: using slow element-wise int" " vector comparison\n", __FUNCTION__); @@ -337,6 +365,8 @@ lp_build_cmp(struct lp_build_context *bld, /** * Return mask ? a : b; + * + * mask is a bitwise mask, composed of 0 or ~0 for each element. */ LLVMValueRef lp_build_select(struct lp_build_context *bld, @@ -351,6 +381,7 @@ lp_build_select(struct lp_build_context *bld, return a; if (type.length == 1) { + mask = LLVMBuildTrunc(bld->builder, mask, LLVMInt1Type(), ""); res = LLVMBuildSelect(bld->builder, mask, a, b, ""); } else { diff --git a/src/gallium/auxiliary/gallivm/lp_bld_logic.h b/src/gallium/auxiliary/gallivm/lp_bld_logic.h index a399ebf39ef..b54ec13b701 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_logic.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_logic.h @@ -37,7 +37,7 @@ #define LP_BLD_LOGIC_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include "pipe/p_defines.h" /* For PIPE_FUNC_xxx */ diff --git a/src/gallium/auxiliary/gallivm/lp_bld_pack.c b/src/gallium/auxiliary/gallivm/lp_bld_pack.c index 4c61d107494..23398f41f99 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_pack.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_pack.c @@ -256,7 +256,7 @@ lp_build_pack2(LLVMBuilderRef builder, LLVMValueRef lo, LLVMValueRef hi) { -#if !(HAVE_LLVM >= 0x0207) +#if HAVE_LLVM < 0x0207 LLVMTypeRef src_vec_type = lp_build_vec_type(src_type); #endif LLVMTypeRef dst_vec_type = lp_build_vec_type(dst_type); diff --git a/src/gallium/auxiliary/gallivm/lp_bld_pack.h b/src/gallium/auxiliary/gallivm/lp_bld_pack.h index fb2a34984a4..346a17d5803 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_pack.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_pack.h @@ -37,7 +37,7 @@ #define LP_BLD_PACK_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct lp_type; diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample.c b/src/gallium/auxiliary/gallivm/lp_bld_sample.c index 311c9f1b9e4..543fd5fea3f 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_sample.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_sample.c @@ -155,14 +155,16 @@ lp_build_gather(LLVMBuilderRef builder, /** * Compute the offset of a pixel. * - * x, y, y_stride are vectors + * x, y, z, y_stride, z_stride are vectors */ LLVMValueRef lp_build_sample_offset(struct lp_build_context *bld, const struct util_format_description *format_desc, LLVMValueRef x, LLVMValueRef y, - LLVMValueRef y_stride) + LLVMValueRef z, + LLVMValueRef y_stride, + LLVMValueRef z_stride) { LLVMValueRef x_stride; LLVMValueRef offset; @@ -178,6 +180,10 @@ lp_build_sample_offset(struct lp_build_context *bld, LLVMValueRef y_offset_lo, y_offset_hi; LLVMValueRef offset_lo, offset_hi; + /* XXX 1D & 3D addressing not done yet */ + assert(!z); + assert(!z_stride); + x_lo = LLVMBuildAnd(bld->builder, x, bld->one, ""); y_lo = LLVMBuildAnd(bld->builder, y, bld->one, ""); @@ -201,13 +207,17 @@ lp_build_sample_offset(struct lp_build_context *bld, offset = lp_build_add(bld, offset_hi, offset_lo); } else { - LLVMValueRef x_offset; - LLVMValueRef y_offset; + offset = lp_build_mul(bld, x, x_stride); - x_offset = lp_build_mul(bld, x, x_stride); - y_offset = lp_build_mul(bld, y, y_stride); + if (y && y_stride) { + LLVMValueRef y_offset = lp_build_mul(bld, y, y_stride); + offset = lp_build_add(bld, offset, y_offset); + } - offset = lp_build_add(bld, x_offset, y_offset); + if (z && z_stride) { + LLVMValueRef z_offset = lp_build_mul(bld, z, z_stride); + offset = lp_build_add(bld, offset, z_offset); + } } return offset; diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample.h b/src/gallium/auxiliary/gallivm/lp_bld_sample.h index 68db91d6fd6..7f08bfaac1f 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_sample.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_sample.h @@ -36,7 +36,7 @@ #define LP_BLD_SAMPLE_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct pipe_texture; struct pipe_sampler_state; @@ -113,9 +113,9 @@ struct lp_sampler_dynamic_state unsigned unit); LLVMValueRef - (*stride)( struct lp_sampler_dynamic_state *state, - LLVMBuilderRef builder, - unsigned unit); + (*row_stride)( struct lp_sampler_dynamic_state *state, + LLVMBuilderRef builder, + unsigned unit); LLVMValueRef (*data_ptr)( struct lp_sampler_dynamic_state *state, @@ -148,7 +148,9 @@ lp_build_sample_offset(struct lp_build_context *bld, const struct util_format_description *format_desc, LLVMValueRef x, LLVMValueRef y, - LLVMValueRef y_stride); + LLVMValueRef z, + LLVMValueRef y_stride, + LLVMValueRef z_stride); void diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c index a965d394f44..72018737a85 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c @@ -48,6 +48,7 @@ #include "lp_bld_logic.h" #include "lp_bld_swizzle.h" #include "lp_bld_pack.h" +#include "lp_bld_flow.h" #include "lp_bld_format.h" #include "lp_bld_sample.h" @@ -139,7 +140,55 @@ lp_build_get_const_mipmap_level(struct lp_build_sample_context *bld, /** - * Gen code to fetch a texel from a texture at int coords (x, y). + * Dereference stride_array[mipmap_level] array to get a stride. + * Return stride as a vector. + */ +static LLVMValueRef +lp_build_get_level_stride_vec(struct lp_build_sample_context *bld, + LLVMValueRef stride_array, LLVMValueRef level) +{ + LLVMValueRef indexes[2], stride; + indexes[0] = LLVMConstInt(LLVMInt32Type(), 0, 0); + indexes[1] = level; + stride = LLVMBuildGEP(bld->builder, stride_array, indexes, 2, ""); + stride = LLVMBuildLoad(bld->builder, stride, ""); + stride = lp_build_broadcast_scalar(&bld->int_coord_bld, stride); + return stride; +} + + +/** Dereference stride_array[0] array to get a stride (as vector). */ +static LLVMValueRef +lp_build_get_const_level_stride_vec(struct lp_build_sample_context *bld, + LLVMValueRef stride_array, int level) +{ + LLVMValueRef lvl = LLVMConstInt(LLVMInt32Type(), level, 0); + return lp_build_get_level_stride_vec(bld, stride_array, lvl); +} + + +static int +texture_dims(enum pipe_texture_target tex) +{ + switch (tex) { + case PIPE_TEXTURE_1D: + return 1; + case PIPE_TEXTURE_2D: + case PIPE_TEXTURE_CUBE: + return 2; + case PIPE_TEXTURE_3D: + return 3; + default: + assert(0 && "bad texture target in texture_dims()"); + return 2; + } +} + + + +/** + * Generate code to fetch a texel from a texture at int coords (x, y, z). + * The computation depends on whether the texture is 1D, 2D or 3D. * The result, texel, will be: * texel[0] = red values * texel[1] = green values @@ -150,12 +199,16 @@ static void lp_build_sample_texel_soa(struct lp_build_sample_context *bld, LLVMValueRef width, LLVMValueRef height, + LLVMValueRef depth, LLVMValueRef x, LLVMValueRef y, + LLVMValueRef z, LLVMValueRef y_stride, + LLVMValueRef z_stride, LLVMValueRef data_ptr, LLVMValueRef *texel) { + const int dims = texture_dims(bld->static_state->target); struct lp_build_context *int_coord_bld = &bld->int_coord_bld; LLVMValueRef offset; LLVMValueRef packed; @@ -169,7 +222,7 @@ lp_build_sample_texel_soa(struct lp_build_sample_context *bld, use_border = LLVMBuildOr(bld->builder, b1, b2, "b1_or_b2"); } - if (wrap_mode_uses_border_color(bld->static_state->wrap_t)) { + if (dims >= 2 && wrap_mode_uses_border_color(bld->static_state->wrap_t)) { LLVMValueRef b1, b2; b1 = lp_build_cmp(int_coord_bld, PIPE_FUNC_LESS, y, int_coord_bld->zero); b2 = lp_build_cmp(int_coord_bld, PIPE_FUNC_GEQUAL, y, height); @@ -182,6 +235,19 @@ lp_build_sample_texel_soa(struct lp_build_sample_context *bld, } } + if (dims == 3 && wrap_mode_uses_border_color(bld->static_state->wrap_r)) { + LLVMValueRef b1, b2; + b1 = lp_build_cmp(int_coord_bld, PIPE_FUNC_LESS, z, int_coord_bld->zero); + b2 = lp_build_cmp(int_coord_bld, PIPE_FUNC_GEQUAL, z, depth); + if (use_border) { + use_border = LLVMBuildOr(bld->builder, use_border, b1, "ub_or_b1"); + use_border = LLVMBuildOr(bld->builder, use_border, b2, "ub_or_b2"); + } + else { + use_border = LLVMBuildOr(bld->builder, b1, b2, "b1_or_b2"); + } + } + /* * Note: if we find an app which frequently samples the texture border * we might want to implement a true conditional here to avoid sampling @@ -197,10 +263,10 @@ lp_build_sample_texel_soa(struct lp_build_sample_context *bld, * the texel color results with the border color. */ - /* convert x,y coords to linear offset from start of texture, in bytes */ + /* convert x,y,z coords to linear offset from start of texture, in bytes */ offset = lp_build_sample_offset(&bld->uint_coord_bld, bld->format_desc, - x, y, y_stride); + x, y, z, y_stride, z_stride); assert(bld->format_desc->block.width == 1); assert(bld->format_desc->block.height == 1); @@ -213,6 +279,8 @@ lp_build_sample_texel_soa(struct lp_build_sample_context *bld, bld->texel_type.width, data_ptr, offset); + texel[0] = texel[1] = texel[2] = texel[3] = NULL; + /* convert texels to float rgba */ lp_build_unpack_rgba_soa(bld->builder, bld->format_desc, @@ -245,7 +313,7 @@ lp_build_sample_packed(struct lp_build_sample_context *bld, offset = lp_build_sample_offset(&bld->uint_coord_bld, bld->format_desc, - x, y, y_stride); + x, y, NULL, y_stride, NULL); assert(bld->format_desc->block.width == 1); assert(bld->format_desc->block.height == 1); @@ -756,24 +824,6 @@ lp_build_minify(struct lp_build_sample_context *bld, } -static int -texture_dims(enum pipe_texture_target tex) -{ - switch (tex) { - case PIPE_TEXTURE_1D: - return 1; - case PIPE_TEXTURE_2D: - case PIPE_TEXTURE_CUBE: - return 2; - case PIPE_TEXTURE_3D: - return 3; - default: - assert(0 && "bad texture target in texture_dims()"); - return 2; - } -} - - /** * Generate code to compute texture level of detail (lambda). * \param s vector of texcoord s values @@ -794,7 +844,6 @@ lp_build_lod_selector(struct lp_build_sample_context *bld, { const int dims = texture_dims(bld->static_state->target); - struct lp_build_context *coord_bld = &bld->coord_bld; struct lp_build_context *float_bld = &bld->float_bld; LLVMValueRef lod_bias = LLVMConstReal(LLVMFloatType(), bld->static_state->lod_bias); LLVMValueRef min_lod = LLVMConstReal(LLVMFloatType(), bld->static_state->min_lod); @@ -921,161 +970,635 @@ lp_build_linear_mip_levels(struct lp_build_sample_context *bld, LLVMValueRef *level1_out, LLVMValueRef *weight_out) { - struct lp_build_context *coord_bld = &bld->coord_bld; - struct lp_build_context *int_coord_bld = &bld->int_coord_bld; + struct lp_build_context *float_bld = &bld->float_bld; + struct lp_build_context *int_bld = &bld->int_bld; LLVMValueRef last_level, level; last_level = bld->dynamic_state->last_level(bld->dynamic_state, bld->builder, unit); /* convert float lod to integer */ - level = lp_build_ifloor(coord_bld, lod); + level = lp_build_ifloor(float_bld, lod); /* compute level 0 and clamp to legal range of levels */ - *level0_out = lp_build_clamp(int_coord_bld, level, - int_coord_bld->zero, + *level0_out = lp_build_clamp(int_bld, level, + int_bld->zero, last_level); /* compute level 1 and clamp to legal range of levels */ - *level1_out = lp_build_add(int_coord_bld, *level0_out, int_coord_bld->one); - *level1_out = lp_build_min(int_coord_bld, *level1_out, int_coord_bld->zero); + *level1_out = lp_build_add(int_bld, *level0_out, int_bld->one); + *level1_out = lp_build_min(int_bld, *level1_out, int_bld->zero); - *weight_out = lp_build_fract(coord_bld, lod); + *weight_out = lp_build_fract(float_bld, lod); } - /** - * Sample 2D texture with nearest filtering, no mipmapping. + * Generate code to sample a mipmap level with nearest filtering. + * If sampling a cube texture, r = cube face in [0,5]. */ static void -lp_build_sample_2d_nearest_soa(struct lp_build_sample_context *bld, - LLVMValueRef s, - LLVMValueRef t, - LLVMValueRef width, - LLVMValueRef height, - LLVMValueRef stride, - LLVMValueRef data_array, - LLVMValueRef *texel) +lp_build_sample_image_nearest(struct lp_build_sample_context *bld, + LLVMValueRef width_vec, + LLVMValueRef height_vec, + LLVMValueRef depth_vec, + LLVMValueRef row_stride_vec, + LLVMValueRef img_stride_vec, + LLVMValueRef data_ptr, + LLVMValueRef s, + LLVMValueRef t, + LLVMValueRef r, + LLVMValueRef colors_out[4]) { - LLVMValueRef x, y; - LLVMValueRef data_ptr; + const int dims = texture_dims(bld->static_state->target); + LLVMValueRef x, y, z; - x = lp_build_sample_wrap_nearest(bld, s, width, + /* + * Compute integer texcoords. + */ + x = lp_build_sample_wrap_nearest(bld, s, width_vec, bld->static_state->pot_width, bld->static_state->wrap_s); - y = lp_build_sample_wrap_nearest(bld, t, height, - bld->static_state->pot_height, - bld->static_state->wrap_t); - lp_build_name(x, "tex.x.wrapped"); - lp_build_name(y, "tex.y.wrapped"); - /* get pointer to mipmap level 0 data */ - data_ptr = lp_build_get_const_mipmap_level(bld, data_array, 0); + if (dims >= 2) { + y = lp_build_sample_wrap_nearest(bld, t, height_vec, + bld->static_state->pot_height, + bld->static_state->wrap_t); + lp_build_name(y, "tex.y.wrapped"); + + if (dims == 3) { + z = lp_build_sample_wrap_nearest(bld, r, depth_vec, + bld->static_state->pot_height, + bld->static_state->wrap_r); + lp_build_name(z, "tex.z.wrapped"); + } + else if (bld->static_state->target == PIPE_TEXTURE_CUBE) { + z = r; + } + else { + z = NULL; + } + } + else { + y = z = NULL; + } - lp_build_sample_texel_soa(bld, width, height, x, y, stride, data_ptr, texel); + /* + * Get texture colors. + */ + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x, y, z, + row_stride_vec, img_stride_vec, + data_ptr, colors_out); } /** - * Sample 2D texture with nearest filtering, nearest mipmap. + * Generate code to sample a mipmap level with linear filtering. + * If sampling a cube texture, r = cube face in [0,5]. */ static void -lp_build_sample_2d_nearest_mip_nearest_soa(struct lp_build_sample_context *bld, - unsigned unit, - LLVMValueRef s, - LLVMValueRef t, - LLVMValueRef width, - LLVMValueRef height, - LLVMValueRef width_vec, - LLVMValueRef height_vec, - LLVMValueRef stride, - LLVMValueRef data_array, - LLVMValueRef *texel) +lp_build_sample_image_linear(struct lp_build_sample_context *bld, + LLVMValueRef width_vec, + LLVMValueRef height_vec, + LLVMValueRef depth_vec, + LLVMValueRef row_stride_vec, + LLVMValueRef img_stride_vec, + LLVMValueRef data_ptr, + LLVMValueRef s, + LLVMValueRef t, + LLVMValueRef r, + LLVMValueRef colors_out[4]) { - LLVMValueRef x, y; - LLVMValueRef lod, ilevel, ilevel_vec; - LLVMValueRef data_ptr; + const int dims = texture_dims(bld->static_state->target); + LLVMValueRef x0, y0, z0, x1, y1, z1; + LLVMValueRef s_fpart, t_fpart, r_fpart; + LLVMValueRef neighbors[2][2][4]; + int chan; - /* compute float LOD */ - lod = lp_build_lod_selector(bld, s, t, NULL, width, height, NULL); + /* + * Compute integer texcoords. + */ + lp_build_sample_wrap_linear(bld, s, width_vec, + bld->static_state->pot_width, + bld->static_state->wrap_s, + &x0, &x1, &s_fpart); + lp_build_name(x0, "tex.x0.wrapped"); + lp_build_name(x1, "tex.x1.wrapped"); + + if (dims >= 2) { + lp_build_sample_wrap_linear(bld, t, height_vec, + bld->static_state->pot_height, + bld->static_state->wrap_t, + &y0, &y1, &t_fpart); + lp_build_name(y0, "tex.y0.wrapped"); + lp_build_name(y1, "tex.y1.wrapped"); + + if (dims == 3) { + lp_build_sample_wrap_linear(bld, r, depth_vec, + bld->static_state->pot_depth, + bld->static_state->wrap_r, + &z0, &z1, &r_fpart); + lp_build_name(z0, "tex.z0.wrapped"); + lp_build_name(z1, "tex.z1.wrapped"); + } + else if (bld->static_state->target == PIPE_TEXTURE_CUBE) { + z0 = z1 = r; /* cube face */ + r_fpart = NULL; + } + else { + z0 = z1 = NULL; + r_fpart = NULL; + } + } + else { + y0 = y1 = t_fpart = NULL; + z0 = z1 = r_fpart = NULL; + } - /* convert LOD to int */ - lp_build_nearest_mip_level(bld, unit, lod, &ilevel); + /* + * Get texture colors. + */ + /* get x0/x1 texels */ + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x0, y0, z0, + row_stride_vec, img_stride_vec, + data_ptr, neighbors[0][0]); + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x1, y0, z0, + row_stride_vec, img_stride_vec, + data_ptr, neighbors[0][1]); + + if (dims == 1) { + /* Interpolate two samples from 1D image to produce one color */ + for (chan = 0; chan < 4; chan++) { + colors_out[chan] = lp_build_lerp(&bld->texel_bld, s_fpart, + neighbors[0][0][chan], + neighbors[0][1][chan]); + } + } + else { + /* 2D/3D texture */ + LLVMValueRef colors0[4]; + + /* get x0/x1 texels at y1 */ + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x0, y1, z0, + row_stride_vec, img_stride_vec, + data_ptr, neighbors[1][0]); + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x1, y1, z0, + row_stride_vec, img_stride_vec, + data_ptr, neighbors[1][1]); + + /* Bilinear interpolate the four samples from the 2D image / 3D slice */ + for (chan = 0; chan < 4; chan++) { + colors0[chan] = lp_build_lerp_2d(&bld->texel_bld, + s_fpart, t_fpart, + neighbors[0][0][chan], + neighbors[0][1][chan], + neighbors[1][0][chan], + neighbors[1][1][chan]); + } - ilevel_vec = lp_build_broadcast_scalar(&bld->int_coord_bld, ilevel); + if (dims == 3) { + LLVMValueRef neighbors1[2][2][4]; + LLVMValueRef colors1[4]; + + /* get x0/x1/y0/y1 texels at z1 */ + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x0, y0, z1, + row_stride_vec, img_stride_vec, + data_ptr, neighbors1[0][0]); + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x1, y0, z1, + row_stride_vec, img_stride_vec, + data_ptr, neighbors1[0][1]); + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x0, y1, z1, + row_stride_vec, img_stride_vec, + data_ptr, neighbors1[1][0]); + lp_build_sample_texel_soa(bld, width_vec, height_vec, depth_vec, + x1, y1, z1, + row_stride_vec, img_stride_vec, + data_ptr, neighbors1[1][1]); + + /* Bilinear interpolate the four samples from the second Z slice */ + for (chan = 0; chan < 4; chan++) { + colors1[chan] = lp_build_lerp_2d(&bld->texel_bld, + s_fpart, t_fpart, + neighbors1[0][0][chan], + neighbors1[0][1][chan], + neighbors1[1][0][chan], + neighbors1[1][1][chan]); + } - /* compute width_vec, height at mipmap level 'ilevel' */ - width_vec = lp_build_minify(bld, width_vec, ilevel_vec); - height_vec = lp_build_minify(bld, height_vec, ilevel_vec); - stride = lp_build_minify(bld, stride, ilevel_vec); + /* Linearly interpolate the two samples from the two 3D slices */ + for (chan = 0; chan < 4; chan++) { + colors_out[chan] = lp_build_lerp(&bld->texel_bld, + r_fpart, + colors0[chan], colors1[chan]); + } + } + else { + /* 2D tex */ + for (chan = 0; chan < 4; chan++) { + colors_out[chan] = colors0[chan]; + } + } + } +} - x = lp_build_sample_wrap_nearest(bld, s, width_vec, - bld->static_state->pot_width, - bld->static_state->wrap_s); - y = lp_build_sample_wrap_nearest(bld, t, height_vec, - bld->static_state->pot_height, - bld->static_state->wrap_t); - lp_build_name(x, "tex.x.wrapped"); - lp_build_name(y, "tex.y.wrapped"); +/** Helper used by lp_build_cube_lookup() */ +static LLVMValueRef +lp_build_cube_ima(struct lp_build_context *coord_bld, LLVMValueRef coord) +{ + /* ima = -0.5 / abs(coord); */ + LLVMValueRef negHalf = lp_build_const_scalar(coord_bld->type, -0.5); + LLVMValueRef absCoord = lp_build_abs(coord_bld, coord); + LLVMValueRef ima = lp_build_mul(coord_bld, negHalf, + lp_build_rcp(coord_bld, absCoord)); + return ima; +} - /* get pointer to mipmap level [ilevel] data */ - if (0) - data_ptr = lp_build_get_mipmap_level(bld, data_array, ilevel); - else - data_ptr = lp_build_get_const_mipmap_level(bld, data_array, 0); - lp_build_sample_texel_soa(bld, width_vec, height_vec, x, y, stride, data_ptr, texel); +/** + * Helper used by lp_build_cube_lookup() + * \param sign scalar +1 or -1 + * \param coord float vector + * \param ima float vector + */ +static LLVMValueRef +lp_build_cube_coord(struct lp_build_context *coord_bld, + LLVMValueRef sign, int negate_coord, + LLVMValueRef coord, LLVMValueRef ima) +{ + /* return negate(coord) * ima * sign + 0.5; */ + LLVMValueRef half = lp_build_const_scalar(coord_bld->type, 0.5); + LLVMValueRef res; + + assert(negate_coord == +1 || negate_coord == -1); + + if (negate_coord == -1) { + coord = lp_build_negate(coord_bld, coord); + } + + res = lp_build_mul(coord_bld, coord, ima); + if (sign) { + sign = lp_build_broadcast_scalar(coord_bld, sign); + res = lp_build_mul(coord_bld, res, sign); + } + res = lp_build_add(coord_bld, res, half); + + return res; } +/** Helper used by lp_build_cube_lookup() + * Return (major_coord >= 0) ? pos_face : neg_face; + */ +static LLVMValueRef +lp_build_cube_face(struct lp_build_sample_context *bld, + LLVMValueRef major_coord, + unsigned pos_face, unsigned neg_face) +{ + LLVMValueRef cmp = LLVMBuildFCmp(bld->builder, LLVMRealUGE, + major_coord, + bld->float_bld.zero, ""); + LLVMValueRef pos = LLVMConstInt(LLVMInt32Type(), pos_face, 0); + LLVMValueRef neg = LLVMConstInt(LLVMInt32Type(), neg_face, 0); + LLVMValueRef res = LLVMBuildSelect(bld->builder, cmp, pos, neg, ""); + return res; +} + + + /** - * Sample 2D texture with bilinear filtering. + * Generate code to do cube face selection and per-face texcoords. */ static void -lp_build_sample_2d_linear_soa(struct lp_build_sample_context *bld, - LLVMValueRef s, - LLVMValueRef t, - LLVMValueRef width, - LLVMValueRef height, - LLVMValueRef stride, - LLVMValueRef data_array, - LLVMValueRef *texel) +lp_build_cube_lookup(struct lp_build_sample_context *bld, + LLVMValueRef s, + LLVMValueRef t, + LLVMValueRef r, + LLVMValueRef *face, + LLVMValueRef *face_s, + LLVMValueRef *face_t) { - LLVMValueRef s_fpart; - LLVMValueRef t_fpart; - LLVMValueRef x0, x1; - LLVMValueRef y0, y1; - LLVMValueRef neighbors[2][2][4]; - LLVMValueRef data_ptr; - unsigned chan; + struct lp_build_context *float_bld = &bld->float_bld; + struct lp_build_context *coord_bld = &bld->coord_bld; + LLVMValueRef rx, ry, rz; + LLVMValueRef arx, ary, arz; + LLVMValueRef c25 = LLVMConstReal(LLVMFloatType(), 0.25); + LLVMValueRef arx_ge_ary, arx_ge_arz; + LLVMValueRef ary_ge_arx, ary_ge_arz; + LLVMValueRef arx_ge_ary_arz, ary_ge_arx_arz; + LLVMValueRef rx_pos, ry_pos, rz_pos; - lp_build_sample_wrap_linear(bld, s, width, bld->static_state->pot_width, - bld->static_state->wrap_s, &x0, &x1, &s_fpart); - lp_build_sample_wrap_linear(bld, t, height, bld->static_state->pot_height, - bld->static_state->wrap_t, &y0, &y1, &t_fpart); + assert(bld->coord_bld.type.length == 4); - /* get pointer to mipmap level 0 data */ - data_ptr = lp_build_get_const_mipmap_level(bld, data_array, 0); + /* + * Use the average of the four pixel's texcoords to choose the face. + */ + rx = lp_build_mul(float_bld, c25, + lp_build_sum_vector(&bld->coord_bld, s)); + ry = lp_build_mul(float_bld, c25, + lp_build_sum_vector(&bld->coord_bld, t)); + rz = lp_build_mul(float_bld, c25, + lp_build_sum_vector(&bld->coord_bld, r)); - lp_build_sample_texel_soa(bld, width, height, x0, y0, stride, data_ptr, neighbors[0][0]); - lp_build_sample_texel_soa(bld, width, height, x1, y0, stride, data_ptr, neighbors[0][1]); - lp_build_sample_texel_soa(bld, width, height, x0, y1, stride, data_ptr, neighbors[1][0]); - lp_build_sample_texel_soa(bld, width, height, x1, y1, stride, data_ptr, neighbors[1][1]); + arx = lp_build_abs(float_bld, rx); + ary = lp_build_abs(float_bld, ry); + arz = lp_build_abs(float_bld, rz); - /* TODO: Don't interpolate missing channels */ - for(chan = 0; chan < 4; ++chan) { - texel[chan] = lp_build_lerp_2d(&bld->texel_bld, - s_fpart, t_fpart, - neighbors[0][0][chan], - neighbors[0][1][chan], - neighbors[1][0][chan], - neighbors[1][1][chan]); + /* + * Compare sign/magnitude of rx,ry,rz to determine face + */ + arx_ge_ary = LLVMBuildFCmp(bld->builder, LLVMRealUGE, arx, ary, ""); + arx_ge_arz = LLVMBuildFCmp(bld->builder, LLVMRealUGE, arx, arz, ""); + ary_ge_arx = LLVMBuildFCmp(bld->builder, LLVMRealUGE, ary, arx, ""); + ary_ge_arz = LLVMBuildFCmp(bld->builder, LLVMRealUGE, ary, arz, ""); + + arx_ge_ary_arz = LLVMBuildAnd(bld->builder, arx_ge_ary, arx_ge_arz, ""); + ary_ge_arx_arz = LLVMBuildAnd(bld->builder, ary_ge_arx, ary_ge_arz, ""); + + rx_pos = LLVMBuildFCmp(bld->builder, LLVMRealUGE, rx, float_bld->zero, ""); + ry_pos = LLVMBuildFCmp(bld->builder, LLVMRealUGE, ry, float_bld->zero, ""); + rz_pos = LLVMBuildFCmp(bld->builder, LLVMRealUGE, rz, float_bld->zero, ""); + + { + struct lp_build_flow_context *flow_ctx; + struct lp_build_if_state if_ctx; + + flow_ctx = lp_build_flow_create(bld->builder); + lp_build_flow_scope_begin(flow_ctx); + + *face_s = bld->coord_bld.undef; + *face_t = bld->coord_bld.undef; + *face = bld->int_bld.undef; + + lp_build_name(*face_s, "face_s"); + lp_build_name(*face_t, "face_t"); + lp_build_name(*face, "face"); + + lp_build_flow_scope_declare(flow_ctx, face_s); + lp_build_flow_scope_declare(flow_ctx, face_t); + lp_build_flow_scope_declare(flow_ctx, face); + + lp_build_if(&if_ctx, flow_ctx, bld->builder, arx_ge_ary_arz); + { + /* +/- X face */ + LLVMValueRef sign = lp_build_sgn(float_bld, rx); + LLVMValueRef ima = lp_build_cube_ima(coord_bld, s); + *face_s = lp_build_cube_coord(coord_bld, sign, +1, r, ima); + *face_t = lp_build_cube_coord(coord_bld, NULL, +1, t, ima); + *face = lp_build_cube_face(bld, rx, + PIPE_TEX_FACE_POS_X, + PIPE_TEX_FACE_NEG_X); + } + lp_build_else(&if_ctx); + { + struct lp_build_flow_context *flow_ctx2; + struct lp_build_if_state if_ctx2; + + LLVMValueRef face_s2 = bld->coord_bld.undef; + LLVMValueRef face_t2 = bld->coord_bld.undef; + LLVMValueRef face2 = bld->int_bld.undef; + + flow_ctx2 = lp_build_flow_create(bld->builder); + lp_build_flow_scope_begin(flow_ctx2); + lp_build_flow_scope_declare(flow_ctx2, &face_s2); + lp_build_flow_scope_declare(flow_ctx2, &face_t2); + lp_build_flow_scope_declare(flow_ctx2, &face2); + + ary_ge_arx_arz = LLVMBuildAnd(bld->builder, ary_ge_arx, ary_ge_arz, ""); + + lp_build_if(&if_ctx2, flow_ctx2, bld->builder, ary_ge_arx_arz); + { + /* +/- Y face */ + LLVMValueRef sign = lp_build_sgn(float_bld, ry); + LLVMValueRef ima = lp_build_cube_ima(coord_bld, t); + face_s2 = lp_build_cube_coord(coord_bld, NULL, -1, s, ima); + face_t2 = lp_build_cube_coord(coord_bld, sign, -1, r, ima); + face2 = lp_build_cube_face(bld, ry, + PIPE_TEX_FACE_POS_Y, + PIPE_TEX_FACE_NEG_Y); + } + lp_build_else(&if_ctx2); + { + /* +/- Z face */ + LLVMValueRef sign = lp_build_sgn(float_bld, rz); + LLVMValueRef ima = lp_build_cube_ima(coord_bld, r); + face_s2 = lp_build_cube_coord(coord_bld, sign, -1, s, ima); + face_t2 = lp_build_cube_coord(coord_bld, NULL, +1, t, ima); + face2 = lp_build_cube_face(bld, rz, + PIPE_TEX_FACE_POS_Z, + PIPE_TEX_FACE_NEG_Z); + } + lp_build_endif(&if_ctx2); + lp_build_flow_scope_end(flow_ctx2); + lp_build_flow_destroy(flow_ctx2); + + *face_s = face_s2; + *face_t = face_t2; + *face = face2; + } + + lp_build_endif(&if_ctx); + lp_build_flow_scope_end(flow_ctx); + lp_build_flow_destroy(flow_ctx); + } +} + + + +/** + * General texture sampling codegen. + * This function handles texture sampling for all texture targets (1D, + * 2D, 3D, cube) and all filtering modes. + */ +static void +lp_build_sample_general(struct lp_build_sample_context *bld, + unsigned unit, + LLVMValueRef s, + LLVMValueRef t, + LLVMValueRef r, + LLVMValueRef width, + LLVMValueRef height, + LLVMValueRef depth, + LLVMValueRef width_vec, + LLVMValueRef height_vec, + LLVMValueRef depth_vec, + LLVMValueRef row_stride_array, + LLVMValueRef img_stride_vec, + LLVMValueRef data_array, + LLVMValueRef *colors_out) +{ + const unsigned mip_filter = bld->static_state->min_mip_filter; + const unsigned min_filter = bld->static_state->min_img_filter; + const unsigned mag_filter = bld->static_state->mag_img_filter; + const int dims = texture_dims(bld->static_state->target); + LLVMValueRef lod, lod_fpart; + LLVMValueRef ilevel0, ilevel1, ilevel0_vec, ilevel1_vec; + LLVMValueRef width0_vec = NULL, height0_vec = NULL, depth0_vec = NULL; + LLVMValueRef width1_vec = NULL, height1_vec = NULL, depth1_vec = NULL; + LLVMValueRef row_stride0_vec = NULL, row_stride1_vec = NULL; + LLVMValueRef img_stride0_vec = NULL, img_stride1_vec = NULL; + LLVMValueRef data_ptr0, data_ptr1; + int chan; + + /* + printf("%s mip %d min %d mag %d\n", __FUNCTION__, + mip_filter, min_filter, mag_filter); + */ + + /* + * Compute the level of detail (mipmap level index(es)). + */ + if (mip_filter == PIPE_TEX_MIPFILTER_NONE) { + /* always use mip level 0 */ + ilevel0 = LLVMConstInt(LLVMInt32Type(), 0, 0); + } + else { + /* compute float LOD */ + lod = lp_build_lod_selector(bld, s, t, r, width, height, depth); + + if (mip_filter == PIPE_TEX_MIPFILTER_NEAREST) { + lp_build_nearest_mip_level(bld, unit, lod, &ilevel0); + } + else { + assert(mip_filter == PIPE_TEX_MIPFILTER_LINEAR); + lp_build_linear_mip_levels(bld, unit, lod, &ilevel0, &ilevel1, + &lod_fpart); + lod_fpart = lp_build_broadcast_scalar(&bld->coord_bld, lod_fpart); + } + } + + /* + * Convert scalar integer mipmap levels into vectors. + */ + ilevel0_vec = lp_build_broadcast_scalar(&bld->int_coord_bld, ilevel0); + if (mip_filter == PIPE_TEX_MIPFILTER_LINEAR) + ilevel1_vec = lp_build_broadcast_scalar(&bld->int_coord_bld, ilevel1); + + /* + * Compute width, height at mipmap level 'ilevel0' + */ + width0_vec = lp_build_minify(bld, width_vec, ilevel0_vec); + if (dims >= 2) { + height0_vec = lp_build_minify(bld, height_vec, ilevel0_vec); + row_stride0_vec = lp_build_get_level_stride_vec(bld, row_stride_array, + ilevel0); + if (dims == 3 || bld->static_state->target == PIPE_TEXTURE_CUBE) { + img_stride0_vec = lp_build_mul(&bld->int_coord_bld, + row_stride0_vec, height0_vec); + if (dims == 3) { + depth0_vec = lp_build_minify(bld, depth_vec, ilevel0_vec); + } + } + } + if (mip_filter == PIPE_TEX_MIPFILTER_LINEAR) { + /* compute width, height, depth for second mipmap level at ilevel1 */ + width1_vec = lp_build_minify(bld, width_vec, ilevel1_vec); + if (dims >= 2) { + height1_vec = lp_build_minify(bld, height_vec, ilevel1_vec); + row_stride1_vec = lp_build_get_level_stride_vec(bld, row_stride_array, + ilevel1); + if (dims == 3 || bld->static_state->target == PIPE_TEXTURE_CUBE) { + img_stride1_vec = lp_build_mul(&bld->int_coord_bld, + row_stride1_vec, height1_vec); + if (dims ==3) { + depth1_vec = lp_build_minify(bld, depth_vec, ilevel1_vec); + } + } + } + } + + /* + * Choose cube face, recompute texcoords. + */ + if (bld->static_state->target == PIPE_TEXTURE_CUBE) { + LLVMValueRef face, face_s, face_t; + lp_build_cube_lookup(bld, s, t, r, &face, &face_s, &face_t); + s = face_s; /* vec */ + t = face_t; /* vec */ + /* use 'r' to indicate cube face */ + r = lp_build_broadcast_scalar(&bld->int_coord_bld, face); /* vec */ + } + + /* + * Get pointer(s) to image data for mipmap level(s). + */ + data_ptr0 = lp_build_get_mipmap_level(bld, data_array, ilevel0); + if (mip_filter == PIPE_TEX_MIPFILTER_LINEAR) { + data_ptr1 = lp_build_get_mipmap_level(bld, data_array, ilevel1); + } + + /* + * Get/interpolate texture colors. + */ + /* XXX temporarily force this path: */ + if (1 /*min_filter == mag_filter*/) { + /* same filter for minification or magnification */ + LLVMValueRef colors0[4], colors1[4]; + + if (min_filter == PIPE_TEX_FILTER_NEAREST) { + lp_build_sample_image_nearest(bld, + width0_vec, height0_vec, depth0_vec, + row_stride0_vec, img_stride0_vec, + data_ptr0, s, t, r, colors0); + + if (mip_filter == PIPE_TEX_MIPFILTER_LINEAR) { + /* sample the second mipmap level, and interp */ + lp_build_sample_image_nearest(bld, + width1_vec, height1_vec, depth1_vec, + row_stride1_vec, img_stride1_vec, + data_ptr1, s, t, r, colors1); + } + } + else { + assert(min_filter == PIPE_TEX_FILTER_LINEAR); + + lp_build_sample_image_linear(bld, + width0_vec, height0_vec, depth0_vec, + row_stride0_vec, img_stride0_vec, + data_ptr0, s, t, r, colors0); + + + if (mip_filter == PIPE_TEX_MIPFILTER_LINEAR) { + /* sample the second mipmap level, and interp */ + lp_build_sample_image_linear(bld, + width1_vec, height1_vec, depth1_vec, + row_stride1_vec, img_stride1_vec, + data_ptr1, s, t, r, colors1); + } + } + + if (mip_filter == PIPE_TEX_MIPFILTER_LINEAR) { + /* interpolate samples from the two mipmap levels */ + for (chan = 0; chan < 4; chan++) { + colors_out[chan] = lp_build_lerp(&bld->texel_bld, lod_fpart, + colors0[chan], colors1[chan]); + } + } + else { + /* use first/only level's colors */ + for (chan = 0; chan < 4; chan++) { + colors_out[chan] = colors0[chan]; + } + } + } + else { + /* emit conditional to choose min image filter or mag image filter + * depending on the lod being >0 or <= 0, respectively. + */ + abort(); } } + static void lp_build_rgba8_to_f32_soa(LLVMBuilderRef builder, struct lp_type dst_type, @@ -1112,7 +1635,7 @@ lp_build_sample_2d_linear_aos(struct lp_build_sample_context *bld, LLVMValueRef t, LLVMValueRef width, LLVMValueRef height, - LLVMValueRef stride, + LLVMValueRef stride_array, LLVMValueRef data_array, LLVMValueRef *texel) { @@ -1129,6 +1652,7 @@ lp_build_sample_2d_linear_aos(struct lp_build_sample_context *bld, LLVMValueRef neighbors_hi[2][2]; LLVMValueRef packed, packed_lo, packed_hi; LLVMValueRef unswizzled[4]; + LLVMValueRef stride; lp_build_context_init(&i32, builder, lp_type_int_vec(32)); lp_build_context_init(&h16, builder, lp_type_ufixed(16)); @@ -1236,6 +1760,8 @@ lp_build_sample_2d_linear_aos(struct lp_build_sample_context *bld, t_fpart_hi = LLVMBuildShuffleVector(builder, t_fpart, h16.undef, shuffle_hi, ""); } + stride = lp_build_get_const_level_stride_vec(bld, stride_array, 0); + /* * Fetch the pixels as 4 x 32bit (rgba order might differ): * @@ -1359,12 +1885,12 @@ lp_build_sample_soa(LLVMBuilderRef builder, struct lp_build_sample_context bld; LLVMValueRef width, width_vec; LLVMValueRef height, height_vec; - LLVMValueRef stride, stride_vec; + LLVMValueRef depth, depth_vec; + LLVMValueRef stride_array; LLVMValueRef data_array; LLVMValueRef s; LLVMValueRef t; LLVMValueRef r; - boolean done = FALSE; (void) lp_build_lod_selector; /* temporary to silence warning */ (void) lp_build_nearest_mip_level; @@ -1395,7 +1921,8 @@ lp_build_sample_soa(LLVMBuilderRef builder, /* Get the dynamic state */ width = dynamic_state->width(dynamic_state, builder, unit); height = dynamic_state->height(dynamic_state, builder, unit); - stride = dynamic_state->stride(dynamic_state, builder, unit); + depth = dynamic_state->depth(dynamic_state, builder, unit); + stride_array = dynamic_state->row_stride(dynamic_state, builder, unit); data_array = dynamic_state->data_ptr(dynamic_state, builder, unit); /* Note that data_array is an array[level] of pointers to texture images */ @@ -1405,58 +1932,26 @@ lp_build_sample_soa(LLVMBuilderRef builder, width_vec = lp_build_broadcast_scalar(&bld.uint_coord_bld, width); height_vec = lp_build_broadcast_scalar(&bld.uint_coord_bld, height); - stride_vec = lp_build_broadcast_scalar(&bld.uint_coord_bld, stride); - - if(static_state->target == PIPE_TEXTURE_1D) - t = bld.coord_bld.zero; - - switch (static_state->min_mip_filter) { - case PIPE_TEX_MIPFILTER_NONE: - break; - case PIPE_TEX_MIPFILTER_NEAREST: - - switch (static_state->min_img_filter) { - case PIPE_TEX_FILTER_NEAREST: - lp_build_sample_2d_nearest_mip_nearest_soa(&bld, unit, - s, t, - width, height, - width_vec, height_vec, - stride_vec, - data_array, texel); - done = TRUE; - break; - } - - break; - case PIPE_TEX_MIPFILTER_LINEAR: - break; - default: - assert(0 && "invalid mip filter"); + depth_vec = lp_build_broadcast_scalar(&bld.uint_coord_bld, depth); + + if (lp_format_is_rgba8(bld.format_desc) && + static_state->target == PIPE_TEXTURE_2D && + static_state->min_img_filter == PIPE_TEX_FILTER_LINEAR && + static_state->mag_img_filter == PIPE_TEX_FILTER_LINEAR && + static_state->min_mip_filter == PIPE_TEX_MIPFILTER_NONE && + is_simple_wrap_mode(static_state->wrap_s) && + is_simple_wrap_mode(static_state->wrap_t)) { + /* special case */ + lp_build_sample_2d_linear_aos(&bld, s, t, width_vec, height_vec, + stride_array, data_array, texel); } - - if (!done) { - switch (static_state->min_img_filter) { - case PIPE_TEX_FILTER_NEAREST: - lp_build_sample_2d_nearest_soa(&bld, s, t, width_vec, height_vec, - stride_vec, data_array, texel); - break; - case PIPE_TEX_FILTER_LINEAR: - if(lp_format_is_rgba8(bld.format_desc) && - is_simple_wrap_mode(static_state->wrap_s) && - is_simple_wrap_mode(static_state->wrap_t)) - lp_build_sample_2d_linear_aos(&bld, s, t, width_vec, height_vec, - stride_vec, data_array, texel); - else - lp_build_sample_2d_linear_soa(&bld, s, t, width_vec, height_vec, - stride_vec, data_array, texel); - break; - default: - assert(0); - } + else { + lp_build_sample_general(&bld, unit, s, t, r, + width, height, depth, + width_vec, height_vec, depth_vec, + stride_array, NULL, data_array, + texel); } - /* FIXME: respect static_state->min_mip_filter */; - /* FIXME: respect static_state->mag_img_filter */; - lp_build_sample_compare(&bld, r, texel); } diff --git a/src/gallium/auxiliary/gallivm/lp_bld_struct.h b/src/gallium/auxiliary/gallivm/lp_bld_struct.h index 740392f5611..34478c10f51 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_struct.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_struct.h @@ -37,7 +37,7 @@ #define LP_BLD_STRUCT_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include <llvm-c/Target.h> #include "util/u_debug.h" diff --git a/src/gallium/auxiliary/gallivm/lp_bld_swizzle.h b/src/gallium/auxiliary/gallivm/lp_bld_swizzle.h index b9472127a63..57b5cc079f2 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_swizzle.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_swizzle.h @@ -37,7 +37,7 @@ #define LP_BLD_SWIZZLE_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct lp_type; diff --git a/src/gallium/auxiliary/gallivm/lp_bld_tgsi.h b/src/gallium/auxiliary/gallivm/lp_bld_tgsi.h index eddb7a83fa2..0f2f8a65b10 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_tgsi.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_tgsi.h @@ -35,7 +35,7 @@ #ifndef LP_BLD_TGSI_H #define LP_BLD_TGSI_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct tgsi_token; diff --git a/src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c b/src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c index fbb664d43a0..5ec59d636cb 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_tgsi_soa.c @@ -41,6 +41,7 @@ #include "util/u_debug.h" #include "util/u_math.h" #include "util/u_memory.h" +#include "tgsi/tgsi_dump.h" #include "tgsi/tgsi_info.h" #include "tgsi/tgsi_parse.h" #include "tgsi/tgsi_util.h" @@ -95,6 +96,19 @@ struct lp_exec_mask { int cond_stack_size; LLVMValueRef cond_mask; + LLVMValueRef break_stack[LP_TGSI_MAX_NESTING]; + int break_stack_size; + LLVMValueRef break_mask; + + LLVMValueRef cont_stack[LP_TGSI_MAX_NESTING]; + int cont_stack_size; + LLVMValueRef cont_mask; + + LLVMBasicBlockRef loop_stack[LP_TGSI_MAX_NESTING]; + int loop_stack_size; + LLVMBasicBlockRef loop_block; + + LLVMValueRef exec_mask; }; @@ -145,14 +159,33 @@ static void lp_exec_mask_init(struct lp_exec_mask *mask, struct lp_build_context mask->bld = bld; mask->has_mask = FALSE; mask->cond_stack_size = 0; + mask->loop_stack_size = 0; + mask->break_stack_size = 0; + mask->cont_stack_size = 0; mask->int_vec_type = lp_build_int_vec_type(mask->bld->type); } static void lp_exec_mask_update(struct lp_exec_mask *mask) { - mask->exec_mask = mask->cond_mask; - mask->has_mask = (mask->cond_stack_size > 0); + if (mask->loop_stack_size) { + /*for loops we need to update the entire mask at + * runtime */ + LLVMValueRef tmp; + tmp = LLVMBuildAnd(mask->bld->builder, + mask->cont_mask, + mask->break_mask, + "maskcb"); + mask->exec_mask = LLVMBuildAnd(mask->bld->builder, + mask->cond_mask, + tmp, + "maskfull"); + } else + mask->exec_mask = mask->cond_mask; + + + mask->has_mask = (mask->cond_stack_size > 0 || + mask->loop_stack_size > 0); } static void lp_exec_mask_cond_push(struct lp_exec_mask *mask, @@ -189,6 +222,89 @@ static void lp_exec_mask_cond_pop(struct lp_exec_mask *mask) lp_exec_mask_update(mask); } +static void lp_exec_bgnloop(struct lp_exec_mask *mask) +{ + + if (mask->cont_stack_size == 0) + mask->cont_mask = LLVMConstAllOnes(mask->int_vec_type); + if (mask->cont_stack_size == 0) + mask->break_mask = LLVMConstAllOnes(mask->int_vec_type); + if (mask->cond_stack_size == 0) + mask->cond_mask = LLVMConstAllOnes(mask->int_vec_type); + mask->loop_stack[mask->loop_stack_size++] = mask->loop_block; + mask->loop_block = lp_build_insert_new_block(mask->bld->builder, "bgnloop"); + LLVMBuildBr(mask->bld->builder, mask->loop_block); + LLVMPositionBuilderAtEnd(mask->bld->builder, mask->loop_block); + + lp_exec_mask_update(mask); +} + +static void lp_exec_break(struct lp_exec_mask *mask) +{ + LLVMValueRef exec_mask = LLVMBuildNot(mask->bld->builder, + mask->exec_mask, + "break"); + + mask->break_stack[mask->break_stack_size++] = mask->break_mask; + if (mask->break_stack_size > 1) { + mask->break_mask = LLVMBuildAnd(mask->bld->builder, + mask->break_mask, + exec_mask, "break_full"); + } else + mask->break_mask = exec_mask; + + lp_exec_mask_update(mask); +} + +static void lp_exec_continue(struct lp_exec_mask *mask) +{ + LLVMValueRef exec_mask = LLVMBuildNot(mask->bld->builder, + mask->exec_mask, + ""); + + mask->cont_stack[mask->cont_stack_size++] = mask->cont_mask; + if (mask->cont_stack_size > 1) { + mask->cont_mask = LLVMBuildAnd(mask->bld->builder, + mask->cont_mask, + exec_mask, ""); + } else + mask->cont_mask = exec_mask; + + lp_exec_mask_update(mask); +} + + +static void lp_exec_endloop(struct lp_exec_mask *mask) +{ + LLVMBasicBlockRef endloop; + LLVMTypeRef reg_type = LLVMIntType(mask->bld->type.width* + mask->bld->type.length); + /* i1cond = (mask == 0) */ + LLVMValueRef i1cond = LLVMBuildICmp( + mask->bld->builder, + LLVMIntNE, + LLVMBuildBitCast(mask->bld->builder, mask->break_mask, reg_type, ""), + LLVMConstNull(reg_type), ""); + + endloop = lp_build_insert_new_block(mask->bld->builder, "endloop"); + + LLVMBuildCondBr(mask->bld->builder, + i1cond, mask->loop_block, endloop); + + LLVMPositionBuilderAtEnd(mask->bld->builder, endloop); + + mask->loop_block = mask->loop_stack[--mask->loop_stack_size]; + /* pop the break mask */ + if (mask->cont_stack_size) { + mask->cont_mask = mask->cont_stack[--mask->cont_stack_size]; + } + if (mask->break_stack_size) { + mask->break_mask = mask->cont_stack[--mask->break_stack_size]; + } + + lp_exec_mask_update(mask); +} + static void lp_exec_mask_store(struct lp_exec_mask *mask, LLVMValueRef val, LLVMValueRef dst) @@ -1363,14 +1479,15 @@ emit_instruction( case TGSI_OPCODE_TXP: emit_tex( bld, inst, FALSE, TRUE, dst0 ); break; - + case TGSI_OPCODE_BRK: - /* FIXME */ - return 0; + lp_exec_break(&bld->exec_mask); break; case TGSI_OPCODE_IF: tmp0 = emit_fetch(bld, inst, 0, CHAN_X); + tmp0 = lp_build_cmp(&bld->base, PIPE_FUNC_NOTEQUAL, + tmp0, bld->base.zero); lp_exec_mask_cond_push(&bld->exec_mask, tmp0); break; @@ -1380,6 +1497,10 @@ emit_instruction( return 0; break; + case TGSI_OPCODE_BGNLOOP: + lp_exec_bgnloop(&bld->exec_mask); + break; + case TGSI_OPCODE_REP: /* deprecated */ assert(0); @@ -1400,6 +1521,10 @@ emit_instruction( return 0; break; + case TGSI_OPCODE_ENDLOOP: + lp_exec_endloop(&bld->exec_mask); + break; + case TGSI_OPCODE_ENDREP: /* deprecated */ assert(0); @@ -1499,8 +1624,7 @@ emit_instruction( break; case TGSI_OPCODE_CONT: - /* FIXME */ - return 0; + lp_exec_continue(&bld->exec_mask); break; case TGSI_OPCODE_EMIT: @@ -1603,7 +1727,14 @@ lp_build_tgsi_soa(LLVMBuilderRef builder, assert( 0 ); } } - + if (0) { + LLVMBasicBlockRef block = LLVMGetInsertBlock(builder); + LLVMValueRef function = LLVMGetBasicBlockParent(block); + debug_printf("11111111111111111111111111111 \n"); + tgsi_dump(tokens, 0); + LLVMDumpValue(function); + debug_printf("2222222222222222222222222222 \n"); + } tgsi_parse_free( &parse ); } diff --git a/src/gallium/auxiliary/gallivm/lp_bld_type.c b/src/gallium/auxiliary/gallivm/lp_bld_type.c index c327ba045a6..796af88caad 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_type.c +++ b/src/gallium/auxiliary/gallivm/lp_bld_type.c @@ -58,7 +58,10 @@ LLVMTypeRef lp_build_vec_type(struct lp_type type) { LLVMTypeRef elem_type = lp_build_elem_type(type); - return LLVMVectorType(elem_type, type.length); + if (type.length == 1) + return elem_type; + else + return LLVMVectorType(elem_type, type.length); } @@ -115,6 +118,9 @@ lp_check_vec_type(struct lp_type type, LLVMTypeRef vec_type) if(!vec_type) return FALSE; + if (type.length == 1) + return lp_check_elem_type(type, vec_type); + if(LLVMGetTypeKind(vec_type) != LLVMVectorTypeKind) return FALSE; @@ -153,7 +159,10 @@ LLVMTypeRef lp_build_int_vec_type(struct lp_type type) { LLVMTypeRef elem_type = lp_build_int_elem_type(type); - return LLVMVectorType(elem_type, type.length); + if (type.length == 1) + return elem_type; + else + return LLVMVectorType(elem_type, type.length); } diff --git a/src/gallium/auxiliary/gallivm/lp_bld_type.h b/src/gallium/auxiliary/gallivm/lp_bld_type.h index 4daa904e633..5b351476ac2 100644 --- a/src/gallium/auxiliary/gallivm/lp_bld_type.h +++ b/src/gallium/auxiliary/gallivm/lp_bld_type.h @@ -37,7 +37,7 @@ #define LP_BLD_TYPE_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include <pipe/p_compiler.h> diff --git a/src/gallium/auxiliary/os/os_llvm.h b/src/gallium/auxiliary/os/os_llvm.h new file mode 100644 index 00000000000..d5edfbfe923 --- /dev/null +++ b/src/gallium/auxiliary/os/os_llvm.h @@ -0,0 +1,47 @@ +/************************************************************************** + * + * Copyright 2010 VMware, Inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. + * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + * + **************************************************************************/ + +/** + * @file + * Wrapper for LLVM header file #includes. + */ + + +#ifndef OS_LLVM_H +#define OS_LLVM_H + + +#include <llvm-c/Core.h> + + +/** Set version to 0 if missing to avoid #ifdef HAVE_LLVM everywhere */ +#ifndef HAVE_LLVM +#define HAVE_LLVM 0x0 +#endif + + +#endif /* OS_LLVM_H */ diff --git a/src/gallium/auxiliary/os/os_time.h b/src/gallium/auxiliary/os/os_time.h index 5b55c1b3747..7e0f67a76b0 100644 --- a/src/gallium/auxiliary/os/os_time.h +++ b/src/gallium/auxiliary/os/os_time.h @@ -71,7 +71,7 @@ os_time_sleep(int64_t usecs); /* * Helper function for detecting time outs, taking in account overflow. * - * Returns true the the current time has elapsed beyond the specified interval. + * Returns true if the current time has elapsed beyond the specified interval. */ static INLINE boolean os_time_timeout(int64_t start, diff --git a/src/gallium/auxiliary/translate/translate_generic.c b/src/gallium/auxiliary/translate/translate_generic.c index c9ec2b32bfe..c3ec9ae3f4b 100644 --- a/src/gallium/auxiliary/translate/translate_generic.c +++ b/src/gallium/auxiliary/translate/translate_generic.c @@ -393,10 +393,10 @@ static fetch_func get_fetch_func( enum pipe_format format ) return &fetch_R8G8B8A8_SSCALED; case PIPE_FORMAT_B8G8R8A8_UNORM: - return &fetch_A8R8G8B8_UNORM; + return &fetch_B8G8R8A8_UNORM; case PIPE_FORMAT_A8R8G8B8_UNORM: - return &fetch_B8G8R8A8_UNORM; + return &fetch_A8R8G8B8_UNORM; case PIPE_FORMAT_R32_FIXED: return &fetch_R32_FIXED; @@ -552,10 +552,10 @@ static emit_func get_emit_func( enum pipe_format format ) return &emit_R8G8B8A8_SSCALED; case PIPE_FORMAT_B8G8R8A8_UNORM: - return &emit_A8R8G8B8_UNORM; + return &emit_B8G8R8A8_UNORM; case PIPE_FORMAT_A8R8G8B8_UNORM: - return &emit_B8G8R8A8_UNORM; + return &emit_A8R8G8B8_UNORM; default: assert(0); diff --git a/src/gallium/auxiliary/translate/translate_sse.c b/src/gallium/auxiliary/translate/translate_sse.c index 03e093c11ea..c13e7427387 100644 --- a/src/gallium/auxiliary/translate/translate_sse.c +++ b/src/gallium/auxiliary/translate/translate_sse.c @@ -336,7 +336,7 @@ static boolean translate_attr( struct translate_sse *p, case PIPE_FORMAT_R32G32B32A32_FLOAT: emit_load_R32G32B32A32(p, dataXMM, srcECX); break; - case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_B8G8R8A8_UNORM: emit_load_R8G8B8A8_UNORM(p, dataXMM, srcECX); emit_swizzle(p, dataXMM, dataXMM, SHUF(Z,Y,X,W)); break; @@ -360,7 +360,7 @@ static boolean translate_attr( struct translate_sse *p, case PIPE_FORMAT_R32G32B32A32_FLOAT: emit_store_R32G32B32A32(p, dstEAX, dataXMM); break; - case PIPE_FORMAT_A8R8G8B8_UNORM: + case PIPE_FORMAT_B8G8R8A8_UNORM: emit_swizzle(p, dataXMM, dataXMM, SHUF(Z,Y,X,W)); emit_store_R8G8B8A8_UNORM(p, dstEAX, dataXMM); break; diff --git a/src/gallium/auxiliary/util/u_blitter.c b/src/gallium/auxiliary/util/u_blitter.c index 71cf9525d39..1692987e8e3 100644 --- a/src/gallium/auxiliary/util/u_blitter.c +++ b/src/gallium/auxiliary/util/u_blitter.c @@ -109,9 +109,9 @@ struct blitter_context_priv struct blitter_context *util_blitter_create(struct pipe_context *pipe) { struct blitter_context_priv *ctx; - struct pipe_blend_state blend = { 0 }; - struct pipe_depth_stencil_alpha_state dsa = { { 0 } }; - struct pipe_rasterizer_state rs_state = { 0 }; + struct pipe_blend_state blend; + struct pipe_depth_stencil_alpha_state dsa; + struct pipe_rasterizer_state rs_state; struct pipe_sampler_state *sampler_state; struct pipe_vertex_element velem[2]; unsigned i; @@ -134,12 +134,14 @@ struct blitter_context *util_blitter_create(struct pipe_context *pipe) ctx->blitter.saved_num_sampler_states = ~0; /* blend state objects */ + memset(&blend, 0, sizeof(blend)); ctx->blend_keep_color = pipe->create_blend_state(pipe, &blend); blend.rt[0].colormask = PIPE_MASK_RGBA; ctx->blend_write_color = pipe->create_blend_state(pipe, &blend); /* depth stencil alpha state objects */ + memset(&dsa, 0, sizeof(dsa)); ctx->dsa_keep_depth_stencil = pipe->create_depth_stencil_alpha_state(pipe, &dsa); diff --git a/src/gallium/auxiliary/util/u_debug.c b/src/gallium/auxiliary/util/u_debug.c index 94be682c4b1..e997cfa8a38 100644 --- a/src/gallium/auxiliary/util/u_debug.c +++ b/src/gallium/auxiliary/util/u_debug.c @@ -421,26 +421,31 @@ void debug_dump_image(const char *prefix, #endif } -void debug_dump_surface(const char *prefix, +void debug_dump_surface(struct pipe_context *pipe, + const char *prefix, struct pipe_surface *surface) { struct pipe_texture *texture; - struct pipe_screen *screen; struct pipe_transfer *transfer; void *data; if (!surface) return; + /* XXX: this doesn't necessarily work, as the driver may be using + * temporary storage for the surface which hasn't been propagated + * back into the texture. Need to nail down the semantics of views + * and transfers a bit better before we can say if extra work needs + * to be done here: + */ texture = surface->texture; - screen = texture->screen; - transfer = screen->get_tex_transfer(screen, texture, surface->face, - surface->level, surface->zslice, - PIPE_TRANSFER_READ, 0, 0, surface->width, - surface->height); + transfer = pipe->get_tex_transfer(pipe, texture, surface->face, + surface->level, surface->zslice, + PIPE_TRANSFER_READ, 0, 0, surface->width, + surface->height); - data = screen->transfer_map(screen, transfer); + data = pipe->transfer_map(pipe, transfer); if(!data) goto error; @@ -452,13 +457,14 @@ void debug_dump_surface(const char *prefix, transfer->stride, data); - screen->transfer_unmap(screen, transfer); + pipe->transfer_unmap(pipe, transfer); error: - screen->tex_transfer_destroy(transfer); + pipe->tex_transfer_destroy(pipe, transfer); } -void debug_dump_texture(const char *prefix, +void debug_dump_texture(struct pipe_context *pipe, + const char *prefix, struct pipe_texture *texture) { struct pipe_surface *surface; @@ -473,7 +479,7 @@ void debug_dump_texture(const char *prefix, surface = screen->get_tex_surface(screen, texture, 0, 0, 0, PIPE_TEXTURE_USAGE_SAMPLER); if (surface) { - debug_dump_surface(prefix, surface); + debug_dump_surface(pipe, prefix, surface); screen->tex_surface_destroy(surface); } } @@ -511,27 +517,28 @@ struct bmp_rgb_quad { }; void -debug_dump_surface_bmp(const char *filename, +debug_dump_surface_bmp(struct pipe_context *pipe, + const char *filename, struct pipe_surface *surface) { #ifndef PIPE_SUBSYSTEM_WINDOWS_MINIPORT struct pipe_transfer *transfer; struct pipe_texture *texture = surface->texture; - struct pipe_screen *screen = texture->screen; - transfer = screen->get_tex_transfer(screen, texture, surface->face, - surface->level, surface->zslice, - PIPE_TRANSFER_READ, 0, 0, surface->width, - surface->height); + transfer = pipe->get_tex_transfer(pipe, texture, surface->face, + surface->level, surface->zslice, + PIPE_TRANSFER_READ, 0, 0, surface->width, + surface->height); - debug_dump_transfer_bmp(filename, transfer); + debug_dump_transfer_bmp(pipe, filename, transfer); - screen->tex_transfer_destroy(transfer); + pipe->tex_transfer_destroy(pipe, transfer); #endif } void -debug_dump_transfer_bmp(const char *filename, +debug_dump_transfer_bmp(struct pipe_context *pipe, + const char *filename, struct pipe_transfer *transfer) { #ifndef PIPE_SUBSYSTEM_WINDOWS_MINIPORT @@ -544,7 +551,7 @@ debug_dump_transfer_bmp(const char *filename, if(!rgba) goto error1; - pipe_get_tile_rgba(transfer, 0, 0, + pipe_get_tile_rgba(pipe, transfer, 0, 0, transfer->width, transfer->height, rgba); diff --git a/src/gallium/auxiliary/util/u_debug.h b/src/gallium/auxiliary/util/u_debug.h index 0f4768f3444..98addeb372e 100644 --- a/src/gallium/auxiliary/util/u_debug.h +++ b/src/gallium/auxiliary/util/u_debug.h @@ -312,6 +312,7 @@ debug_memory_end(unsigned long beginning); #ifdef DEBUG +struct pipe_context; struct pipe_surface; struct pipe_transfer; struct pipe_texture; @@ -321,21 +322,25 @@ void debug_dump_image(const char *prefix, unsigned width, unsigned height, unsigned stride, const void *data); -void debug_dump_surface(const char *prefix, +void debug_dump_surface(struct pipe_context *pipe, + const char *prefix, struct pipe_surface *surface); -void debug_dump_texture(const char *prefix, +void debug_dump_texture(struct pipe_context *pipe, + const char *prefix, struct pipe_texture *texture); -void debug_dump_surface_bmp(const char *filename, +void debug_dump_surface_bmp(struct pipe_context *pipe, + const char *filename, struct pipe_surface *surface); -void debug_dump_transfer_bmp(const char *filename, +void debug_dump_transfer_bmp(struct pipe_context *pipe, + const char *filename, struct pipe_transfer *transfer); void debug_dump_float_rgba_bmp(const char *filename, unsigned width, unsigned height, float *rgba, unsigned stride); #else #define debug_dump_image(prefix, format, cpp, width, height, stride, data) ((void)0) -#define debug_dump_surface(prefix, surface) ((void)0) -#define debug_dump_surface_bmp(filename, surface) ((void)0) +#define debug_dump_surface(pipe, prefix, surface) ((void)0) +#define debug_dump_surface_bmp(pipe, filename, surface) ((void)0) #define debug_dump_transfer_bmp(filename, transfer) ((void)0) #define debug_dump_float_rgba_bmp(filename, width, height, rgba, stride) ((void)0) #endif diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c index d421bee8efe..5c51b53d7bd 100644 --- a/src/gallium/auxiliary/util/u_gen_mipmap.c +++ b/src/gallium/auxiliary/util/u_gen_mipmap.c @@ -1119,7 +1119,6 @@ make_1d_mipmap(struct gen_mipmap_state *ctx, uint face, uint baseLevel, uint lastLevel) { struct pipe_context *pipe = ctx->pipe; - struct pipe_screen *screen = pipe->screen; const uint zslice = 0; uint dstLevel; @@ -1128,27 +1127,27 @@ make_1d_mipmap(struct gen_mipmap_state *ctx, struct pipe_transfer *srcTrans, *dstTrans; void *srcMap, *dstMap; - srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice, + srcTrans = pipe->get_tex_transfer(pipe, pt, face, srcLevel, zslice, PIPE_TRANSFER_READ, 0, 0, u_minify(pt->width0, srcLevel), u_minify(pt->height0, srcLevel)); - dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice, + dstTrans = pipe->get_tex_transfer(pipe, pt, face, dstLevel, zslice, PIPE_TRANSFER_WRITE, 0, 0, u_minify(pt->width0, dstLevel), u_minify(pt->height0, dstLevel)); - srcMap = (ubyte *) screen->transfer_map(screen, srcTrans); - dstMap = (ubyte *) screen->transfer_map(screen, dstTrans); + srcMap = (ubyte *) pipe->transfer_map(pipe, srcTrans); + dstMap = (ubyte *) pipe->transfer_map(pipe, dstTrans); reduce_1d(pt->format, srcTrans->width, srcMap, dstTrans->width, dstMap); - screen->transfer_unmap(screen, srcTrans); - screen->transfer_unmap(screen, dstTrans); + pipe->transfer_unmap(pipe, srcTrans); + pipe->transfer_unmap(pipe, dstTrans); - screen->tex_transfer_destroy(srcTrans); - screen->tex_transfer_destroy(dstTrans); + pipe->tex_transfer_destroy(pipe, srcTrans); + pipe->tex_transfer_destroy(pipe, dstTrans); } } @@ -1159,7 +1158,6 @@ make_2d_mipmap(struct gen_mipmap_state *ctx, uint face, uint baseLevel, uint lastLevel) { struct pipe_context *pipe = ctx->pipe; - struct pipe_screen *screen = pipe->screen; const uint zslice = 0; uint dstLevel; @@ -1171,17 +1169,17 @@ make_2d_mipmap(struct gen_mipmap_state *ctx, struct pipe_transfer *srcTrans, *dstTrans; ubyte *srcMap, *dstMap; - srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice, + srcTrans = pipe->get_tex_transfer(pipe, pt, face, srcLevel, zslice, PIPE_TRANSFER_READ, 0, 0, u_minify(pt->width0, srcLevel), u_minify(pt->height0, srcLevel)); - dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice, + dstTrans = pipe->get_tex_transfer(pipe, pt, face, dstLevel, zslice, PIPE_TRANSFER_WRITE, 0, 0, u_minify(pt->width0, dstLevel), u_minify(pt->height0, dstLevel)); - srcMap = (ubyte *) screen->transfer_map(screen, srcTrans); - dstMap = (ubyte *) screen->transfer_map(screen, dstTrans); + srcMap = (ubyte *) pipe->transfer_map(pipe, srcTrans); + dstMap = (ubyte *) pipe->transfer_map(pipe, dstTrans); reduce_2d(pt->format, srcTrans->width, srcTrans->height, @@ -1189,11 +1187,11 @@ make_2d_mipmap(struct gen_mipmap_state *ctx, dstTrans->width, dstTrans->height, dstTrans->stride, dstMap); - screen->transfer_unmap(screen, srcTrans); - screen->transfer_unmap(screen, dstTrans); + pipe->transfer_unmap(pipe, srcTrans); + pipe->transfer_unmap(pipe, dstTrans); - screen->tex_transfer_destroy(srcTrans); - screen->tex_transfer_destroy(dstTrans); + pipe->tex_transfer_destroy(pipe, srcTrans); + pipe->tex_transfer_destroy(pipe, dstTrans); } } @@ -1216,17 +1214,17 @@ make_3d_mipmap(struct gen_mipmap_state *ctx, struct pipe_transfer *srcTrans, *dstTrans; ubyte *srcMap, *dstMap; - srcTrans = screen->get_tex_transfer(screen, pt, face, srcLevel, zslice, + srcTrans = pipe->get_tex_transfer(pipe, pt, face, srcLevel, zslice, PIPE_TRANSFER_READ, 0, 0, u_minify(pt->width0, srcLevel), u_minify(pt->height0, srcLevel)); - dstTrans = screen->get_tex_transfer(screen, pt, face, dstLevel, zslice, + dstTrans = pipe->get_tex_transfer(pipe, pt, face, dstLevel, zslice, PIPE_TRANSFER_WRITE, 0, 0, u_minify(pt->width0, dstLevel), u_minify(pt->height0, dstLevel)); - srcMap = (ubyte *) screen->transfer_map(screen, srcTrans); - dstMap = (ubyte *) screen->transfer_map(screen, dstTrans); + srcMap = (ubyte *) pipe->transfer_map(pipe, srcTrans); + dstMap = (ubyte *) pipe->transfer_map(pipe, dstTrans); reduce_3d(pt->format, srcTrans->width, srcTrans->height, @@ -1234,11 +1232,11 @@ make_3d_mipmap(struct gen_mipmap_state *ctx, dstTrans->width, dstTrans->height, dstTrans->stride, dstMap); - screen->transfer_unmap(screen, srcTrans); - screen->transfer_unmap(screen, dstTrans); + pipe->transfer_unmap(pipe, srcTrans); + pipe->transfer_unmap(pipe, dstTrans); - screen->tex_transfer_destroy(srcTrans); - screen->tex_transfer_destroy(dstTrans); + pipe->tex_transfer_destroy(pipe, srcTrans); + pipe->tex_transfer_destroy(pipe, dstTrans); } #else (void) reduce_3d; diff --git a/src/gallium/auxiliary/util/u_inlines.h b/src/gallium/auxiliary/util/u_inlines.h index 84d849aa901..e22ab188e11 100644 --- a/src/gallium/auxiliary/util/u_inlines.h +++ b/src/gallium/auxiliary/util/u_inlines.h @@ -274,24 +274,24 @@ pipe_buffer_read(struct pipe_screen *screen, } static INLINE void * -pipe_transfer_map( struct pipe_transfer *transf ) +pipe_transfer_map( struct pipe_context *context, + struct pipe_transfer *transf ) { - struct pipe_screen *screen = transf->texture->screen; - return screen->transfer_map(screen, transf); + return context->transfer_map(context, transf); } static INLINE void -pipe_transfer_unmap( struct pipe_transfer *transf ) +pipe_transfer_unmap( struct pipe_context *context, + struct pipe_transfer *transf ) { - struct pipe_screen *screen = transf->texture->screen; - screen->transfer_unmap(screen, transf); + context->transfer_unmap(context, transf); } static INLINE void -pipe_transfer_destroy( struct pipe_transfer *transf ) +pipe_transfer_destroy( struct pipe_context *context, + struct pipe_transfer *transfer ) { - struct pipe_screen *screen = transf->texture->screen; - screen->tex_transfer_destroy(transf); + context->tex_transfer_destroy(context, transfer); } static INLINE unsigned diff --git a/src/gallium/auxiliary/util/u_rect.c b/src/gallium/auxiliary/util/u_rect.c index 8479161c744..e73797f1b7e 100644 --- a/src/gallium/auxiliary/util/u_rect.c +++ b/src/gallium/auxiliary/util/u_rect.c @@ -169,7 +169,6 @@ util_surface_copy(struct pipe_context *pipe, unsigned src_x, unsigned src_y, unsigned w, unsigned h) { - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *src_trans, *dst_trans; void *dst_map; const void *src_map; @@ -182,7 +181,7 @@ util_surface_copy(struct pipe_context *pipe, src_format = src->texture->format; dst_format = dst->texture->format; - src_trans = screen->get_tex_transfer(screen, + src_trans = pipe->get_tex_transfer(pipe, src->texture, src->face, src->level, @@ -190,7 +189,7 @@ util_surface_copy(struct pipe_context *pipe, PIPE_TRANSFER_READ, src_x, src_y, w, h); - dst_trans = screen->get_tex_transfer(screen, + dst_trans = pipe->get_tex_transfer(pipe, dst->texture, dst->face, dst->level, @@ -202,8 +201,8 @@ util_surface_copy(struct pipe_context *pipe, assert(util_format_get_blockwidth(dst_format) == util_format_get_blockwidth(src_format)); assert(util_format_get_blockheight(dst_format) == util_format_get_blockheight(src_format)); - src_map = pipe->screen->transfer_map(screen, src_trans); - dst_map = pipe->screen->transfer_map(screen, dst_trans); + src_map = pipe->transfer_map(pipe, src_trans); + dst_map = pipe->transfer_map(pipe, dst_trans); assert(src_map); assert(dst_map); @@ -221,11 +220,11 @@ util_surface_copy(struct pipe_context *pipe, do_flip ? h - 1 : 0); } - pipe->screen->transfer_unmap(pipe->screen, src_trans); - pipe->screen->transfer_unmap(pipe->screen, dst_trans); + pipe->transfer_unmap(pipe, src_trans); + pipe->transfer_unmap(pipe, dst_trans); - screen->tex_transfer_destroy(src_trans); - screen->tex_transfer_destroy(dst_trans); + pipe->tex_transfer_destroy(pipe, src_trans); + pipe->tex_transfer_destroy(pipe, dst_trans); } @@ -243,14 +242,13 @@ util_surface_fill(struct pipe_context *pipe, unsigned dstx, unsigned dsty, unsigned width, unsigned height, unsigned value) { - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *dst_trans; void *dst_map; assert(dst->texture); if (!dst->texture) return; - dst_trans = screen->get_tex_transfer(screen, + dst_trans = pipe->get_tex_transfer(pipe, dst->texture, dst->face, dst->level, @@ -258,7 +256,7 @@ util_surface_fill(struct pipe_context *pipe, PIPE_TRANSFER_WRITE, dstx, dsty, width, height); - dst_map = pipe->screen->transfer_map(screen, dst_trans); + dst_map = pipe->transfer_map(pipe, dst_trans); assert(dst_map); @@ -302,6 +300,6 @@ util_surface_fill(struct pipe_context *pipe, } } - pipe->screen->transfer_unmap(pipe->screen, dst_trans); - screen->tex_transfer_destroy(dst_trans); + pipe->transfer_unmap(pipe, dst_trans); + pipe->tex_transfer_destroy(pipe, dst_trans); } diff --git a/src/gallium/auxiliary/util/u_tile.c b/src/gallium/auxiliary/util/u_tile.c index 8a36d4d9d18..82e44192aaf 100644 --- a/src/gallium/auxiliary/util/u_tile.c +++ b/src/gallium/auxiliary/util/u_tile.c @@ -45,11 +45,11 @@ * Move raw block of pixels from transfer object to user memory. */ void -pipe_get_tile_raw(struct pipe_transfer *pt, +pipe_get_tile_raw(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, void *dst, int dst_stride) { - struct pipe_screen *screen = pt->texture->screen; const void *src; if (dst_stride == 0) @@ -58,14 +58,14 @@ pipe_get_tile_raw(struct pipe_transfer *pt, if (pipe_clip_tile(x, y, &w, &h, pt)) return; - src = screen->transfer_map(screen, pt); + src = pipe->transfer_map(pipe, pt); assert(src); if(!src) return; util_copy_rect(dst, pt->texture->format, dst_stride, 0, 0, w, h, src, pt->stride, x, y); - screen->transfer_unmap(screen, pt); + pipe->transfer_unmap(pipe, pt); } @@ -73,11 +73,11 @@ pipe_get_tile_raw(struct pipe_transfer *pt, * Move raw block of pixels from user memory to transfer object. */ void -pipe_put_tile_raw(struct pipe_transfer *pt, +pipe_put_tile_raw(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, const void *src, int src_stride) { - struct pipe_screen *screen = pt->texture->screen; void *dst; enum pipe_format format = pt->texture->format; @@ -87,14 +87,14 @@ pipe_put_tile_raw(struct pipe_transfer *pt, if (pipe_clip_tile(x, y, &w, &h, pt)) return; - dst = screen->transfer_map(screen, pt); + dst = pipe->transfer_map(pipe, pt); assert(dst); if(!dst) return; util_copy_rect(dst, format, pt->stride, x, y, w, h, src, src_stride, 0, 0); - screen->transfer_unmap(screen, pt); + pipe->transfer_unmap(pipe, pt); } @@ -1246,7 +1246,8 @@ pipe_tile_raw_to_rgba(enum pipe_format format, void -pipe_get_tile_rgba(struct pipe_transfer *pt, +pipe_get_tile_rgba(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, float *p) { @@ -1265,7 +1266,7 @@ pipe_get_tile_rgba(struct pipe_transfer *pt, if(format == PIPE_FORMAT_UYVY || format == PIPE_FORMAT_YUYV) assert((x & 1) == 0); - pipe_get_tile_raw(pt, x, y, w, h, packed, 0); + pipe_get_tile_raw(pipe, pt, x, y, w, h, packed, 0); pipe_tile_raw_to_rgba(format, packed, w, h, p, dst_stride); @@ -1274,7 +1275,8 @@ pipe_get_tile_rgba(struct pipe_transfer *pt, void -pipe_get_tile_swizzle(struct pipe_transfer *pt, +pipe_get_tile_swizzle(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, @@ -1304,7 +1306,7 @@ pipe_get_tile_swizzle(struct pipe_transfer *pt, assert((x & 1) == 0); } - pipe_get_tile_raw(pt, x, y, w, h, packed, 0); + pipe_get_tile_raw(pipe, pt, x, y, w, h, packed, 0); pipe_tile_raw_to_rgba(format, packed, w, h, p, dst_stride); @@ -1336,7 +1338,8 @@ pipe_get_tile_swizzle(struct pipe_transfer *pt, void -pipe_put_tile_rgba(struct pipe_transfer *pt, +pipe_put_tile_rgba(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, const float *p) { @@ -1425,7 +1428,7 @@ pipe_put_tile_rgba(struct pipe_transfer *pt, 0, 0, w, h); } - pipe_put_tile_raw(pt, x, y, w, h, packed, 0); + pipe_put_tile_raw(pipe, pt, x, y, w, h, packed, 0); FREE(packed); } @@ -1435,11 +1438,11 @@ pipe_put_tile_rgba(struct pipe_transfer *pt, * Get a block of Z values, converted to 32-bit range. */ void -pipe_get_tile_z(struct pipe_transfer *pt, +pipe_get_tile_z(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, uint *z) { - struct pipe_screen *screen = pt->texture->screen; const uint dstStride = w; ubyte *map; uint *pDest = z; @@ -1449,7 +1452,7 @@ pipe_get_tile_z(struct pipe_transfer *pt, if (pipe_clip_tile(x, y, &w, &h, pt)) return; - map = (ubyte *)screen->transfer_map(screen, pt); + map = (ubyte *)pipe->transfer_map(pipe, pt); if (!map) { assert(0); return; @@ -1515,16 +1518,16 @@ pipe_get_tile_z(struct pipe_transfer *pt, assert(0); } - screen->transfer_unmap(screen, pt); + pipe->transfer_unmap(pipe, pt); } void -pipe_put_tile_z(struct pipe_transfer *pt, +pipe_put_tile_z(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, const uint *zSrc) { - struct pipe_screen *screen = pt->texture->screen; const uint srcStride = w; const uint *ptrc = zSrc; ubyte *map; @@ -1534,7 +1537,7 @@ pipe_put_tile_z(struct pipe_transfer *pt, if (pipe_clip_tile(x, y, &w, &h, pt)) return; - map = (ubyte *)screen->transfer_map(screen, pt); + map = (ubyte *)pipe->transfer_map(pipe, pt); if (!map) { assert(0); return; @@ -1622,7 +1625,7 @@ pipe_put_tile_z(struct pipe_transfer *pt, assert(0); } - screen->transfer_unmap(screen, pt); + pipe->transfer_unmap(pipe, pt); } diff --git a/src/gallium/auxiliary/util/u_tile.h b/src/gallium/auxiliary/util/u_tile.h index d665fdb1bb1..1d8ce7d8cbc 100644 --- a/src/gallium/auxiliary/util/u_tile.h +++ b/src/gallium/auxiliary/util/u_tile.h @@ -56,23 +56,27 @@ extern "C" { #endif void -pipe_get_tile_raw(struct pipe_transfer *pt, +pipe_get_tile_raw(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, void *p, int dst_stride); void -pipe_put_tile_raw(struct pipe_transfer *pt, +pipe_put_tile_raw(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, const void *p, int src_stride); void -pipe_get_tile_rgba(struct pipe_transfer *pt, +pipe_get_tile_rgba(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, float *p); void -pipe_get_tile_swizzle(struct pipe_transfer *pt, +pipe_get_tile_swizzle(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, @@ -85,18 +89,21 @@ pipe_get_tile_swizzle(struct pipe_transfer *pt, float *p); void -pipe_put_tile_rgba(struct pipe_transfer *pt, +pipe_put_tile_rgba(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, const float *p); void -pipe_get_tile_z(struct pipe_transfer *pt, +pipe_get_tile_z(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, uint *z); void -pipe_put_tile_z(struct pipe_transfer *pt, +pipe_put_tile_z(struct pipe_context *pipe, + struct pipe_transfer *pt, uint x, uint y, uint w, uint h, const uint *z); diff --git a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c index 0763b5bb0e4..beb4722901e 100644 --- a/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c +++ b/src/gallium/auxiliary/vl/vl_mpeg12_mc_renderer.c @@ -680,14 +680,14 @@ xfer_buffers_map(struct vl_mpeg12_mc_renderer *r) assert(r); for (i = 0; i < 3; ++i) { - r->tex_transfer[i] = r->pipe->screen->get_tex_transfer + r->tex_transfer[i] = r->pipe->get_tex_transfer ( - r->pipe->screen, r->textures.all[i], + r->pipe, r->textures.all[i], 0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0, r->textures.all[i]->width0, r->textures.all[i]->height0 ); - r->texels[i] = r->pipe->screen->transfer_map(r->pipe->screen, r->tex_transfer[i]); + r->texels[i] = r->pipe->transfer_map(r->pipe, r->tex_transfer[i]); } } @@ -699,8 +699,8 @@ xfer_buffers_unmap(struct vl_mpeg12_mc_renderer *r) assert(r); for (i = 0; i < 3; ++i) { - r->pipe->screen->transfer_unmap(r->pipe->screen, r->tex_transfer[i]); - r->pipe->screen->tex_transfer_destroy(r->tex_transfer[i]); + r->pipe->transfer_unmap(r->pipe, r->tex_transfer[i]); + r->pipe->tex_transfer_destroy(r->pipe, r->tex_transfer[i]); } } diff --git a/src/gallium/drivers/cell/ppu/cell_context.c b/src/gallium/drivers/cell/ppu/cell_context.c index afcea616d59..f6cb1fc9be7 100644 --- a/src/gallium/drivers/cell/ppu/cell_context.c +++ b/src/gallium/drivers/cell/ppu/cell_context.c @@ -158,6 +158,7 @@ cell_create_context(struct pipe_screen *screen, cell_init_shader_functions(cell); cell_init_surface_functions(cell); cell_init_vertex_functions(cell); + cell_init_texture_transfer_funcs(cell); cell->draw = cell_draw_create(cell); diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c index 576d514741d..c54576b3c32 100644 --- a/src/gallium/drivers/cell/ppu/cell_gen_fragment.c +++ b/src/gallium/drivers/cell/ppu/cell_gen_fragment.c @@ -1352,7 +1352,7 @@ gen_stencil_values(struct spe_function *f, */ ASSERT(fbS_reg != newS_reg); - /* The code also assumes the the stencil_max_value is of the form + /* The code also assumes that the stencil_max_value is of the form * 2^n-1 and can therefore be used as a mask for the valid bits in * addition to a maximum. Make sure this is the case as well. * The clever math below exploits the fact that incrementing a diff --git a/src/gallium/drivers/cell/ppu/cell_spu.c b/src/gallium/drivers/cell/ppu/cell_spu.c index 28e5e6d706d..39284f3a5d1 100644 --- a/src/gallium/drivers/cell/ppu/cell_spu.c +++ b/src/gallium/drivers/cell/ppu/cell_spu.c @@ -135,7 +135,7 @@ cell_thread_function(void *arg) /** * Create the SPU threads. This is done once during driver initialization. - * This involves setting the the "init" message which is sent to each SPU. + * This involves setting the "init" message which is sent to each SPU. * The init message specifies an SPU id, total number of SPUs, location * and number of batch buffers, etc. */ diff --git a/src/gallium/drivers/cell/ppu/cell_state_vertex.c b/src/gallium/drivers/cell/ppu/cell_state_vertex.c index d3efb8ecea2..9510ea9ac2b 100644 --- a/src/gallium/drivers/cell/ppu/cell_state_vertex.c +++ b/src/gallium/drivers/cell/ppu/cell_state_vertex.c @@ -62,7 +62,8 @@ cell_bind_vertex_elements_state(struct pipe_context *pipe, cell->dirty |= CELL_NEW_VERTEX; - draw_set_vertex_elements(cell->draw, cell_velems->count, cell_velems->velem); + if (cell_velems) + draw_set_vertex_elements(cell->draw, cell_velems->count, cell_velems->velem); } void diff --git a/src/gallium/drivers/cell/ppu/cell_texture.c b/src/gallium/drivers/cell/ppu/cell_texture.c index c8a1acd86a9..c65c3b4f885 100644 --- a/src/gallium/drivers/cell/ppu/cell_texture.c +++ b/src/gallium/drivers/cell/ppu/cell_texture.c @@ -355,7 +355,7 @@ cell_tex_surface_destroy(struct pipe_surface *surf) * back out for glGetTexImage). */ static struct pipe_transfer * -cell_get_tex_transfer(struct pipe_screen *screen, +cell_get_tex_transfer(struct pipe_context *ctx, struct pipe_texture *texture, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, @@ -402,7 +402,7 @@ cell_get_tex_transfer(struct pipe_screen *screen, static void -cell_tex_transfer_destroy(struct pipe_transfer *t) +cell_tex_transfer_destroy(struct pipe_context *ctx, struct pipe_transfer *t) { struct cell_transfer *transfer = cell_transfer(t); /* Effectively do the texture_update work here - if texture images @@ -419,7 +419,7 @@ cell_tex_transfer_destroy(struct pipe_transfer *t) * Return pointer to texture image data in linear layout. */ static void * -cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer) +cell_transfer_map(struct pipe_context *ctx, struct pipe_transfer *transfer) { struct cell_transfer *ctrans = cell_transfer(transfer); struct pipe_texture *pt = transfer->texture; @@ -471,7 +471,7 @@ cell_transfer_map(struct pipe_screen *screen, struct pipe_transfer *transfer) * to tiled data. */ static void -cell_transfer_unmap(struct pipe_screen *screen, +cell_transfer_unmap(struct pipe_context *ctx, struct pipe_transfer *transfer) { struct cell_transfer *ctrans = cell_transfer(transfer); @@ -560,11 +560,14 @@ cell_init_screen_texture_funcs(struct pipe_screen *screen) screen->get_tex_surface = cell_get_tex_surface; screen->tex_surface_destroy = cell_tex_surface_destroy; - screen->get_tex_transfer = cell_get_tex_transfer; - screen->tex_transfer_destroy = cell_tex_transfer_destroy; - - screen->transfer_map = cell_transfer_map; - screen->transfer_unmap = cell_transfer_unmap; - screen->flush_frontbuffer = cell_flush_frontbuffer; } + +void +cell_init_texture_transfer_funcs(struct cell_context *cell) +{ + cell->pipe.get_tex_transfer = cell_get_tex_transfer; + cell->pipe.tex_transfer_destroy = cell_tex_transfer_destroy; + cell->pipe.transfer_map = cell_transfer_map; + cell->pipe.transfer_unmap = cell_transfer_unmap; +} diff --git a/src/gallium/drivers/cell/ppu/cell_texture.h b/src/gallium/drivers/cell/ppu/cell_texture.h index 2be05793122..ac0b9167750 100644 --- a/src/gallium/drivers/cell/ppu/cell_texture.h +++ b/src/gallium/drivers/cell/ppu/cell_texture.h @@ -95,5 +95,7 @@ cell_transfer(struct pipe_transfer *pt) extern void cell_init_screen_texture_funcs(struct pipe_screen *screen); +extern void +cell_init_texture_transfer_funcs(struct cell_context *cell); #endif /* CELL_TEXTURE_H */ diff --git a/src/gallium/drivers/i915/i915_context.c b/src/gallium/drivers/i915/i915_context.c index 3d45a22b7e7..130519ffa50 100644 --- a/src/gallium/drivers/i915/i915_context.c +++ b/src/gallium/drivers/i915/i915_context.c @@ -221,6 +221,7 @@ i915_create_context(struct pipe_screen *screen, void *priv) i915_init_surface_functions(i915); i915_init_state_functions(i915); i915_init_flush_functions(i915); + i915_init_texture_functions(i915); draw_install_aaline_stage(i915->draw, &i915->base); draw_install_aapoint_stage(i915->draw, &i915->base); diff --git a/src/gallium/drivers/i915/i915_context.h b/src/gallium/drivers/i915/i915_context.h index 9d03ca37fda..5348f62ca74 100644 --- a/src/gallium/drivers/i915/i915_context.h +++ b/src/gallium/drivers/i915/i915_context.h @@ -350,6 +350,12 @@ struct pipe_context *i915_create_context(struct pipe_screen *screen, /*********************************************************************** + * i915_texture.c + */ +void i915_init_texture_functions(struct i915_context *i915 ); + + +/*********************************************************************** * Inline conversion functions. These are better-typed than the * macros used previously: */ diff --git a/src/gallium/drivers/i915/i915_state.c b/src/gallium/drivers/i915/i915_state.c index e54997736f9..0f7395246cc 100644 --- a/src/gallium/drivers/i915/i915_state.c +++ b/src/gallium/drivers/i915/i915_state.c @@ -797,7 +797,10 @@ i915_bind_vertex_elements_state(struct pipe_context *pipe, draw_flush(i915->draw); /* pass-through to draw module */ - draw_set_vertex_elements(i915->draw, i915_velems->count, i915_velems->velem); + if (i915_velems) { + draw_set_vertex_elements(i915->draw, + i915_velems->count, i915_velems->velem); + } } static void diff --git a/src/gallium/drivers/i915/i915_texture.c b/src/gallium/drivers/i915/i915_texture.c index 3ce52cdcdbd..b252fb5330c 100644 --- a/src/gallium/drivers/i915/i915_texture.c +++ b/src/gallium/drivers/i915/i915_texture.c @@ -795,12 +795,12 @@ i915_tex_surface_destroy(struct pipe_surface *surf) /* - * Screen transfer functions + * Texture transfer functions */ -static struct pipe_transfer* -i915_get_tex_transfer(struct pipe_screen *screen, +static struct pipe_transfer * +i915_get_tex_transfer(struct pipe_context *pipe, struct pipe_texture *texture, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, unsigned x, unsigned y, @@ -837,7 +837,7 @@ i915_get_tex_transfer(struct pipe_screen *screen, } static void * -i915_transfer_map(struct pipe_screen *screen, +i915_transfer_map(struct pipe_context *pipe, struct pipe_transfer *transfer) { struct i915_texture *tex = (struct i915_texture *)transfer->texture; @@ -859,7 +859,7 @@ i915_transfer_map(struct pipe_screen *screen, } static void -i915_transfer_unmap(struct pipe_screen *screen, +i915_transfer_unmap(struct pipe_context *pipe, struct pipe_transfer *transfer) { struct i915_texture *tex = (struct i915_texture *)transfer->texture; @@ -868,7 +868,8 @@ i915_transfer_unmap(struct pipe_screen *screen, } static void -i915_tex_transfer_destroy(struct pipe_transfer *trans) +i915_tex_transfer_destroy(struct pipe_context *pipe, + struct pipe_transfer *trans) { pipe_texture_reference(&trans->texture, NULL); FREE(trans); @@ -879,6 +880,14 @@ i915_tex_transfer_destroy(struct pipe_transfer *trans) * Other texture functions */ +void +i915_init_texture_functions(struct i915_context *i915 ) +{ + i915->base.get_tex_transfer = i915_get_tex_transfer; + i915->base.transfer_map = i915_transfer_map; + i915->base.transfer_unmap = i915_transfer_unmap; + i915->base.tex_transfer_destroy = i915_tex_transfer_destroy; +} void i915_init_screen_texture_functions(struct i915_screen *is) @@ -889,8 +898,4 @@ i915_init_screen_texture_functions(struct i915_screen *is) is->base.texture_destroy = i915_texture_destroy; is->base.get_tex_surface = i915_get_tex_surface; is->base.tex_surface_destroy = i915_tex_surface_destroy; - is->base.get_tex_transfer = i915_get_tex_transfer; - is->base.transfer_map = i915_transfer_map; - is->base.transfer_unmap = i915_transfer_unmap; - is->base.tex_transfer_destroy = i915_tex_transfer_destroy; } diff --git a/src/gallium/drivers/i965/brw_screen.h b/src/gallium/drivers/i965/brw_screen.h index e0f3cd2a9fe..e3a7c64d489 100644 --- a/src/gallium/drivers/i965/brw_screen.h +++ b/src/gallium/drivers/i965/brw_screen.h @@ -181,6 +181,10 @@ void brw_update_texture( struct brw_screen *brw_screen, struct brw_texture *tex ); +/* brw_screen_texture.h + */ +struct brw_context; +void brw_tex_init( struct brw_context *brw ); void brw_screen_tex_init( struct brw_screen *brw_screen ); void brw_screen_tex_surface_init( struct brw_screen *brw_screen ); diff --git a/src/gallium/drivers/i965/brw_screen_texture.c b/src/gallium/drivers/i965/brw_screen_texture.c index cc79bfc7715..cadcb7cee2a 100644 --- a/src/gallium/drivers/i965/brw_screen_texture.c +++ b/src/gallium/drivers/i965/brw_screen_texture.c @@ -37,6 +37,8 @@ #include "brw_defines.h" #include "brw_structs.h" #include "brw_winsys.h" +#include "brw_context.h" + @@ -479,7 +481,7 @@ boolean brw_is_texture_referenced_by_bo( struct brw_screen *brw_screen, */ static struct pipe_transfer* -brw_get_tex_transfer(struct pipe_screen *screen, +brw_get_tex_transfer(struct pipe_context *pipe, struct pipe_texture *texture, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, unsigned x, unsigned y, @@ -514,11 +516,11 @@ brw_get_tex_transfer(struct pipe_screen *screen, } static void * -brw_transfer_map(struct pipe_screen *screen, +brw_transfer_map(struct pipe_context *pipe, struct pipe_transfer *transfer) { struct brw_texture *tex = brw_texture(transfer->texture); - struct brw_winsys_screen *sws = brw_screen(screen)->sws; + struct brw_winsys_screen *sws = brw_screen(pipe->screen)->sws; char *map; unsigned usage = transfer->usage; @@ -541,23 +543,32 @@ brw_transfer_map(struct pipe_screen *screen, } static void -brw_transfer_unmap(struct pipe_screen *screen, +brw_transfer_unmap(struct pipe_context *pipe, struct pipe_transfer *transfer) { struct brw_texture *tex = brw_texture(transfer->texture); - struct brw_winsys_screen *sws = brw_screen(screen)->sws; + struct brw_winsys_screen *sws = brw_screen(pipe->screen)->sws; sws->bo_unmap(tex->bo); } static void -brw_tex_transfer_destroy(struct pipe_transfer *trans) +brw_tex_transfer_destroy(struct pipe_context *pipe, + struct pipe_transfer *trans) { pipe_texture_reference(&trans->texture, NULL); FREE(trans); } +void brw_tex_init( struct brw_context *brw ) +{ + brw->base.get_tex_transfer = brw_get_tex_transfer; + brw->base.transfer_map = brw_transfer_map; + brw->base.transfer_unmap = brw_transfer_unmap; + brw->base.tex_transfer_destroy = brw_tex_transfer_destroy; +} + void brw_screen_tex_init( struct brw_screen *brw_screen ) { brw_screen->base.is_format_supported = brw_is_format_supported; @@ -565,8 +576,4 @@ void brw_screen_tex_init( struct brw_screen *brw_screen ) brw_screen->base.texture_from_handle = brw_texture_from_handle; brw_screen->base.texture_get_handle = brw_texture_get_handle; brw_screen->base.texture_destroy = brw_texture_destroy; - brw_screen->base.get_tex_transfer = brw_get_tex_transfer; - brw_screen->base.transfer_map = brw_transfer_map; - brw_screen->base.transfer_unmap = brw_transfer_unmap; - brw_screen->base.tex_transfer_destroy = brw_tex_transfer_destroy; } diff --git a/src/gallium/drivers/identity/id_context.c b/src/gallium/drivers/identity/id_context.c index ef5b428161c..00a542215ad 100644 --- a/src/gallium/drivers/identity/id_context.c +++ b/src/gallium/drivers/identity/id_context.c @@ -747,6 +747,75 @@ identity_sampler_view_destroy(struct pipe_context *pipe, free(view); } + +static struct pipe_transfer * +identity_context_get_tex_transfer(struct pipe_context *_context, + struct pipe_texture *_texture, + unsigned face, + unsigned level, + unsigned zslice, + enum pipe_transfer_usage usage, + unsigned x, + unsigned y, + unsigned w, + unsigned h) +{ + struct identity_context *id_context = identity_context(_context); + struct identity_texture *id_texture = identity_texture(_texture); + struct pipe_context *context = id_context->pipe; + struct pipe_texture *texture = id_texture->texture; + struct pipe_transfer *result; + + result = context->get_tex_transfer(context, + texture, + face, + level, + zslice, + usage, + x, + y, + w, + h); + + if (result) + return identity_transfer_create(id_context, id_texture, result); + return NULL; +} + +static void +identity_context_tex_transfer_destroy(struct pipe_context *_pipe, + struct pipe_transfer *_transfer) +{ + identity_transfer_destroy(identity_context(_pipe), + identity_transfer(_transfer)); +} + +static void * +identity_context_transfer_map(struct pipe_context *_context, + struct pipe_transfer *_transfer) +{ + struct identity_context *id_context = identity_context(_context); + struct identity_transfer *id_transfer = identity_transfer(_transfer); + struct pipe_context *context = id_context->pipe; + struct pipe_transfer *transfer = id_transfer->transfer; + + return context->transfer_map(context, + transfer); +} + +static void +identity_context_transfer_unmap(struct pipe_context *_context, + struct pipe_transfer *_transfer) +{ + struct identity_context *id_context = identity_context(_context); + struct identity_transfer *id_transfer = identity_transfer(_transfer); + struct pipe_context *context = id_context->pipe; + struct pipe_transfer *transfer = id_transfer->transfer; + + context->transfer_unmap(context, + transfer); +} + struct pipe_context * identity_context_create(struct pipe_screen *_screen, struct pipe_context *pipe) { @@ -813,6 +882,10 @@ identity_context_create(struct pipe_screen *_screen, struct pipe_context *pipe) id_pipe->base.is_buffer_referenced = identity_is_buffer_referenced; id_pipe->base.create_sampler_view = identity_create_sampler_view; id_pipe->base.sampler_view_destroy = identity_sampler_view_destroy; + id_pipe->base.get_tex_transfer = identity_context_get_tex_transfer; + id_pipe->base.tex_transfer_destroy = identity_context_tex_transfer_destroy; + id_pipe->base.transfer_map = identity_context_transfer_map; + id_pipe->base.transfer_unmap = identity_context_transfer_unmap; id_pipe->pipe = pipe; diff --git a/src/gallium/drivers/identity/id_objects.c b/src/gallium/drivers/identity/id_objects.c index 2b1a60c1bf1..d37fb0042e5 100644 --- a/src/gallium/drivers/identity/id_objects.c +++ b/src/gallium/drivers/identity/id_objects.c @@ -30,6 +30,7 @@ #include "id_screen.h" #include "id_objects.h" +#include "id_context.h" struct pipe_buffer * identity_buffer_create(struct identity_screen *id_screen, @@ -142,7 +143,8 @@ identity_surface_destroy(struct identity_surface *id_surface) struct pipe_transfer * -identity_transfer_create(struct identity_texture *id_texture, +identity_transfer_create(struct identity_context *id_context, + struct identity_texture *id_texture, struct pipe_transfer *transfer) { struct identity_transfer *id_transfer; @@ -159,25 +161,25 @@ identity_transfer_create(struct identity_texture *id_texture, memcpy(&id_transfer->base, transfer, sizeof(struct pipe_transfer)); id_transfer->base.texture = NULL; - pipe_texture_reference(&id_transfer->base.texture, &id_texture->base); id_transfer->transfer = transfer; + + pipe_texture_reference(&id_transfer->base.texture, &id_texture->base); assert(id_transfer->base.texture == &id_texture->base); return &id_transfer->base; error: - transfer->texture->screen->tex_transfer_destroy(transfer); + id_context->pipe->tex_transfer_destroy(id_context->pipe, transfer); return NULL; } void -identity_transfer_destroy(struct identity_transfer *id_transfer) +identity_transfer_destroy(struct identity_context *id_context, + struct identity_transfer *id_transfer) { - struct identity_screen *id_screen = identity_screen(id_transfer->base.texture->screen); - struct pipe_screen *screen = id_screen->screen; - pipe_texture_reference(&id_transfer->base.texture, NULL); - screen->tex_transfer_destroy(id_transfer->transfer); + id_context->pipe->tex_transfer_destroy(id_context->pipe, + id_transfer->transfer); FREE(id_transfer); } diff --git a/src/gallium/drivers/identity/id_objects.h b/src/gallium/drivers/identity/id_objects.h index b48df83b3ed..9a07ebe8d72 100644 --- a/src/gallium/drivers/identity/id_objects.h +++ b/src/gallium/drivers/identity/id_objects.h @@ -35,6 +35,7 @@ #include "id_screen.h" +struct identity_context; struct identity_buffer { @@ -72,6 +73,7 @@ struct identity_transfer { struct pipe_transfer base; + struct pipe_context *pipe; struct pipe_transfer *transfer; }; @@ -203,11 +205,13 @@ void identity_surface_destroy(struct identity_surface *id_surface); struct pipe_transfer * -identity_transfer_create(struct identity_texture *id_texture, +identity_transfer_create(struct identity_context *id_context, + struct identity_texture *id_texture, struct pipe_transfer *transfer); void -identity_transfer_destroy(struct identity_transfer *id_transfer); +identity_transfer_destroy(struct identity_context *id_context, + struct identity_transfer *id_transfer); struct pipe_video_surface * identity_video_surface_create(struct identity_screen *id_screen, diff --git a/src/gallium/drivers/identity/id_screen.c b/src/gallium/drivers/identity/id_screen.c index b9d0f003d74..419b1465787 100644 --- a/src/gallium/drivers/identity/id_screen.c +++ b/src/gallium/drivers/identity/id_screen.c @@ -207,71 +207,6 @@ identity_screen_tex_surface_destroy(struct pipe_surface *_surface) identity_surface_destroy(identity_surface(_surface)); } -static struct pipe_transfer * -identity_screen_get_tex_transfer(struct pipe_screen *_screen, - struct pipe_texture *_texture, - unsigned face, - unsigned level, - unsigned zslice, - enum pipe_transfer_usage usage, - unsigned x, - unsigned y, - unsigned w, - unsigned h) -{ - struct identity_screen *id_screen = identity_screen(_screen); - struct identity_texture *id_texture = identity_texture(_texture); - struct pipe_screen *screen = id_screen->screen; - struct pipe_texture *texture = id_texture->texture; - struct pipe_transfer *result; - - result = screen->get_tex_transfer(screen, - texture, - face, - level, - zslice, - usage, - x, - y, - w, - h); - - if (result) - return identity_transfer_create(id_texture, result); - return NULL; -} - -static void -identity_screen_tex_transfer_destroy(struct pipe_transfer *_transfer) -{ - identity_transfer_destroy(identity_transfer(_transfer)); -} - -static void * -identity_screen_transfer_map(struct pipe_screen *_screen, - struct pipe_transfer *_transfer) -{ - struct identity_screen *id_screen = identity_screen(_screen); - struct identity_transfer *id_transfer = identity_transfer(_transfer); - struct pipe_screen *screen = id_screen->screen; - struct pipe_transfer *transfer = id_transfer->transfer; - - return screen->transfer_map(screen, - transfer); -} - -static void -identity_screen_transfer_unmap(struct pipe_screen *_screen, - struct pipe_transfer *_transfer) -{ - struct identity_screen *id_screen = identity_screen(_screen); - struct identity_transfer *id_transfer = identity_transfer(_transfer); - struct pipe_screen *screen = id_screen->screen; - struct pipe_transfer *transfer = id_transfer->transfer; - - screen->transfer_unmap(screen, - transfer); -} static struct pipe_buffer * identity_screen_buffer_create(struct pipe_screen *_screen, @@ -488,10 +423,6 @@ identity_screen_create(struct pipe_screen *screen) id_screen->base.texture_destroy = identity_screen_texture_destroy; id_screen->base.get_tex_surface = identity_screen_get_tex_surface; id_screen->base.tex_surface_destroy = identity_screen_tex_surface_destroy; - id_screen->base.get_tex_transfer = identity_screen_get_tex_transfer; - id_screen->base.tex_transfer_destroy = identity_screen_tex_transfer_destroy; - id_screen->base.transfer_map = identity_screen_transfer_map; - id_screen->base.transfer_unmap = identity_screen_transfer_unmap; id_screen->base.buffer_create = identity_screen_buffer_create; id_screen->base.user_buffer_create = identity_screen_user_buffer_create; if (screen->buffer_map) diff --git a/src/gallium/drivers/llvmpipe/Makefile b/src/gallium/drivers/llvmpipe/Makefile index 41ac1cee72d..89c06ea3ad7 100644 --- a/src/gallium/drivers/llvmpipe/Makefile +++ b/src/gallium/drivers/llvmpipe/Makefile @@ -55,7 +55,7 @@ testprogs := lp_test_format \ LIBS += $(GL_LIB_DEPS) -L. -lllvmpipe -L../../auxiliary/ -lgallium -$(testprogs): lp_test_% : lp_test_%.o lp_test_main.o libllvmpipe.a - $(LD) $(filter %.o,$^) -o $@ -Wl,--start-group $(LIBS) -Wl,--end-group +#$(testprogs): lp_test_% : lp_test_%.o lp_test_main.o libllvmpipe.a +# $(LD) $(filter %.o,$^) -o $@ -Wl,--start-group $(LIBS) -Wl,--end-group -default: $(testprogs) +#default: $(testprogs) diff --git a/src/gallium/drivers/llvmpipe/lp_buffer.c b/src/gallium/drivers/llvmpipe/lp_buffer.c index dab20cb6397..6e0f37393e9 100644 --- a/src/gallium/drivers/llvmpipe/lp_buffer.c +++ b/src/gallium/drivers/llvmpipe/lp_buffer.c @@ -33,7 +33,6 @@ #include "lp_screen.h" #include "lp_buffer.h" -#include "state_tracker/sw_winsys.h" static void * llvmpipe_buffer_map(struct pipe_screen *screen, diff --git a/src/gallium/drivers/llvmpipe/lp_context.c b/src/gallium/drivers/llvmpipe/lp_context.c index bf7ed118095..951a695f964 100644 --- a/src/gallium/drivers/llvmpipe/lp_context.c +++ b/src/gallium/drivers/llvmpipe/lp_context.c @@ -175,6 +175,7 @@ llvmpipe_create_context( struct pipe_screen *screen, void *priv ) llvmpipe->pipe.is_buffer_referenced = llvmpipe_is_buffer_referenced; llvmpipe_init_query_funcs( llvmpipe ); + llvmpipe_init_context_texture_funcs( &llvmpipe->pipe ); /* * Create drawing context and plug our rendering stage into it. diff --git a/src/gallium/drivers/llvmpipe/lp_context.h b/src/gallium/drivers/llvmpipe/lp_context.h index f83e5ffacf8..71f991049e5 100644 --- a/src/gallium/drivers/llvmpipe/lp_context.h +++ b/src/gallium/drivers/llvmpipe/lp_context.h @@ -45,7 +45,7 @@ struct draw_stage; struct lp_fragment_shader; struct lp_vertex_shader; struct lp_blend_state; -struct setup_context; +struct lp_setup_context; struct lp_velems_state; struct llvmpipe_context { @@ -98,7 +98,7 @@ struct llvmpipe_context { int psize_slot; /** The tiling engine */ - struct setup_context *setup; + struct lp_setup_context *setup; /** The primitive drawing context */ struct draw_context *draw; diff --git a/src/gallium/drivers/llvmpipe/lp_flush.c b/src/gallium/drivers/llvmpipe/lp_flush.c index bf832433be1..1b4e8899359 100644 --- a/src/gallium/drivers/llvmpipe/lp_flush.c +++ b/src/gallium/drivers/llvmpipe/lp_flush.c @@ -79,12 +79,12 @@ llvmpipe_flush( struct pipe_context *pipe, for (i = 0; i < llvmpipe->framebuffer.nr_cbufs; i++) { util_snprintf(filename, sizeof(filename), "cbuf%u_%u", i, frame_no); - debug_dump_surface(filename, llvmpipe->framebuffer.cbufs[i]); + debug_dump_surface_bmp(&llvmpipe->pipe, filename, llvmpipe->framebuffer.cbufs[0]); } if (0) { util_snprintf(filename, sizeof(filename), "zsbuf_%u", frame_no); - debug_dump_surface(filename, llvmpipe->framebuffer.zsbuf); + debug_dump_surface_bmp(&llvmpipe->pipe, filename, llvmpipe->framebuffer.zsbuf); } ++frame_no; diff --git a/src/gallium/drivers/llvmpipe/lp_jit.c b/src/gallium/drivers/llvmpipe/lp_jit.c index 08c8f93794d..5887613120d 100644 --- a/src/gallium/drivers/llvmpipe/lp_jit.c +++ b/src/gallium/drivers/llvmpipe/lp_jit.c @@ -57,7 +57,8 @@ lp_jit_init_globals(struct llvmpipe_screen *screen) elem_types[LP_JIT_TEXTURE_HEIGHT] = LLVMInt32Type(); elem_types[LP_JIT_TEXTURE_DEPTH] = LLVMInt32Type(); elem_types[LP_JIT_TEXTURE_LAST_LEVEL] = LLVMInt32Type(); - elem_types[LP_JIT_TEXTURE_STRIDE] = LLVMInt32Type(); + elem_types[LP_JIT_TEXTURE_ROW_STRIDE] = + LLVMArrayType(LLVMInt32Type(), LP_MAX_TEXTURE_2D_LEVELS); elem_types[LP_JIT_TEXTURE_DATA] = LLVMArrayType(LLVMPointerType(LLVMInt8Type(), 0), LP_MAX_TEXTURE_2D_LEVELS); @@ -76,9 +77,9 @@ lp_jit_init_globals(struct llvmpipe_screen *screen) LP_CHECK_MEMBER_OFFSET(struct lp_jit_texture, last_level, screen->target, texture_type, LP_JIT_TEXTURE_LAST_LEVEL); - LP_CHECK_MEMBER_OFFSET(struct lp_jit_texture, stride, + LP_CHECK_MEMBER_OFFSET(struct lp_jit_texture, row_stride, screen->target, texture_type, - LP_JIT_TEXTURE_STRIDE); + LP_JIT_TEXTURE_ROW_STRIDE); LP_CHECK_MEMBER_OFFSET(struct lp_jit_texture, data, screen->target, texture_type, LP_JIT_TEXTURE_DATA); diff --git a/src/gallium/drivers/llvmpipe/lp_jit.h b/src/gallium/drivers/llvmpipe/lp_jit.h index 5cc7a12c03b..13167ae3bf4 100644 --- a/src/gallium/drivers/llvmpipe/lp_jit.h +++ b/src/gallium/drivers/llvmpipe/lp_jit.h @@ -51,7 +51,7 @@ struct lp_jit_texture uint32_t height; uint32_t depth; uint32_t last_level; - uint32_t stride; + uint32_t row_stride[LP_MAX_TEXTURE_2D_LEVELS]; const void *data[LP_MAX_TEXTURE_2D_LEVELS]; }; @@ -61,7 +61,7 @@ enum { LP_JIT_TEXTURE_HEIGHT, LP_JIT_TEXTURE_DEPTH, LP_JIT_TEXTURE_LAST_LEVEL, - LP_JIT_TEXTURE_STRIDE, + LP_JIT_TEXTURE_ROW_STRIDE, LP_JIT_TEXTURE_DATA }; diff --git a/src/gallium/drivers/llvmpipe/lp_rast.h b/src/gallium/drivers/llvmpipe/lp_rast.h index dc5fc5fc7d6..303f6e3f7e4 100644 --- a/src/gallium/drivers/llvmpipe/lp_rast.h +++ b/src/gallium/drivers/llvmpipe/lp_rast.h @@ -95,7 +95,7 @@ struct lp_rast_shader_inputs { * Rasterization information for a triangle known to be in this bin, * plus inputs to run the shader: * These fields are tile- and bin-independent. - * Objects of this type are put into the setup_context::data buffer. + * Objects of this type are put into the lp_setup_context::data buffer. */ struct lp_rast_triangle { #ifdef DEBUG diff --git a/src/gallium/drivers/llvmpipe/lp_scene.c b/src/gallium/drivers/llvmpipe/lp_scene.c index 72492c0f0ca..505cb21503a 100644 --- a/src/gallium/drivers/llvmpipe/lp_scene.c +++ b/src/gallium/drivers/llvmpipe/lp_scene.c @@ -397,7 +397,7 @@ end: static boolean lp_scene_map_buffers( struct lp_scene *scene ) { - struct pipe_screen *screen = scene->pipe->screen; + struct pipe_context *pipe = scene->pipe; struct pipe_surface *cbuf, *zsbuf; int i; @@ -409,7 +409,7 @@ lp_scene_map_buffers( struct lp_scene *scene ) for (i = 0; i < scene->fb.nr_cbufs; i++) { cbuf = scene->fb.cbufs[i]; if (cbuf) { - scene->cbuf_transfer[i] = screen->get_tex_transfer(screen, + scene->cbuf_transfer[i] = pipe->get_tex_transfer(pipe, cbuf->texture, cbuf->face, cbuf->level, @@ -421,7 +421,7 @@ lp_scene_map_buffers( struct lp_scene *scene ) if (!scene->cbuf_transfer[i]) goto fail; - scene->cbuf_map[i] = screen->transfer_map(screen, + scene->cbuf_map[i] = pipe->transfer_map(pipe, scene->cbuf_transfer[i]); if (!scene->cbuf_map[i]) goto fail; @@ -432,7 +432,7 @@ lp_scene_map_buffers( struct lp_scene *scene ) */ zsbuf = scene->fb.zsbuf; if (zsbuf) { - scene->zsbuf_transfer = screen->get_tex_transfer(screen, + scene->zsbuf_transfer = pipe->get_tex_transfer(pipe, zsbuf->texture, zsbuf->face, zsbuf->level, @@ -444,7 +444,7 @@ lp_scene_map_buffers( struct lp_scene *scene ) if (!scene->zsbuf_transfer) goto fail; - scene->zsbuf_map = screen->transfer_map(screen, + scene->zsbuf_map = pipe->transfer_map(pipe, scene->zsbuf_transfer); if (!scene->zsbuf_map) goto fail; @@ -469,25 +469,25 @@ fail: static void lp_scene_unmap_buffers( struct lp_scene *scene ) { - struct pipe_screen *screen = scene->pipe->screen; + struct pipe_context *pipe = scene->pipe; unsigned i; for (i = 0; i < scene->fb.nr_cbufs; i++) { if (scene->cbuf_map[i]) - screen->transfer_unmap(screen, scene->cbuf_transfer[i]); + pipe->transfer_unmap(pipe, scene->cbuf_transfer[i]); if (scene->cbuf_transfer[i]) - screen->tex_transfer_destroy(scene->cbuf_transfer[i]); + pipe->tex_transfer_destroy(pipe, scene->cbuf_transfer[i]); scene->cbuf_transfer[i] = NULL; scene->cbuf_map[i] = NULL; } if (scene->zsbuf_map) - screen->transfer_unmap(screen, scene->zsbuf_transfer); + pipe->transfer_unmap(pipe, scene->zsbuf_transfer); if (scene->zsbuf_transfer) - screen->tex_transfer_destroy(scene->zsbuf_transfer); + pipe->tex_transfer_destroy(pipe, scene->zsbuf_transfer); scene->zsbuf_transfer = NULL; scene->zsbuf_map = NULL; diff --git a/src/gallium/drivers/llvmpipe/lp_screen.h b/src/gallium/drivers/llvmpipe/lp_screen.h index f4e62cbf08e..d977f98cfaa 100644 --- a/src/gallium/drivers/llvmpipe/lp_screen.h +++ b/src/gallium/drivers/llvmpipe/lp_screen.h @@ -34,9 +34,7 @@ #ifndef LP_SCREEN_H #define LP_SCREEN_H -#include <llvm-c/Core.h> -#include <llvm-c/Analysis.h> -#include <llvm-c/Target.h> +#include "os/os_llvm.h" #include <llvm-c/ExecutionEngine.h> #include "pipe/p_screen.h" diff --git a/src/gallium/drivers/llvmpipe/lp_setup.c b/src/gallium/drivers/llvmpipe/lp_setup.c index 059584f2a77..cd16b6b2d38 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup.c +++ b/src/gallium/drivers/llvmpipe/lp_setup.c @@ -52,11 +52,11 @@ #include "draw/draw_vbuf.h" -static void set_scene_state( struct setup_context *, unsigned ); +static void set_scene_state( struct lp_setup_context *, unsigned ); struct lp_scene * -lp_setup_get_current_scene(struct setup_context *setup) +lp_setup_get_current_scene(struct lp_setup_context *setup) { if (!setup->scene) { @@ -74,7 +74,7 @@ lp_setup_get_current_scene(struct setup_context *setup) static void -first_triangle( struct setup_context *setup, +first_triangle( struct lp_setup_context *setup, const float (*v0)[4], const float (*v1)[4], const float (*v2)[4]) @@ -85,7 +85,7 @@ first_triangle( struct setup_context *setup, } static void -first_line( struct setup_context *setup, +first_line( struct lp_setup_context *setup, const float (*v0)[4], const float (*v1)[4]) { @@ -95,7 +95,7 @@ first_line( struct setup_context *setup, } static void -first_point( struct setup_context *setup, +first_point( struct lp_setup_context *setup, const float (*v0)[4]) { set_scene_state( setup, SETUP_ACTIVE ); @@ -103,7 +103,7 @@ first_point( struct setup_context *setup, setup->point( setup, v0 ); } -static void reset_context( struct setup_context *setup ) +static void reset_context( struct lp_setup_context *setup ) { LP_DBG(DEBUG_SETUP, "%s\n", __FUNCTION__); @@ -131,7 +131,7 @@ static void reset_context( struct setup_context *setup ) /** Rasterize all scene's bins */ static void -lp_setup_rasterize_scene( struct setup_context *setup, +lp_setup_rasterize_scene( struct lp_setup_context *setup, boolean write_depth ) { struct lp_scene *scene = lp_setup_get_current_scene(setup); @@ -148,7 +148,7 @@ lp_setup_rasterize_scene( struct setup_context *setup, static void -begin_binning( struct setup_context *setup ) +begin_binning( struct lp_setup_context *setup ) { struct lp_scene *scene = lp_setup_get_current_scene(setup); @@ -184,7 +184,7 @@ begin_binning( struct setup_context *setup ) * TODO: fast path for fullscreen clears and no triangles. */ static void -execute_clears( struct setup_context *setup ) +execute_clears( struct lp_setup_context *setup ) { LP_DBG(DEBUG_SETUP, "%s\n", __FUNCTION__); @@ -194,7 +194,7 @@ execute_clears( struct setup_context *setup ) static void -set_scene_state( struct setup_context *setup, +set_scene_state( struct lp_setup_context *setup, unsigned new_state ) { unsigned old_state = setup->state; @@ -229,7 +229,7 @@ set_scene_state( struct setup_context *setup, void -lp_setup_flush( struct setup_context *setup, +lp_setup_flush( struct lp_setup_context *setup, unsigned flags ) { LP_DBG(DEBUG_SETUP, "%s\n", __FUNCTION__); @@ -239,7 +239,7 @@ lp_setup_flush( struct setup_context *setup, void -lp_setup_bind_framebuffer( struct setup_context *setup, +lp_setup_bind_framebuffer( struct lp_setup_context *setup, const struct pipe_framebuffer_state *fb ) { LP_DBG(DEBUG_SETUP, "%s\n", __FUNCTION__); @@ -256,7 +256,7 @@ lp_setup_bind_framebuffer( struct setup_context *setup, void -lp_setup_clear( struct setup_context *setup, +lp_setup_clear( struct lp_setup_context *setup, const float *color, double depth, unsigned stencil, @@ -314,7 +314,7 @@ lp_setup_clear( struct setup_context *setup, * Emit a fence. */ struct pipe_fence_handle * -lp_setup_fence( struct setup_context *setup ) +lp_setup_fence( struct lp_setup_context *setup ) { struct lp_scene *scene = lp_setup_get_current_scene(setup); const unsigned rank = lp_scene_get_num_bins( scene ); /* xxx */ @@ -334,10 +334,11 @@ lp_setup_fence( struct setup_context *setup ) void -lp_setup_set_triangle_state( struct setup_context *setup, +lp_setup_set_triangle_state( struct lp_setup_context *setup, unsigned cull_mode, boolean ccw_is_frontface, - boolean scissor ) + boolean scissor, + boolean gl_rasterization_rules) { LP_DBG(DEBUG_SETUP, "%s\n", __FUNCTION__); @@ -345,12 +346,13 @@ lp_setup_set_triangle_state( struct setup_context *setup, setup->cullmode = cull_mode; setup->triangle = first_triangle; setup->scissor_test = scissor; + setup->pixel_offset = gl_rasterization_rules ? 0.5f : 0.0f; } void -lp_setup_set_fs_inputs( struct setup_context *setup, +lp_setup_set_fs_inputs( struct lp_setup_context *setup, const struct lp_shader_input *input, unsigned nr ) { @@ -361,7 +363,7 @@ lp_setup_set_fs_inputs( struct setup_context *setup, } void -lp_setup_set_fs_functions( struct setup_context *setup, +lp_setup_set_fs_functions( struct lp_setup_context *setup, lp_jit_frag_func jit_function0, lp_jit_frag_func jit_function1, boolean opaque ) @@ -376,7 +378,7 @@ lp_setup_set_fs_functions( struct setup_context *setup, } void -lp_setup_set_fs_constants(struct setup_context *setup, +lp_setup_set_fs_constants(struct lp_setup_context *setup, struct pipe_buffer *buffer) { LP_DBG(DEBUG_SETUP, "%s %p\n", __FUNCTION__, (void *) buffer); @@ -388,7 +390,7 @@ lp_setup_set_fs_constants(struct setup_context *setup, void -lp_setup_set_alpha_ref_value( struct setup_context *setup, +lp_setup_set_alpha_ref_value( struct lp_setup_context *setup, float alpha_ref_value ) { LP_DBG(DEBUG_SETUP, "%s %f\n", __FUNCTION__, alpha_ref_value); @@ -400,7 +402,7 @@ lp_setup_set_alpha_ref_value( struct setup_context *setup, } void -lp_setup_set_blend_color( struct setup_context *setup, +lp_setup_set_blend_color( struct lp_setup_context *setup, const struct pipe_blend_color *blend_color ) { LP_DBG(DEBUG_SETUP, "%s\n", __FUNCTION__); @@ -415,7 +417,7 @@ lp_setup_set_blend_color( struct setup_context *setup, void -lp_setup_set_scissor( struct setup_context *setup, +lp_setup_set_scissor( struct lp_setup_context *setup, const struct pipe_scissor_state *scissor ) { LP_DBG(DEBUG_SETUP, "%s\n", __FUNCTION__); @@ -430,7 +432,7 @@ lp_setup_set_scissor( struct setup_context *setup, void -lp_setup_set_flatshade_first( struct setup_context *setup, +lp_setup_set_flatshade_first( struct lp_setup_context *setup, boolean flatshade_first ) { setup->flatshade_first = flatshade_first; @@ -438,7 +440,7 @@ lp_setup_set_flatshade_first( struct setup_context *setup, void -lp_setup_set_vertex_info( struct setup_context *setup, +lp_setup_set_vertex_info( struct lp_setup_context *setup, struct vertex_info *vertex_info ) { /* XXX: just silently holding onto the pointer: @@ -451,7 +453,7 @@ lp_setup_set_vertex_info( struct setup_context *setup, * Called during state validation when LP_NEW_SAMPLER_VIEW is set. */ void -lp_setup_set_fragment_sampler_views(struct setup_context *setup, +lp_setup_set_fragment_sampler_views(struct lp_setup_context *setup, unsigned num, struct pipe_sampler_view **views) { @@ -473,13 +475,13 @@ lp_setup_set_fragment_sampler_views(struct setup_context *setup, jit_tex->height = tex->height0; jit_tex->depth = tex->depth0; jit_tex->last_level = tex->last_level; - jit_tex->stride = lp_tex->stride[0]; if (!lp_tex->dt) { /* regular texture - setup array of mipmap level pointers */ int j; - for (j = 0; j < LP_MAX_TEXTURE_2D_LEVELS; j++) { + for (j = 0; j <= tex->last_level; j++) { jit_tex->data[j] = (ubyte *) lp_tex->data + lp_tex->level_offset[j]; + jit_tex->row_stride[j] = lp_tex->stride[j]; } } else { @@ -492,6 +494,7 @@ lp_setup_set_fragment_sampler_views(struct setup_context *setup, struct sw_winsys *winsys = screen->winsys; jit_tex->data[0] = winsys->displaytarget_map(winsys, lp_tex->dt, PIPE_BUFFER_USAGE_CPU_READ); + jit_tex->row_stride[0] = lp_tex->stride[0]; assert(jit_tex->data[0]); } @@ -513,7 +516,7 @@ lp_setup_set_fragment_sampler_views(struct setup_context *setup, * being rendered and the current scene being built. */ unsigned -lp_setup_is_texture_referenced( const struct setup_context *setup, +lp_setup_is_texture_referenced( const struct lp_setup_context *setup, const struct pipe_texture *texture ) { unsigned i; @@ -542,7 +545,7 @@ lp_setup_is_texture_referenced( const struct setup_context *setup, * Called by vbuf code when we're about to draw something. */ void -lp_setup_update_state( struct setup_context *setup ) +lp_setup_update_state( struct lp_setup_context *setup ) { struct lp_scene *scene = lp_setup_get_current_scene(setup); @@ -660,7 +663,7 @@ lp_setup_update_state( struct setup_context *setup ) /* Only caller is lp_setup_vbuf_destroy() */ void -lp_setup_destroy( struct setup_context *setup ) +lp_setup_destroy( struct lp_setup_context *setup ) { reset_context( setup ); @@ -685,12 +688,12 @@ lp_setup_destroy( struct setup_context *setup ) * the draw module. Currently also creates a rasterizer to use with * it. */ -struct setup_context * +struct lp_setup_context * lp_setup_create( struct pipe_context *pipe, struct draw_context *draw ) { unsigned i; - struct setup_context *setup = CALLOC_STRUCT(setup_context); + struct lp_setup_context *setup = CALLOC_STRUCT(lp_setup_context); if (!setup) return NULL; diff --git a/src/gallium/drivers/llvmpipe/lp_setup.h b/src/gallium/drivers/llvmpipe/lp_setup.h index 72116b8c6c4..414eaec98d1 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup.h +++ b/src/gallium/drivers/llvmpipe/lp_setup.h @@ -61,79 +61,80 @@ struct pipe_framebuffer_state; struct lp_fragment_shader; struct lp_jit_context; -struct setup_context * +struct lp_setup_context * lp_setup_create( struct pipe_context *pipe, struct draw_context *draw ); void -lp_setup_clear(struct setup_context *setup, +lp_setup_clear(struct lp_setup_context *setup, const float *clear_color, double clear_depth, unsigned clear_stencil, unsigned flags); struct pipe_fence_handle * -lp_setup_fence( struct setup_context *setup ); +lp_setup_fence( struct lp_setup_context *setup ); void -lp_setup_flush( struct setup_context *setup, +lp_setup_flush( struct lp_setup_context *setup, unsigned flags ); void -lp_setup_bind_framebuffer( struct setup_context *setup, +lp_setup_bind_framebuffer( struct lp_setup_context *setup, const struct pipe_framebuffer_state *fb ); void -lp_setup_set_triangle_state( struct setup_context *setup, +lp_setup_set_triangle_state( struct lp_setup_context *setup, unsigned cullmode, boolean front_is_ccw, - boolean scissor ); + boolean scissor, + boolean gl_rasterization_rules ); void -lp_setup_set_fs_inputs( struct setup_context *setup, +lp_setup_set_fs_inputs( struct lp_setup_context *setup, const struct lp_shader_input *interp, unsigned nr ); void -lp_setup_set_fs_functions( struct setup_context *setup, +lp_setup_set_fs_functions( struct lp_setup_context *setup, lp_jit_frag_func jit_function0, lp_jit_frag_func jit_function1, boolean opaque ); void -lp_setup_set_fs_constants(struct setup_context *setup, +lp_setup_set_fs_constants(struct lp_setup_context *setup, struct pipe_buffer *buffer); void -lp_setup_set_alpha_ref_value( struct setup_context *setup, +lp_setup_set_alpha_ref_value( struct lp_setup_context *setup, float alpha_ref_value ); void -lp_setup_set_blend_color( struct setup_context *setup, +lp_setup_set_blend_color( struct lp_setup_context *setup, const struct pipe_blend_color *blend_color ); void -lp_setup_set_scissor( struct setup_context *setup, +lp_setup_set_scissor( struct lp_setup_context *setup, const struct pipe_scissor_state *scissor ); void -lp_setup_set_fragment_sampler_views(struct setup_context *setup, +lp_setup_set_fragment_sampler_views(struct lp_setup_context *setup, unsigned num, struct pipe_sampler_view **views); unsigned -lp_setup_is_texture_referenced( const struct setup_context *setup, +lp_setup_is_texture_referenced( const struct lp_setup_context *setup, const struct pipe_texture *texture ); void -lp_setup_set_flatshade_first( struct setup_context *setup, +lp_setup_set_flatshade_first( struct lp_setup_context *setup, boolean flatshade_first ); void -lp_setup_set_vertex_info( struct setup_context *setup, +lp_setup_set_vertex_info( struct lp_setup_context *setup, struct vertex_info *info ); diff --git a/src/gallium/drivers/llvmpipe/lp_setup_context.h b/src/gallium/drivers/llvmpipe/lp_setup_context.h index a5fc34e54a2..464fb369840 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup_context.h +++ b/src/gallium/drivers/llvmpipe/lp_setup_context.h @@ -65,7 +65,7 @@ struct lp_scene_queue; * Subclass of vbuf_render, plugged directly into the draw module as * the rendering backend. */ -struct setup_context +struct lp_setup_context { struct vbuf_render base; @@ -89,6 +89,7 @@ struct setup_context boolean ccw_is_frontface; boolean scissor_test; unsigned cullmode; + float pixel_offset; struct pipe_framebuffer_state fb; @@ -131,29 +132,29 @@ struct setup_context unsigned dirty; /**< bitmask of LP_SETUP_NEW_x bits */ - void (*point)( struct setup_context *, + void (*point)( struct lp_setup_context *, const float (*v0)[4]); - void (*line)( struct setup_context *, + void (*line)( struct lp_setup_context *, const float (*v0)[4], const float (*v1)[4]); - void (*triangle)( struct setup_context *, + void (*triangle)( struct lp_setup_context *, const float (*v0)[4], const float (*v1)[4], const float (*v2)[4]); }; -void lp_setup_choose_triangle( struct setup_context *setup ); -void lp_setup_choose_line( struct setup_context *setup ); -void lp_setup_choose_point( struct setup_context *setup ); +void lp_setup_choose_triangle( struct lp_setup_context *setup ); +void lp_setup_choose_line( struct lp_setup_context *setup ); +void lp_setup_choose_point( struct lp_setup_context *setup ); -struct lp_scene *lp_setup_get_current_scene(struct setup_context *setup); +struct lp_scene *lp_setup_get_current_scene(struct lp_setup_context *setup); -void lp_setup_init_vbuf(struct setup_context *setup); +void lp_setup_init_vbuf(struct lp_setup_context *setup); -void lp_setup_update_state( struct setup_context *setup ); +void lp_setup_update_state( struct lp_setup_context *setup ); -void lp_setup_destroy( struct setup_context *setup ); +void lp_setup_destroy( struct lp_setup_context *setup ); #endif diff --git a/src/gallium/drivers/llvmpipe/lp_setup_line.c b/src/gallium/drivers/llvmpipe/lp_setup_line.c index feea79d3943..be41c44e6f5 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup_line.c +++ b/src/gallium/drivers/llvmpipe/lp_setup_line.c @@ -31,7 +31,7 @@ #include "lp_setup_context.h" -static void line_nop( struct setup_context *setup, +static void line_nop( struct lp_setup_context *setup, const float (*v0)[4], const float (*v1)[4] ) { @@ -39,7 +39,7 @@ static void line_nop( struct setup_context *setup, void -lp_setup_choose_line( struct setup_context *setup ) +lp_setup_choose_line( struct lp_setup_context *setup ) { setup->line = line_nop; } diff --git a/src/gallium/drivers/llvmpipe/lp_setup_point.c b/src/gallium/drivers/llvmpipe/lp_setup_point.c index f03ca729b24..9f69e6c5ce2 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup_point.c +++ b/src/gallium/drivers/llvmpipe/lp_setup_point.c @@ -31,14 +31,14 @@ #include "lp_setup_context.h" -static void point_nop( struct setup_context *setup, +static void point_nop( struct lp_setup_context *setup, const float (*v0)[4] ) { } void -lp_setup_choose_point( struct setup_context *setup ) +lp_setup_choose_point( struct lp_setup_context *setup ) { setup->point = point_nop; } diff --git a/src/gallium/drivers/llvmpipe/lp_setup_tri.c b/src/gallium/drivers/llvmpipe/lp_setup_tri.c index e75412ac9aa..ac6264dc73e 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup_tri.c +++ b/src/gallium/drivers/llvmpipe/lp_setup_tri.c @@ -41,7 +41,8 @@ /** * Compute a0 for a constant-valued coefficient (GL_FLAT shading). */ -static void constant_coef( struct lp_rast_triangle *tri, +static void constant_coef( struct lp_setup_context *setup, + struct lp_rast_triangle *tri, unsigned slot, const float value, unsigned i ) @@ -56,7 +57,8 @@ static void constant_coef( struct lp_rast_triangle *tri, * Compute a0, dadx and dady for a linearly interpolated coefficient, * for a triangle. */ -static void linear_coef( struct lp_rast_triangle *tri, +static void linear_coef( struct lp_setup_context *setup, + struct lp_rast_triangle *tri, float oneoverarea, unsigned slot, const float (*v1)[4], @@ -90,8 +92,8 @@ static void linear_coef( struct lp_rast_triangle *tri, * instead - i'll switch to this later. */ tri->inputs.a0[slot][i] = (a1 - - (dadx * (v1[0][0] - 0.5f) + - dady * (v1[0][1] - 0.5f))); + (dadx * (v1[0][0] - setup->pixel_offset) + + dady * (v1[0][1] - setup->pixel_offset))); } @@ -103,7 +105,8 @@ static void linear_coef( struct lp_rast_triangle *tri, * Later, when we compute the value at a particular fragment position we'll * divide the interpolated value by the interpolated W at that fragment. */ -static void perspective_coef( struct lp_rast_triangle *tri, +static void perspective_coef( struct lp_setup_context *setup, + struct lp_rast_triangle *tri, float oneoverarea, unsigned slot, const float (*v1)[4], @@ -125,8 +128,8 @@ static void perspective_coef( struct lp_rast_triangle *tri, tri->inputs.dadx[slot][i] = dadx; tri->inputs.dady[slot][i] = dady; tri->inputs.a0[slot][i] = (a1 - - (dadx * (v1[0][0] - 0.5f) + - dady * (v1[0][1] - 0.5f))); + (dadx * (v1[0][0] - setup->pixel_offset) + + dady * (v1[0][1] - setup->pixel_offset))); } @@ -137,7 +140,8 @@ static void perspective_coef( struct lp_rast_triangle *tri, * We could do a bit less work if we'd examine gl_FragCoord's swizzle mask. */ static void -setup_fragcoord_coef(struct lp_rast_triangle *tri, +setup_fragcoord_coef(struct lp_setup_context *setup, + struct lp_rast_triangle *tri, float oneoverarea, unsigned slot, const float (*v1)[4], @@ -153,27 +157,28 @@ setup_fragcoord_coef(struct lp_rast_triangle *tri, tri->inputs.dadx[slot][1] = 0.0; tri->inputs.dady[slot][1] = 1.0; /*Z*/ - linear_coef(tri, oneoverarea, slot, v1, v2, v3, 0, 2); + linear_coef(setup, tri, oneoverarea, slot, v1, v2, v3, 0, 2); /*W*/ - linear_coef(tri, oneoverarea, slot, v1, v2, v3, 0, 3); + linear_coef(setup, tri, oneoverarea, slot, v1, v2, v3, 0, 3); } -static void setup_facing_coef( struct lp_rast_triangle *tri, +static void setup_facing_coef( struct lp_setup_context *setup, + struct lp_rast_triangle *tri, unsigned slot, boolean frontface ) { - constant_coef( tri, slot, 1.0f - frontface, 0 ); - constant_coef( tri, slot, 0.0f, 1 ); /* wasted */ - constant_coef( tri, slot, 0.0f, 2 ); /* wasted */ - constant_coef( tri, slot, 0.0f, 3 ); /* wasted */ + constant_coef( setup, tri, slot, 1.0f - frontface, 0 ); + constant_coef( setup, tri, slot, 0.0f, 1 ); /* wasted */ + constant_coef( setup, tri, slot, 0.0f, 2 ); /* wasted */ + constant_coef( setup, tri, slot, 0.0f, 3 ); /* wasted */ } /** * Compute the tri->coef[] array dadx, dady, a0 values. */ -static void setup_tri_coefficients( struct setup_context *setup, +static void setup_tri_coefficients( struct lp_setup_context *setup, struct lp_rast_triangle *tri, float oneoverarea, const float (*v1)[4], @@ -185,7 +190,7 @@ static void setup_tri_coefficients( struct setup_context *setup, /* The internal position input is in slot zero: */ - setup_fragcoord_coef(tri, oneoverarea, 0, v1, v2, v3); + setup_fragcoord_coef(setup, tri, oneoverarea, 0, v1, v2, v3); /* setup interpolation for all the remaining attributes: */ @@ -196,27 +201,27 @@ static void setup_tri_coefficients( struct setup_context *setup, switch (setup->fs.input[slot].interp) { case LP_INTERP_CONSTANT: for (i = 0; i < NUM_CHANNELS; i++) - constant_coef(tri, slot+1, v3[vert_attr][i], i); + constant_coef(setup, tri, slot+1, v3[vert_attr][i], i); break; case LP_INTERP_LINEAR: for (i = 0; i < NUM_CHANNELS; i++) - linear_coef(tri, oneoverarea, slot+1, v1, v2, v3, vert_attr, i); + linear_coef(setup, tri, oneoverarea, slot+1, v1, v2, v3, vert_attr, i); break; case LP_INTERP_PERSPECTIVE: for (i = 0; i < NUM_CHANNELS; i++) - perspective_coef(tri, oneoverarea, slot+1, v1, v2, v3, vert_attr, i); + perspective_coef(setup, tri, oneoverarea, slot+1, v1, v2, v3, vert_attr, i); break; case LP_INTERP_POSITION: /* XXX: fix me - duplicates the values in slot zero. */ - setup_fragcoord_coef(tri, oneoverarea, slot+1, v1, v2, v3); + setup_fragcoord_coef(setup, tri, oneoverarea, slot+1, v1, v2, v3); break; case LP_INTERP_FACING: - setup_facing_coef(tri, slot+1, frontface); + setup_facing_coef(setup, tri, slot+1, frontface); break; default: @@ -274,19 +279,19 @@ alloc_triangle(struct lp_scene *scene, unsigned nr_inputs, unsigned *tri_size) * bins for the tiles which we overlap. */ static void -do_triangle_ccw(struct setup_context *setup, +do_triangle_ccw(struct lp_setup_context *setup, const float (*v1)[4], const float (*v2)[4], const float (*v3)[4], boolean frontfacing ) { /* x/y positions in fixed point */ - const int x1 = subpixel_snap(v1[0][0]); - const int x2 = subpixel_snap(v2[0][0]); - const int x3 = subpixel_snap(v3[0][0]); - const int y1 = subpixel_snap(v1[0][1]); - const int y2 = subpixel_snap(v2[0][1]); - const int y3 = subpixel_snap(v3[0][1]); + const int x1 = subpixel_snap(v1[0][0] + 0.5 - setup->pixel_offset); + const int x2 = subpixel_snap(v2[0][0] + 0.5 - setup->pixel_offset); + const int x3 = subpixel_snap(v3[0][0] + 0.5 - setup->pixel_offset); + const int y1 = subpixel_snap(v1[0][1] + 0.5 - setup->pixel_offset); + const int y2 = subpixel_snap(v2[0][1] + 0.5 - setup->pixel_offset); + const int y3 = subpixel_snap(v3[0][1] + 0.5 - setup->pixel_offset); struct lp_scene *scene = lp_setup_get_current_scene(setup); struct lp_rast_triangle *tri; @@ -565,7 +570,7 @@ do_triangle_ccw(struct setup_context *setup, } -static void triangle_cw( struct setup_context *setup, +static void triangle_cw( struct lp_setup_context *setup, const float (*v0)[4], const float (*v1)[4], const float (*v2)[4] ) @@ -574,7 +579,7 @@ static void triangle_cw( struct setup_context *setup, } -static void triangle_ccw( struct setup_context *setup, +static void triangle_ccw( struct lp_setup_context *setup, const float (*v0)[4], const float (*v1)[4], const float (*v2)[4] ) @@ -583,7 +588,7 @@ static void triangle_ccw( struct setup_context *setup, } -static void triangle_both( struct setup_context *setup, +static void triangle_both( struct lp_setup_context *setup, const float (*v0)[4], const float (*v1)[4], const float (*v2)[4] ) @@ -602,7 +607,7 @@ static void triangle_both( struct setup_context *setup, } -static void triangle_nop( struct setup_context *setup, +static void triangle_nop( struct lp_setup_context *setup, const float (*v0)[4], const float (*v1)[4], const float (*v2)[4] ) @@ -611,7 +616,7 @@ static void triangle_nop( struct setup_context *setup, void -lp_setup_choose_triangle( struct setup_context *setup ) +lp_setup_choose_triangle( struct lp_setup_context *setup ) { switch (setup->cullmode) { case PIPE_WINDING_NONE: diff --git a/src/gallium/drivers/llvmpipe/lp_setup_vbuf.c b/src/gallium/drivers/llvmpipe/lp_setup_vbuf.c index 671e74465c0..d7336d82b21 100644 --- a/src/gallium/drivers/llvmpipe/lp_setup_vbuf.c +++ b/src/gallium/drivers/llvmpipe/lp_setup_vbuf.c @@ -48,10 +48,10 @@ /** cast wrapper */ -static struct setup_context * -setup_context(struct vbuf_render *vbr) +static struct lp_setup_context * +lp_setup_context(struct vbuf_render *vbr) { - return (struct setup_context *) vbr; + return (struct lp_setup_context *) vbr; } @@ -59,7 +59,7 @@ setup_context(struct vbuf_render *vbr) static const struct vertex_info * lp_setup_get_vertex_info(struct vbuf_render *vbr) { - struct setup_context *setup = setup_context(vbr); + struct lp_setup_context *setup = lp_setup_context(vbr); return setup->vertex_info; } @@ -68,7 +68,7 @@ static boolean lp_setup_allocate_vertices(struct vbuf_render *vbr, ushort vertex_size, ushort nr_vertices) { - struct setup_context *setup = setup_context(vbr); + struct lp_setup_context *setup = lp_setup_context(vbr); unsigned size = vertex_size * nr_vertices; if (setup->vertex_buffer_size < size) { @@ -92,7 +92,7 @@ lp_setup_release_vertices(struct vbuf_render *vbr) static void * lp_setup_map_vertices(struct vbuf_render *vbr) { - struct setup_context *setup = setup_context(vbr); + struct lp_setup_context *setup = lp_setup_context(vbr); return setup->vertex_buffer; } @@ -101,7 +101,7 @@ lp_setup_unmap_vertices(struct vbuf_render *vbr, ushort min_index, ushort max_index ) { - struct setup_context *setup = setup_context(vbr); + struct lp_setup_context *setup = lp_setup_context(vbr); assert( setup->vertex_buffer_size >= (max_index+1) * setup->vertex_size ); /* do nothing */ } @@ -110,7 +110,7 @@ lp_setup_unmap_vertices(struct vbuf_render *vbr, static boolean lp_setup_set_primitive(struct vbuf_render *vbr, unsigned prim) { - setup_context(vbr)->prim = prim; + lp_setup_context(vbr)->prim = prim; return TRUE; } @@ -129,7 +129,7 @@ static INLINE const_float4_ptr get_vert( const void *vertex_buffer, static void lp_setup_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) { - struct setup_context *setup = setup_context(vbr); + struct lp_setup_context *setup = lp_setup_context(vbr); const unsigned stride = setup->vertex_info->size * sizeof(float); const void *vertex_buffer = setup->vertex_buffer; unsigned i; @@ -284,7 +284,7 @@ lp_setup_draw(struct vbuf_render *vbr, const ushort *indices, uint nr) static void lp_setup_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) { - struct setup_context *setup = setup_context(vbr); + struct lp_setup_context *setup = lp_setup_context(vbr); const unsigned stride = setup->vertex_info->size * sizeof(float); const void *vertex_buffer = (void *) get_vert(setup->vertex_buffer, start, stride); @@ -436,7 +436,7 @@ lp_setup_draw_arrays(struct vbuf_render *vbr, uint start, uint nr) static void lp_setup_vbuf_destroy(struct vbuf_render *vbr) { - lp_setup_destroy(setup_context(vbr)); + lp_setup_destroy(lp_setup_context(vbr)); } @@ -444,7 +444,7 @@ lp_setup_vbuf_destroy(struct vbuf_render *vbr) * Create the post-transform vertex handler for the given context. */ void -lp_setup_init_vbuf(struct setup_context *setup) +lp_setup_init_vbuf(struct lp_setup_context *setup) { setup->base.max_indices = LP_MAX_VBUF_INDEXES; setup->base.max_vertex_buffer_bytes = LP_MAX_VBUF_SIZE; diff --git a/src/gallium/drivers/llvmpipe/lp_state.h b/src/gallium/drivers/llvmpipe/lp_state.h index 9b87cd202e7..13cd934b13f 100644 --- a/src/gallium/drivers/llvmpipe/lp_state.h +++ b/src/gallium/drivers/llvmpipe/lp_state.h @@ -31,7 +31,7 @@ #ifndef LP_STATE_H #define LP_STATE_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include "pipe/p_state.h" #include "tgsi/tgsi_scan.h" diff --git a/src/gallium/drivers/llvmpipe/lp_state_fs.c b/src/gallium/drivers/llvmpipe/lp_state_fs.c index 3e681b561aa..b75ca282067 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_fs.c +++ b/src/gallium/drivers/llvmpipe/lp_state_fs.c @@ -40,7 +40,7 @@ * - depth/stencil test (stencil TBI) * - blending * - * This file has only the glue to assembly the fragment pipeline. The actual + * This file has only the glue to assemble the fragment pipeline. The actual * plumbing of converting Gallium state into LLVM IR is done elsewhere, in the * lp_bld_*.[ch] files, and in a complete generic and reusable way. Here we * muster the LLVM JIT execution engine to create a function that follows an @@ -95,6 +95,9 @@ #include "lp_tex_sample.h" +#include <llvm-c/Analysis.h> + + static const unsigned char quad_offset_x[4] = {0, 1, 0, 1}; static const unsigned char quad_offset_y[4] = {0, 0, 1, 1}; diff --git a/src/gallium/drivers/llvmpipe/lp_state_rasterizer.c b/src/gallium/drivers/llvmpipe/lp_state_rasterizer.c index feb012816c9..6df3ef25b0e 100644 --- a/src/gallium/drivers/llvmpipe/lp_state_rasterizer.c +++ b/src/gallium/drivers/llvmpipe/lp_state_rasterizer.c @@ -62,7 +62,8 @@ void llvmpipe_bind_rasterizer_state(struct pipe_context *pipe, lp_setup_set_triangle_state( llvmpipe->setup, llvmpipe->rasterizer->cull_mode, llvmpipe->rasterizer->front_winding == PIPE_WINDING_CCW, - llvmpipe->rasterizer->scissor); + llvmpipe->rasterizer->scissor, + llvmpipe->rasterizer->gl_rasterization_rules); } llvmpipe->dirty |= LP_NEW_RASTERIZER; diff --git a/src/gallium/drivers/llvmpipe/lp_test.h b/src/gallium/drivers/llvmpipe/lp_test.h index a9b99945f92..1df98978988 100644 --- a/src/gallium/drivers/llvmpipe/lp_test.h +++ b/src/gallium/drivers/llvmpipe/lp_test.h @@ -41,7 +41,7 @@ #include <stdio.h> #include <float.h> -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include <llvm-c/Analysis.h> #include <llvm-c/ExecutionEngine.h> #include <llvm-c/Target.h> diff --git a/src/gallium/drivers/llvmpipe/lp_test_format.c b/src/gallium/drivers/llvmpipe/lp_test_format.c index d05157991bb..2c4d7fb6e14 100644 --- a/src/gallium/drivers/llvmpipe/lp_test_format.c +++ b/src/gallium/drivers/llvmpipe/lp_test_format.c @@ -29,7 +29,7 @@ #include <stdlib.h> #include <stdio.h> -#include <llvm-c/Core.h> +#include "os/os_llvm.h" #include <llvm-c/Analysis.h> #include <llvm-c/ExecutionEngine.h> #include <llvm-c/Target.h> diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample.h b/src/gallium/drivers/llvmpipe/lp_tex_sample.h index cb59a94464a..799df182b6a 100644 --- a/src/gallium/drivers/llvmpipe/lp_tex_sample.h +++ b/src/gallium/drivers/llvmpipe/lp_tex_sample.h @@ -29,7 +29,7 @@ #define LP_TEX_SAMPLE_H -#include <llvm-c/Core.h> +#include "os/os_llvm.h" struct lp_sampler_static_state; diff --git a/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c b/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c index 5a3cf37d6d8..662508af61a 100644 --- a/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c +++ b/src/gallium/drivers/llvmpipe/lp_tex_sample_llvm.c @@ -147,7 +147,7 @@ LP_LLVM_TEXTURE_MEMBER(width, LP_JIT_TEXTURE_WIDTH, TRUE) LP_LLVM_TEXTURE_MEMBER(height, LP_JIT_TEXTURE_HEIGHT, TRUE) LP_LLVM_TEXTURE_MEMBER(depth, LP_JIT_TEXTURE_DEPTH, TRUE) LP_LLVM_TEXTURE_MEMBER(last_level, LP_JIT_TEXTURE_LAST_LEVEL, TRUE) -LP_LLVM_TEXTURE_MEMBER(stride, LP_JIT_TEXTURE_STRIDE, TRUE) +LP_LLVM_TEXTURE_MEMBER(row_stride, LP_JIT_TEXTURE_ROW_STRIDE, FALSE) LP_LLVM_TEXTURE_MEMBER(data_ptr, LP_JIT_TEXTURE_DATA, FALSE) @@ -204,7 +204,7 @@ lp_llvm_sampler_soa_create(const struct lp_sampler_static_state *static_state, sampler->dynamic_state.base.height = lp_llvm_texture_height; sampler->dynamic_state.base.depth = lp_llvm_texture_depth; sampler->dynamic_state.base.last_level = lp_llvm_texture_last_level; - sampler->dynamic_state.base.stride = lp_llvm_texture_stride; + sampler->dynamic_state.base.row_stride = lp_llvm_texture_row_stride; sampler->dynamic_state.base.data_ptr = lp_llvm_texture_data_ptr; sampler->dynamic_state.static_state = static_state; sampler->dynamic_state.context_ptr = context_ptr; diff --git a/src/gallium/drivers/llvmpipe/lp_texture.c b/src/gallium/drivers/llvmpipe/lp_texture.c index 74b7b4ec5e1..f2c6dbd088a 100644 --- a/src/gallium/drivers/llvmpipe/lp_texture.c +++ b/src/gallium/drivers/llvmpipe/lp_texture.c @@ -243,7 +243,7 @@ llvmpipe_tex_surface_destroy(struct pipe_surface *surf) static struct pipe_transfer * -llvmpipe_get_tex_transfer(struct pipe_screen *screen, +llvmpipe_get_tex_transfer(struct pipe_context *pipe, struct pipe_texture *texture, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, @@ -294,7 +294,8 @@ llvmpipe_get_tex_transfer(struct pipe_screen *screen, static void -llvmpipe_tex_transfer_destroy(struct pipe_transfer *transfer) +llvmpipe_tex_transfer_destroy(struct pipe_context *pipe, + struct pipe_transfer *transfer) { /* Effectively do the texture_update work here - if texture images * needed post-processing to put them into hardware layout, this is @@ -307,10 +308,10 @@ llvmpipe_tex_transfer_destroy(struct pipe_transfer *transfer) static void * -llvmpipe_transfer_map( struct pipe_screen *_screen, +llvmpipe_transfer_map( struct pipe_context *pipe, struct pipe_transfer *transfer ) { - struct llvmpipe_screen *screen = llvmpipe_screen(_screen); + struct llvmpipe_screen *screen = llvmpipe_screen(pipe->screen); ubyte *map, *xfer_map; struct llvmpipe_texture *lpt; enum pipe_format format; @@ -351,10 +352,10 @@ llvmpipe_transfer_map( struct pipe_screen *_screen, static void -llvmpipe_transfer_unmap(struct pipe_screen *screen, - struct pipe_transfer *transfer) +llvmpipe_transfer_unmap(struct pipe_context *pipe, + struct pipe_transfer *transfer) { - struct llvmpipe_screen *lp_screen = llvmpipe_screen(screen); + struct llvmpipe_screen *lp_screen = llvmpipe_screen(pipe->screen); struct llvmpipe_texture *lpt; assert(transfer->texture); @@ -376,9 +377,14 @@ llvmpipe_init_screen_texture_funcs(struct pipe_screen *screen) screen->get_tex_surface = llvmpipe_get_tex_surface; screen->tex_surface_destroy = llvmpipe_tex_surface_destroy; +} + - screen->get_tex_transfer = llvmpipe_get_tex_transfer; - screen->tex_transfer_destroy = llvmpipe_tex_transfer_destroy; - screen->transfer_map = llvmpipe_transfer_map; - screen->transfer_unmap = llvmpipe_transfer_unmap; +void +llvmpipe_init_context_texture_funcs(struct pipe_context *pipe) +{ + pipe->get_tex_transfer = llvmpipe_get_tex_transfer; + pipe->tex_transfer_destroy = llvmpipe_tex_transfer_destroy; + pipe->transfer_map = llvmpipe_transfer_map; + pipe->transfer_unmap = llvmpipe_transfer_unmap; } diff --git a/src/gallium/drivers/llvmpipe/lp_texture.h b/src/gallium/drivers/llvmpipe/lp_texture.h index b23f929b167..94b667abf31 100644 --- a/src/gallium/drivers/llvmpipe/lp_texture.h +++ b/src/gallium/drivers/llvmpipe/lp_texture.h @@ -98,5 +98,7 @@ llvmpipe_transfer(struct pipe_transfer *pt) extern void llvmpipe_init_screen_texture_funcs(struct pipe_screen *screen); +extern void +llvmpipe_init_context_texture_funcs(struct pipe_context *pipe); #endif /* LP_TEXTURE_H */ diff --git a/src/gallium/drivers/nouveau/nv04_surface_2d.c b/src/gallium/drivers/nouveau/nv04_surface_2d.c index b074547c4da..93114465d5e 100644 --- a/src/gallium/drivers/nouveau/nv04_surface_2d.c +++ b/src/gallium/drivers/nouveau/nv04_surface_2d.c @@ -518,7 +518,6 @@ nv04_surface_wrap_for_render(struct pipe_screen *pscreen, struct nv04_surface_2d ns->base.usage = PIPE_BUFFER_USAGE_GPU_WRITE | NOUVEAU_BUFFER_USAGE_NO_RENDER | PIPE_BUFFER_USAGE_GPU_READ; } - struct nv40_screen* screen = (struct nv40_screen*)pscreen; ns->base.usage = PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE; struct pipe_texture templ; diff --git a/src/gallium/drivers/nv30/nv30_context.c b/src/gallium/drivers/nv30/nv30_context.c index 279b74445ca..825c167b01b 100644 --- a/src/gallium/drivers/nv30/nv30_context.c +++ b/src/gallium/drivers/nv30/nv30_context.c @@ -74,6 +74,7 @@ nv30_create(struct pipe_screen *pscreen, void *priv) nv30_init_query_functions(nv30); nv30_init_surface_functions(nv30); nv30_init_state_functions(nv30); + nv30_init_transfer_functions(nv30); /* Create, configure, and install fallback swtnl path */ nv30->draw = draw_create(); diff --git a/src/gallium/drivers/nv30/nv30_context.h b/src/gallium/drivers/nv30/nv30_context.h index 34b5953ccf1..38a17a44552 100644 --- a/src/gallium/drivers/nv30/nv30_context.h +++ b/src/gallium/drivers/nv30/nv30_context.h @@ -168,6 +168,7 @@ struct nv30_state_entry { extern void nv30_init_state_functions(struct nv30_context *nv30); extern void nv30_init_surface_functions(struct nv30_context *nv30); extern void nv30_init_query_functions(struct nv30_context *nv30); +extern void nv30_init_transfer_functions(struct nv30_context *nv30); extern void nv30_screen_init_miptree_functions(struct pipe_screen *pscreen); diff --git a/src/gallium/drivers/nv30/nv30_screen.c b/src/gallium/drivers/nv30/nv30_screen.c index 85433d20953..db24335b7c1 100644 --- a/src/gallium/drivers/nv30/nv30_screen.c +++ b/src/gallium/drivers/nv30/nv30_screen.c @@ -214,7 +214,6 @@ nv30_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) pscreen->context_create = nv30_create; nv30_screen_init_miptree_functions(pscreen); - nv30_screen_init_transfer_functions(pscreen); /* 3D object */ switch (dev->chipset & 0xf0) { diff --git a/src/gallium/drivers/nv30/nv30_screen.h b/src/gallium/drivers/nv30/nv30_screen.h index 8591cd31cab..b7856cdf005 100644 --- a/src/gallium/drivers/nv30/nv30_screen.h +++ b/src/gallium/drivers/nv30/nv30_screen.h @@ -35,7 +35,4 @@ nv30_screen(struct pipe_screen *screen) return (struct nv30_screen *)screen; } -void -nv30_screen_init_transfer_functions(struct pipe_screen *pscreen); - #endif diff --git a/src/gallium/drivers/nv30/nv30_transfer.c b/src/gallium/drivers/nv30/nv30_transfer.c index 3aeda51ea19..cfc109bb740 100644 --- a/src/gallium/drivers/nv30/nv30_transfer.c +++ b/src/gallium/drivers/nv30/nv30_transfer.c @@ -33,11 +33,12 @@ nv30_compatible_transfer_tex(struct pipe_texture *pt, unsigned width, unsigned h } static struct pipe_transfer * -nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, +nv30_transfer_new(struct pipe_context *pcontext, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, unsigned x, unsigned y, unsigned w, unsigned h) { + struct pipe_screen *pscreen = pcontext->screen; struct nv30_miptree *mt = (struct nv30_miptree *)pt; struct nv30_transfer *tx; struct pipe_texture tx_tex_template, *tx_tex; @@ -117,12 +118,13 @@ nv30_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, } static void -nv30_transfer_del(struct pipe_transfer *ptx) +nv30_transfer_del(struct pipe_context *pcontext, + struct pipe_transfer *ptx) { struct nv30_transfer *tx = (struct nv30_transfer *)ptx; if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) { - struct pipe_screen *pscreen = ptx->texture->screen; + struct pipe_screen *pscreen = pcontext->screen; struct nv30_screen *nvscreen = nv30_screen(pscreen); struct pipe_surface *dst; @@ -145,8 +147,9 @@ nv30_transfer_del(struct pipe_transfer *ptx) } static void * -nv30_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) +nv30_transfer_map(struct pipe_context *pcontext, struct pipe_transfer *ptx) { + struct pipe_screen *pscreen = pcontext->screen; struct nv30_transfer *tx = (struct nv30_transfer *)ptx; struct nv04_surface *ns = (struct nv04_surface *)tx->surface; struct nv30_miptree *mt = (struct nv30_miptree *)tx->surface->texture; @@ -160,8 +163,9 @@ nv30_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) } static void -nv30_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx) +nv30_transfer_unmap(struct pipe_context *pcontext, struct pipe_transfer *ptx) { + struct pipe_screen *pscreen = pcontext->screen; struct nv30_transfer *tx = (struct nv30_transfer *)ptx; struct nv30_miptree *mt = (struct nv30_miptree *)tx->surface->texture; @@ -169,10 +173,10 @@ nv30_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx) } void -nv30_screen_init_transfer_functions(struct pipe_screen *pscreen) +nv30_init_transfer_functions(struct nv30_context *nv30) { - pscreen->get_tex_transfer = nv30_transfer_new; - pscreen->tex_transfer_destroy = nv30_transfer_del; - pscreen->transfer_map = nv30_transfer_map; - pscreen->transfer_unmap = nv30_transfer_unmap; + nv30->pipe.get_tex_transfer = nv30_transfer_new; + nv30->pipe.tex_transfer_destroy = nv30_transfer_del; + nv30->pipe.transfer_map = nv30_transfer_map; + nv30->pipe.transfer_unmap = nv30_transfer_unmap; } diff --git a/src/gallium/drivers/nv40/nv40_context.c b/src/gallium/drivers/nv40/nv40_context.c index 65dc73e88b3..e828f17643b 100644 --- a/src/gallium/drivers/nv40/nv40_context.c +++ b/src/gallium/drivers/nv40/nv40_context.c @@ -74,6 +74,7 @@ nv40_create(struct pipe_screen *pscreen, void *priv) nv40_init_query_functions(nv40); nv40_init_surface_functions(nv40); nv40_init_state_functions(nv40); + nv40_init_transfer_functions(nv40); /* Create, configure, and install fallback swtnl path */ nv40->draw = draw_create(); diff --git a/src/gallium/drivers/nv40/nv40_context.h b/src/gallium/drivers/nv40/nv40_context.h index 4d2ffd9772d..1e1d64ee498 100644 --- a/src/gallium/drivers/nv40/nv40_context.h +++ b/src/gallium/drivers/nv40/nv40_context.h @@ -184,6 +184,7 @@ struct nv40_state_entry { extern void nv40_init_state_functions(struct nv40_context *nv40); extern void nv40_init_surface_functions(struct nv40_context *nv40); extern void nv40_init_query_functions(struct nv40_context *nv40); +extern void nv40_init_transfer_functions(struct nv40_context *nv40); extern void nv40_screen_init_miptree_functions(struct pipe_screen *pscreen); diff --git a/src/gallium/drivers/nv40/nv40_screen.c b/src/gallium/drivers/nv40/nv40_screen.c index b216c5e38c9..dbcc33d8d9e 100644 --- a/src/gallium/drivers/nv40/nv40_screen.c +++ b/src/gallium/drivers/nv40/nv40_screen.c @@ -201,7 +201,6 @@ nv40_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) pscreen->context_create = nv40_create; nv40_screen_init_miptree_functions(pscreen); - nv40_screen_init_transfer_functions(pscreen); /* 3D object */ switch (dev->chipset & 0xf0) { diff --git a/src/gallium/drivers/nv40/nv40_screen.h b/src/gallium/drivers/nv40/nv40_screen.h index 9437aa050d4..2765ab764ae 100644 --- a/src/gallium/drivers/nv40/nv40_screen.h +++ b/src/gallium/drivers/nv40/nv40_screen.h @@ -34,7 +34,4 @@ nv40_screen(struct pipe_screen *screen) return (struct nv40_screen *)screen; } -void -nv40_screen_init_transfer_functions(struct pipe_screen *pscreen); - #endif diff --git a/src/gallium/drivers/nv40/nv40_transfer.c b/src/gallium/drivers/nv40/nv40_transfer.c index 0462a042c38..c552a681138 100644 --- a/src/gallium/drivers/nv40/nv40_transfer.c +++ b/src/gallium/drivers/nv40/nv40_transfer.c @@ -33,11 +33,12 @@ nv40_compatible_transfer_tex(struct pipe_texture *pt, unsigned width, unsigned h } static struct pipe_transfer * -nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, +nv40_transfer_new(struct pipe_context *pcontext, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, unsigned x, unsigned y, unsigned w, unsigned h) { + struct pipe_screen *pscreen = pcontext->screen; struct nv40_miptree *mt = (struct nv40_miptree *)pt; struct nv40_transfer *tx; struct pipe_texture tx_tex_template, *tx_tex; @@ -117,12 +118,12 @@ nv40_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, } static void -nv40_transfer_del(struct pipe_transfer *ptx) +nv40_transfer_del(struct pipe_context *pcontext, struct pipe_transfer *ptx) { struct nv40_transfer *tx = (struct nv40_transfer *)ptx; if (!tx->direct && (ptx->usage & PIPE_TRANSFER_WRITE)) { - struct pipe_screen *pscreen = ptx->texture->screen; + struct pipe_screen *pscreen = pcontext->screen; struct nv40_screen *nvscreen = nv40_screen(pscreen); struct pipe_surface *dst; @@ -145,8 +146,9 @@ nv40_transfer_del(struct pipe_transfer *ptx) } static void * -nv40_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) +nv40_transfer_map(struct pipe_context *pcontext, struct pipe_transfer *ptx) { + struct pipe_screen *pscreen = pcontext->screen; struct nv40_transfer *tx = (struct nv40_transfer *)ptx; struct nv04_surface *ns = (struct nv04_surface *)tx->surface; struct nv40_miptree *mt = (struct nv40_miptree *)tx->surface->texture; @@ -160,8 +162,9 @@ nv40_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) } static void -nv40_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx) +nv40_transfer_unmap(struct pipe_context *pcontext, struct pipe_transfer *ptx) { + struct pipe_screen *pscreen = pcontext->screen; struct nv40_transfer *tx = (struct nv40_transfer *)ptx; struct nv40_miptree *mt = (struct nv40_miptree *)tx->surface->texture; @@ -169,10 +172,10 @@ nv40_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx) } void -nv40_screen_init_transfer_functions(struct pipe_screen *pscreen) +nv40_init_transfer_functions(struct nv40_context *nv40) { - pscreen->get_tex_transfer = nv40_transfer_new; - pscreen->tex_transfer_destroy = nv40_transfer_del; - pscreen->transfer_map = nv40_transfer_map; - pscreen->transfer_unmap = nv40_transfer_unmap; + nv40->pipe.get_tex_transfer = nv40_transfer_new; + nv40->pipe.tex_transfer_destroy = nv40_transfer_del; + nv40->pipe.transfer_map = nv40_transfer_map; + nv40->pipe.transfer_unmap = nv40_transfer_unmap; } diff --git a/src/gallium/drivers/nv40/nv40_vertprog.c b/src/gallium/drivers/nv40/nv40_vertprog.c index b289eef0fc2..c93c5d127c4 100644 --- a/src/gallium/drivers/nv40/nv40_vertprog.c +++ b/src/gallium/drivers/nv40/nv40_vertprog.c @@ -742,7 +742,7 @@ nv40_vertprog_translate(struct nv40_context *nv40, } /* Redirect post-transform vertex position to a temp if user clip - * planes are enabled. We need to append code the the vtxprog + * planes are enabled. We need to append code to the vtxprog * to handle clip planes later. */ if (vp->ucp.nr) { diff --git a/src/gallium/drivers/nv50/nv50_context.c b/src/gallium/drivers/nv50/nv50_context.c index 0eb42f323ff..aa14e17872d 100644 --- a/src/gallium/drivers/nv50/nv50_context.c +++ b/src/gallium/drivers/nv50/nv50_context.c @@ -97,6 +97,7 @@ nv50_create(struct pipe_screen *pscreen, void *priv) nv50_init_surface_functions(nv50); nv50_init_state_functions(nv50); nv50_init_query_functions(nv50); + nv50_init_transfer_functions(nv50); nv50->draw = draw_create(); assert(nv50->draw); diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h index f35bc411feb..bc7831d9aca 100644 --- a/src/gallium/drivers/nv50/nv50_context.h +++ b/src/gallium/drivers/nv50/nv50_context.h @@ -186,6 +186,7 @@ nv50_context(struct pipe_context *pipe) extern void nv50_init_surface_functions(struct nv50_context *nv50); extern void nv50_init_state_functions(struct nv50_context *nv50); extern void nv50_init_query_functions(struct nv50_context *nv50); +extern void nv50_init_transfer_functions(struct nv50_context *nv50); extern void nv50_screen_init_miptree_functions(struct pipe_screen *pscreen); diff --git a/src/gallium/drivers/nv50/nv50_screen.c b/src/gallium/drivers/nv50/nv50_screen.c index 8283524b638..d7f5863fb71 100644 --- a/src/gallium/drivers/nv50/nv50_screen.c +++ b/src/gallium/drivers/nv50/nv50_screen.c @@ -234,7 +234,6 @@ nv50_screen_create(struct pipe_winsys *ws, struct nouveau_device *dev) pscreen->context_create = nv50_create; nv50_screen_init_miptree_functions(pscreen); - nv50_transfer_init_screen_functions(pscreen); /* DMA engine object */ ret = nouveau_grobj_alloc(chan, 0xbeef5039, diff --git a/src/gallium/drivers/nv50/nv50_screen.h b/src/gallium/drivers/nv50/nv50_screen.h index d1bc80cb9ed..ec19ea655b1 100644 --- a/src/gallium/drivers/nv50/nv50_screen.h +++ b/src/gallium/drivers/nv50/nv50_screen.h @@ -38,6 +38,4 @@ nv50_screen(struct pipe_screen *screen) return (struct nv50_screen *)screen; } -void nv50_transfer_init_screen_functions(struct pipe_screen *); - #endif diff --git a/src/gallium/drivers/nv50/nv50_transfer.c b/src/gallium/drivers/nv50/nv50_transfer.c index 7c360e9e73a..9eb223eca65 100644 --- a/src/gallium/drivers/nv50/nv50_transfer.c +++ b/src/gallium/drivers/nv50/nv50_transfer.c @@ -121,11 +121,12 @@ nv50_transfer_rect_m2mf(struct pipe_screen *pscreen, } static struct pipe_transfer * -nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, +nv50_transfer_new(struct pipe_context *pcontext, struct pipe_texture *pt, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, unsigned x, unsigned y, unsigned w, unsigned h) { + struct pipe_screen *pscreen = pcontext->screen; struct nouveau_device *dev = nouveau_screen(pscreen)->device; struct nv50_miptree *mt = nv50_miptree(pt); struct nv50_miptree_level *lvl = &mt->level[level]; @@ -186,7 +187,7 @@ nv50_transfer_new(struct pipe_screen *pscreen, struct pipe_texture *pt, } static void -nv50_transfer_del(struct pipe_transfer *ptx) +nv50_transfer_del(struct pipe_context *pcontext, struct pipe_transfer *ptx) { struct nv50_transfer *tx = (struct nv50_transfer *)ptx; struct nv50_miptree *mt = nv50_miptree(ptx->texture); @@ -196,7 +197,7 @@ nv50_transfer_del(struct pipe_transfer *ptx) unsigned ny = util_format_get_nblocksy(pt->format, tx->base.height); if (ptx->usage & PIPE_TRANSFER_WRITE) { - struct pipe_screen *pscreen = pt->screen; + struct pipe_screen *pscreen = pcontext->screen; nv50_transfer_rect_m2mf(pscreen, tx->bo, 0, tx->base.stride, tx->bo->tile_mode, @@ -218,7 +219,7 @@ nv50_transfer_del(struct pipe_transfer *ptx) } static void * -nv50_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) +nv50_transfer_map(struct pipe_context *pcontext, struct pipe_transfer *ptx) { struct nv50_transfer *tx = (struct nv50_transfer *)ptx; unsigned flags = 0; @@ -236,7 +237,7 @@ nv50_transfer_map(struct pipe_screen *pscreen, struct pipe_transfer *ptx) } static void -nv50_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx) +nv50_transfer_unmap(struct pipe_context *pcontext, struct pipe_transfer *ptx) { struct nv50_transfer *tx = (struct nv50_transfer *)ptx; @@ -244,12 +245,12 @@ nv50_transfer_unmap(struct pipe_screen *pscreen, struct pipe_transfer *ptx) } void -nv50_transfer_init_screen_functions(struct pipe_screen *pscreen) +nv50_init_transfer_functions(struct nv50_context *nv50) { - pscreen->get_tex_transfer = nv50_transfer_new; - pscreen->tex_transfer_destroy = nv50_transfer_del; - pscreen->transfer_map = nv50_transfer_map; - pscreen->transfer_unmap = nv50_transfer_unmap; + nv50->pipe.get_tex_transfer = nv50_transfer_new; + nv50->pipe.tex_transfer_destroy = nv50_transfer_del; + nv50->pipe.transfer_map = nv50_transfer_map; + nv50->pipe.transfer_unmap = nv50_transfer_unmap; } void diff --git a/src/gallium/drivers/r300/r300_context.c b/src/gallium/drivers/r300/r300_context.c index 923e1e541ff..8606c0004e5 100644 --- a/src/gallium/drivers/r300/r300_context.c +++ b/src/gallium/drivers/r300/r300_context.c @@ -34,6 +34,7 @@ #include "r300_screen.h" #include "r300_state_invariant.h" #include "r300_texture.h" +#include "r300_transfer.h" #include "radeon_winsys.h" @@ -209,6 +210,8 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen, r300_init_query_functions(r300); + r300_init_transfer_functions(r300); + /* r300_init_surface_functions(r300); */ r300_init_state_functions(r300); diff --git a/src/gallium/drivers/r300/r300_context.h b/src/gallium/drivers/r300/r300_context.h index 3b70bcda827..7b86669888b 100644 --- a/src/gallium/drivers/r300/r300_context.h +++ b/src/gallium/drivers/r300/r300_context.h @@ -405,6 +405,7 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen, struct draw_stage* r300_draw_stage(struct r300_context* r300); void r300_init_state_functions(struct r300_context* r300); void r300_init_surface_functions(struct r300_context* r300); +void r300_init_tex_functions( struct pipe_context *pipe ); static INLINE boolean CTX_DBG_ON(struct r300_context * ctx, unsigned flags) { diff --git a/src/gallium/drivers/r300/r300_reg.h b/src/gallium/drivers/r300/r300_reg.h index a249e8b36be..c67cc868713 100644 --- a/src/gallium/drivers/r300/r300_reg.h +++ b/src/gallium/drivers/r300/r300_reg.h @@ -540,7 +540,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. # define R300_PVS_FIRST_INST(x) ((x) << 0) # define R300_PVS_XYZW_VALID_INST(x) ((x) << 10) # define R300_PVS_LAST_INST(x) ((x) << 20) -/* Addresses are relative the the vertex program parameters area. */ +/* Addresses are relative to the vertex program parameters area. */ #define R300_VAP_PVS_CONST_CNTL 0x22D4 # define R300_PVS_CONST_BASE_OFFSET_SHIFT 0 # define R300_PVS_MAX_CONST_ADDR_SHIFT 16 @@ -1857,7 +1857,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. * The destination register index is in FPI1 (color) and FPI3 (alpha) * together with enable bits. * There are separate enable bits for writing into temporary registers - * (DSTC_REG_* /DSTA_REG) and and program output registers (DSTC_OUTPUT_* + * (DSTC_REG_* /DSTA_REG) and program output registers (DSTC_OUTPUT_* * /DSTA_OUTPUT). You can write to both at once, or not write at all (the * same index must be used for both). * diff --git a/src/gallium/drivers/r300/r300_render.c b/src/gallium/drivers/r300/r300_render.c index 9c001ae186d..ccf39876a59 100644 --- a/src/gallium/drivers/r300/r300_render.c +++ b/src/gallium/drivers/r300/r300_render.c @@ -274,14 +274,14 @@ static void r300_emit_draw_elements(struct r300_context *r300, #endif CS_LOCALS(r300); - assert((start * indexSize) % 4 == 0); + assert((start * indexSize) % 4 == 0); assert(count < (1 << 24)); + maxIndex = MIN3(maxIndex, r300->vertex_buffer_max_index, count - minIndex); + DBG(r300, DBG_DRAW, "r300: Indexbuf of %u indices, min %u max %u\n", count, minIndex, maxIndex); - maxIndex = MIN2(maxIndex, r300->vertex_buffer_max_index); - if (alt_num_verts) { BEGIN_CS(15); OUT_CS_REG(R500_VAP_ALT_NUM_VERTICES, count); diff --git a/src/gallium/drivers/r300/r300_screen.c b/src/gallium/drivers/r300/r300_screen.c index 5880eecd5fe..64d1909a382 100644 --- a/src/gallium/drivers/r300/r300_screen.c +++ b/src/gallium/drivers/r300/r300_screen.c @@ -27,7 +27,6 @@ #include "r300_context.h" #include "r300_texture.h" -#include "r300_transfer.h" #include "radeon_winsys.h" #include "r300_winsys.h" @@ -290,7 +289,6 @@ struct pipe_screen* r300_create_screen(struct radeon_winsys* radeon_winsys) r300screen->screen.context_create = r300_create_context; r300_init_screen_texture_functions(&r300screen->screen); - r300_init_screen_transfer_functions(&r300screen->screen); u_simple_screen_init(&r300screen->screen); return &r300screen->screen; diff --git a/src/gallium/drivers/r300/r300_state.c b/src/gallium/drivers/r300/r300_state.c index d73ec78fda8..d4b7376f126 100644 --- a/src/gallium/drivers/r300/r300_state.c +++ b/src/gallium/drivers/r300/r300_state.c @@ -1061,7 +1061,7 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe, const struct pipe_vertex_buffer* buffers) { struct r300_context* r300 = r300_context(pipe); - unsigned i, max_index = ~0; + unsigned i, max_index = (1 << 24) - 1; memcpy(r300->vertex_buffer, buffers, sizeof(struct pipe_vertex_buffer) * count); diff --git a/src/gallium/drivers/r300/r300_transfer.c b/src/gallium/drivers/r300/r300_transfer.c index ec89681a3c0..495e3dee767 100644 --- a/src/gallium/drivers/r300/r300_transfer.c +++ b/src/gallium/drivers/r300/r300_transfer.c @@ -118,15 +118,15 @@ static void r300_copy_into_tiled_texture(struct pipe_context *ctx, } static struct pipe_transfer* -r300_get_tex_transfer(struct pipe_screen *screen, +r300_get_tex_transfer(struct pipe_context *ctx, struct pipe_texture *texture, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, unsigned x, unsigned y, unsigned w, unsigned h) { struct r300_texture *tex = (struct r300_texture *)texture; + struct r300_screen *r300screen = r300_screen(ctx->screen); struct r300_transfer *trans; - struct r300_screen *r300screen = r300_screen(screen); struct pipe_texture template; trans = CALLOC_STRUCT(r300_transfer); @@ -136,7 +136,7 @@ r300_get_tex_transfer(struct pipe_screen *screen, trans->transfer.usage = usage; trans->transfer.width = w; trans->transfer.height = h; - trans->ctx = r300screen->ctx; + trans->ctx = ctx; trans->x = x; trans->y = y; trans->level = level; @@ -174,8 +174,10 @@ r300_get_tex_transfer(struct pipe_screen *screen, } /* Create the temporary texture. */ - trans->detiled_texture = - (struct r300_texture*)screen->texture_create(screen, &template); + trans->detiled_texture = (struct r300_texture*) + ctx->screen->texture_create(ctx->screen, + &template); + assert(!trans->detiled_texture->microtile && !trans->detiled_texture->macrotile); @@ -187,7 +189,7 @@ r300_get_tex_transfer(struct pipe_screen *screen, if (usage & PIPE_TRANSFER_READ) { /* We cannot map a tiled texture directly because the data is * in a different order, therefore we do detiling using a blit. */ - r300_copy_from_tiled_texture(r300screen->ctx, trans); + r300_copy_from_tiled_texture(ctx, trans); } } else { trans->transfer.x = x; @@ -203,7 +205,8 @@ r300_get_tex_transfer(struct pipe_screen *screen, return &trans->transfer; } -static void r300_tex_transfer_destroy(struct pipe_transfer *trans) +static void r300_tex_transfer_destroy(struct pipe_context *ctx, + struct pipe_transfer *trans) { struct r300_transfer *r300transfer = r300_transfer(trans); @@ -219,7 +222,7 @@ static void r300_tex_transfer_destroy(struct pipe_transfer *trans) FREE(trans); } -static void* r300_transfer_map(struct pipe_screen *screen, +static void* r300_transfer_map(struct pipe_context *ctx, struct pipe_transfer *transfer) { struct r300_transfer *r300transfer = r300_transfer(transfer); @@ -230,12 +233,12 @@ static void* r300_transfer_map(struct pipe_screen *screen, if (r300transfer->detiled_texture) { /* The detiled texture is of the same size as the region being mapped * (no offset needed). */ - return pipe_buffer_map(screen, + return pipe_buffer_map(ctx->screen, r300transfer->detiled_texture->buffer, pipe_transfer_buffer_flags(transfer)); } else { /* Tiling is disabled. */ - map = pipe_buffer_map(screen, tex->buffer, + map = pipe_buffer_map(ctx->screen, tex->buffer, pipe_transfer_buffer_flags(transfer)); if (!map) { @@ -248,23 +251,26 @@ static void* r300_transfer_map(struct pipe_screen *screen, } } -static void r300_transfer_unmap(struct pipe_screen *screen, +static void r300_transfer_unmap(struct pipe_context *ctx, struct pipe_transfer *transfer) { struct r300_transfer *r300transfer = r300_transfer(transfer); struct r300_texture *tex = (struct r300_texture*)transfer->texture; if (r300transfer->detiled_texture) { - pipe_buffer_unmap(screen, r300transfer->detiled_texture->buffer); + pipe_buffer_unmap(ctx->screen, r300transfer->detiled_texture->buffer); } else { - pipe_buffer_unmap(screen, tex->buffer); + pipe_buffer_unmap(ctx->screen, tex->buffer); } } -void r300_init_screen_transfer_functions(struct pipe_screen *screen) + +void r300_init_transfer_functions( struct r300_context *r300ctx ) { - screen->get_tex_transfer = r300_get_tex_transfer; - screen->tex_transfer_destroy = r300_tex_transfer_destroy; - screen->transfer_map = r300_transfer_map; - screen->transfer_unmap = r300_transfer_unmap; + struct pipe_context *ctx = &r300ctx->context; + + ctx->get_tex_transfer = r300_get_tex_transfer; + ctx->tex_transfer_destroy = r300_tex_transfer_destroy; + ctx->transfer_map = r300_transfer_map; + ctx->transfer_unmap = r300_transfer_unmap; } diff --git a/src/gallium/drivers/r300/r300_transfer.h b/src/gallium/drivers/r300/r300_transfer.h index 60d1d3dc85c..79baf6d0480 100644 --- a/src/gallium/drivers/r300/r300_transfer.h +++ b/src/gallium/drivers/r300/r300_transfer.h @@ -26,6 +26,8 @@ #include "pipe/p_screen.h" -void r300_init_screen_transfer_functions(struct pipe_screen *screen); +struct r300_context; + +void r300_init_transfer_functions(struct r300_context *r300ctx); #endif diff --git a/src/gallium/drivers/softpipe/Makefile b/src/gallium/drivers/softpipe/Makefile index 1c6e4ae0768..239655d628b 100644 --- a/src/gallium/drivers/softpipe/Makefile +++ b/src/gallium/drivers/softpipe/Makefile @@ -8,6 +8,7 @@ C_SOURCES = \ sp_fs_sse.c \ sp_buffer.c \ sp_clear.c \ + sp_fence.c \ sp_flush.c \ sp_query.c \ sp_context.c \ diff --git a/src/gallium/drivers/softpipe/SConscript b/src/gallium/drivers/softpipe/SConscript index 27ab00b036e..9949a53adfd 100644 --- a/src/gallium/drivers/softpipe/SConscript +++ b/src/gallium/drivers/softpipe/SConscript @@ -11,6 +11,7 @@ softpipe = env.ConvenienceLibrary( 'sp_clear.c', 'sp_context.c', 'sp_draw_arrays.c', + 'sp_fence.c', 'sp_flush.c', 'sp_prim_vbuf.c', 'sp_setup.c', diff --git a/src/gallium/drivers/softpipe/sp_context.c b/src/gallium/drivers/softpipe/sp_context.c index 891a903e4f1..937a573092f 100644 --- a/src/gallium/drivers/softpipe/sp_context.c +++ b/src/gallium/drivers/softpipe/sp_context.c @@ -44,6 +44,7 @@ #include "sp_surface.h" #include "sp_tile_cache.h" #include "sp_tex_tile_cache.h" +#include "sp_texture.h" #include "sp_query.h" @@ -277,6 +278,7 @@ softpipe_create_context( struct pipe_screen *screen, softpipe->pipe.is_buffer_referenced = softpipe_is_buffer_referenced; softpipe_init_query_funcs( softpipe ); + softpipe_init_texture_funcs( &softpipe->pipe ); softpipe->pipe.render_condition = softpipe_render_condition; @@ -285,13 +287,13 @@ softpipe_create_context( struct pipe_screen *screen, * Must be before quad stage setup! */ for (i = 0; i < PIPE_MAX_COLOR_BUFS; i++) - softpipe->cbuf_cache[i] = sp_create_tile_cache( screen ); - softpipe->zsbuf_cache = sp_create_tile_cache( screen ); + softpipe->cbuf_cache[i] = sp_create_tile_cache( &softpipe->pipe ); + softpipe->zsbuf_cache = sp_create_tile_cache( &softpipe->pipe ); for (i = 0; i < PIPE_MAX_SAMPLERS; i++) - softpipe->tex_cache[i] = sp_create_tex_tile_cache( screen ); + softpipe->tex_cache[i] = sp_create_tex_tile_cache( &softpipe->pipe ); for (i = 0; i < PIPE_MAX_VERTEX_SAMPLERS; i++) { - softpipe->vertex_tex_cache[i] = sp_create_tex_tile_cache(screen); + softpipe->vertex_tex_cache[i] = sp_create_tex_tile_cache( &softpipe->pipe ); } /* setup quad rendering stages */ diff --git a/src/gallium/drivers/softpipe/sp_fence.c b/src/gallium/drivers/softpipe/sp_fence.c new file mode 100644 index 00000000000..66c52141132 --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_fence.c @@ -0,0 +1,70 @@ +/************************************************************************** + * + * Copyright 2010 VMware, Inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + **************************************************************************/ + + +#include "pipe/p_screen.h" +#include "util/u_debug.h" +#include "sp_fence.h" + + +static void +softpipe_fence_reference(struct pipe_screen *screen, + struct pipe_fence_handle **ptr, + struct pipe_fence_handle *fence) +{ + assert(!*ptr); + assert(!fence); +} + + +static int +softpipe_fence_signalled(struct pipe_screen *screen, + struct pipe_fence_handle *fence, + unsigned flags) +{ + assert(!fence); + return 0; +} + + +static int +softpipe_fence_finish(struct pipe_screen *screen, + struct pipe_fence_handle *fence, + unsigned flags) +{ + assert(!fence); + return 0; +} + + +void +softpipe_init_screen_fence_funcs(struct pipe_screen *screen) +{ + screen->fence_reference = softpipe_fence_reference; + screen->fence_finish = softpipe_fence_finish; + screen->fence_signalled = softpipe_fence_signalled; +} diff --git a/src/gallium/drivers/softpipe/sp_fence.h b/src/gallium/drivers/softpipe/sp_fence.h new file mode 100644 index 00000000000..39c33243bd5 --- /dev/null +++ b/src/gallium/drivers/softpipe/sp_fence.h @@ -0,0 +1,40 @@ +/************************************************************************** + * + * Copyright 2010 VMware, Inc. + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + **************************************************************************/ + + +#ifndef SP_FENCE_H_ +#define SP_FENCE_H_ + + +struct pipe_screen; + + +void +softpipe_init_screen_fence_funcs(struct pipe_screen *screen); + + +#endif /* SP_FENCE_H_ */ diff --git a/src/gallium/drivers/softpipe/sp_flush.c b/src/gallium/drivers/softpipe/sp_flush.c index 38dea13c66b..508fe8f764d 100644 --- a/src/gallium/drivers/softpipe/sp_flush.c +++ b/src/gallium/drivers/softpipe/sp_flush.c @@ -93,9 +93,9 @@ softpipe_flush( struct pipe_context *pipe, static unsigned frame_no = 1; static char filename[256]; util_snprintf(filename, sizeof(filename), "cbuf_%u.bmp", frame_no); - debug_dump_surface_bmp(filename, softpipe->framebuffer.cbufs[0]); + debug_dump_surface_bmp(softpipe, filename, softpipe->framebuffer.cbufs[0]); util_snprintf(filename, sizeof(filename), "zsbuf_%u.bmp", frame_no); - debug_dump_surface_bmp(filename, softpipe->framebuffer.zsbuf); + debug_dump_surface_bmp(softpipe, filename, softpipe->framebuffer.zsbuf); ++frame_no; } #endif diff --git a/src/gallium/drivers/softpipe/sp_fs_exec.c b/src/gallium/drivers/softpipe/sp_fs_exec.c index 27fa126b7c3..67e2c8f8bc4 100644 --- a/src/gallium/drivers/softpipe/sp_fs_exec.c +++ b/src/gallium/drivers/softpipe/sp_fs_exec.c @@ -145,8 +145,13 @@ exec_run( const struct sp_fragment_shader *base, case TGSI_SEMANTIC_COLOR: { uint cbuf = sem_index[i]; + + assert(sizeof(quad->output.color[cbuf]) == + sizeof(machine->Outputs[i])); + + /* copy float[4][4] result */ memcpy(quad->output.color[cbuf], - &machine->Outputs[i].xyzw[0].f[0], + &machine->Outputs[i], sizeof(quad->output.color[0]) ); } break; diff --git a/src/gallium/drivers/softpipe/sp_fs_sse.c b/src/gallium/drivers/softpipe/sp_fs_sse.c index acee2136706..daa158df7c4 100644 --- a/src/gallium/drivers/softpipe/sp_fs_sse.c +++ b/src/gallium/drivers/softpipe/sp_fs_sse.c @@ -156,8 +156,13 @@ fs_sse_run( const struct sp_fragment_shader *base, case TGSI_SEMANTIC_COLOR: { uint cbuf = sem_index[i]; + + assert(sizeof(quad->output.color[cbuf]) == + sizeof(machine->Outputs[i])); + + /* copy float[4][4] result */ memcpy(quad->output.color[cbuf], - &machine->Outputs[i].xyzw[0].f[0], + &machine->Outputs[i], sizeof(quad->output.color[0]) ); } break; diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c index e0a2ef604e9..d62bfa3d633 100644 --- a/src/gallium/drivers/softpipe/sp_screen.c +++ b/src/gallium/drivers/softpipe/sp_screen.c @@ -36,6 +36,7 @@ #include "sp_screen.h" #include "sp_context.h" #include "sp_buffer.h" +#include "sp_fence.h" #include "sp_public.h" @@ -239,6 +240,7 @@ softpipe_create_screen(struct sw_winsys *winsys) softpipe_init_screen_texture_funcs(&screen->base); softpipe_init_screen_buffer_funcs(&screen->base); + softpipe_init_screen_fence_funcs(&screen->base); return &screen->base; } diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c b/src/gallium/drivers/softpipe/sp_tex_sample.c index ef7ccf41898..fa9e19b282b 100644 --- a/src/gallium/drivers/softpipe/sp_tex_sample.c +++ b/src/gallium/drivers/softpipe/sp_tex_sample.c @@ -1614,7 +1614,6 @@ sample_cube(struct tgsi_sampler *tgsi_sampler, struct sp_sampler_varient *samp = sp_sampler_varient(tgsi_sampler); unsigned j; float ssss[4], tttt[4]; - unsigned face; /* major axis @@ -1628,7 +1627,8 @@ sample_cube(struct tgsi_sampler *tgsi_sampler, -rz TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT -rx -ry rz */ - /* First choose the cube face. + /* Choose the cube face and compute new s/t coords for the 2D face. + * * Use the same cube face for all four pixels in the quad. * * This isn't ideal, but if we want to use a different cube face @@ -1647,85 +1647,37 @@ sample_cube(struct tgsi_sampler *tgsi_sampler, const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz); if (arx >= ary && arx >= arz) { - if (rx >= 0.0F) { - face = PIPE_TEX_FACE_POS_X; - } - else { - face = PIPE_TEX_FACE_NEG_X; + float sign = (rx >= 0.0F) ? 1.0F : -1.0F; + uint face = (rx >= 0.0F) ? PIPE_TEX_FACE_POS_X : PIPE_TEX_FACE_NEG_X; + for (j = 0; j < QUAD_SIZE; j++) { + const float ima = -0.5F / fabsf(s[j]); + ssss[j] = sign * p[j] * ima + 0.5F; + tttt[j] = t[j] * ima + 0.5F; + samp->faces[j] = face; } } else if (ary >= arx && ary >= arz) { - if (ry >= 0.0F) { - face = PIPE_TEX_FACE_POS_Y; - } - else { - face = PIPE_TEX_FACE_NEG_Y; + float sign = (ry >= 0.0F) ? 1.0F : -1.0F; + uint face = (ry >= 0.0F) ? PIPE_TEX_FACE_POS_Y : PIPE_TEX_FACE_NEG_Y; + for (j = 0; j < QUAD_SIZE; j++) { + const float ima = -0.5F / fabsf(t[j]); + ssss[j] = -s[j] * ima + 0.5F; + tttt[j] = sign * -p[j] * ima + 0.5F; + samp->faces[j] = face; } } else { - if (rz > 0.0F) { - face = PIPE_TEX_FACE_POS_Z; - } - else { - face = PIPE_TEX_FACE_NEG_Z; + float sign = (rz >= 0.0F) ? 1.0F : -1.0F; + uint face = (rz >= 0.0F) ? PIPE_TEX_FACE_POS_Z : PIPE_TEX_FACE_NEG_Z; + for (j = 0; j < QUAD_SIZE; j++) { + const float ima = -0.5 / fabsf(p[j]); + ssss[j] = sign * -s[j] * ima + 0.5F; + tttt[j] = t[j] * ima + 0.5F; + samp->faces[j] = face; } } } - /* Now compute the 2D _face_ texture coords from the - * 3D _cube_ texture coords. - */ - for (j = 0; j < QUAD_SIZE; j++) { - const float rx = s[j], ry = t[j], rz = p[j]; - const float arx = fabsf(rx), ary = fabsf(ry), arz = fabsf(rz); - float sc, tc, ma; - - switch (face) { - case PIPE_TEX_FACE_POS_X: - sc = -rz; - tc = -ry; - ma = arx; - break; - case PIPE_TEX_FACE_NEG_X: - sc = rz; - tc = -ry; - ma = arx; - break; - case PIPE_TEX_FACE_POS_Y: - sc = rx; - tc = rz; - ma = ary; - break; - case PIPE_TEX_FACE_NEG_Y: - sc = rx; - tc = -rz; - ma = ary; - break; - case PIPE_TEX_FACE_POS_Z: - sc = rx; - tc = -ry; - ma = arz; - break; - case PIPE_TEX_FACE_NEG_Z: - sc = -rx; - tc = -ry; - ma = arz; - break; - default: - assert(0 && "bad cube face"); - sc = 0.0F; - tc = 0.0F; - ma = 0.0F; - } - - { - const float ima = 1.0 / ma; - ssss[j] = ( sc * ima + 1.0F ) * 0.5F; - tttt[j] = ( tc * ima + 1.0F ) * 0.5F; - samp->faces[j] = face; - } - } - /* In our little pipeline, the compare stage is next. If compare * is not active, this will point somewhere deeper into the * pipeline, eg. to mip_filter or even img_filter. diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c index dfa002a79b4..6594514c38f 100644 --- a/src/gallium/drivers/softpipe/sp_tex_tile_cache.c +++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.c @@ -43,14 +43,14 @@ struct softpipe_tex_tile_cache * -sp_create_tex_tile_cache( struct pipe_screen *screen ) +sp_create_tex_tile_cache( struct pipe_context *pipe ) { struct softpipe_tex_tile_cache *tc; uint pos; tc = CALLOC_STRUCT( softpipe_tex_tile_cache ); if (tc) { - tc->screen = screen; + tc->pipe = pipe; for (pos = 0; pos < NUM_ENTRIES; pos++) { tc->entries[pos].addr.bits.invalid = 1; } @@ -63,19 +63,16 @@ sp_create_tex_tile_cache( struct pipe_screen *screen ) void sp_destroy_tex_tile_cache(struct softpipe_tex_tile_cache *tc) { - struct pipe_screen *screen; uint pos; for (pos = 0; pos < NUM_ENTRIES; pos++) { /*assert(tc->entries[pos].x < 0);*/ } if (tc->transfer) { - screen = tc->transfer->texture->screen; - screen->tex_transfer_destroy(tc->transfer); + tc->pipe->tex_transfer_destroy(tc->pipe, tc->transfer); } if (tc->tex_trans) { - screen = tc->tex_trans->texture->screen; - screen->tex_transfer_destroy(tc->tex_trans); + tc->pipe->tex_transfer_destroy(tc->pipe, tc->tex_trans); } FREE( tc ); @@ -88,7 +85,7 @@ void sp_tex_tile_cache_map_transfers(struct softpipe_tex_tile_cache *tc) { if (tc->tex_trans && !tc->tex_trans_map) - tc->tex_trans_map = tc->screen->transfer_map(tc->screen, tc->tex_trans); + tc->tex_trans_map = tc->pipe->transfer_map(tc->pipe, tc->tex_trans); } @@ -96,7 +93,7 @@ void sp_tex_tile_cache_unmap_transfers(struct softpipe_tex_tile_cache *tc) { if (tc->tex_trans_map) { - tc->screen->transfer_unmap(tc->screen, tc->tex_trans); + tc->pipe->transfer_unmap(tc->pipe, tc->tex_trans); tc->tex_trans_map = NULL; } } @@ -134,14 +131,12 @@ sp_tex_tile_cache_set_sampler_view(struct softpipe_tex_tile_cache *tc, pipe_texture_reference(&tc->texture, texture); if (tc->tex_trans) { - struct pipe_screen *screen = tc->tex_trans->texture->screen; - if (tc->tex_trans_map) { - screen->transfer_unmap(screen, tc->tex_trans); + tc->pipe->transfer_unmap(tc->pipe, tc->tex_trans); tc->tex_trans_map = NULL; } - screen->tex_transfer_destroy(tc->tex_trans); + tc->pipe->tex_transfer_destroy(tc->pipe, tc->tex_trans); tc->tex_trans = NULL; } @@ -213,7 +208,6 @@ const struct softpipe_tex_cached_tile * sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, union tex_tile_address addr ) { - struct pipe_screen *screen = tc->screen; struct softpipe_tex_cached_tile *tile; tile = tc->entries + tex_cache_pos( addr ); @@ -241,16 +235,16 @@ sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, if (tc->tex_trans) { if (tc->tex_trans_map) { - tc->screen->transfer_unmap(tc->screen, tc->tex_trans); + tc->pipe->transfer_unmap(tc->pipe, tc->tex_trans); tc->tex_trans_map = NULL; } - screen->tex_transfer_destroy(tc->tex_trans); + tc->pipe->tex_transfer_destroy(tc->pipe, tc->tex_trans); tc->tex_trans = NULL; } tc->tex_trans = - screen->get_tex_transfer(screen, tc->texture, + tc->pipe->get_tex_transfer(tc->pipe, tc->texture, addr.bits.face, addr.bits.level, addr.bits.z, @@ -258,7 +252,7 @@ sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, u_minify(tc->texture->width0, addr.bits.level), u_minify(tc->texture->height0, addr.bits.level)); - tc->tex_trans_map = screen->transfer_map(screen, tc->tex_trans); + tc->tex_trans_map = tc->pipe->transfer_map(tc->pipe, tc->tex_trans); tc->tex_face = addr.bits.face; tc->tex_level = addr.bits.level; @@ -266,7 +260,8 @@ sp_find_cached_tile_tex(struct softpipe_tex_tile_cache *tc, } /* get tile from the transfer (view into texture) */ - pipe_get_tile_swizzle(tc->tex_trans, + pipe_get_tile_swizzle(tc->pipe, + tc->tex_trans, addr.bits.x * TILE_SIZE, addr.bits.y * TILE_SIZE, TILE_SIZE, diff --git a/src/gallium/drivers/softpipe/sp_tex_tile_cache.h b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h index f8770409d87..12ae7ba12d6 100644 --- a/src/gallium/drivers/softpipe/sp_tex_tile_cache.h +++ b/src/gallium/drivers/softpipe/sp_tex_tile_cache.h @@ -70,7 +70,7 @@ struct softpipe_tex_cached_tile struct softpipe_tex_tile_cache { - struct pipe_screen *screen; + struct pipe_context *pipe; struct pipe_transfer *transfer; void *transfer_map; @@ -94,7 +94,7 @@ struct softpipe_tex_tile_cache extern struct softpipe_tex_tile_cache * -sp_create_tex_tile_cache( struct pipe_screen *screen ); +sp_create_tex_tile_cache( struct pipe_context *pipe ); extern void sp_destroy_tex_tile_cache(struct softpipe_tex_tile_cache *tc); diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c index 11d184effb9..da8529c154e 100644 --- a/src/gallium/drivers/softpipe/sp_texture.c +++ b/src/gallium/drivers/softpipe/sp_texture.c @@ -256,7 +256,7 @@ softpipe_tex_surface_destroy(struct pipe_surface *surf) * \param height height of region to read/write */ static struct pipe_transfer * -softpipe_get_tex_transfer(struct pipe_screen *screen, +softpipe_get_tex_transfer(struct pipe_context *pipe, struct pipe_texture *texture, unsigned face, unsigned level, unsigned zslice, enum pipe_transfer_usage usage, @@ -310,7 +310,8 @@ softpipe_get_tex_transfer(struct pipe_screen *screen, * softpipe_get_tex_transfer(). */ static void -softpipe_tex_transfer_destroy(struct pipe_transfer *transfer) +softpipe_tex_transfer_destroy(struct pipe_context *pipe, + struct pipe_transfer *transfer) { /* Effectively do the texture_update work here - if texture images * needed post-processing to put them into hardware layout, this is @@ -326,7 +327,7 @@ softpipe_tex_transfer_destroy(struct pipe_transfer *transfer) * Create memory mapping for given pipe_transfer object. */ static void * -softpipe_transfer_map( struct pipe_screen *screen, +softpipe_transfer_map( struct pipe_context *pipe, struct pipe_transfer *transfer ) { ubyte *map, *xfer_map; @@ -339,7 +340,7 @@ softpipe_transfer_map( struct pipe_screen *screen, if (spt->dt) { /* display target */ - struct sw_winsys *winsys = softpipe_screen(screen)->winsys; + struct sw_winsys *winsys = softpipe_screen(pipe->screen)->winsys; map = winsys->displaytarget_map(winsys, spt->dt, pipe_transfer_buffer_flags(transfer)); @@ -359,7 +360,7 @@ softpipe_transfer_map( struct pipe_screen *screen, /* Do something to notify sharing contexts of a texture change. * In softpipe, that would mean flushing the texture cache. */ - softpipe_screen(screen)->timestamp++; + softpipe_screen(pipe->screen)->timestamp++; } xfer_map = map + softpipe_transfer(transfer)->offset + @@ -374,7 +375,7 @@ softpipe_transfer_map( struct pipe_screen *screen, * Unmap memory mapping for given pipe_transfer object. */ static void -softpipe_transfer_unmap(struct pipe_screen *screen, +softpipe_transfer_unmap(struct pipe_context *pipe, struct pipe_transfer *transfer) { struct softpipe_texture *spt; @@ -384,7 +385,7 @@ softpipe_transfer_unmap(struct pipe_screen *screen, if (spt->dt) { /* display target */ - struct sw_winsys *winsys = softpipe_screen(screen)->winsys; + struct sw_winsys *winsys = softpipe_screen(pipe->screen)->winsys; winsys->displaytarget_unmap(winsys, spt->dt); } @@ -448,6 +449,15 @@ softpipe_video_surface_destroy(struct pipe_video_surface *vsfc) void +softpipe_init_texture_funcs(struct pipe_context *pipe) +{ + pipe->get_tex_transfer = softpipe_get_tex_transfer; + pipe->tex_transfer_destroy = softpipe_tex_transfer_destroy; + pipe->transfer_map = softpipe_transfer_map; + pipe->transfer_unmap = softpipe_transfer_unmap; +} + +void softpipe_init_screen_texture_funcs(struct pipe_screen *screen) { screen->texture_create = softpipe_texture_create; @@ -456,11 +466,6 @@ softpipe_init_screen_texture_funcs(struct pipe_screen *screen) screen->get_tex_surface = softpipe_get_tex_surface; screen->tex_surface_destroy = softpipe_tex_surface_destroy; - screen->get_tex_transfer = softpipe_get_tex_transfer; - screen->tex_transfer_destroy = softpipe_tex_transfer_destroy; - screen->transfer_map = softpipe_transfer_map; - screen->transfer_unmap = softpipe_transfer_unmap; - screen->video_surface_create = softpipe_video_surface_create; screen->video_surface_destroy = softpipe_video_surface_destroy; } diff --git a/src/gallium/drivers/softpipe/sp_texture.h b/src/gallium/drivers/softpipe/sp_texture.h index 1c8636d1d56..c0e6ba8a869 100644 --- a/src/gallium/drivers/softpipe/sp_texture.h +++ b/src/gallium/drivers/softpipe/sp_texture.h @@ -107,5 +107,8 @@ softpipe_video_surface(struct pipe_video_surface *pvs) extern void softpipe_init_screen_texture_funcs(struct pipe_screen *screen); +void +softpipe_init_texture_funcs(struct pipe_context *pipe); + #endif /* SP_TEXTURE */ diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.c b/src/gallium/drivers/softpipe/sp_tile_cache.c index aedfdf1b469..1c3c2667d73 100644 --- a/src/gallium/drivers/softpipe/sp_tile_cache.c +++ b/src/gallium/drivers/softpipe/sp_tile_cache.c @@ -79,20 +79,20 @@ clear_clear_flag(uint *bitvec, union tile_address addr) struct softpipe_tile_cache * -sp_create_tile_cache( struct pipe_screen *screen ) +sp_create_tile_cache( struct pipe_context *pipe ) { struct softpipe_tile_cache *tc; uint pos; int maxLevels, maxTexSize; /* sanity checking: max sure MAX_WIDTH/HEIGHT >= largest texture image */ - maxLevels = screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS); + maxLevels = pipe->screen->get_param(pipe->screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS); maxTexSize = 1 << (maxLevels - 1); assert(MAX_WIDTH >= maxTexSize); tc = CALLOC_STRUCT( softpipe_tile_cache ); if (tc) { - tc->screen = screen; + tc->pipe = pipe; for (pos = 0; pos < NUM_ENTRIES; pos++) { tc->entries[pos].addr.bits.invalid = 1; } @@ -115,15 +115,13 @@ sp_create_tile_cache( struct pipe_screen *screen ) void sp_destroy_tile_cache(struct softpipe_tile_cache *tc) { - struct pipe_screen *screen; uint pos; for (pos = 0; pos < NUM_ENTRIES; pos++) { /*assert(tc->entries[pos].x < 0);*/ } if (tc->transfer) { - screen = tc->transfer->texture->screen; - screen->tex_transfer_destroy(tc->transfer); + tc->pipe->tex_transfer_destroy(tc->pipe, tc->transfer); } FREE( tc ); @@ -137,27 +135,25 @@ void sp_tile_cache_set_surface(struct softpipe_tile_cache *tc, struct pipe_surface *ps) { - if (tc->transfer) { - struct pipe_screen *screen = tc->transfer->texture->screen; + struct pipe_context *pipe = tc->pipe; + if (tc->transfer) { if (ps == tc->surface) return; if (tc->transfer_map) { - screen->transfer_unmap(screen, tc->transfer); + pipe->transfer_unmap(pipe, tc->transfer); tc->transfer_map = NULL; } - screen->tex_transfer_destroy(tc->transfer); + pipe->tex_transfer_destroy(pipe, tc->transfer); tc->transfer = NULL; } tc->surface = ps; if (ps) { - struct pipe_screen *screen = ps->texture->screen; - - tc->transfer = screen->get_tex_transfer(screen, ps->texture, ps->face, + tc->transfer = pipe->get_tex_transfer(pipe, ps->texture, ps->face, ps->level, ps->zslice, PIPE_TRANSFER_READ_WRITE, 0, 0, ps->width, ps->height); @@ -187,7 +183,7 @@ void sp_tile_cache_map_transfers(struct softpipe_tile_cache *tc) { if (tc->transfer && !tc->transfer_map) - tc->transfer_map = tc->screen->transfer_map(tc->screen, tc->transfer); + tc->transfer_map = tc->pipe->transfer_map(tc->pipe, tc->transfer); } @@ -195,7 +191,7 @@ void sp_tile_cache_unmap_transfers(struct softpipe_tile_cache *tc) { if (tc->transfer_map) { - tc->screen->transfer_unmap(tc->screen, tc->transfer); + tc->pipe->transfer_unmap(tc->pipe, tc->transfer); tc->transfer_map = NULL; } } @@ -295,7 +291,8 @@ sp_tile_cache_flush_clear(struct softpipe_tile_cache *tc) union tile_address addr = tile_address(x, y); if (is_clear_flag_set(tc->clear_flags, addr)) { - pipe_put_tile_raw(pt, + pipe_put_tile_raw(tc->pipe, + pt, x, y, TILE_SIZE, TILE_SIZE, tc->tile.data.color32, 0/*STRIDE*/); @@ -329,14 +326,14 @@ sp_flush_tile_cache(struct softpipe_tile_cache *tc) struct softpipe_cached_tile *tile = tc->entries + pos; if (!tile->addr.bits.invalid) { if (tc->depth_stencil) { - pipe_put_tile_raw(pt, + pipe_put_tile_raw(tc->pipe, pt, tile->addr.bits.x * TILE_SIZE, tile->addr.bits.y * TILE_SIZE, TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { - pipe_put_tile_rgba(pt, + pipe_put_tile_rgba(tc->pipe, pt, tile->addr.bits.x * TILE_SIZE, tile->addr.bits.y * TILE_SIZE, TILE_SIZE, TILE_SIZE, @@ -379,14 +376,14 @@ sp_find_cached_tile(struct softpipe_tile_cache *tc, if (tile->addr.bits.invalid == 0) { /* put dirty tile back in framebuffer */ if (tc->depth_stencil) { - pipe_put_tile_raw(pt, + pipe_put_tile_raw(tc->pipe, pt, tile->addr.bits.x * TILE_SIZE, tile->addr.bits.y * TILE_SIZE, TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { - pipe_put_tile_rgba(pt, + pipe_put_tile_rgba(tc->pipe, pt, tile->addr.bits.x * TILE_SIZE, tile->addr.bits.y * TILE_SIZE, TILE_SIZE, TILE_SIZE, @@ -409,14 +406,14 @@ sp_find_cached_tile(struct softpipe_tile_cache *tc, else { /* get new tile data from transfer */ if (tc->depth_stencil) { - pipe_get_tile_raw(pt, + pipe_get_tile_raw(tc->pipe, pt, tile->addr.bits.x * TILE_SIZE, tile->addr.bits.y * TILE_SIZE, TILE_SIZE, TILE_SIZE, tile->data.depth32, 0/*STRIDE*/); } else { - pipe_get_tile_rgba(pt, + pipe_get_tile_rgba(tc->pipe, pt, tile->addr.bits.x * TILE_SIZE, tile->addr.bits.y * TILE_SIZE, TILE_SIZE, TILE_SIZE, diff --git a/src/gallium/drivers/softpipe/sp_tile_cache.h b/src/gallium/drivers/softpipe/sp_tile_cache.h index a12092702a6..753d8c0daac 100644 --- a/src/gallium/drivers/softpipe/sp_tile_cache.h +++ b/src/gallium/drivers/softpipe/sp_tile_cache.h @@ -80,7 +80,7 @@ struct softpipe_cached_tile struct softpipe_tile_cache { - struct pipe_screen *screen; + struct pipe_context *pipe; struct pipe_surface *surface; /**< the surface we're caching */ struct pipe_transfer *transfer; void *transfer_map; @@ -98,7 +98,7 @@ struct softpipe_tile_cache extern struct softpipe_tile_cache * -sp_create_tile_cache( struct pipe_screen *screen ); +sp_create_tile_cache( struct pipe_context *pipe ); extern void sp_destroy_tile_cache(struct softpipe_tile_cache *tc); diff --git a/src/gallium/drivers/svga/svga_context.c b/src/gallium/drivers/svga/svga_context.c index d499ae6acc9..adb7840182b 100644 --- a/src/gallium/drivers/svga/svga_context.c +++ b/src/gallium/drivers/svga/svga_context.c @@ -164,6 +164,8 @@ struct pipe_context *svga_context_create( struct pipe_screen *screen, svga_init_constbuffer_functions(svga); svga_init_query_functions(svga); + svga_init_texture_functions(&svga->pipe); + /* debug */ svga->debug.no_swtnl = debug_get_bool_option("SVGA_NO_SWTNL", FALSE); svga->debug.force_swtnl = debug_get_bool_option("SVGA_FORCE_SWTNL", FALSE); diff --git a/src/gallium/drivers/svga/svga_screen_texture.c b/src/gallium/drivers/svga/svga_screen_texture.c index 5b581debfc7..107e4a39620 100644 --- a/src/gallium/drivers/svga/svga_screen_texture.c +++ b/src/gallium/drivers/svga/svga_screen_texture.c @@ -783,15 +783,17 @@ svga_surface_needs_propagation(struct pipe_surface *surf) return s->dirty && s->handle != tex->handle; } - +/* XXX: Still implementing this as if it was a screen function, but + * can now modify it to queue transfers on the context. + */ static struct pipe_transfer * -svga_get_tex_transfer(struct pipe_screen *screen, - struct pipe_texture *texture, - unsigned face, unsigned level, unsigned zslice, - enum pipe_transfer_usage usage, unsigned x, unsigned y, - unsigned w, unsigned h) +svga_get_tex_transfer(struct pipe_context *pipe, + struct pipe_texture *texture, + unsigned face, unsigned level, unsigned zslice, + enum pipe_transfer_usage usage, unsigned x, unsigned y, + unsigned w, unsigned h) { - struct svga_screen *ss = svga_screen(screen); + struct svga_screen *ss = svga_screen(pipe->screen); struct svga_winsys_screen *sws = ss->sws; struct svga_transfer *st; unsigned nblocksx = util_format_get_nblocksx(texture->format, w); @@ -859,11 +861,14 @@ no_hwbuf: } +/* XXX: Still implementing this as if it was a screen function, but + * can now modify it to queue transfers on the context. + */ static void * -svga_transfer_map( struct pipe_screen *screen, +svga_transfer_map( struct pipe_context *pipe, struct pipe_transfer *transfer ) { - struct svga_screen *ss = svga_screen(screen); + struct svga_screen *ss = svga_screen(pipe->screen); struct svga_winsys_screen *sws = ss->sws; struct svga_transfer *st = svga_transfer(transfer); @@ -877,11 +882,14 @@ svga_transfer_map( struct pipe_screen *screen, } +/* XXX: Still implementing this as if it was a screen function, but + * can now modify it to queue transfers on the context. + */ static void -svga_transfer_unmap(struct pipe_screen *screen, +svga_transfer_unmap(struct pipe_context *pipe, struct pipe_transfer *transfer) { - struct svga_screen *ss = svga_screen(screen); + struct svga_screen *ss = svga_screen(pipe->screen); struct svga_winsys_screen *sws = ss->sws; struct svga_transfer *st = svga_transfer(transfer); @@ -891,10 +899,11 @@ svga_transfer_unmap(struct pipe_screen *screen, static void -svga_tex_transfer_destroy(struct pipe_transfer *transfer) +svga_tex_transfer_destroy(struct pipe_context *pipe, + struct pipe_transfer *transfer) { struct svga_texture *tex = svga_texture(transfer->texture); - struct svga_screen *ss = svga_screen(transfer->texture->screen); + struct svga_screen *ss = svga_screen(pipe->screen); struct svga_winsys_screen *sws = ss->sws; struct svga_transfer *st = svga_transfer(transfer); @@ -911,6 +920,17 @@ svga_tex_transfer_destroy(struct pipe_transfer *transfer) FREE(st); } + +void +svga_init_texture_functions(struct pipe_context *pipe) +{ + pipe->get_tex_transfer = svga_get_tex_transfer; + pipe->transfer_map = svga_transfer_map; + pipe->transfer_unmap = svga_transfer_unmap; + pipe->tex_transfer_destroy = svga_tex_transfer_destroy; +} + + void svga_screen_init_texture_functions(struct pipe_screen *screen) { @@ -920,10 +940,6 @@ svga_screen_init_texture_functions(struct pipe_screen *screen) screen->texture_destroy = svga_texture_destroy; screen->get_tex_surface = svga_get_tex_surface; screen->tex_surface_destroy = svga_tex_surface_destroy; - screen->get_tex_transfer = svga_get_tex_transfer; - screen->transfer_map = svga_transfer_map; - screen->transfer_unmap = svga_transfer_unmap; - screen->tex_transfer_destroy = svga_tex_transfer_destroy; } /*********************************************************************** diff --git a/src/gallium/drivers/svga/svga_screen_texture.h b/src/gallium/drivers/svga/svga_screen_texture.h index ca6602b4369..96d035b12d8 100644 --- a/src/gallium/drivers/svga/svga_screen_texture.h +++ b/src/gallium/drivers/svga/svga_screen_texture.h @@ -186,6 +186,9 @@ svga_surface_needs_propagation(struct pipe_surface *surf); extern void svga_screen_init_texture_functions(struct pipe_screen *screen); +void +svga_init_texture_functions(struct pipe_context *pipe); + enum SVGA3dSurfaceFormat svga_translate_format(enum pipe_format format); diff --git a/src/gallium/drivers/svga/svgadump/svga_shader_dump.c b/src/gallium/drivers/svga/svgadump/svga_shader_dump.c index 705ca29e8f5..4ee1bf2c353 100644 --- a/src/gallium/drivers/svga/svgadump/svga_shader_dump.c +++ b/src/gallium/drivers/svga/svgadump/svga_shader_dump.c @@ -360,7 +360,9 @@ dump_dstreg(struct sh_dstreg dstreg, union { struct sh_reg reg; struct sh_dstreg dstreg; - } u = { { 0 } }; + } u; + + memset(&u, 0, sizeof(u)); assert( (dstreg.modifier & (SVGA3DDSTMOD_SATURATE | SVGA3DDSTMOD_PARTIALPRECISION)) == dstreg.modifier ); diff --git a/src/gallium/drivers/trace/tr_context.c b/src/gallium/drivers/trace/tr_context.c index f9555fb922c..5c24bd1f7df 100644 --- a/src/gallium/drivers/trace/tr_context.c +++ b/src/gallium/drivers/trace/tr_context.c @@ -28,7 +28,9 @@ #include "util/u_inlines.h" #include "util/u_memory.h" #include "util/u_simple_list.h" +#include "util/u_format.h" +#include "pipe/p_format.h" #include "pipe/p_screen.h" #include "tr_dump.h" @@ -1340,6 +1342,136 @@ trace_is_buffer_referenced( struct pipe_context *_pipe, return referenced; } + +/******************************************************************** + * transfer + */ + + +static struct pipe_transfer * +trace_context_get_tex_transfer(struct pipe_context *_context, + struct pipe_texture *_texture, + unsigned face, unsigned level, + unsigned zslice, + enum pipe_transfer_usage usage, + unsigned x, unsigned y, unsigned w, unsigned h) +{ + struct trace_context *tr_context = trace_context(_context); + struct trace_texture *tr_tex = trace_texture(_texture); + struct pipe_context *context = tr_context->pipe; + struct pipe_texture *texture = tr_tex->texture; + struct pipe_transfer *result = NULL; + + assert(texture->screen == context->screen); + + trace_dump_call_begin("pipe_context", "get_tex_transfer"); + + trace_dump_arg(ptr, context); + trace_dump_arg(ptr, texture); + trace_dump_arg(uint, face); + trace_dump_arg(uint, level); + trace_dump_arg(uint, zslice); + trace_dump_arg(uint, usage); + + trace_dump_arg(uint, x); + trace_dump_arg(uint, y); + trace_dump_arg(uint, w); + trace_dump_arg(uint, h); + + result = context->get_tex_transfer(context, texture, face, level, zslice, usage, + x, y, w, h); + + trace_dump_ret(ptr, result); + + trace_dump_call_end(); + + if (result) + result = trace_transfer_create(tr_context, tr_tex, result); + + return result; +} + + +static void +trace_context_tex_transfer_destroy(struct pipe_context *_context, + struct pipe_transfer *_transfer) +{ + struct trace_context *tr_context = trace_context(_context); + struct trace_transfer *tr_trans = trace_transfer(_transfer); + struct pipe_context *context = tr_context->pipe; + struct pipe_transfer *transfer = tr_trans->transfer; + + trace_dump_call_begin("pipe_context", "tex_transfer_destroy"); + + trace_dump_arg(ptr, context); + trace_dump_arg(ptr, transfer); + + trace_dump_call_end(); + + trace_transfer_destroy(tr_context, tr_trans); +} + + +static void * +trace_context_transfer_map(struct pipe_context *_context, + struct pipe_transfer *_transfer) +{ + struct trace_context *tr_context = trace_context(_context); + struct trace_transfer *tr_trans = trace_transfer(_transfer); + struct pipe_context *context = tr_context->pipe; + struct pipe_transfer *transfer = tr_trans->transfer; + void *map; + + map = context->transfer_map(context, transfer); + if(map) { + if(transfer->usage & PIPE_TRANSFER_WRITE) { + assert(!tr_trans->map); + tr_trans->map = map; + } + } + + return map; +} + + +static void +trace_context_transfer_unmap(struct pipe_context *_context, + struct pipe_transfer *_transfer) +{ + struct trace_context *tr_ctx = trace_context(_context); + struct trace_transfer *tr_trans = trace_transfer(_transfer); + struct pipe_context *context = tr_ctx->pipe; + struct pipe_transfer *transfer = tr_trans->transfer; + + if(tr_trans->map) { + size_t size = util_format_get_nblocksy(transfer->texture->format, transfer->height) * transfer->stride; + + trace_dump_call_begin("pipe_context", "transfer_write"); + + trace_dump_arg(ptr, context); + + trace_dump_arg(ptr, transfer); + + trace_dump_arg_begin("stride"); + trace_dump_uint(transfer->stride); + trace_dump_arg_end(); + + trace_dump_arg_begin("data"); + trace_dump_bytes(tr_trans->map, size); + trace_dump_arg_end(); + + trace_dump_arg_begin("size"); + trace_dump_uint(size); + trace_dump_arg_end(); + + trace_dump_call_end(); + + tr_trans->map = NULL; + } + + context->transfer_unmap(context, transfer); +} + static const struct debug_named_value rbug_blocker_flags[] = { {"before", 1}, {"after", 2}, @@ -1426,6 +1558,11 @@ trace_context_create(struct trace_screen *tr_scr, tr_ctx->base.is_texture_referenced = trace_is_texture_referenced; tr_ctx->base.is_buffer_referenced = trace_is_buffer_referenced; + tr_ctx->base.get_tex_transfer = trace_context_get_tex_transfer; + tr_ctx->base.tex_transfer_destroy = trace_context_tex_transfer_destroy; + tr_ctx->base.transfer_map = trace_context_transfer_map; + tr_ctx->base.transfer_unmap = trace_context_transfer_unmap; + tr_ctx->pipe = pipe; trace_screen_add_to_list(tr_scr, contexts, tr_ctx); diff --git a/src/gallium/drivers/trace/tr_rbug.c b/src/gallium/drivers/trace/tr_rbug.c index 07f9de253ec..53ab8c686d8 100644 --- a/src/gallium/drivers/trace/tr_rbug.c +++ b/src/gallium/drivers/trace/tr_rbug.c @@ -219,7 +219,7 @@ trace_rbug_texture_read(struct trace_rbug *tr_rbug, struct rbug_header *header, struct trace_texture *tr_tex = NULL; struct tr_list *ptr; - struct pipe_screen *screen = tr_scr->screen; + struct pipe_context *context = tr_scr->private_context; struct pipe_texture *tex; struct pipe_transfer *t; @@ -239,12 +239,12 @@ trace_rbug_texture_read(struct trace_rbug *tr_rbug, struct rbug_header *header, } tex = tr_tex->texture; - t = screen->get_tex_transfer(tr_scr->screen, tex, - gptr->face, gptr->level, gptr->zslice, - PIPE_TRANSFER_READ, - gptr->x, gptr->y, gptr->w, gptr->h); + t = context->get_tex_transfer(context, tex, + gptr->face, gptr->level, gptr->zslice, + PIPE_TRANSFER_READ, + gptr->x, gptr->y, gptr->w, gptr->h); - map = screen->transfer_map(screen, t); + map = context->transfer_map(context, t); rbug_send_texture_read_reply(tr_rbug->con, serial, t->texture->format, @@ -256,8 +256,8 @@ trace_rbug_texture_read(struct trace_rbug *tr_rbug, struct rbug_header *header, t->stride, NULL); - screen->transfer_unmap(screen, t); - screen->tex_transfer_destroy(t); + context->transfer_unmap(context, t); + context->tex_transfer_destroy(context, t); pipe_mutex_unlock(tr_scr->list_mutex); diff --git a/src/gallium/drivers/trace/tr_screen.c b/src/gallium/drivers/trace/tr_screen.c index 86ddb995405..25990bdac7f 100644 --- a/src/gallium/drivers/trace/tr_screen.c +++ b/src/gallium/drivers/trace/tr_screen.c @@ -353,133 +353,7 @@ trace_screen_tex_surface_destroy(struct pipe_surface *_surface) } -/******************************************************************** - * transfer - */ - - -static struct pipe_transfer * -trace_screen_get_tex_transfer(struct pipe_screen *_screen, - struct pipe_texture *_texture, - unsigned face, unsigned level, - unsigned zslice, - enum pipe_transfer_usage usage, - unsigned x, unsigned y, unsigned w, unsigned h) -{ - struct trace_screen *tr_scr = trace_screen(_screen); - struct trace_texture *tr_tex = trace_texture(_texture); - struct pipe_screen *screen = tr_scr->screen; - struct pipe_texture *texture = tr_tex->texture; - struct pipe_transfer *result = NULL; - - assert(texture->screen == screen); - - trace_dump_call_begin("pipe_screen", "get_tex_transfer"); - - trace_dump_arg(ptr, screen); - trace_dump_arg(ptr, texture); - trace_dump_arg(uint, face); - trace_dump_arg(uint, level); - trace_dump_arg(uint, zslice); - trace_dump_arg(uint, usage); - - trace_dump_arg(uint, x); - trace_dump_arg(uint, y); - trace_dump_arg(uint, w); - trace_dump_arg(uint, h); - - result = screen->get_tex_transfer(screen, texture, face, level, zslice, usage, - x, y, w, h); - - trace_dump_ret(ptr, result); - - trace_dump_call_end(); - - if (result) - result = trace_transfer_create(tr_tex, result); - - return result; -} - - -static void -trace_screen_tex_transfer_destroy(struct pipe_transfer *_transfer) -{ - struct trace_screen *tr_scr = trace_screen(_transfer->texture->screen); - struct trace_transfer *tr_trans = trace_transfer(_transfer); - struct pipe_screen *screen = tr_scr->screen; - struct pipe_transfer *transfer = tr_trans->transfer; - - trace_dump_call_begin("pipe_screen", "tex_transfer_destroy"); - - trace_dump_arg(ptr, screen); - trace_dump_arg(ptr, transfer); - - trace_dump_call_end(); - - trace_transfer_destroy(tr_trans); -} - - -static void * -trace_screen_transfer_map(struct pipe_screen *_screen, - struct pipe_transfer *_transfer) -{ - struct trace_screen *tr_scr = trace_screen(_screen); - struct trace_transfer *tr_trans = trace_transfer(_transfer); - struct pipe_screen *screen = tr_scr->screen; - struct pipe_transfer *transfer = tr_trans->transfer; - void *map; - - map = screen->transfer_map(screen, transfer); - if(map) { - if(transfer->usage & PIPE_TRANSFER_WRITE) { - assert(!tr_trans->map); - tr_trans->map = map; - } - } - - return map; -} - - -static void -trace_screen_transfer_unmap(struct pipe_screen *_screen, - struct pipe_transfer *_transfer) -{ - struct trace_screen *tr_scr = trace_screen(_screen); - struct trace_transfer *tr_trans = trace_transfer(_transfer); - struct pipe_screen *screen = tr_scr->screen; - struct pipe_transfer *transfer = tr_trans->transfer; - - if(tr_trans->map) { - size_t size = util_format_get_nblocksy(transfer->texture->format, transfer->height) * transfer->stride; - - trace_dump_call_begin("pipe_screen", "transfer_write"); - - trace_dump_arg(ptr, screen); - - trace_dump_arg(ptr, transfer); - - trace_dump_arg_begin("stride"); - trace_dump_uint(transfer->stride); - trace_dump_arg_end(); - - trace_dump_arg_begin("data"); - trace_dump_bytes(tr_trans->map, size); - trace_dump_arg_end(); - trace_dump_arg_begin("size"); - trace_dump_uint(size); - trace_dump_arg_end(); - - trace_dump_call_end(); - - tr_trans->map = NULL; - } - - screen->transfer_unmap(screen, transfer); -} /******************************************************************** @@ -901,10 +775,6 @@ trace_screen_create(struct pipe_screen *screen) tr_scr->base.texture_destroy = trace_screen_texture_destroy; tr_scr->base.get_tex_surface = trace_screen_get_tex_surface; tr_scr->base.tex_surface_destroy = trace_screen_tex_surface_destroy; - tr_scr->base.get_tex_transfer = trace_screen_get_tex_transfer; - tr_scr->base.tex_transfer_destroy = trace_screen_tex_transfer_destroy; - tr_scr->base.transfer_map = trace_screen_transfer_map; - tr_scr->base.transfer_unmap = trace_screen_transfer_unmap; tr_scr->base.buffer_create = trace_screen_buffer_create; tr_scr->base.user_buffer_create = trace_screen_user_buffer_create; if (screen->buffer_map) @@ -920,7 +790,11 @@ trace_screen_create(struct pipe_screen *screen) tr_scr->base.fence_signalled = trace_screen_fence_signalled; tr_scr->base.fence_finish = trace_screen_fence_finish; tr_scr->base.flush_frontbuffer = trace_screen_flush_frontbuffer; + tr_scr->screen = screen; + tr_scr->private_context = screen->context_create(screen, NULL); + if (tr_scr->private_context == NULL) + goto error3; trace_dump_ret(ptr, screen); trace_dump_call_end(); @@ -930,10 +804,8 @@ trace_screen_create(struct pipe_screen *screen) return &tr_scr->base; -#if 0 error3: FREE(tr_scr); -#endif error2: trace_dump_ret(ptr, screen); trace_dump_call_end(); diff --git a/src/gallium/drivers/trace/tr_screen.h b/src/gallium/drivers/trace/tr_screen.h index 597e2fc2650..9bfbe72e2c0 100644 --- a/src/gallium/drivers/trace/tr_screen.h +++ b/src/gallium/drivers/trace/tr_screen.h @@ -56,6 +56,7 @@ struct trace_screen struct pipe_screen base; struct pipe_screen *screen; + struct pipe_context *private_context; /* remote debugger */ struct trace_rbug *rbug; diff --git a/src/gallium/drivers/trace/tr_texture.c b/src/gallium/drivers/trace/tr_texture.c index 5321d68ec0c..d818e21bb82 100644 --- a/src/gallium/drivers/trace/tr_texture.c +++ b/src/gallium/drivers/trace/tr_texture.c @@ -31,6 +31,7 @@ #include "util/u_simple_list.h" #include "tr_screen.h" +#include "tr_context.h" #include "tr_texture.h" @@ -124,8 +125,9 @@ trace_surface_destroy(struct trace_surface *tr_surf) struct pipe_transfer * -trace_transfer_create(struct trace_texture *tr_tex, - struct pipe_transfer *transfer) +trace_transfer_create(struct trace_context *tr_ctx, + struct trace_texture *tr_tex, + struct pipe_transfer *transfer) { struct trace_screen *tr_scr = trace_screen(tr_tex->base.screen); struct trace_transfer *tr_trans; @@ -142,8 +144,9 @@ trace_transfer_create(struct trace_texture *tr_tex, memcpy(&tr_trans->base, transfer, sizeof(struct pipe_transfer)); tr_trans->base.texture = NULL; - pipe_texture_reference(&tr_trans->base.texture, &tr_tex->base); tr_trans->transfer = transfer; + + pipe_texture_reference(&tr_trans->base.texture, &tr_tex->base); assert(tr_trans->base.texture == &tr_tex->base); trace_screen_add_to_list(tr_scr, transfers, tr_trans); @@ -151,21 +154,23 @@ trace_transfer_create(struct trace_texture *tr_tex, return &tr_trans->base; error: - transfer->texture->screen->tex_transfer_destroy(transfer); + tr_ctx->pipe->tex_transfer_destroy(tr_ctx->pipe, transfer); return NULL; } void -trace_transfer_destroy(struct trace_transfer *tr_trans) +trace_transfer_destroy(struct trace_context *tr_context, + struct trace_transfer *tr_trans) { - struct trace_screen *tr_scr = trace_screen(tr_trans->base.texture->screen); - struct pipe_screen *screen = tr_trans->transfer->texture->screen; + struct trace_screen *tr_scr = trace_screen(tr_context->base.screen); + struct pipe_context *context = tr_context->pipe; + struct pipe_transfer *transfer = tr_trans->transfer; trace_screen_remove_from_list(tr_scr, transfers, tr_trans); pipe_texture_reference(&tr_trans->base.texture, NULL); - screen->tex_transfer_destroy(tr_trans->transfer); + context->tex_transfer_destroy(context, transfer); FREE(tr_trans); } diff --git a/src/gallium/drivers/trace/tr_texture.h b/src/gallium/drivers/trace/tr_texture.h index 3a99dcdaece..66250465e44 100644 --- a/src/gallium/drivers/trace/tr_texture.h +++ b/src/gallium/drivers/trace/tr_texture.h @@ -34,6 +34,7 @@ #include "tr_screen.h" +struct trace_context; struct trace_texture { @@ -68,6 +69,7 @@ struct trace_transfer struct pipe_transfer base; struct pipe_transfer *transfer; + struct pipe_context *pipe; struct tr_list list; @@ -129,11 +131,13 @@ void trace_surface_destroy(struct trace_surface *tr_surf); struct pipe_transfer * -trace_transfer_create(struct trace_texture *tr_tex, - struct pipe_transfer *transfer); +trace_transfer_create(struct trace_context *tr_ctx, + struct trace_texture *tr_tex, + struct pipe_transfer *transfer); void -trace_transfer_destroy(struct trace_transfer *tr_trans); +trace_transfer_destroy(struct trace_context *tr_ctx, + struct trace_transfer *tr_trans); #endif /* TR_TEXTURE_H_ */ diff --git a/src/gallium/include/pipe/p_context.h b/src/gallium/include/pipe/p_context.h index 17fad1fa243..d1b734a9f9a 100644 --- a/src/gallium/include/pipe/p_context.h +++ b/src/gallium/include/pipe/p_context.h @@ -316,6 +316,32 @@ struct pipe_context { void (*sampler_view_destroy)(struct pipe_context *ctx, struct pipe_sampler_view *view); + + + /** + * Get a transfer object for transferring data to/from a texture. + * + * Transfers are (by default) context-private and allow uploads to be + * interleaved with + */ + struct pipe_transfer *(*get_tex_transfer)(struct pipe_context *, + struct pipe_texture *texture, + unsigned face, unsigned level, + unsigned zslice, + enum pipe_transfer_usage usage, + unsigned x, unsigned y, + unsigned w, unsigned h); + + void (*tex_transfer_destroy)(struct pipe_context *, + struct pipe_transfer *); + + void *(*transfer_map)( struct pipe_context *, + struct pipe_transfer *transfer ); + + void (*transfer_unmap)( struct pipe_context *, + struct pipe_transfer *transfer ); + + }; diff --git a/src/gallium/include/pipe/p_screen.h b/src/gallium/include/pipe/p_screen.h index 690455f7222..b7cb83abbe5 100644 --- a/src/gallium/include/pipe/p_screen.h +++ b/src/gallium/include/pipe/p_screen.h @@ -142,23 +142,6 @@ struct pipe_screen { void (*tex_surface_destroy)(struct pipe_surface *); - /** Get a transfer object for transferring data to/from a texture */ - struct pipe_transfer *(*get_tex_transfer)(struct pipe_screen *, - struct pipe_texture *texture, - unsigned face, unsigned level, - unsigned zslice, - enum pipe_transfer_usage usage, - unsigned x, unsigned y, - unsigned w, unsigned h); - - void (*tex_transfer_destroy)(struct pipe_transfer *); - - void *(*transfer_map)( struct pipe_screen *, - struct pipe_transfer *transfer ); - - void (*transfer_unmap)( struct pipe_screen *, - struct pipe_transfer *transfer ); - /** * Create a new buffer. @@ -265,6 +248,7 @@ struct pipe_screen { /** * Do any special operations to ensure buffer size is correct + * \param context_private the private data of the calling context */ void (*update_buffer)( struct pipe_screen *ws, void *context_private ); @@ -272,10 +256,12 @@ struct pipe_screen { /** * Do any special operations to ensure frontbuffer contents are * displayed, eg copy fake frontbuffer. + * \param winsys_drawable_handle an opaque handle that the calling context + * gets out-of-band */ void (*flush_frontbuffer)( struct pipe_screen *screen, struct pipe_surface *surf, - void *context_private ); + void *winsys_drawable_handle ); diff --git a/src/gallium/state_trackers/dri/dri_extensions.c b/src/gallium/state_trackers/dri/dri_extensions.c index 1259813a412..800677a2d1e 100644 --- a/src/gallium/state_trackers/dri/dri_extensions.c +++ b/src/gallium/state_trackers/dri/dri_extensions.c @@ -33,110 +33,14 @@ #include "dri_context.h" #include "state_tracker/st_context.h" -#define need_GL_ARB_map_buffer_range -#define need_GL_ARB_multisample -#define need_GL_ARB_occlusion_query -#define need_GL_ARB_point_parameters -#define need_GL_ARB_provoking_vertex -#define need_GL_ARB_shader_objects -#define need_GL_ARB_texture_compression -#define need_GL_ARB_vertex_array_object -#define need_GL_ARB_vertex_buffer_object -#define need_GL_ARB_vertex_program -#define need_GL_ARB_vertex_shader -#define need_GL_ARB_window_pos -#define need_GL_EXT_blend_color -#define need_GL_EXT_blend_equation_separate -#define need_GL_EXT_blend_func_separate -#define need_GL_EXT_blend_minmax -#define need_GL_EXT_cull_vertex -#define need_GL_EXT_draw_buffers2 -#define need_GL_EXT_fog_coord -#define need_GL_EXT_framebuffer_object -#define need_GL_EXT_multi_draw_arrays -#define need_GL_EXT_provoking_vertex -#define need_GL_EXT_secondary_color -#define need_GL_EXT_stencil_two_side -#define need_GL_APPLE_vertex_array_object -#define need_GL_NV_vertex_program -#define need_GL_VERSION_2_0 -#define need_GL_VERSION_2_1 -#include "main/remap_helper.h" #include "utils.h" -/** - * Extension strings exported by the driver. - */ -static const struct dri_extension card_extensions[] = { - {"GL_ARB_fragment_shader", NULL}, - {"GL_ARB_map_buffer_range", GL_ARB_map_buffer_range_functions}, - {"GL_ARB_multisample", GL_ARB_multisample_functions}, - {"GL_ARB_multitexture", NULL}, - {"GL_ARB_occlusion_query", GL_ARB_occlusion_query_functions}, - {"GL_ARB_pixel_buffer_object", NULL}, - {"GL_ARB_provoking_vertex", GL_ARB_provoking_vertex_functions}, - {"GL_ARB_point_parameters", GL_ARB_point_parameters_functions}, - {"GL_ARB_shading_language_100", GL_VERSION_2_0_functions }, - {"GL_ARB_shading_language_120", GL_VERSION_2_1_functions }, - {"GL_ARB_shader_objects", GL_ARB_shader_objects_functions}, - {"GL_ARB_texture_border_clamp", NULL}, - {"GL_ARB_texture_compression", GL_ARB_texture_compression_functions}, - {"GL_ARB_texture_cube_map", NULL}, - {"GL_ARB_texture_env_add", NULL}, - {"GL_ARB_texture_env_combine", NULL}, - {"GL_ARB_texture_env_dot3", NULL}, - {"GL_ARB_texture_mirrored_repeat", NULL}, - {"GL_ARB_texture_non_power_of_two", NULL}, - {"GL_ARB_texture_rectangle", NULL}, - {"GL_ARB_vertex_array_object", GL_ARB_vertex_array_object_functions}, - {"GL_ARB_vertex_buffer_object", GL_ARB_vertex_buffer_object_functions}, - {"GL_ARB_vertex_shader", GL_ARB_vertex_shader_functions}, - {"GL_ARB_vertex_program", GL_ARB_vertex_program_functions}, - {"GL_ARB_window_pos", GL_ARB_window_pos_functions}, - {"GL_EXT_blend_color", GL_EXT_blend_color_functions}, - {"GL_EXT_blend_equation_separate", GL_EXT_blend_equation_separate_functions}, - {"GL_EXT_blend_func_separate", GL_EXT_blend_func_separate_functions}, - {"GL_EXT_blend_minmax", GL_EXT_blend_minmax_functions}, - {"GL_EXT_blend_subtract", NULL}, - {"GL_EXT_cull_vertex", GL_EXT_cull_vertex_functions}, - {"GL_EXT_draw_buffers2", GL_EXT_draw_buffers2_functions}, - {"GL_EXT_fog_coord", GL_EXT_fog_coord_functions}, - {"GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions}, - {"GL_EXT_multi_draw_arrays", GL_EXT_multi_draw_arrays_functions}, - {"GL_EXT_packed_depth_stencil", NULL}, - {"GL_EXT_pixel_buffer_object", NULL}, - {"GL_EXT_provoking_vertex", GL_EXT_provoking_vertex_functions}, - {"GL_EXT_secondary_color", GL_EXT_secondary_color_functions}, - {"GL_EXT_stencil_two_side", GL_EXT_stencil_two_side_functions}, - {"GL_EXT_stencil_wrap", NULL}, - {"GL_EXT_texture_edge_clamp", NULL}, - {"GL_EXT_texture_env_combine", NULL}, - {"GL_EXT_texture_env_dot3", NULL}, - {"GL_EXT_texture_filter_anisotropic", NULL}, - {"GL_EXT_texture_lod_bias", NULL}, - {"GL_3DFX_texture_compression_FXT1", NULL}, - {"GL_APPLE_client_storage", NULL}, - {"GL_APPLE_vertex_array_object", GL_APPLE_vertex_array_object_functions}, - {"GL_MESA_pack_invert", NULL}, - {"GL_MESA_ycbcr_texture", NULL}, - {"GL_NV_blend_square", NULL}, - {"GL_NV_vertex_program", GL_NV_vertex_program_functions}, - {"GL_NV_vertex_program1_1", NULL}, - {"GL_SGIS_generate_mipmap", NULL}, - {NULL, NULL} -}; - void dri_init_extensions(struct dri_context *ctx) { - /* The card_extensions list should be pruned according to the - * capabilities of the pipe_screen. This is actually something - * that can/should be done inside st_create_context(). - * XXX Not pruning is very bogus. Always all these extensions above - * will be advertized, regardless what st_init_extensions - * (which depends on the pipe cap bits) does. - */ - driInitExtensions(ctx->st->ctx, card_extensions, GL_TRUE); + /* New extensions should be added in mesa/state_tracker/st_extensions.c + * and not in this file. */ + driInitExtensions(ctx->st->ctx, NULL, GL_FALSE); } /* vim: set sw=3 ts=8 sts=3 expandtab: */ diff --git a/src/gallium/state_trackers/dri/dri_screen.c b/src/gallium/state_trackers/dri/dri_screen.c index 60bc560049c..7ccad8f5dd6 100644 --- a/src/gallium/state_trackers/dri/dri_screen.c +++ b/src/gallium/state_trackers/dri/dri_screen.c @@ -90,6 +90,9 @@ dri_fill_in_modes(struct dri_screen *screen, unsigned pixel_bits) { __DRIconfig **configs = NULL; + __DRIconfig **configs_r5g6b5 = NULL; + __DRIconfig **configs_a8r8g8b8 = NULL; + __DRIconfig **configs_x8r8g8b8 = NULL; unsigned num_modes; uint8_t depth_bits_array[5]; uint8_t stencil_bits_array[5]; @@ -127,25 +130,23 @@ dri_fill_in_modes(struct dri_screen *screen, pf_x8r8g8b8 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_B8G8R8X8_UNORM, PIPE_TEXTURE_2D, PIPE_TEXTURE_USAGE_RENDER_TARGET, 0); - - /* we support buffers with different depths only if we can tell the driver - * the actual depth of each of them. */ - if (screen->sPriv->dri2.loader - && (screen->sPriv->dri2.loader->base.version > 2) - && (screen->sPriv->dri2.loader->getBuffersWithFormat != NULL)) { + pf_r5g6b5 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_B5G6R5_UNORM, + PIPE_TEXTURE_2D, + PIPE_TEXTURE_USAGE_RENDER_TARGET, 0); + + /* We can only get a 16 or 32 bit depth buffer with getBuffersWithFormat */ + if (screen->sPriv->dri2.loader && + (screen->sPriv->dri2.loader->base.version > 2) && + (screen->sPriv->dri2.loader->getBuffersWithFormat != NULL)) { pf_z16 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_Z16_UNORM, PIPE_TEXTURE_2D, PIPE_TEXTURE_USAGE_DEPTH_STENCIL, 0); pf_z32 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_Z32_UNORM, PIPE_TEXTURE_2D, PIPE_TEXTURE_USAGE_DEPTH_STENCIL, 0); - pf_r5g6b5 = p_screen->is_format_supported(p_screen, PIPE_FORMAT_B5G6R5_UNORM, - PIPE_TEXTURE_2D, - PIPE_TEXTURE_USAGE_RENDER_TARGET, 0); } else { pf_z16 = FALSE; pf_z32 = FALSE; - pf_r5g6b5 = FALSE; } if (pf_z16) { @@ -175,46 +176,48 @@ dri_fill_in_modes(struct dri_screen *screen, num_modes = depth_buffer_factor * back_buffer_factor * msaa_samples_factor * 4; - if (pixel_bits == 16 && pf_r5g6b5) { - configs = driCreateConfigs(GL_RGB, GL_UNSIGNED_SHORT_5_6_5, - depth_bits_array, stencil_bits_array, - depth_buffer_factor, back_buffer_modes, - back_buffer_factor, - msaa_samples_array, msaa_samples_factor, - GL_TRUE); + if (pf_r5g6b5) + configs_r5g6b5 = driCreateConfigs(GL_RGB, GL_UNSIGNED_SHORT_5_6_5, + depth_bits_array, stencil_bits_array, + depth_buffer_factor, back_buffer_modes, + back_buffer_factor, + msaa_samples_array, msaa_samples_factor, + GL_TRUE); + + if (pf_a8r8g8b8) + configs_a8r8g8b8 = driCreateConfigs(GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV, + depth_bits_array, + stencil_bits_array, + depth_buffer_factor, + back_buffer_modes, + back_buffer_factor, + msaa_samples_array, + msaa_samples_factor, + GL_TRUE); + + if (pf_x8r8g8b8) + configs_x8r8g8b8 = driCreateConfigs(GL_BGR, GL_UNSIGNED_INT_8_8_8_8_REV, + depth_bits_array, + stencil_bits_array, + depth_buffer_factor, + back_buffer_modes, + back_buffer_factor, + msaa_samples_array, + msaa_samples_factor, + GL_TRUE); + + if (pixel_bits == 16) { + configs = configs_r5g6b5; + if (configs_a8r8g8b8) + configs = configs ? driConcatConfigs(configs, configs_a8r8g8b8) : configs_a8r8g8b8; + if (configs_x8r8g8b8) + configs = configs ? driConcatConfigs(configs, configs_x8r8g8b8) : configs_x8r8g8b8; } else { - __DRIconfig **configs_a8r8g8b8 = NULL; - __DRIconfig **configs_x8r8g8b8 = NULL; - - if (pf_a8r8g8b8) - configs_a8r8g8b8 = driCreateConfigs(GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV, - depth_bits_array, - stencil_bits_array, - depth_buffer_factor, - back_buffer_modes, - back_buffer_factor, - msaa_samples_array, - msaa_samples_factor, - GL_TRUE); - if (pf_x8r8g8b8) - configs_x8r8g8b8 = driCreateConfigs(GL_BGR, GL_UNSIGNED_INT_8_8_8_8_REV, - depth_bits_array, - stencil_bits_array, - depth_buffer_factor, - back_buffer_modes, - back_buffer_factor, - msaa_samples_array, - msaa_samples_factor, - GL_TRUE); - - if (configs_a8r8g8b8 && configs_x8r8g8b8) - configs = driConcatConfigs(configs_x8r8g8b8, configs_a8r8g8b8); - else if (configs_a8r8g8b8) - configs = configs_a8r8g8b8; - else if (configs_x8r8g8b8) - configs = configs_x8r8g8b8; - else - configs = NULL; + configs = configs_a8r8g8b8; + if (configs_x8r8g8b8) + configs = configs ? driConcatConfigs(configs, configs_x8r8g8b8) : configs_x8r8g8b8; + if (configs_r5g6b5) + configs = configs ? driConcatConfigs(configs, configs_r5g6b5) : configs_r5g6b5; } if (configs == NULL) { diff --git a/src/gallium/state_trackers/egl/x11/native_ximage.c b/src/gallium/state_trackers/egl/x11/native_ximage.c index e0d12acabe8..c6b16354f9b 100644 --- a/src/gallium/state_trackers/egl/x11/native_ximage.c +++ b/src/gallium/state_trackers/egl/x11/native_ximage.c @@ -82,6 +82,8 @@ struct ximage_surface { int width, height; struct ximage_buffer buffers[NUM_NATIVE_ATTACHMENTS]; uint valid_mask; + + struct pipe_surface *draw_surface; }; struct ximage_config { @@ -266,15 +268,19 @@ ximage_surface_draw_buffer(struct native_surface *nsurf, assert(xsurf->drawable && xbuf->texture); - /* what's the cost of surface creation? */ - psurf = screen->get_tex_surface(screen, - xbuf->texture, 0, 0, 0, PIPE_BUFFER_USAGE_CPU_READ); - if (!psurf) - return FALSE; + psurf = xsurf->draw_surface; + if (!psurf || psurf->texture != xbuf->texture) { + pipe_surface_reference(&xsurf->draw_surface, NULL); - screen->flush_frontbuffer(screen, psurf, &xbuf->xdraw); + psurf = screen->get_tex_surface(screen, + xbuf->texture, 0, 0, 0, PIPE_BUFFER_USAGE_CPU_READ); + if (!psurf) + return FALSE; - pipe_surface_reference(&psurf, NULL); + xsurf->draw_surface = psurf; + } + + screen->flush_frontbuffer(screen, psurf, &xbuf->xdraw); return TRUE; } @@ -371,6 +377,8 @@ ximage_surface_destroy(struct native_surface *nsurf) struct ximage_surface *xsurf = ximage_surface(nsurf); int i; + pipe_surface_reference(&xsurf->draw_surface, NULL); + for (i = 0; i < NUM_NATIVE_ATTACHMENTS; i++) ximage_surface_free_buffer(&xsurf->base, i); diff --git a/src/gallium/state_trackers/python/SConscript b/src/gallium/state_trackers/python/SConscript index 781f54bf2b9..d0d141fd249 100644 --- a/src/gallium/state_trackers/python/SConscript +++ b/src/gallium/state_trackers/python/SConscript @@ -3,7 +3,8 @@ import os.path Import('*') -if 'python' in env['statetrackers']: +if 'python' in env['statetrackers'] and 0: + # FIXME: Disable python state tracker until transfers are done by contexts env = env.Clone() diff --git a/src/gallium/state_trackers/vega/api_filters.c b/src/gallium/state_trackers/vega/api_filters.c index 02248ad4337..18e2cc1f250 100644 --- a/src/gallium/state_trackers/vega/api_filters.c +++ b/src/gallium/state_trackers/vega/api_filters.c @@ -78,14 +78,14 @@ static INLINE struct pipe_texture *create_texture_1d(struct vg_context *ctx, { /* upload color_data */ struct pipe_transfer *transfer = - screen->get_tex_transfer(screen, tex, - 0, 0, 0, - PIPE_TRANSFER_READ_WRITE , - 0, 0, tex->width0, tex->height0); - void *map = screen->transfer_map(screen, transfer); + pipe->get_tex_transfer(pipe, tex, + 0, 0, 0, + PIPE_TRANSFER_READ_WRITE , + 0, 0, tex->width0, tex->height0); + void *map = pipe->transfer_map(pipe, transfer); memcpy(map, color_data, sizeof(VGint)*color_data_len); - screen->transfer_unmap(screen, transfer); - screen->tex_transfer_destroy(transfer); + pipe->transfer_unmap(pipe, transfer); + pipe->tex_transfer_destroy(pipe, transfer); } return tex; diff --git a/src/gallium/state_trackers/vega/api_images.c b/src/gallium/state_trackers/vega/api_images.c index 015241498ed..fec473d9d23 100644 --- a/src/gallium/state_trackers/vega/api_images.c +++ b/src/gallium/state_trackers/vega/api_images.c @@ -397,7 +397,6 @@ void vgReadPixels(void * data, VGint dataStride, { struct vg_context *ctx = vg_current_context(); struct pipe_context *pipe = ctx->pipe; - struct pipe_screen *screen = pipe->screen; struct st_framebuffer *stfb = ctx->draw_buffer; struct st_renderbuffer *strb = stfb->strb; @@ -442,7 +441,7 @@ void vgReadPixels(void * data, VGint dataStride, { struct pipe_transfer *transfer; - transfer = screen->get_tex_transfer(screen, strb->texture, 0, 0, 0, + transfer = pipe->get_tex_transfer(pipe, strb->texture, 0, 0, 0, PIPE_TRANSFER_READ, 0, 0, width, height); @@ -451,14 +450,14 @@ void vgReadPixels(void * data, VGint dataStride, #if 0 debug_printf("%d-%d == %d\n", sy, height, y); #endif - pipe_get_tile_rgba(transfer, sx, y, width, 1, df); + pipe_get_tile_rgba(pipe, transfer, sx, y, width, 1, df); y += yStep; _vega_pack_rgba_span_float(ctx, width, temp, dataFormat, dst + yoffset + xoffset); dst += dataStride; } - screen->tex_transfer_destroy(transfer); + pipe->tex_transfer_destroy(pipe, transfer); } } diff --git a/src/gallium/state_trackers/vega/image.c b/src/gallium/state_trackers/vega/image.c index 41c979bfecf..a71579cd264 100644 --- a/src/gallium/state_trackers/vega/image.c +++ b/src/gallium/state_trackers/vega/image.c @@ -378,7 +378,7 @@ void image_sub_data(struct vg_image *image, VGfloat *df = (VGfloat*)temp; VGint i; struct vg_context *ctx = vg_current_context(); - struct pipe_screen *screen = ctx->pipe->screen; + struct pipe_context *pipe = ctx->pipe; struct pipe_texture *texture = image_texture(image); VGint xoffset = 0, yoffset = 0; @@ -412,17 +412,17 @@ void image_sub_data(struct vg_image *image, } { /* upload color_data */ - struct pipe_transfer *transfer = screen->get_tex_transfer( - screen, texture, 0, 0, 0, + struct pipe_transfer *transfer = pipe->get_tex_transfer( + pipe, texture, 0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0, texture->width0, texture->height0); src += (dataStride * yoffset); for (i = 0; i < height; i++) { _vega_unpack_float_span_rgba(ctx, width, xoffset, src, dataFormat, temp); - pipe_put_tile_rgba(transfer, x+image->x, y+image->y, width, 1, df); + pipe_put_tile_rgba(pipe, transfer, x+image->x, y+image->y, width, 1, df); y += yStep; src += dataStride; } - screen->tex_transfer_destroy(transfer); + pipe->tex_transfer_destroy(pipe, transfer); } } @@ -435,7 +435,6 @@ void image_get_sub_data(struct vg_image * image, { struct vg_context *ctx = vg_current_context(); struct pipe_context *pipe = ctx->pipe; - struct pipe_screen *screen = pipe->screen; VGfloat temp[VEGA_MAX_IMAGE_WIDTH][4]; VGfloat *df = (VGfloat*)temp; VGint y = 0, yStep = 1; @@ -444,7 +443,7 @@ void image_get_sub_data(struct vg_image * image, { struct pipe_transfer *transfer = - screen->get_tex_transfer(screen, + pipe->get_tex_transfer(pipe, image->texture, 0, 0, 0, PIPE_TRANSFER_READ, 0, 0, @@ -455,13 +454,13 @@ void image_get_sub_data(struct vg_image * image, #if 0 debug_printf("%d-%d == %d\n", sy, height, y); #endif - pipe_get_tile_rgba(transfer, sx+image->x, y, width, 1, df); + pipe_get_tile_rgba(pipe, transfer, sx+image->x, y, width, 1, df); y += yStep; _vega_pack_rgba_span_float(ctx, width, temp, dataFormat, dst); dst += dataStride; } - screen->tex_transfer_destroy(transfer); + pipe->tex_transfer_destroy(pipe, transfer); } } diff --git a/src/gallium/state_trackers/vega/paint.c b/src/gallium/state_trackers/vega/paint.c index cdb87d3bf68..dc56b8c5f3b 100644 --- a/src/gallium/state_trackers/vega/paint.c +++ b/src/gallium/state_trackers/vega/paint.c @@ -164,10 +164,10 @@ static INLINE struct pipe_texture *create_gradient_texture(struct vg_paint *p) struct pipe_transfer *transfer = st_no_flush_get_tex_transfer(p->base.ctx, tex, 0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0, 1024, 1); - void *map = screen->transfer_map(screen, transfer); + void *map = pipe->transfer_map(pipe, transfer); memcpy(map, p->gradient.color_data, sizeof(VGint)*1024); - screen->transfer_unmap(screen, transfer); - screen->tex_transfer_destroy(transfer); + pipe->transfer_unmap(pipe, transfer); + pipe->tex_transfer_destroy(pipe, transfer); } return tex; diff --git a/src/gallium/state_trackers/vega/st_inlines.h b/src/gallium/state_trackers/vega/st_inlines.h index 419151c3aee..4d12a4efdd6 100644 --- a/src/gallium/state_trackers/vega/st_inlines.h +++ b/src/gallium/state_trackers/vega/st_inlines.h @@ -51,7 +51,6 @@ st_cond_flush_get_tex_transfer(struct vg_context *st, unsigned int x, unsigned int y, unsigned int w, unsigned int h) { - struct pipe_screen *screen = st->pipe->screen; struct pipe_context *pipe = st->pipe; unsigned referenced = pipe->is_texture_referenced(pipe, pt, face, level); @@ -60,7 +59,7 @@ st_cond_flush_get_tex_transfer(struct vg_context *st, (usage & PIPE_TRANSFER_WRITE))) vgFlush(); - return screen->get_tex_transfer(screen, pt, face, level, zslice, usage, + return pipe->get_tex_transfer(pipe, pt, face, level, zslice, usage, x, y, w, h); } @@ -74,10 +73,10 @@ st_no_flush_get_tex_transfer(struct vg_context *st, unsigned int x, unsigned int y, unsigned int w, unsigned int h) { - struct pipe_screen *screen = st->pipe->screen; + struct pipe_context *pipe = st->pipe; - return screen->get_tex_transfer(screen, pt, face, level, - zslice, usage, x, y, w, h); + return pipe->get_tex_transfer(pipe, pt, face, level, + zslice, usage, x, y, w, h); } static INLINE void * diff --git a/src/gallium/state_trackers/vega/vg_tracker.c b/src/gallium/state_trackers/vega/vg_tracker.c index a002e50faf7..ea5c2ce41f6 100644 --- a/src/gallium/state_trackers/vega/vg_tracker.c +++ b/src/gallium/state_trackers/vega/vg_tracker.c @@ -380,16 +380,8 @@ boolean st_make_current(struct vg_context *st, void *winsys_drawable_handle) { vg_set_current_context(st); - if (st) { + if (st) st->draw_buffer = draw; - - /* VG state tracker doesn't seem to do front-buffer rendering - * (no calls to flush_frontbuffer). If it ever did start doing - * that, it would need to pass this value down in the - * flush_frontbuffer call: - */ - st->pipe->priv = winsys_drawable_handle; - } return VG_TRUE; } diff --git a/src/gallium/state_trackers/wgl/stw_framebuffer.h b/src/gallium/state_trackers/wgl/stw_framebuffer.h index 08cc4973bce..e61e9bf9c26 100644 --- a/src/gallium/state_trackers/wgl/stw_framebuffer.h +++ b/src/gallium/state_trackers/wgl/stw_framebuffer.h @@ -45,7 +45,7 @@ struct stw_framebuffer /** * This mutex has two purposes: * - protect the access to the mutable data members below - * - prevent the the framebuffer from being deleted while being accessed. + * - prevent the framebuffer from being deleted while being accessed. * * It is OK to lock this mutex while holding the stw_device::fb_mutex lock, * but the opposite must never happen. diff --git a/src/gallium/state_trackers/xorg/xorg_composite.c b/src/gallium/state_trackers/xorg/xorg_composite.c index c50873c1508..715a5e7b943 100644 --- a/src/gallium/state_trackers/xorg/xorg_composite.c +++ b/src/gallium/state_trackers/xorg/xorg_composite.c @@ -359,14 +359,6 @@ bind_samplers(struct exa_context *exa, int op, exa->num_bound_samplers = 0; -#if 0 - if ((pSrc && (exa->pipe->is_texture_referenced(exa->pipe, pSrc->tex, 0, 0) & - PIPE_REFERENCED_FOR_WRITE)) || - (pMask && (exa->pipe->is_texture_referenced(exa->pipe, pMask->tex, 0, 0) & - PIPE_REFERENCED_FOR_WRITE))) - xorg_exa_flush(exa, PIPE_FLUSH_RENDER_CACHE, NULL); -#endif - memset(&src_sampler, 0, sizeof(struct pipe_sampler_state)); memset(&mask_sampler, 0, sizeof(struct pipe_sampler_state)); diff --git a/src/gallium/state_trackers/xorg/xorg_crtc.c b/src/gallium/state_trackers/xorg/xorg_crtc.c index 44f7da0f960..eef428232b0 100644 --- a/src/gallium/state_trackers/xorg/xorg_crtc.c +++ b/src/gallium/state_trackers/xorg/xorg_crtc.c @@ -39,6 +39,7 @@ #include <xf86.h> #include <xf86i2c.h> #include <xf86Crtc.h> +#include <cursorstr.h> #include "xorg_tracker.h" #include "xf86Modes.h" @@ -219,16 +220,16 @@ crtc_load_cursor_argb_ga3d(xf86CrtcPtr crtc, CARD32 * image) crtcp->cursor_handle = whandle.handle; } - transfer = ms->screen->get_tex_transfer(ms->screen, crtcp->cursor_tex, - 0, 0, 0, - PIPE_TRANSFER_WRITE, - 0, 0, 64, 64); - ptr = ms->screen->transfer_map(ms->screen, transfer); + transfer = ms->ctx->get_tex_transfer(ms->ctx, crtcp->cursor_tex, + 0, 0, 0, + PIPE_TRANSFER_WRITE, + 0, 0, 64, 64); + ptr = ms->ctx->transfer_map(ms->ctx, transfer); util_copy_rect(ptr, crtcp->cursor_tex->format, transfer->stride, 0, 0, 64, 64, (void*)image, 64 * 4, 0, 0); - ms->screen->transfer_unmap(ms->screen, transfer); - ms->screen->tex_transfer_destroy(transfer); + ms->ctx->transfer_unmap(ms->ctx, transfer); + ms->ctx->tex_transfer_destroy(ms->ctx, transfer); } #if HAVE_LIBKMS @@ -276,7 +277,21 @@ err_bo_destroy: static void crtc_load_cursor_argb(xf86CrtcPtr crtc, CARD32 * image) { + xf86CrtcConfigPtr config = XF86_CRTC_CONFIG_PTR(crtc->scrn); modesettingPtr ms = modesettingPTR(crtc->scrn); + + /* Older X servers have cursor reference counting bugs leading to use of + * freed memory and consequently random crashes. Should be fixed as of + * xserver 1.8, but this workaround shouldn't hurt anyway. + */ + if (config->cursor) + config->cursor->refcnt++; + + if (ms->cursor) + FreeCursor(ms->cursor, None); + + ms->cursor = config->cursor; + if (ms->screen) crtc_load_cursor_argb_ga3d(crtc, image); #ifdef HAVE_LIBKMS diff --git a/src/gallium/state_trackers/xorg/xorg_driver.c b/src/gallium/state_trackers/xorg/xorg_driver.c index 004a28f00e1..8ac5179545a 100644 --- a/src/gallium/state_trackers/xorg/xorg_driver.c +++ b/src/gallium/state_trackers/xorg/xorg_driver.c @@ -922,6 +922,11 @@ drv_close_screen(int scrnIndex, ScreenPtr pScreen) drv_leave_vt(scrnIndex, 0); } + if (ms->cursor) { + FreeCursor(ms->cursor, None); + ms->cursor = NULL; + } + if (cust && cust->winsys_screen_close) cust->winsys_screen_close(cust); diff --git a/src/gallium/state_trackers/xorg/xorg_exa.c b/src/gallium/state_trackers/xorg/xorg_exa.c index a7ffe3f4992..bdec0e254fa 100644 --- a/src/gallium/state_trackers/xorg/xorg_exa.c +++ b/src/gallium/state_trackers/xorg/xorg_exa.c @@ -188,11 +188,7 @@ ExaDownloadFromScreen(PixmapPtr pPix, int x, int y, int w, int h, char *dst, if (!priv || !priv->tex) return FALSE; - if (exa->pipe->is_texture_referenced(exa->pipe, priv->tex, 0, 0) & - PIPE_REFERENCED_FOR_WRITE) - exa->pipe->flush(exa->pipe, 0, NULL); - - transfer = exa->scrn->get_tex_transfer(exa->scrn, priv->tex, 0, 0, 0, + transfer = exa->pipe->get_tex_transfer(exa->pipe, priv->tex, 0, 0, 0, PIPE_TRANSFER_READ, x, y, w, h); if (!transfer) return FALSE; @@ -203,11 +199,11 @@ ExaDownloadFromScreen(PixmapPtr pPix, int x, int y, int w, int h, char *dst, #endif util_copy_rect((unsigned char*)dst, priv->tex->format, dst_pitch, 0, 0, - w, h, exa->scrn->transfer_map(exa->scrn, transfer), + w, h, exa->pipe->transfer_map(exa->pipe, transfer), transfer->stride, 0, 0); - exa->scrn->transfer_unmap(exa->scrn, transfer); - exa->scrn->tex_transfer_destroy(transfer); + exa->pipe->transfer_unmap(exa->pipe, transfer); + exa->pipe->tex_transfer_destroy(exa->pipe, transfer); return TRUE; } @@ -226,12 +222,7 @@ ExaUploadToScreen(PixmapPtr pPix, int x, int y, int w, int h, char *src, if (!priv || !priv->tex) return FALSE; - /* make sure that any pending operations are flushed to hardware */ - if (exa->pipe->is_texture_referenced(exa->pipe, priv->tex, 0, 0) & - (PIPE_REFERENCED_FOR_READ | PIPE_REFERENCED_FOR_WRITE)) - xorg_exa_flush(exa, 0, NULL); - - transfer = exa->scrn->get_tex_transfer(exa->scrn, priv->tex, 0, 0, 0, + transfer = exa->pipe->get_tex_transfer(exa->pipe, priv->tex, 0, 0, 0, PIPE_TRANSFER_WRITE, x, y, w, h); if (!transfer) return FALSE; @@ -241,12 +232,12 @@ ExaUploadToScreen(PixmapPtr pPix, int x, int y, int w, int h, char *src, x, y, w, h, src_pitch); #endif - util_copy_rect(exa->scrn->transfer_map(exa->scrn, transfer), + util_copy_rect(exa->pipe->transfer_map(exa->pipe, transfer), priv->tex->format, transfer->stride, 0, 0, w, h, (unsigned char*)src, src_pitch, 0, 0); - exa->scrn->transfer_unmap(exa->scrn, transfer); - exa->scrn->tex_transfer_destroy(transfer); + exa->pipe->transfer_unmap(exa->pipe, transfer); + exa->pipe->tex_transfer_destroy(exa->pipe, transfer); return TRUE; } @@ -270,15 +261,11 @@ ExaPrepareAccess(PixmapPtr pPix, int index) if (priv->map_count == 0) { - if (exa->pipe->is_texture_referenced(exa->pipe, priv->tex, 0, 0) & - PIPE_REFERENCED_FOR_WRITE) - exa->pipe->flush(exa->pipe, 0, NULL); - assert(pPix->drawable.width <= priv->tex->width0); assert(pPix->drawable.height <= priv->tex->height0); priv->map_transfer = - exa->scrn->get_tex_transfer(exa->scrn, priv->tex, 0, 0, 0, + exa->pipe->get_tex_transfer(exa->pipe, priv->tex, 0, 0, 0, #ifdef EXA_MIXED_PIXMAPS PIPE_TRANSFER_MAP_DIRECTLY | #endif @@ -294,7 +281,7 @@ ExaPrepareAccess(PixmapPtr pPix, int index) #endif pPix->devPrivate.ptr = - exa->scrn->transfer_map(exa->scrn, priv->map_transfer); + exa->pipe->transfer_map(exa->pipe, priv->map_transfer); pPix->devKind = priv->map_transfer->stride; } @@ -321,8 +308,8 @@ ExaFinishAccess(PixmapPtr pPix, int index) if (--priv->map_count == 0) { assert(priv->map_transfer); - exa->scrn->transfer_unmap(exa->scrn, priv->map_transfer); - exa->scrn->tex_transfer_destroy(priv->map_transfer); + exa->pipe->transfer_unmap(exa->pipe, priv->map_transfer); + exa->pipe->tex_transfer_destroy(exa->pipe, priv->map_transfer); priv->map_transfer = NULL; pPix->devPrivate.ptr = NULL; } diff --git a/src/gallium/state_trackers/xorg/xorg_tracker.h b/src/gallium/state_trackers/xorg/xorg_tracker.h index 2f5cc64d9c5..c1884ebd115 100644 --- a/src/gallium/state_trackers/xorg/xorg_tracker.h +++ b/src/gallium/state_trackers/xorg/xorg_tracker.h @@ -90,6 +90,7 @@ typedef struct _modesettingRec Bool noAccel; Bool SWCursor; + CursorPtr cursor; CloseScreenProcPtr CloseScreen; /* Broken-out options. */ diff --git a/src/gallium/state_trackers/xorg/xorg_xv.c b/src/gallium/state_trackers/xorg/xorg_xv.c index e37a1c39596..5a195cb482d 100644 --- a/src/gallium/state_trackers/xorg/xorg_xv.c +++ b/src/gallium/state_trackers/xorg/xorg_xv.c @@ -275,28 +275,28 @@ copy_packed_data(ScrnInfoPtr pScrn, int i, j; struct pipe_texture **dst = port->yuv[port->current_set]; struct pipe_transfer *ytrans, *utrans, *vtrans; - struct pipe_screen *screen = port->r->pipe->screen; + struct pipe_context *pipe = port->r->pipe; char *ymap, *vmap, *umap; unsigned char y1, y2, u, v; int yidx, uidx, vidx; int y_array_size = w * h; - ytrans = screen->get_tex_transfer(screen, dst[0], - 0, 0, 0, - PIPE_TRANSFER_WRITE, - left, top, w, h); - utrans = screen->get_tex_transfer(screen, dst[1], - 0, 0, 0, - PIPE_TRANSFER_WRITE, - left, top, w, h); - vtrans = screen->get_tex_transfer(screen, dst[2], - 0, 0, 0, - PIPE_TRANSFER_WRITE, - left, top, w, h); - - ymap = (char*)screen->transfer_map(screen, ytrans); - umap = (char*)screen->transfer_map(screen, utrans); - vmap = (char*)screen->transfer_map(screen, vtrans); + ytrans = pipe->get_tex_transfer(pipe, dst[0], + 0, 0, 0, + PIPE_TRANSFER_WRITE, + left, top, w, h); + utrans = pipe->get_tex_transfer(pipe, dst[1], + 0, 0, 0, + PIPE_TRANSFER_WRITE, + left, top, w, h); + vtrans = pipe->get_tex_transfer(pipe, dst[2], + 0, 0, 0, + PIPE_TRANSFER_WRITE, + left, top, w, h); + + ymap = (char*)pipe->transfer_map(pipe, ytrans); + umap = (char*)pipe->transfer_map(pipe, utrans); + vmap = (char*)pipe->transfer_map(pipe, vtrans); yidx = uidx = vidx = 0; @@ -362,12 +362,12 @@ copy_packed_data(ScrnInfoPtr pScrn, break; } - screen->transfer_unmap(screen, ytrans); - screen->transfer_unmap(screen, utrans); - screen->transfer_unmap(screen, vtrans); - screen->tex_transfer_destroy(ytrans); - screen->tex_transfer_destroy(utrans); - screen->tex_transfer_destroy(vtrans); + pipe->transfer_unmap(pipe, ytrans); + pipe->transfer_unmap(pipe, utrans); + pipe->transfer_unmap(pipe, vtrans); + pipe->tex_transfer_destroy(pipe, ytrans); + pipe->tex_transfer_destroy(pipe, utrans); + pipe->tex_transfer_destroy(pipe, vtrans); } diff --git a/src/gallium/winsys/drm/vmware/dri/SConscript b/src/gallium/winsys/drm/vmware/dri/SConscript index 84319f91ff1..d26d0cd7483 100644 --- a/src/gallium/winsys/drm/vmware/dri/SConscript +++ b/src/gallium/winsys/drm/vmware/dri/SConscript @@ -48,6 +48,7 @@ if env['platform'] == 'linux': svgadrm, svga, mesa, + glsl, gallium, ]) diff --git a/src/gallium/winsys/xlib/xlib_sw_winsys.c b/src/gallium/winsys/xlib/xlib_sw_winsys.c index 29c642706a4..cecfa4a53d4 100644 --- a/src/gallium/winsys/xlib/xlib_sw_winsys.c +++ b/src/gallium/winsys/xlib/xlib_sw_winsys.c @@ -363,7 +363,7 @@ xm_displaytarget_create(struct sw_winsys *winsys, unsigned alignment, unsigned *stride) { - struct xm_displaytarget *xm_dt = CALLOC_STRUCT(xm_displaytarget); + struct xm_displaytarget *xm_dt; unsigned nblocksy, size; xm_dt = CALLOC_STRUCT(xm_displaytarget); diff --git a/src/glu/sgi/libnurbs/nurbtess/partitionY.h b/src/glu/sgi/libnurbs/nurbtess/partitionY.h index 8dda409de1a..5570c183d79 100644 --- a/src/glu/sgi/libnurbs/nurbtess/partitionY.h +++ b/src/glu/sgi/libnurbs/nurbtess/partitionY.h @@ -39,7 +39,7 @@ *or both at or below v. In addition, at least one of the ajacent verteces is *strictly below or above v. * A vertex is a relex vertex if the internals angle is strictly greater than - *180. In other words, if the the signed area is negative: + *180. In other words, if the signed area is negative: *(x1, y1), (x2, y2), (x3, y3) are the three vertices along a polygon, the *order is such that left hand side is inside the polygon. Then (x2,y2) is *reflex if: diff --git a/src/glu/sgi/libtess/normal.c b/src/glu/sgi/libtess/normal.c index 0a2494be343..7ab83167bbd 100644 --- a/src/glu/sgi/libtess/normal.c +++ b/src/glu/sgi/libtess/normal.c @@ -142,7 +142,7 @@ static void CheckOrientation( GLUtesselator *tess ) GLUhalfEdge *e; /* When we compute the normal automatically, we choose the orientation - * so that the the sum of the signed areas of all contours is non-negative. + * so that the sum of the signed areas of all contours is non-negative. */ area = 0; for( f = fHead->next; f != fHead; f = f->next ) { diff --git a/src/glx/single2.c b/src/glx/single2.c index 9ecf589ffff..a1461956b99 100644 --- a/src/glx/single2.c +++ b/src/glx/single2.c @@ -44,6 +44,9 @@ #include <X11/Xlib-xcb.h> #endif /* USE_XCB */ +#if !defined(__GNUC__) +# define __builtin_expect(x, y) x +#endif /* Used for GL_ARB_transpose_matrix */ static void diff --git a/src/mesa/drivers/dri/common/dri_util.c b/src/mesa/drivers/dri/common/dri_util.c index 890ae513397..75c98825b79 100644 --- a/src/mesa/drivers/dri/common/dri_util.c +++ b/src/mesa/drivers/dri/common/dri_util.c @@ -698,7 +698,7 @@ setupLoaderExtensions(__DRIscreen *psp, * \param drm_version Version of the kernel DRM. * \param frame_buffer Data describing the location and layout of the * framebuffer. - * \param pSAREA Pointer the the SAREA. + * \param pSAREA Pointer to the SAREA. * \param fd Device handle for the DRM. * \param extensions ?? * \param driver_modes Returns modes suppoted by the driver diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile index a242580273f..842d4b7aa10 100644 --- a/src/mesa/drivers/dri/i965/Makefile +++ b/src/mesa/drivers/dri/i965/Makefile @@ -54,6 +54,7 @@ DRIVER_SOURCES = \ brw_gs_emit.c \ brw_gs_state.c \ brw_misc_state.c \ + brw_optimize.c \ brw_program.c \ brw_queryobj.c \ brw_sf.c \ diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c index a512896f315..241193c3579 100644 --- a/src/mesa/drivers/dri/i965/brw_context.c +++ b/src/mesa/drivers/dri/i965/brw_context.c @@ -156,6 +156,7 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis, brw->has_surface_tile_offset = GL_TRUE; brw->has_compr4 = GL_TRUE; brw->has_aa_line_parameters = GL_TRUE; + brw->has_pln = GL_TRUE; } else { brw->CMD_VF_STATISTICS = CMD_VF_STATISTICS_965; brw->CMD_PIPELINE_SELECT = CMD_PIPELINE_SELECT_965; diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h index d6fc37e4d89..2855c93ea66 100644 --- a/src/mesa/drivers/dri/i965/brw_context.h +++ b/src/mesa/drivers/dri/i965/brw_context.h @@ -446,6 +446,7 @@ struct brw_context GLboolean has_compr4; GLboolean has_negative_rhw_bug; GLboolean has_aa_line_parameters; + GLboolean has_pln; ; struct { struct brw_state_flags dirty; diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h index bb1b5f5ef03..984e56d00c8 100644 --- a/src/mesa/drivers/dri/i965/brw_defines.h +++ b/src/mesa/drivers/dri/i965/brw_defines.h @@ -550,6 +550,7 @@ #define BRW_OPCODE_DP2 87 #define BRW_OPCODE_DPA2 88 #define BRW_OPCODE_LINE 89 +#define BRW_OPCODE_PLN 90 #define BRW_OPCODE_NOP 126 #define BRW_PREDICATE_NONE 0 diff --git a/src/mesa/drivers/dri/i965/brw_disasm.c b/src/mesa/drivers/dri/i965/brw_disasm.c index 54699cf8d34..ad61770212c 100644 --- a/src/mesa/drivers/dri/i965/brw_disasm.c +++ b/src/mesa/drivers/dri/i965/brw_disasm.c @@ -50,6 +50,7 @@ struct { [BRW_OPCODE_MAC] = { .name = "mac", .nsrc = 2, .ndst = 1 }, [BRW_OPCODE_MACH] = { .name = "mach", .nsrc = 2, .ndst = 1 }, [BRW_OPCODE_LINE] = { .name = "line", .nsrc = 2, .ndst = 1 }, + [BRW_OPCODE_PLN] = { .name = "pln", .nsrc = 2, .ndst = 1 }, [BRW_OPCODE_SAD2] = { .name = "sad2", .nsrc = 2, .ndst = 1 }, [BRW_OPCODE_SADA2] = { .name = "sada2", .nsrc = 2, .ndst = 1 }, [BRW_OPCODE_DP4] = { .name = "dp4", .nsrc = 2, .ndst = 1 }, @@ -73,7 +74,7 @@ struct { [BRW_OPCODE_NOP] = { .name = "nop", .nsrc = 0, .ndst = 0 }, [BRW_OPCODE_JMPI] = { .name = "jmpi", .nsrc = 1, .ndst = 0 }, [BRW_OPCODE_IF] = { .name = "if", .nsrc = 2, .ndst = 0 }, - [BRW_OPCODE_IFF] = { .name = "iff", .nsrc = 1, .ndst = 01 }, + [BRW_OPCODE_IFF] = { .name = "iff", .nsrc = 2, .ndst = 1 }, [BRW_OPCODE_WHILE] = { .name = "while", .nsrc = 2, .ndst = 0 }, [BRW_OPCODE_ELSE] = { .name = "else", .nsrc = 2, .ndst = 0 }, [BRW_OPCODE_BREAK] = { .name = "break", .nsrc = 2, .ndst = 0 }, diff --git a/src/mesa/drivers/dri/i965/brw_eu.h b/src/mesa/drivers/dri/i965/brw_eu.h index 39eb88d7c2b..4f55158e8f3 100644 --- a/src/mesa/drivers/dri/i965/brw_eu.h +++ b/src/mesa/drivers/dri/i965/brw_eu.h @@ -795,6 +795,7 @@ ALU2(DPH) ALU2(DP3) ALU2(DP2) ALU2(LINE) +ALU2(PLN) #undef ALU1 #undef ALU2 @@ -965,4 +966,9 @@ void brw_math_invert( struct brw_compile *p, void brw_set_src1( struct brw_instruction *insn, struct brw_reg reg ); + + +/* brw_optimize.c */ +void brw_optimize(struct brw_compile *p); + #endif diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c index f69d5296137..d2395dec288 100644 --- a/src/mesa/drivers/dri/i965/brw_eu_emit.c +++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c @@ -573,7 +573,7 @@ ALU2(DPH) ALU2(DP3) ALU2(DP2) ALU2(LINE) - +ALU2(PLN) @@ -1290,7 +1290,7 @@ void brw_SAMPLE(struct brw_compile *p, GLuint simd_mode) { GLboolean need_stall = 0; - + if (writemask == 0) { /*printf("%s: zero writemask??\n", __FUNCTION__); */ return; @@ -1327,8 +1327,14 @@ void brw_SAMPLE(struct brw_compile *p, /* printf("need stall %x %x\n", newmask , writemask); */ } else { + GLboolean dispatch_16 = GL_FALSE; + struct brw_reg m1 = brw_message_reg(msg_reg_nr); - + + guess_execution_size(p->current, dest); + if (p->current->header.execution_size == BRW_EXECUTE_16) + dispatch_16 = GL_TRUE; + newmask = ~newmask & WRITEMASK_XYZW; brw_push_insn_state(p); @@ -1343,7 +1349,13 @@ void brw_SAMPLE(struct brw_compile *p, src0 = retype(brw_null_reg(), BRW_REGISTER_TYPE_UW); dest = offset(dest, dst_offset); - response_length = len * 2; + + /* For 16-wide dispatch, masked channels are skipped in the + * response. For 8-wide, masked channels still take up slots, + * and are just not written to. + */ + if (dispatch_16) + response_length = len * 2; } } diff --git a/src/mesa/drivers/dri/i965/brw_optimize.c b/src/mesa/drivers/dri/i965/brw_optimize.c new file mode 100644 index 00000000000..57df9ea1151 --- /dev/null +++ b/src/mesa/drivers/dri/i965/brw_optimize.c @@ -0,0 +1,115 @@ +/* + * Copyright © 2010 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + * Authors: + * Eric Anholt <eric@anholt.net> + * + */ + +#include "main/macros.h" +#include "shader/program.h" +#include "shader/prog_parameter.h" +#include "shader/prog_print.h" +#include "brw_context.h" +#include "brw_defines.h" +#include "brw_eu.h" + +static GLboolean +is_single_channel_dp4(struct brw_instruction *insn) +{ + if (insn->header.opcode != BRW_OPCODE_DP4 || + insn->header.execution_size != BRW_EXECUTE_8 || + insn->header.access_mode != BRW_ALIGN_16 || + insn->bits1.da1.dest_reg_file != BRW_GENERAL_REGISTER_FILE) + return GL_FALSE; + + if (!is_power_of_two(insn->bits1.da16.dest_writemask)) + return GL_FALSE; + + return GL_TRUE; +} + +/** + * Sets the dependency control fields on DP4 instructions. + * + * The hardware only tracks dependencies on a register basis, so when + * you do: + * + * DP4 dst.x src1 src2 + * DP4 dst.y src1 src3 + * DP4 dst.z src1 src4 + * DP4 dst.w src1 src5 + * + * It will wait to do the DP4 dst.y until the dst.x is resolved, etc. + * We can examine our instruction stream and set the dependency + * control fields to tell the hardware when to do it. + * + * We may want to extend this to other instructions that are used to + * fill in a channel at a time of the destination register. + */ +static void +brw_set_dp4_dependency_control(struct brw_compile *p) +{ + int i; + + for (i = 1; i < p->nr_insn; i++) { + struct brw_instruction *insn = &p->store[i]; + struct brw_instruction *prev = &p->store[i - 1]; + + if (!is_single_channel_dp4(prev)) + continue; + + if (!is_single_channel_dp4(insn)) { + i++; + continue; + } + + /* Only avoid hw dep control if the write masks are different + * channels of one reg. + */ + if (insn->bits1.da16.dest_writemask == prev->bits1.da16.dest_writemask) + continue; + if (insn->bits1.da16.dest_reg_nr != prev->bits1.da16.dest_reg_nr) + continue; + + /* Check if the second instruction depends on the previous one + * for a src. + */ + if (insn->bits1.da1.src0_reg_file == BRW_GENERAL_REGISTER_FILE && + (insn->bits2.da1.src0_address_mode != BRW_ADDRESS_DIRECT || + insn->bits2.da1.src0_reg_nr == insn->bits1.da16.dest_reg_nr)) + continue; + if (insn->bits1.da1.src1_reg_file == BRW_GENERAL_REGISTER_FILE && + (insn->bits3.da1.src1_address_mode != BRW_ADDRESS_DIRECT || + insn->bits3.da1.src1_reg_nr == insn->bits1.da16.dest_reg_nr)) + continue; + + prev->header.dependency_control |= BRW_DEPENDENCY_NOTCLEARED; + insn->header.dependency_control |= BRW_DEPENDENCY_NOTCHECKED; + } +} + +void +brw_optimize(struct brw_compile *p) +{ + brw_set_dp4_dependency_control(p); +} diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c index a48804a660f..d16e916832e 100644 --- a/src/mesa/drivers/dri/i965/brw_vs_emit.c +++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c @@ -384,9 +384,8 @@ static void emit_sop( struct brw_vs_compile *c, { struct brw_compile *p = &c->func; - brw_MOV(p, dst, brw_imm_f(0.0f)); - brw_CMP(p, brw_null_reg(), cond, arg0, arg1); - brw_MOV(p, dst, brw_imm_f(1.0f)); + brw_CMP(p, brw_null_reg(), cond, arg1, arg0); + brw_SEL(p, dst, brw_null_reg(), brw_imm_f(1.0f)); brw_set_predicate_control_flag_value(p, 0xff); } @@ -1825,6 +1824,8 @@ void brw_vs_emit(struct brw_vs_compile *c ) post_vs_emit(c, end_inst, last_inst); + brw_optimize(p); + if (INTEL_DEBUG & DEBUG_VS) { int i; diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h index 88d84ee82fe..47b764d24d1 100644 --- a/src/mesa/drivers/dri/i965/brw_wm.h +++ b/src/mesa/drivers/dri/i965/brw_wm.h @@ -328,6 +328,12 @@ void emit_cinterp(struct brw_compile *p, const struct brw_reg *dst, GLuint mask, const struct brw_reg *arg0); +void emit_cmp(struct brw_compile *p, + const struct brw_reg *dst, + GLuint mask, + const struct brw_reg *arg0, + const struct brw_reg *arg1, + const struct brw_reg *arg2); void emit_ddxy(struct brw_compile *p, const struct brw_reg *dst, GLuint mask, diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c index 9315bca3156..05e464d4b61 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_emit.c +++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c @@ -34,6 +34,23 @@ #include "brw_context.h" #include "brw_wm.h" +static GLboolean can_do_pln(struct intel_context *intel, + const struct brw_reg *deltas) +{ + struct brw_context *brw = brw_context(&intel->ctx); + + if (!brw->has_pln) + return GL_FALSE; + + if (deltas[1].nr != deltas[0].nr + 1) + return GL_FALSE; + + if (intel->gen < 6 && ((deltas[0].nr & 1) != 0)) + return GL_FALSE; + + return GL_TRUE; +} + /* Not quite sure how correct this is - need to understand horiz * vs. vertical strides a little better. */ @@ -45,7 +62,13 @@ static INLINE struct brw_reg sechalf( struct brw_reg reg ) } -/* Payload R0: +/** + * Computes the screen-space x,y position of the pixels. + * + * This will be used by emit_delta_xy() or emit_wpos_xy() for + * interpolation of attributes.. + * + * Payload R0: * * R0.0 -- pixel mask, one bit for each of 4 pixels in 4 tiles, * corresponding to each of the 16 execution channels. @@ -60,7 +83,6 @@ static INLINE struct brw_reg sechalf( struct brw_reg reg ) * R1.7 -- ? * R1.8 -- ? */ - void emit_pixel_xy(struct brw_wm_compile *c, const struct brw_reg *dst, GLuint mask) @@ -100,7 +122,14 @@ void emit_pixel_xy(struct brw_wm_compile *c, brw_pop_insn_state(p); } - +/** + * Computes the screen-space x,y distance of the pixels from the start + * vertex. + * + * This will be used in linterp or pinterp with the start vertex value + * and the Cx, Cy, and C0 coefficients passed in from the setup engine + * to produce interpolated attribute values. + */ void emit_delta_xy(struct brw_compile *p, const struct brw_reg *dst, GLuint mask, @@ -108,25 +137,27 @@ void emit_delta_xy(struct brw_compile *p, { struct brw_reg r1 = brw_vec1_grf(1, 0); - /* Calc delta X,Y by subtracting origin in r1 from the pixel - * centers. - */ - if (mask & WRITEMASK_X) { - brw_ADD(p, - dst[0], - retype(arg0[0], BRW_REGISTER_TYPE_UW), - negate(r1)); - } + if (mask == 0) + return; - if (mask & WRITEMASK_Y) { - brw_ADD(p, - dst[1], - retype(arg0[1], BRW_REGISTER_TYPE_UW), - negate(suboffset(r1,1))); + assert(mask == WRITEMASK_XY); - } + /* Calc delta X,Y by subtracting origin in r1 from the pixel + * centers produced by emit_pixel_xy(). + */ + brw_ADD(p, + dst[0], + retype(arg0[0], BRW_REGISTER_TYPE_UW), + negate(r1)); + brw_ADD(p, + dst[1], + retype(arg0[1], BRW_REGISTER_TYPE_UW), + negate(suboffset(r1,1))); } +/** + * Computes the pixel offset from the window origin for gl_FragCoord(). + */ void emit_wpos_xy(struct brw_wm_compile *c, const struct brw_reg *dst, GLuint mask, @@ -134,9 +165,6 @@ void emit_wpos_xy(struct brw_wm_compile *c, { struct brw_compile *p = &c->func; - /* Calculate the pixel offset from window bottom left into destination - * X and Y channels. - */ if (mask & WRITEMASK_X) { if (c->fp->program.PixelCenterInteger) { /* X' = X */ @@ -186,6 +214,7 @@ void emit_pixel_w(struct brw_wm_compile *c, const struct brw_reg *deltas) { struct brw_compile *p = &c->func; + struct intel_context *intel = &p->brw->intel; /* Don't need this if all you are doing is interpolating color, for * instance. @@ -196,8 +225,12 @@ void emit_pixel_w(struct brw_wm_compile *c, /* Calc 1/w - just linterp wpos[3] optimized by putting the * result straight into a message reg. */ - brw_LINE(p, brw_null_reg(), interp3, deltas[0]); - brw_MAC(p, brw_message_reg(2), suboffset(interp3, 1), deltas[1]); + if (can_do_pln(intel, deltas)) { + brw_PLN(p, brw_message_reg(2), interp3, deltas[0]); + } else { + brw_LINE(p, brw_null_reg(), interp3, deltas[0]); + brw_MAC(p, brw_message_reg(2), suboffset(interp3, 1), deltas[1]); + } /* Calc w */ if (c->dispatch_width == 16) { @@ -224,6 +257,7 @@ void emit_linterp(struct brw_compile *p, const struct brw_reg *arg0, const struct brw_reg *deltas) { + struct intel_context *intel = &p->brw->intel; struct brw_reg interp[4]; GLuint nr = arg0[0].nr; GLuint i; @@ -235,8 +269,12 @@ void emit_linterp(struct brw_compile *p, for (i = 0; i < 4; i++) { if (mask & (1<<i)) { - brw_LINE(p, brw_null_reg(), interp[i], deltas[0]); - brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]); + if (can_do_pln(intel, deltas)) { + brw_PLN(p, dst[i], interp[i], deltas[0]); + } else { + brw_LINE(p, brw_null_reg(), interp[i], deltas[0]); + brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]); + } } } } @@ -249,6 +287,7 @@ void emit_pinterp(struct brw_compile *p, const struct brw_reg *deltas, const struct brw_reg *w) { + struct intel_context *intel = &p->brw->intel; struct brw_reg interp[4]; GLuint nr = arg0[0].nr; GLuint i; @@ -260,8 +299,12 @@ void emit_pinterp(struct brw_compile *p, for (i = 0; i < 4; i++) { if (mask & (1<<i)) { - brw_LINE(p, brw_null_reg(), interp[i], deltas[0]); - brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]); + if (can_do_pln(intel, deltas)) { + brw_PLN(p, dst[i], interp[i], deltas[0]); + } else { + brw_LINE(p, brw_null_reg(), interp[i], deltas[0]); + brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]); + } } } for (i = 0; i < 4; i++) { @@ -502,11 +545,8 @@ void emit_sop(struct brw_compile *p, for (i = 0; i < 4; i++) { if (mask & (1<<i)) { brw_push_insn_state(p); - brw_CMP(p, brw_null_reg(), cond, arg0[i], arg1[i]); - brw_set_predicate_control(p, BRW_PREDICATE_NONE); - brw_MOV(p, dst[i], brw_imm_f(0)); - brw_set_predicate_control(p, BRW_PREDICATE_NORMAL); - brw_MOV(p, dst[i], brw_imm_f(1.0)); + brw_CMP(p, brw_null_reg(), cond, arg1[i], arg0[i]); + brw_SEL(p, dst[i], brw_null_reg(), brw_imm_f(1.0)); brw_pop_insn_state(p); } } @@ -566,12 +606,12 @@ static void emit_sne( struct brw_compile *p, emit_sop(p, dst, mask, BRW_CONDITIONAL_NEQ, arg0, arg1); } -static void emit_cmp( struct brw_compile *p, - const struct brw_reg *dst, - GLuint mask, - const struct brw_reg *arg0, - const struct brw_reg *arg1, - const struct brw_reg *arg2 ) +void emit_cmp(struct brw_compile *p, + const struct brw_reg *dst, + GLuint mask, + const struct brw_reg *arg0, + const struct brw_reg *arg1, + const struct brw_reg *arg2) { GLuint i; @@ -601,14 +641,10 @@ void emit_max(struct brw_compile *p, for (i = 0; i < 4; i++) { if (mask & (1<<i)) { - brw_set_saturate(p, (mask & SATURATE) ? 1 : 0); - brw_MOV(p, dst[i], arg0[i]); - brw_set_saturate(p, 0); - - brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0[i], arg1[i]); + brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_GE, arg0[i], arg1[i]); brw_set_saturate(p, (mask & SATURATE) ? 1 : 0); - brw_MOV(p, dst[i], arg1[i]); + brw_SEL(p, dst[i], arg0[i], arg1[i]); brw_set_saturate(p, 0); brw_set_predicate_control_flag_value(p, 0xff); } @@ -625,14 +661,10 @@ void emit_min(struct brw_compile *p, for (i = 0; i < 4; i++) { if (mask & (1<<i)) { - brw_set_saturate(p, (mask & SATURATE) ? 1 : 0); - brw_MOV(p, dst[i], arg1[i]); - brw_set_saturate(p, 0); - brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0[i], arg1[i]); brw_set_saturate(p, (mask & SATURATE) ? 1 : 0); - brw_MOV(p, dst[i], arg0[i]); + brw_SEL(p, dst[i], arg0[i], arg1[i]); brw_set_saturate(p, 0); brw_set_predicate_control_flag_value(p, 0xff); } @@ -1086,11 +1118,19 @@ static void emit_kil( struct brw_wm_compile *c, { struct brw_compile *p = &c->func; struct brw_reg r0uw = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW); - GLuint i; - - /* XXX - usually won't need 4 compares! - */ + GLuint i, j; + for (i = 0; i < 4; i++) { + /* Check if we've already done the comparison for this reg + * -- common when someone does KIL TEMP.wwww. + */ + for (j = 0; j < i; j++) { + if (memcmp(&arg0[j], &arg0[i], sizeof(arg0[0])) == 0) + break; + } + if (j != i) + continue; + brw_push_insn_state(p); brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_GE, arg0[i], brw_imm_f(0)); brw_set_predicate_control_flag_value(p, 0xff); diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c index ea3c2405af9..0b66cc6c9f3 100644 --- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c +++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c @@ -289,6 +289,7 @@ reclaim_temps(struct brw_wm_compile *c) */ static void prealloc_reg(struct brw_wm_compile *c) { + struct intel_context *intel = &c->func.brw->intel; int i, j; struct brw_reg reg; int urb_read_length = 0; @@ -413,6 +414,43 @@ static void prealloc_reg(struct brw_wm_compile *c) } } + for (i = 0; i < c->nr_fp_insns; i++) { + const struct prog_instruction *inst = &c->prog_instructions[i]; + + switch (inst->Opcode) { + case WM_DELTAXY: + /* Allocate WM_DELTAXY destination on G45/GM45 to an + * even-numbered GRF if possible so that we can use the PLN + * instruction. + */ + if (inst->DstReg.WriteMask == WRITEMASK_XY && + !c->wm_regs[inst->DstReg.File][inst->DstReg.Index][0].inited && + !c->wm_regs[inst->DstReg.File][inst->DstReg.Index][1].inited && + (IS_G4X(intel->intelScreen->deviceID) || intel->gen == 5)) { + int grf; + + for (grf = c->first_free_grf & ~1; + grf < BRW_WM_MAX_GRF; + grf += 2) + { + if (!c->used_grf[grf] && !c->used_grf[grf + 1]) { + c->used_grf[grf] = GL_TRUE; + c->used_grf[grf + 1] = GL_TRUE; + c->first_free_grf = grf + 2; /* a guess */ + + set_reg(c, inst->DstReg.File, inst->DstReg.Index, 0, + brw_vec8_grf(grf, 0)); + set_reg(c, inst->DstReg.File, inst->DstReg.Index, 1, + brw_vec8_grf(grf + 1, 0)); + break; + } + } + } + default: + break; + } + } + /* An instruction may reference up to three constants. * They'll be found in these registers. * XXX alloc these on demand! @@ -1869,6 +1907,9 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) case OPCODE_LG2: emit_math1(c, BRW_MATH_FUNCTION_LOG, dst, dst_flags, args[0]); break; + case OPCODE_CMP: + emit_cmp(p, dst, dst_flags, args[0], args[1], args[2]); + break; case OPCODE_MIN: emit_min(p, dst, dst_flags, args[0], args[1]); break; @@ -2026,8 +2067,9 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c) } break; default: - printf("unsupported IR in fragment shader %d\n", - inst->Opcode); + printf("unsupported opcode %d (%s) in fragment shader\n", + inst->Opcode, inst->Opcode < MAX_OPCODE ? + _mesa_opcode_string(inst->Opcode) : "unknown"); } /* Release temporaries containing any unaliased source regs. */ diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.c b/src/mesa/drivers/dri/nouveau/nouveau_context.c index 502e01255cb..be57d48b8dd 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_context.c +++ b/src/mesa/drivers/dri/nouveau/nouveau_context.c @@ -69,8 +69,8 @@ nouveau_channel_flush_notify(struct nouveau_channel *chan) struct nouveau_context *nctx = chan->user_private; GLcontext *ctx = &nctx->base; - if (nctx->fallback < SWRAST && ctx->DrawBuffer) - nouveau_state_emit(&nctx->base); + if (nctx->fallback < SWRAST) + nouveau_bo_state_emit(ctx); } GLboolean diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.h b/src/mesa/drivers/dri/nouveau/nouveau_context.h index 682f8a414e3..fe64fec930b 100644 --- a/src/mesa/drivers/dri/nouveau/nouveau_context.h +++ b/src/mesa/drivers/dri/nouveau/nouveau_context.h @@ -85,6 +85,8 @@ struct nouveau_context { BITSET_SET(to_nouveau_context(ctx)->dirty, NOUVEAU_STATE_##s) #define context_dirty_i(ctx, s, i) \ BITSET_SET(to_nouveau_context(ctx)->dirty, NOUVEAU_STATE_##s##0 + i) +#define context_emit(ctx, s) \ + context_drv(ctx)->emit[NOUVEAU_STATE_##s](ctx, NOUVEAU_STATE_##s) GLboolean nouveau_context_create(const __GLcontextModes *visual, __DRIcontext *dri_ctx, diff --git a/src/mesa/drivers/dri/nouveau/nv04_context.c b/src/mesa/drivers/dri/nouveau/nv04_context.c index a442425e448..3624b3af921 100644 --- a/src/mesa/drivers/dri/nouveau/nv04_context.c +++ b/src/mesa/drivers/dri/nouveau/nv04_context.c @@ -75,18 +75,16 @@ nv04_channel_flush_notify(struct nouveau_channel *chan) struct nouveau_context *nctx = chan->user_private; GLcontext *ctx = &nctx->base; - if (nctx->fallback < SWRAST && ctx->DrawBuffer) { - GLcontext *ctx = &nctx->base; - + if (nctx->fallback < SWRAST) { /* Flushing seems to clobber the engine context. */ - context_dirty_i(ctx, TEX_OBJ, 0); - context_dirty_i(ctx, TEX_OBJ, 1); - context_dirty_i(ctx, TEX_ENV, 0); - context_dirty_i(ctx, TEX_ENV, 1); - context_dirty(ctx, CONTROL); - context_dirty(ctx, BLEND); - - nouveau_state_emit(ctx); + context_emit(ctx, TEX_OBJ0); + context_emit(ctx, TEX_OBJ1); + context_emit(ctx, TEX_ENV0); + context_emit(ctx, TEX_ENV1); + context_emit(ctx, CONTROL); + context_emit(ctx, BLEND); + + nouveau_bo_state_emit(ctx); } } @@ -200,9 +198,9 @@ nv04_context_create(struct nouveau_screen *screen, const GLvisual *visual, if (ret) goto fail; + init_dummy_texture(ctx); nv04_hwctx_init(ctx); nv04_render_init(ctx); - init_dummy_texture(ctx); return ctx; diff --git a/src/mesa/drivers/dri/r128/r128_tex.c b/src/mesa/drivers/dri/r128/r128_tex.c index 24fbf8f5194..4ec4be9a47b 100644 --- a/src/mesa/drivers/dri/r128/r128_tex.c +++ b/src/mesa/drivers/dri/r128/r128_tex.c @@ -468,7 +468,7 @@ static void r128TexEnv( GLcontext *ctx, GLenum target, * certain point. It is better than completely ignoring the LOD * bias. Unfortunately there isn't much range in the bias, the * spec mentions strides that vary between 0.5 and 2.0 but these - * numbers don't seem to relate the the GL LOD bias value at all. + * numbers don't seem to relate to the GL LOD bias value at all. */ if ( param[0] >= 1.0 ) { bias = -128; diff --git a/src/mesa/drivers/dri/r200/Makefile b/src/mesa/drivers/dri/r200/Makefile index 9ea81fd5059..c9c1346c3a2 100644 --- a/src/mesa/drivers/dri/r200/Makefile +++ b/src/mesa/drivers/dri/r200/Makefile @@ -34,7 +34,6 @@ DRIVER_SOURCES = r200_context.c \ r200_state.c \ r200_state_init.c \ r200_cmdbuf.c \ - r200_pixel.c \ r200_tex.c \ r200_texstate.c \ r200_tcl.c \ diff --git a/src/mesa/drivers/dri/r200/r200_blit.c b/src/mesa/drivers/dri/r200/r200_blit.c index 30757600934..e187fc0f61e 100644 --- a/src/mesa/drivers/dri/r200/r200_blit.c +++ b/src/mesa/drivers/dri/r200/r200_blit.c @@ -48,6 +48,11 @@ unsigned r200_check_blit(gl_format mesa_format) case MESA_FORMAT_ARGB4444: case MESA_FORMAT_ARGB1555: case MESA_FORMAT_A8: + case MESA_FORMAT_L8: + case MESA_FORMAT_I8: + /* swizzled */ + case MESA_FORMAT_RGBA8888: + case MESA_FORMAT_RGBA8888_REV: break; default: return 0; @@ -86,7 +91,8 @@ static inline void emit_vtx_state(struct r200_context *r200) } static void inline emit_tx_setup(struct r200_context *r200, - gl_format mesa_format, + gl_format src_mesa_format, + gl_format dst_mesa_format, struct radeon_bo *bo, intptr_t offset, unsigned width, @@ -101,10 +107,16 @@ static void inline emit_tx_setup(struct r200_context *r200, assert(offset % 32 == 0); /* XXX others? BE/LE? */ - switch (mesa_format) { + switch (src_mesa_format) { case MESA_FORMAT_ARGB8888: txformat |= R200_TXFORMAT_ARGB8888 | R200_TXFORMAT_ALPHA_IN_MAP; break; + case MESA_FORMAT_RGBA8888: + txformat |= R200_TXFORMAT_RGBA8888 | R200_TXFORMAT_ALPHA_IN_MAP; + break; + case MESA_FORMAT_RGBA8888_REV: + txformat |= R200_TXFORMAT_ABGR8888 | R200_TXFORMAT_ALPHA_IN_MAP; + break; case MESA_FORMAT_XRGB8888: txformat |= R200_TXFORMAT_ARGB8888; break; @@ -118,26 +130,143 @@ static void inline emit_tx_setup(struct r200_context *r200, txformat |= R200_TXFORMAT_ARGB1555 | R200_TXFORMAT_ALPHA_IN_MAP; break; case MESA_FORMAT_A8: + case MESA_FORMAT_I8: txformat |= R200_TXFORMAT_I8 | R200_TXFORMAT_ALPHA_IN_MAP; break; + case MESA_FORMAT_L8: + txformat |= R200_TXFORMAT_I8; + break; + case MESA_FORMAT_AL88: + txformat |= R200_TXFORMAT_AI88 | R200_TXFORMAT_ALPHA_IN_MAP; + break; + default: + break; + } + + switch (dst_mesa_format) { + case MESA_FORMAT_ARGB8888: + case MESA_FORMAT_XRGB8888: + case MESA_FORMAT_RGB565: + case MESA_FORMAT_ARGB4444: + case MESA_FORMAT_ARGB1555: + case MESA_FORMAT_A8: + case MESA_FORMAT_L8: + case MESA_FORMAT_I8: default: + /* no swizzle required */ + BEGIN_BATCH(10); + OUT_BATCH_REGVAL(RADEON_PP_CNTL, (RADEON_TEX_0_ENABLE | + RADEON_TEX_BLEND_0_ENABLE)); + OUT_BATCH_REGVAL(R200_PP_TXCBLEND_0, (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_0, (R200_TXC_CLAMP_0_1 | + R200_TXC_OUTPUT_REG_R0)); + OUT_BATCH_REGVAL(R200_PP_TXABLEND_0, (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXABLEND2_0, (R200_TXA_CLAMP_0_1 | + R200_TXA_OUTPUT_REG_R0)); + END_BATCH(); + break; + case MESA_FORMAT_RGBA8888: + BEGIN_BATCH(10); + OUT_BATCH_REGVAL(RADEON_PP_CNTL, (RADEON_TEX_0_ENABLE | + RADEON_TEX_BLEND_0_ENABLE)); + OUT_BATCH_REGVAL(R200_PP_TXCBLEND_0, (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_0, (R200_TXC_CLAMP_0_1 | + R200_TXC_OUTPUT_ROTATE_GBA | + R200_TXC_OUTPUT_REG_R0)); + OUT_BATCH_REGVAL(R200_PP_TXABLEND_0, (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXABLEND2_0, (R200_TXA_CLAMP_0_1 | + (R200_TXA_REPL_RED << R200_TXA_REPL_ARG_C_SHIFT) | + R200_TXA_OUTPUT_REG_R0)); + END_BATCH(); + break; + case MESA_FORMAT_RGBA8888_REV: + BEGIN_BATCH(34); + OUT_BATCH_REGVAL(RADEON_PP_CNTL, (RADEON_TEX_0_ENABLE | + RADEON_TEX_BLEND_0_ENABLE | + RADEON_TEX_BLEND_1_ENABLE | + RADEON_TEX_BLEND_2_ENABLE | + RADEON_TEX_BLEND_3_ENABLE)); + /* r1.r = r0.b */ + OUT_BATCH_REGVAL(R200_PP_TXCBLEND_0, (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_0, (R200_TXC_CLAMP_0_1 | + R200_TXC_OUTPUT_MASK_R | + (R200_TXC_REPL_BLUE << R200_TXC_REPL_ARG_C_SHIFT) | + R200_TXC_OUTPUT_REG_R1)); + /* r1.a = r0.a */ + OUT_BATCH_REGVAL(R200_PP_TXABLEND_0, (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXABLEND2_0, (R200_TXA_CLAMP_0_1 | + R200_TXA_OUTPUT_REG_R1)); + /* r1.g = r0.g */ + OUT_BATCH_REGVAL(R200_PP_TXCBLEND_1, (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_1, (R200_TXC_CLAMP_0_1 | + R200_TXC_OUTPUT_MASK_G | + (R200_TXC_REPL_GREEN << R200_TXC_REPL_ARG_C_SHIFT) | + R200_TXC_OUTPUT_REG_R1)); + /* r1.a = r0.a */ + OUT_BATCH_REGVAL(R200_PP_TXABLEND_1, (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXABLEND2_1, (R200_TXA_CLAMP_0_1 | + R200_TXA_OUTPUT_REG_R1)); + /* r1.b = r0.r */ + OUT_BATCH_REGVAL(R200_PP_TXCBLEND_2, (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_2, (R200_TXC_CLAMP_0_1 | + R200_TXC_OUTPUT_MASK_B | + (R200_TXC_REPL_RED << R200_TXC_REPL_ARG_C_SHIFT) | + R200_TXC_OUTPUT_REG_R1)); + /* r1.a = r0.a */ + OUT_BATCH_REGVAL(R200_PP_TXABLEND_2, (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXABLEND2_2, (R200_TXA_CLAMP_0_1 | + R200_TXA_OUTPUT_REG_R1)); + /* r0.rgb = r1.rgb */ + OUT_BATCH_REGVAL(R200_PP_TXCBLEND_3, (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R1_COLOR | + R200_TXC_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_3, (R200_TXC_CLAMP_0_1 | + R200_TXC_OUTPUT_REG_R0)); + /* r0.a = r1.a */ + OUT_BATCH_REGVAL(R200_PP_TXABLEND_3, (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R1_ALPHA | + R200_TXA_OP_MADD)); + OUT_BATCH_REGVAL(R200_PP_TXABLEND2_3, (R200_TXA_CLAMP_0_1 | + R200_TXA_OUTPUT_REG_R0)); + END_BATCH(); break; } - BEGIN_BATCH(28); - OUT_BATCH_REGVAL(RADEON_PP_CNTL, RADEON_TEX_0_ENABLE | RADEON_TEX_BLEND_0_ENABLE); + BEGIN_BATCH(18); OUT_BATCH_REGVAL(R200_PP_CNTL_X, 0); OUT_BATCH_REGVAL(R200_PP_TXMULTI_CTL_0, 0); - OUT_BATCH_REGVAL(R200_PP_TXCBLEND_0, (R200_TXC_ARG_A_ZERO | - R200_TXC_ARG_B_ZERO | - R200_TXC_ARG_C_R0_COLOR | - R200_TXC_OP_MADD)); - OUT_BATCH_REGVAL(R200_PP_TXCBLEND2_0, R200_TXC_CLAMP_0_1 | R200_TXC_OUTPUT_REG_R0); - OUT_BATCH_REGVAL(R200_PP_TXABLEND_0, (R200_TXA_ARG_A_ZERO | - R200_TXA_ARG_B_ZERO | - R200_TXA_ARG_C_R0_ALPHA | - R200_TXA_OP_MADD)); - OUT_BATCH_REGVAL(R200_PP_TXABLEND2_0, R200_TXA_CLAMP_0_1 | R200_TXA_OUTPUT_REG_R0); OUT_BATCH_REGVAL(R200_PP_TXFILTER_0, (R200_CLAMP_S_CLAMP_LAST | R200_CLAMP_T_CLAMP_LAST | R200_MAG_FILTER_NEAREST | @@ -146,7 +275,7 @@ static void inline emit_tx_setup(struct r200_context *r200, OUT_BATCH_REGVAL(R200_PP_TXFORMAT_X_0, 0); OUT_BATCH_REGVAL(R200_PP_TXSIZE_0, ((width - 1) | ((height - 1) << RADEON_TEX_VSIZE_SHIFT))); - OUT_BATCH_REGVAL(R200_PP_TXPITCH_0, pitch * _mesa_get_format_bytes(mesa_format) - 32); + OUT_BATCH_REGVAL(R200_PP_TXPITCH_0, pitch * _mesa_get_format_bytes(src_mesa_format) - 32); OUT_BATCH_REGSEQ(R200_PP_TXOFFSET_0, 1); OUT_BATCH_RELOC(0, bo, 0, RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0); @@ -170,6 +299,8 @@ static inline void emit_cb_setup(struct r200_context *r200, switch (mesa_format) { case MESA_FORMAT_ARGB8888: case MESA_FORMAT_XRGB8888: + case MESA_FORMAT_RGBA8888: + case MESA_FORMAT_RGBA8888_REV: dst_format = RADEON_COLOR_FORMAT_ARGB8888; break; case MESA_FORMAT_RGB565: @@ -182,6 +313,8 @@ static inline void emit_cb_setup(struct r200_context *r200, dst_format = RADEON_COLOR_FORMAT_ARGB1555; break; case MESA_FORMAT_A8: + case MESA_FORMAT_L8: + case MESA_FORMAT_I8: dst_format = RADEON_COLOR_FORMAT_RGB8; break; default: @@ -384,15 +517,15 @@ unsigned r200_blit(GLcontext *ctx, /* Flush is needed to make sure that source buffer has correct data */ radeonFlush(r200->radeon.glCtx); - rcommonEnsureCmdBufSpace(&r200->radeon, 78, __FUNCTION__); + rcommonEnsureCmdBufSpace(&r200->radeon, 102, __FUNCTION__); if (!validate_buffers(r200, src_bo, dst_bo)) return GL_FALSE; /* 14 */ emit_vtx_state(r200); - /* 28 */ - emit_tx_setup(r200, src_mesaformat, src_bo, src_offset, src_width, src_height, src_pitch); + /* 52 */ + emit_tx_setup(r200, src_mesaformat, dst_mesaformat, src_bo, src_offset, src_width, src_height, src_pitch); /* 22 */ emit_cb_setup(r200, dst_bo, dst_offset, dst_mesaformat, dst_pitch, dst_width, dst_height); /* 14 */ diff --git a/src/mesa/drivers/dri/r200/r200_context.c b/src/mesa/drivers/dri/r200/r200_context.c index 4f1a56658cc..36a29350ccc 100644 --- a/src/mesa/drivers/dri/r200/r200_context.c +++ b/src/mesa/drivers/dri/r200/r200_context.c @@ -51,7 +51,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "r200_context.h" #include "r200_ioctl.h" #include "r200_state.h" -#include "r200_pixel.h" #include "r200_tex.h" #include "r200_swtcl.h" #include "r200_tcl.h" @@ -325,7 +324,7 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual, _mesa_init_driver_functions(&functions); r200InitDriverFuncs(&functions); r200InitIoctlFuncs(&functions); - r200InitStateFuncs(&functions); + r200InitStateFuncs(&rmesa->radeon, &functions); r200InitTextureFuncs(&rmesa->radeon, &functions); r200InitShaderFuncs(&functions); radeonInitQueryObjFunctions(&functions); @@ -474,7 +473,6 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual, /* XXX these should really go right after _mesa_init_driver_functions() */ radeon_fbo_init(&rmesa->radeon); radeonInitSpanFuncs( ctx ); - r200InitPixelFuncs( ctx ); r200InitTnlFuncs( ctx ); r200InitState( rmesa ); r200InitSwtcl( ctx ); diff --git a/src/mesa/drivers/dri/r200/r200_pixel.c b/src/mesa/drivers/dri/r200/r200_pixel.c deleted file mode 100644 index bfb7e2a2ed3..00000000000 --- a/src/mesa/drivers/dri/r200/r200_pixel.c +++ /dev/null @@ -1,494 +0,0 @@ -/* -Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. - -The Weather Channel (TM) funded Tungsten Graphics to develop the -initial release of the Radeon 8500 driver under the XFree86 license. -This notice must be preserved. - -Permission is hereby granted, free of charge, to any person obtaining -a copy of this software and associated documentation files (the -"Software"), to deal in the Software without restriction, including -without limitation the rights to use, copy, modify, merge, publish, -distribute, sublicense, and/or sell copies of the Software, and to -permit persons to whom the Software is furnished to do so, subject to -the following conditions: - -The above copyright notice and this permission notice (including the -next paragraph) shall be included in all copies or substantial -portions of the Software. - -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, -EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF -MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. -IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE -LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION -OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION -WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. -*/ - -/* - * Authors: - * Keith Whitwell <keith@tungstengraphics.com> - */ - -#include "main/glheader.h" -#include "main/enums.h" -#include "main/mtypes.h" -#include "main/macros.h" -#include "swrast/swrast.h" - -#include "r200_context.h" -#include "r200_ioctl.h" -#include "r200_pixel.h" -#include "r200_swtcl.h" - -#include "drirenderbuffer.h" - - -static GLboolean -check_color( const GLcontext *ctx, GLenum type, GLenum format, - const struct gl_pixelstore_attrib *packing, - const void *pixels, GLint sz, GLint pitch ) -{ - r200ContextPtr rmesa = R200_CONTEXT(ctx); - GLuint cpp = rmesa->radeon.radeonScreen->cpp; - - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s\n", __FUNCTION__); - - if ( (pitch & 63) || - ctx->_ImageTransferState || - packing->SwapBytes || - packing->LsbFirst) { - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s: failed 1\n", __FUNCTION__); - return GL_FALSE; - } - - if ( type == GL_UNSIGNED_INT_8_8_8_8_REV && - cpp == 4 && - format == GL_BGRA ) { - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s: passed 2\n", __FUNCTION__); - return GL_TRUE; - } - - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s: failed\n", __FUNCTION__); - - return GL_FALSE; -} - -static GLboolean -check_color_per_fragment_ops( const GLcontext *ctx ) -{ - int result; - result = (!( ctx->Color.AlphaEnabled || - ctx->Depth.Test || - ctx->Fog.Enabled || - ctx->Scissor.Enabled || - ctx->Stencil._Enabled || - !ctx->Color.ColorMask[0][0] || - !ctx->Color.ColorMask[0][1] || - !ctx->Color.ColorMask[0][2] || - !ctx->Color.ColorMask[0][3] || - ctx->Color.ColorLogicOpEnabled || - ctx->Texture._EnabledUnits - ) && - ctx->Current.RasterPosValid); - - return result; -} - - -#if 0 -static GLboolean -clip_pixelrect( const GLcontext *ctx, - const GLframebuffer *buffer, - GLint *x, GLint *y, - GLsizei *width, GLsizei *height, - GLint *size ) -{ - r200ContextPtr rmesa = R200_CONTEXT(ctx); - - /* left clipping */ - if (*x < buffer->_Xmin) { - *width -= (buffer->_Xmin - *x); - *x = buffer->_Xmin; - } - - /* right clipping */ - if (*x + *width > buffer->_Xmax) - *width -= (*x + *width - buffer->_Xmax - 1); - - if (*width <= 0) - return GL_FALSE; - - /* bottom clipping */ - if (*y < buffer->_Ymin) { - *height -= (buffer->_Ymin - *y); - *y = buffer->_Ymin; - } - - /* top clipping */ - if (*y + *height > buffer->_Ymax) - *height -= (*y + *height - buffer->_Ymax - 1); - - if (*height <= 0) - return GL_FALSE; - - *size = ((*y + *height - 1) * rmesa->radeon.radeonScreen->frontPitch + - (*x + *width - 1) * rmesa->radeon.radeonScreen->cpp); - - return GL_TRUE; -} -#endif - -static GLboolean -r200TryReadPixels( GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *pack, - GLvoid *pixels ) -{ - return GL_FALSE; -#if 0 - r200ContextPtr rmesa = R200_CONTEXT(ctx); - GLint pitch = pack->RowLength ? pack->RowLength : width; - GLint blit_format; - GLuint cpp = rmesa->radeon.radeonScreen->cpp; - GLint size = width * height * cpp; - - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s\n", __FUNCTION__); - - /* Only accelerate reading to GART buffers. - */ - if ( !r200IsGartMemory(rmesa, pixels, - pitch * height * rmesa->radeon.radeonScreen->cpp ) ) { - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s: dest not GART\n", __FUNCTION__); - } - - /* Need GL_PACK_INVERT_MESA to cope with upsidedown results from - * blitter: - */ - if (!pack->Invert) { - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s: MESA_PACK_INVERT not set\n", __FUNCTION__); - return GL_FALSE; - } - - if (!check_color(ctx, type, format, pack, pixels, size, pitch)) - return GL_FALSE; - - switch ( rmesa->radeon.radeonScreen->cpp ) { - case 4: - blit_format = R200_CP_COLOR_FORMAT_ARGB8888; - break; - default: - return GL_FALSE; - } - - - /* Although the blits go on the command buffer, need to do this and - * fire with lock held to guarentee cliprects and drawOffset are - * correct. - * - * This is an unusual situation however, as the code which flushes - * a full command buffer expects to be called unlocked. As a - * workaround, immediately flush the buffer on aquiring the lock. - */ - LOCK_HARDWARE( &rmesa->radeon ); - - if (rmesa->store.cmd_used) - rcommonFlushCmdBufLocked( &rmesa->radeon, __FUNCTION__ ); - - if (!clip_pixelrect(ctx, ctx->ReadBuffer, &x, &y, &width, &height, - &size)) { - UNLOCK_HARDWARE( &rmesa->radeon ); - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s totally clipped -- nothing to do\n", - __FUNCTION__); - return GL_TRUE; - } - - { - __DRIdrawable *dPriv = rmesa->radeon.dri.drawable; - driRenderbuffer *drb = (driRenderbuffer *) ctx->ReadBuffer->_ColorReadBuffer; - int nbox = dPriv->numClipRects; - int src_offset = drb->offset - + rmesa->radeon.radeonScreen->fbLocation; - int src_pitch = drb->pitch * drb->cpp; - int dst_offset = r200GartOffsetFromVirtual( rmesa, pixels ); - int dst_pitch = pitch * rmesa->radeon.radeonScreen->cpp; - drm_clip_rect_t *box = dPriv->pClipRects; - int i; - - r200EmitWait( rmesa, RADEON_WAIT_3D ); - - y = dPriv->h - y - height; - x += dPriv->x; - y += dPriv->y; - - - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "readpixel blit src_pitch %d dst_pitch %d\n", - src_pitch, dst_pitch); - - for (i = 0 ; i < nbox ; i++) - { - GLint bx = box[i].x1; - GLint by = box[i].y1; - GLint bw = box[i].x2 - bx; - GLint bh = box[i].y2 - by; - - if (bx < x) bw -= x - bx, bx = x; - if (by < y) bh -= y - by, by = y; - if (bx + bw > x + width) bw = x + width - bx; - if (by + bh > y + height) bh = y + height - by; - if (bw <= 0) continue; - if (bh <= 0) continue; - - r200EmitBlit( rmesa, - blit_format, - src_pitch, src_offset, - dst_pitch, dst_offset, - bx, by, - bx - x, by - y, - bw, bh ); - } - - rcommonFlushCmdBufLocked( &rmesa->radeon, __FUNCTION__ ); - } - UNLOCK_HARDWARE( &rmesa->radeon ); - - radeonFinish( ctx ); /* required by GL */ -#endif - return GL_TRUE; -} - -static void -r200ReadPixels( GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *pack, - GLvoid *pixels ) -{ - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s\n", __FUNCTION__); - - if (!r200TryReadPixels( ctx, x, y, width, height, format, type, pack, - pixels)) - _swrast_ReadPixels( ctx, x, y, width, height, format, type, pack, - pixels); -} - - - - -static void do_draw_pix( GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLint pitch, - const void *pixels, - GLuint planemask) -{ - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s\n", __FUNCTION__); - -#if 0 - r200ContextPtr rmesa = R200_CONTEXT(ctx); - __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon); - drm_clip_rect_t *box = dPriv->pClipRects; - struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorDrawBuffers[0]; - driRenderbuffer *drb = (driRenderbuffer *) rb; - int nbox = dPriv->numClipRects; - int i; - int blit_format; - int size; - int src_offset = r200GartOffsetFromVirtual( rmesa, pixels ); - int src_pitch = pitch * rmesa->radeon.radeonScreen->cpp; - - switch ( rmesa->radeon.radeonScreen->cpp ) { - case 2: - blit_format = R200_CP_COLOR_FORMAT_RGB565; - break; - case 4: - blit_format = R200_CP_COLOR_FORMAT_ARGB8888; - break; - default: - return; - } - - - LOCK_HARDWARE( &rmesa->radeon ); - - if (rmesa->store.cmd_used) - rcommonFlushCmdBufLocked( &rmesa->radeon, __FUNCTION__ ); - - y -= height; /* cope with pixel zoom */ - - if (!clip_pixelrect(ctx, ctx->DrawBuffer, - &x, &y, &width, &height, - &size)) { - UNLOCK_HARDWARE( &rmesa->radeon ); - return; - } - - y = dPriv->h - y - height; /* convert from gl to hardware coords */ - x += dPriv->x; - y += dPriv->y; - - - r200EmitWait( rmesa, RADEON_WAIT_3D ); - - for (i = 0 ; i < nbox ; i++ ) - { - GLint bx = box[i].x1; - GLint by = box[i].y1; - GLint bw = box[i].x2 - bx; - GLint bh = box[i].y2 - by; - - if (bx < x) bw -= x - bx, bx = x; - if (by < y) bh -= y - by, by = y; - if (bx + bw > x + width) bw = x + width - bx; - if (by + bh > y + height) bh = y + height - by; - if (bw <= 0) continue; - if (bh <= 0) continue; - - r200EmitBlit( rmesa, - blit_format, - src_pitch, src_offset, - drb->pitch * drb->cpp, - drb->offset + rmesa->radeon.radeonScreen->fbLocation, - bx - x, by - y, - bx, by, - bw, bh ); - } - - rcommonFlushCmdBufLocked( &rmesa->radeon, __FUNCTION__ ); - radeonWaitForIdleLocked( &rmesa->radeon ); /* required by GL */ - UNLOCK_HARDWARE( &rmesa->radeon ); -#endif -} - - - - -static GLboolean -r200TryDrawPixels( GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *unpack, - const GLvoid *pixels ) -{ - r200ContextPtr rmesa = R200_CONTEXT(ctx); - GLint pitch = unpack->RowLength ? unpack->RowLength : width; - GLuint planemask; - GLuint cpp = rmesa->radeon.radeonScreen->cpp; - GLint size = height * pitch * cpp; - - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s\n", __FUNCTION__); - - /* check that we're drawing to exactly one color buffer */ - if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) - return GL_FALSE; - - switch (format) { - case GL_RGB: - case GL_RGBA: - case GL_BGRA: - planemask = radeonPackColor(cpp, - ctx->Color.ColorMask[0][RCOMP], - ctx->Color.ColorMask[0][GCOMP], - ctx->Color.ColorMask[0][BCOMP], - ctx->Color.ColorMask[0][ACOMP]); - - if (cpp == 2) - planemask |= planemask << 16; - - if (planemask != ~0) - return GL_FALSE; /* fix me -- should be possible */ - - /* Can't do conversions on GART reads/draws. - */ - if ( !r200IsGartMemory( rmesa, pixels, size ) ) { - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s: not GART memory\n", __FUNCTION__); - return GL_FALSE; - } - - if (!check_color(ctx, type, format, unpack, pixels, size, pitch)) { - return GL_FALSE; - } - if (!check_color_per_fragment_ops(ctx)) { - return GL_FALSE; - } - - if (ctx->Pixel.ZoomX != 1.0F || - ctx->Pixel.ZoomY != -1.0F) - return GL_FALSE; - break; - - default: - return GL_FALSE; - } - - if (0)// r200IsGartMemory(rmesa, pixels, size) ) - { - do_draw_pix( ctx, x, y, width, height, pitch, pixels, planemask ); - return GL_TRUE; - } - else if (0) - { - /* Pixels is in regular memory -- get dma buffers and perform - * upload through them. - */ - } - else - return GL_FALSE; -} - -static void -r200DrawPixels( GLcontext *ctx, - GLint x, GLint y, GLsizei width, GLsizei height, - GLenum format, GLenum type, - const struct gl_pixelstore_attrib *unpack, - const GLvoid *pixels ) -{ - if (R200_DEBUG & RADEON_PIXEL) - fprintf(stderr, "%s\n", __FUNCTION__); - - if (!r200TryDrawPixels( ctx, x, y, width, height, format, type, - unpack, pixels )) - _swrast_DrawPixels( ctx, x, y, width, height, format, type, - unpack, pixels ); -} - - -static void -r200Bitmap( GLcontext *ctx, GLint px, GLint py, - GLsizei width, GLsizei height, - const struct gl_pixelstore_attrib *unpack, - const GLubyte *bitmap ) -{ - r200ContextPtr rmesa = R200_CONTEXT(ctx); - - if (rmesa->radeon.Fallback) - _swrast_Bitmap( ctx, px, py, width, height, unpack, bitmap ); - else - r200PointsBitmap( ctx, px, py, width, height, unpack, bitmap ); -} - - - -void r200InitPixelFuncs( GLcontext *ctx ) -{ - if (!getenv("R200_NO_BLITS")) { - ctx->Driver.ReadPixels = r200ReadPixels; - ctx->Driver.DrawPixels = r200DrawPixels; - if (getenv("R200_HW_BITMAP")) - ctx->Driver.Bitmap = r200Bitmap; - } -} diff --git a/src/mesa/drivers/dri/r200/r200_pixel.h b/src/mesa/drivers/dri/r200/r200_pixel.h deleted file mode 100644 index e62aa05d749..00000000000 --- a/src/mesa/drivers/dri/r200/r200_pixel.h +++ /dev/null @@ -1,39 +0,0 @@ -/* -Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. - -The Weather Channel (TM) funded Tungsten Graphics to develop the -initial release of the Radeon 8500 driver under the XFree86 license. -This notice must be preserved. - -Permission is hereby granted, free of charge, to any person obtaining -a copy of this software and associated documentation files (the -"Software"), to deal in the Software without restriction, including -without limitation the rights to use, copy, modify, merge, publish, -distribute, sublicense, and/or sell copies of the Software, and to -permit persons to whom the Software is furnished to do so, subject to -the following conditions: - -The above copyright notice and this permission notice (including the -next paragraph) shall be included in all copies or substantial -portions of the Software. - -THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, -EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF -MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. -IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE -LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION -OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION -WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. -*/ - -/* - * Authors: - * Keith Whitwell <keith@tungstengraphics.com> - */ - -#ifndef __R200_PIXEL_H__ -#define __R200_PIXEL_H__ - -extern void r200InitPixelFuncs( GLcontext *ctx ); - -#endif diff --git a/src/mesa/drivers/dri/r200/r200_reg.h b/src/mesa/drivers/dri/r200/r200_reg.h index 59115212cee..e331be223b8 100644 --- a/src/mesa/drivers/dri/r200/r200_reg.h +++ b/src/mesa/drivers/dri/r200/r200_reg.h @@ -690,7 +690,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. # define R200_PVS_CNTL_1_PROGRAM_START_SHIFT 0 # define R200_PVS_CNTL_1_POS_END_SHIFT 10 # define R200_PVS_CNTL_1_PROGRAM_END_SHIFT 20 -/* Addresses are relative the the vertex program parameters area. */ +/* Addresses are relative to the vertex program parameters area. */ #define R200_VAP_PVS_CNTL_2 0x22d4 # define R200_PVS_CNTL_2_PARAM_OFFSET_SHIFT 0 # define R200_PVS_CNTL_2_PARAM_COUNT_SHIFT 16 @@ -1265,6 +1265,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #define R200_TXC_OUTPUT_MASK_G (5 << 20) #define R200_TXC_OUTPUT_MASK_B (6 << 20) #define R200_TXC_OUTPUT_MASK_NONE (7 << 20) +#define R200_TXC_OUTPUT_ROTATE_RGB (0 << 24) +#define R200_TXC_OUTPUT_ROTATE_ARG (1 << 24) +#define R200_TXC_OUTPUT_ROTATE_GBA (2 << 24) +#define R200_TXC_OUTPUT_ROTATE_RGA (3 << 24) #define R200_TXC_REPL_NORMAL 0 #define R200_TXC_REPL_RED 1 #define R200_TXC_REPL_GREEN 2 diff --git a/src/mesa/drivers/dri/r200/r200_state.c b/src/mesa/drivers/dri/r200/r200_state.c index 050e5aa8770..9c2ac05ad6c 100644 --- a/src/mesa/drivers/dri/r200/r200_state.c +++ b/src/mesa/drivers/dri/r200/r200_state.c @@ -46,6 +46,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "tnl/tnl.h" #include "tnl/t_pipeline.h" #include "swrast_setup/swrast_setup.h" +#include "drivers/common/meta.h" #include "radeon_common.h" #include "radeon_mipmap_tree.h" @@ -2487,7 +2488,7 @@ static void r200PolygonStipple( GLcontext *ctx, const GLubyte *mask ) } /* Initialize the driver's state functions. */ -void r200InitStateFuncs( struct dd_function_table *functions ) +void r200InitStateFuncs( radeonContextPtr radeon, struct dd_function_table *functions ) { functions->UpdateState = r200InvalidateState; functions->LightingSpaceChange = r200LightingSpaceChange; @@ -2495,6 +2496,12 @@ void r200InitStateFuncs( struct dd_function_table *functions ) functions->DrawBuffer = radeonDrawBuffer; functions->ReadBuffer = radeonReadBuffer; + if (radeon->radeonScreen->kernel_mm) { + functions->CopyPixels = _mesa_meta_CopyPixels; + functions->DrawPixels = _mesa_meta_DrawPixels; + functions->ReadPixels = radeonReadPixels; + } + functions->AlphaFunc = r200AlphaFunc; functions->BlendColor = r200BlendColor; functions->BlendEquationSeparate = r200BlendEquationSeparate; diff --git a/src/mesa/drivers/dri/r200/r200_state.h b/src/mesa/drivers/dri/r200/r200_state.h index 7b9b0c106aa..327ba837e25 100644 --- a/src/mesa/drivers/dri/r200/r200_state.h +++ b/src/mesa/drivers/dri/r200/r200_state.h @@ -38,7 +38,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "r200_context.h" extern void r200InitState( r200ContextPtr rmesa ); -extern void r200InitStateFuncs( struct dd_function_table *functions ); +extern void r200InitStateFuncs( radeonContextPtr radeon, struct dd_function_table *functions ); extern void r200InitTnlFuncs( GLcontext *ctx ); extern void r200UpdateMaterial( GLcontext *ctx ); diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c index 364e0ba6b61..cfeb5407e91 100644 --- a/src/mesa/drivers/dri/r300/r300_context.c +++ b/src/mesa/drivers/dri/r300/r300_context.c @@ -500,7 +500,7 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual, _mesa_init_driver_functions(&functions); r300InitIoctlFuncs(&functions); - r300InitStateFuncs(&functions); + r300InitStateFuncs(&r300->radeon, &functions); r300InitTextureFuncs(&r300->radeon, &functions); r300InitShaderFuncs(&functions); radeonInitQueryObjFunctions(&functions); diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h index d18ebab8ff2..ac93563ed9e 100644 --- a/src/mesa/drivers/dri/r300/r300_reg.h +++ b/src/mesa/drivers/dri/r300/r300_reg.h @@ -482,7 +482,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. # define R300_PVS_FIRST_INST_SHIFT 0 # define R300_PVS_XYZW_VALID_INST_SHIFT 10 # define R300_PVS_LAST_INST_SHIFT 20 -/* Addresses are relative the the vertex program parameters area. */ +/* Addresses are relative to the vertex program parameters area. */ #define R300_VAP_PVS_CONST_CNTL 0x22D4 # define R300_PVS_CONST_BASE_OFFSET_SHIFT 0 # define R300_PVS_MAX_CONST_ADDR_SHIFT 16 @@ -1760,7 +1760,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE. * The destination register index is in FPI1 (color) and FPI3 (alpha) * together with enable bits. * There are separate enable bits for writing into temporary registers - * (DSTC_REG_* /DSTA_REG) and and program output registers (DSTC_OUTPUT_* + * (DSTC_REG_* /DSTA_REG) and program output registers (DSTC_OUTPUT_* * /DSTA_OUTPUT). You can write to both at once, or not write at all (the * same index must be used for both). * diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c index 87489412419..749a2464e7c 100644 --- a/src/mesa/drivers/dri/r300/r300_state.c +++ b/src/mesa/drivers/dri/r300/r300_state.c @@ -590,7 +590,7 @@ static void r300SetDepthState(GLcontext * ctx) R500_STENCIL_REFMASK_FRONT_BACK); r300->hw.zs.cmd[R300_ZS_CNTL_1] &= ~(R300_ZS_MASK << R300_Z_FUNC_SHIFT); - if (ctx->Depth.Test) { + if (ctx->Depth.Test && ctx->DrawBuffer->_DepthBuffer) { r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_Z_ENABLE; if (ctx->Depth.Mask) r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_Z_WRITE_ENABLE; @@ -2354,7 +2354,7 @@ static void r300RenderMode(GLcontext * ctx, GLenum mode) /** * Initialize driver's state callback functions */ -void r300InitStateFuncs(struct dd_function_table *functions) +void r300InitStateFuncs(radeonContextPtr radeon, struct dd_function_table *functions) { functions->UpdateState = r300InvalidateState; @@ -2396,9 +2396,11 @@ void r300InitStateFuncs(struct dd_function_table *functions) functions->DrawBuffer = radeonDrawBuffer; functions->ReadBuffer = radeonReadBuffer; - functions->CopyPixels = _mesa_meta_CopyPixels; - functions->DrawPixels = _mesa_meta_DrawPixels; - functions->ReadPixels = radeonReadPixels; + if (radeon->radeonScreen->kernel_mm) { + functions->CopyPixels = _mesa_meta_CopyPixels; + functions->DrawPixels = _mesa_meta_DrawPixels; + functions->ReadPixels = radeonReadPixels; + } } void r300InitShaderFunctions(r300ContextPtr r300) diff --git a/src/mesa/drivers/dri/r300/r300_state.h b/src/mesa/drivers/dri/r300/r300_state.h index d46bf9f1796..e70f84f4e4b 100644 --- a/src/mesa/drivers/dri/r300/r300_state.h +++ b/src/mesa/drivers/dri/r300/r300_state.h @@ -55,7 +55,7 @@ void r300UpdateDrawBuffer (GLcontext * ctx); void r300UpdateShaders (r300ContextPtr rmesa); void r300UpdateShaderStates (r300ContextPtr rmesa); void r300InitState (r300ContextPtr r300); -void r300InitStateFuncs (struct dd_function_table *functions); +void r300InitStateFuncs (radeonContextPtr radeon, struct dd_function_table *functions); void r300VapCntl(r300ContextPtr rmesa, GLuint input_count, GLuint output_count, GLuint temp_count); void r300SetupVAP(GLcontext *ctx, GLuint InputsRead, GLuint OutputsWritten); diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c index 76d5027649e..fddac2f9bdc 100644 --- a/src/mesa/drivers/dri/r600/r600_context.c +++ b/src/mesa/drivers/dri/r600/r600_context.c @@ -384,7 +384,7 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual, */ _mesa_init_driver_functions(&functions); - r700InitStateFuncs(&functions); + r700InitStateFuncs(&r600->radeon, &functions); r600InitTextureFuncs(&r600->radeon, &functions); r700InitShaderFuncs(&functions); radeonInitQueryObjFunctions(&functions); diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c index 6f156b54096..2953ffd0288 100644 --- a/src/mesa/drivers/dri/r600/r700_state.c +++ b/src/mesa/drivers/dri/r600/r700_state.c @@ -39,6 +39,7 @@ #include "swrast_setup/swrast_setup.h" #include "main/api_arrayelt.h" #include "main/framebuffer.h" +#include "drivers/common/meta.h" #include "shader/prog_parameter.h" #include "shader/prog_statevars.h" @@ -1816,7 +1817,7 @@ void r700InitState(GLcontext * ctx) //------------------- } -void r700InitStateFuncs(struct dd_function_table *functions) //----------------- +void r700InitStateFuncs(radeonContextPtr radeon, struct dd_function_table *functions) { functions->UpdateState = r700InvalidateState; functions->AlphaFunc = r700AlphaFunc; @@ -1857,8 +1858,13 @@ void r700InitStateFuncs(struct dd_function_table *functions) //----------------- functions->Scissor = radeonScissor; - functions->DrawBuffer = radeonDrawBuffer; - functions->ReadBuffer = radeonReadBuffer; + functions->DrawBuffer = radeonDrawBuffer; + functions->ReadBuffer = radeonReadBuffer; + if (radeon->radeonScreen->kernel_mm) { + functions->CopyPixels = _mesa_meta_CopyPixels; + functions->DrawPixels = _mesa_meta_DrawPixels; + functions->ReadPixels = radeonReadPixels; + } } diff --git a/src/mesa/drivers/dri/r600/r700_state.h b/src/mesa/drivers/dri/r600/r700_state.h index 60c6a7f23ca..56885e0b154 100644 --- a/src/mesa/drivers/dri/r600/r700_state.h +++ b/src/mesa/drivers/dri/r600/r700_state.h @@ -40,7 +40,7 @@ extern void r700UpdateShaderStates(GLcontext * ctx); extern void r700UpdateViewportOffset(GLcontext * ctx); extern void r700InitState (GLcontext * ctx); -extern void r700InitStateFuncs (struct dd_function_table *functions); +extern void r700InitStateFuncs (radeonContextPtr radeon, struct dd_function_table *functions); extern void r700SetScissor(context_t *context); diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c index 07e0adc8905..05c65164d60 100644 --- a/src/mesa/drivers/dri/r600/r700_vertprog.c +++ b/src/mesa/drivers/dri/r600/r700_vertprog.c @@ -42,7 +42,7 @@ #include "radeon_debug.h" #include "r600_context.h" #include "r600_cmdbuf.h" -#include "shader/programopt.c" +#include "shader/programopt.h" #include "r700_debug.h" #include "r700_vertprog.h" diff --git a/src/mesa/drivers/dri/radeon/radeon_blit.c b/src/mesa/drivers/dri/radeon/radeon_blit.c index e1e1f215508..143822361e1 100644 --- a/src/mesa/drivers/dri/radeon/radeon_blit.c +++ b/src/mesa/drivers/dri/radeon/radeon_blit.c @@ -48,6 +48,8 @@ unsigned r100_check_blit(gl_format mesa_format) case MESA_FORMAT_ARGB4444: case MESA_FORMAT_ARGB1555: case MESA_FORMAT_A8: + case MESA_FORMAT_L8: + case MESA_FORMAT_I8: break; default: return 0; @@ -103,6 +105,9 @@ static void inline emit_tx_setup(struct r100_context *r100, case MESA_FORMAT_ARGB8888: txformat |= RADEON_TXFORMAT_ARGB8888 | RADEON_TXFORMAT_ALPHA_IN_MAP; break; + case MESA_FORMAT_RGBA8888: + txformat |= RADEON_TXFORMAT_RGBA8888 | RADEON_TXFORMAT_ALPHA_IN_MAP; + break; case MESA_FORMAT_XRGB8888: txformat |= RADEON_TXFORMAT_ARGB8888; break; @@ -116,8 +121,15 @@ static void inline emit_tx_setup(struct r100_context *r100, txformat |= RADEON_TXFORMAT_ARGB1555 | RADEON_TXFORMAT_ALPHA_IN_MAP; break; case MESA_FORMAT_A8: + case MESA_FORMAT_I8: txformat |= RADEON_TXFORMAT_I8 | RADEON_TXFORMAT_ALPHA_IN_MAP; break; + case MESA_FORMAT_L8: + txformat |= RADEON_TXFORMAT_I8; + break; + case MESA_FORMAT_AL88: + txformat |= RADEON_TXFORMAT_AI88 | RADEON_TXFORMAT_ALPHA_IN_MAP; + break; default: break; } @@ -177,6 +189,8 @@ static inline void emit_cb_setup(struct r100_context *r100, dst_format = RADEON_COLOR_FORMAT_ARGB1555; break; case MESA_FORMAT_A8: + case MESA_FORMAT_L8: + case MESA_FORMAT_I8: dst_format = RADEON_COLOR_FORMAT_RGB8; break; default: diff --git a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c index 78c5f5dd572..7f5fb99fa4f 100644 --- a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c +++ b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c @@ -553,6 +553,8 @@ static radeon_mipmap_tree * get_biggest_matching_miptree(radeonTexObj *texObj, } if (mtCount == 0) { + free(mtSizes); + free(mts); return NULL; } diff --git a/src/mesa/drivers/dri/radeon/radeon_pixel_read.c b/src/mesa/drivers/dri/radeon/radeon_pixel_read.c index 27841938e66..b180c1d9a5c 100644 --- a/src/mesa/drivers/dri/radeon/radeon_pixel_read.c +++ b/src/mesa/drivers/dri/radeon/radeon_pixel_read.c @@ -96,7 +96,7 @@ do_blit_readpixels(GLcontext * ctx, return GL_FALSE; } - if (ctx->_ImageTransferState) { + if (ctx->_ImageTransferState || ctx->Color._LogicOpEnabled) { return GL_FALSE; } diff --git a/src/mesa/drivers/dri/radeon/radeon_state.c b/src/mesa/drivers/dri/radeon/radeon_state.c index 0ce97e86972..0afbc19c127 100644 --- a/src/mesa/drivers/dri/radeon/radeon_state.c +++ b/src/mesa/drivers/dri/radeon/radeon_state.c @@ -45,6 +45,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. #include "tnl/tnl.h" #include "tnl/t_pipeline.h" #include "swrast_setup/swrast_setup.h" +#include "drivers/common/meta.h" #include "radeon_context.h" #include "radeon_mipmap_tree.h" @@ -1900,7 +1901,7 @@ void radeonUploadTexMatrix( r100ContextPtr rmesa, So: if we need the q coord in the end (solely determined by the texture target, i.e. 2d / 1d / texrect targets) we swap the third and 4th row. Additionally, if we don't have texgen but 4 tex coords submitted, we swap - column 3 and 4 (for the 2d / 1d / texrect targets) since the the q coord + column 3 and 4 (for the 2d / 1d / texrect targets) since the q coord will get submitted in the "wrong", i.e. 3rd, slot. If an app submits 3 coords for 2d targets, we assume it is saving on vertex size and using the texture matrix to swap the r and q coords around (ut2k3 @@ -2248,6 +2249,11 @@ void radeonInitStateFuncs( GLcontext *ctx , GLboolean dri2 ) ctx->Driver.DrawBuffer = radeonDrawBuffer; ctx->Driver.ReadBuffer = radeonReadBuffer; + if (dri2) { + ctx->Driver.CopyPixels = _mesa_meta_CopyPixels; + ctx->Driver.DrawPixels = _mesa_meta_DrawPixels; + ctx->Driver.ReadPixels = radeonReadPixels; + } ctx->Driver.AlphaFunc = radeonAlphaFunc; ctx->Driver.BlendEquationSeparate = radeonBlendEquationSeparate; diff --git a/src/mesa/drivers/dri/radeon/radeon_tex_copy.c b/src/mesa/drivers/dri/radeon/radeon_tex_copy.c index a4bb03d5d39..e57d77e7ef2 100644 --- a/src/mesa/drivers/dri/radeon/radeon_tex_copy.c +++ b/src/mesa/drivers/dri/radeon/radeon_tex_copy.c @@ -52,15 +52,18 @@ do_copy_texsubimage(GLcontext *ctx, gl_format dst_mesaformat; unsigned src_width; unsigned dst_width; + unsigned flip_y; if (!radeon->vtbl.blit) { return GL_FALSE; } if (_mesa_get_format_bits(timg->base.TexFormat, GL_DEPTH_BITS) > 0) { - rrb = radeon_get_depthbuffer(radeon); + rrb = radeon_renderbuffer(ctx->ReadBuffer->_DepthBuffer); + flip_y = ctx->ReadBuffer->Attachment[BUFFER_DEPTH].Type == GL_NONE; } else { - rrb = radeon_get_colorbuffer(radeon); + rrb = radeon_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer); + flip_y = ctx->ReadBuffer->Attachment[BUFFER_COLOR0].Type == GL_NONE; } if (!timg->mt) { @@ -93,6 +96,10 @@ do_copy_texsubimage(GLcontext *ctx, src_bpp = _mesa_get_format_bytes(src_mesaformat); dst_bpp = _mesa_get_format_bytes(dst_mesaformat); if (!radeon->vtbl.check_blit(dst_mesaformat)) { + /* depth formats tend to be special */ + if (_mesa_get_format_bits(dst_mesaformat, GL_DEPTH_BITS) > 0) + return GL_FALSE; + if (src_bpp != dst_bpp) return GL_FALSE; @@ -120,7 +127,7 @@ do_copy_texsubimage(GLcontext *ctx, timg->mt->bo, dst_offset, dst_mesaformat, timg->mt->levels[level].rowstride / dst_bpp, dst_width, timg->base.Height, - dstx, dsty, width, height, 1); + dstx, dsty, width, height, flip_y); } void diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c index 91ddc3615a1..22b0c46b4f7 100644 --- a/src/mesa/drivers/windows/gdi/wmesa.c +++ b/src/mesa/drivers/windows/gdi/wmesa.c @@ -248,16 +248,6 @@ static void wmesa_flush(GLcontext *ctx) */ /* - * Set the color index used to clear the color buffer. - */ -static void clear_index(GLcontext *ctx, GLuint index) -{ - WMesaContext pwc = wmesa_context(ctx); - /* Note that indexed mode is not supported yet */ - pwc->clearColorRef = RGB(0,0,0); -} - -/* * Set the color used to clear the color buffer. */ static void clear_color(GLcontext *ctx, const GLfloat color[4]) @@ -482,7 +472,7 @@ static void write_rgba_span_front(const GLcontext *ctx, }; } BGRA; BGRA *bgra, c; - int i; + GLuint i; if (n < 16) { // the value 16 is just guessed y=FLIP(y); @@ -827,9 +817,9 @@ static void read_rgba_span_32(const GLcontext *ctx, lpdw = ((LPDWORD)(pwfb->pbPixels + pwfb->ScanWidth * y)) + x; for (i=0; i<n; i++) { pixel = lpdw[i]; - rgba[i][RCOMP] = (pixel & 0x00ff0000) >> 16; - rgba[i][GCOMP] = (pixel & 0x0000ff00) >> 8; - rgba[i][BCOMP] = (pixel & 0x000000ff); + rgba[i][RCOMP] = (GLubyte)((pixel & 0x00ff0000) >> 16); + rgba[i][GCOMP] = (GLubyte)((pixel & 0x0000ff00) >> 8); + rgba[i][BCOMP] = (GLubyte)(pixel & 0x000000ff); rgba[i][ACOMP] = 255; } } @@ -851,9 +841,9 @@ static void read_rgba_pixels_32(const GLcontext *ctx, GLint y2 = FLIP(y[i]); lpdw = ((LPDWORD)(pwfb->pbPixels + pwfb->ScanWidth * y2)) + x[i]; pixel = *lpdw; - rgba[i][RCOMP] = (pixel & 0x00ff0000) >> 16; - rgba[i][GCOMP] = (pixel & 0x0000ff00) >> 8; - rgba[i][BCOMP] = (pixel & 0x000000ff); + rgba[i][RCOMP] = (GLubyte)((pixel & 0x00ff0000) >> 16); + rgba[i][GCOMP] = (GLubyte)((pixel & 0x0000ff00) >> 8); + rgba[i][BCOMP] = (GLubyte)(pixel & 0x000000ff); rgba[i][ACOMP] = 255; } } @@ -1271,7 +1261,7 @@ wmesa_renderbuffer_storage(GLcontext *ctx, * on if we're drawing to the front or back color buffer. */ void wmesa_set_renderbuffer_funcs(struct gl_renderbuffer *rb, int pixelformat, - BYTE cColorBits, int double_buffer) + int cColorBits, int double_buffer) { if (double_buffer) { /* back buffer */ @@ -1483,7 +1473,6 @@ WMesaContext WMesaCreateContext(HDC hDC, functions.GetBufferSize = wmesa_get_buffer_size; functions.Flush = wmesa_flush; functions.Clear = clear; - functions.ClearIndex = clear_index; functions.ClearColor = clear_color; functions.ResizeBuffers = wmesa_resize_buffers; functions.Viewport = wmesa_viewport; diff --git a/src/mesa/drivers/windows/gdi/wmesadef.h b/src/mesa/drivers/windows/gdi/wmesadef.h index 83a42e60824..1c0e2451114 100644 --- a/src/mesa/drivers/windows/gdi/wmesadef.h +++ b/src/mesa/drivers/windows/gdi/wmesadef.h @@ -27,7 +27,7 @@ struct wmesa_framebuffer HDC hDC; int pixelformat; GLuint ScanWidth; - BYTE cColorBits; + int cColorBits; /* back buffer DIB fields */ HDC dib_hDC; BITMAPINFO bmi; diff --git a/src/mesa/drivers/x11/xmesa.h b/src/mesa/drivers/x11/xmesa.h index 98139af8336..f63626a9702 100644 --- a/src/mesa/drivers/x11/xmesa.h +++ b/src/mesa/drivers/x11/xmesa.h @@ -287,7 +287,7 @@ extern void XMesaCopySubBuffer( XMesaBuffer b, /* - * Return a pointer to the the Pixmap or XImage being used as the back + * Return a pointer to the Pixmap or XImage being used as the back * color buffer of an XMesaBuffer. This function is a way to get "under * the hood" of X/Mesa so one can manipulate the back buffer directly. * Input: b - the XMesaBuffer diff --git a/src/mesa/drivers/x11/xmesaP.h b/src/mesa/drivers/x11/xmesaP.h index 3ffd7661e35..e0a6908228d 100644 --- a/src/mesa/drivers/x11/xmesaP.h +++ b/src/mesa/drivers/x11/xmesaP.h @@ -431,7 +431,7 @@ extern const int xmesa_kernel8[DITH_DY * DITH_DX]; * If pixelformat==PF_HPCR: * * HP Color Recovery dithering (ad@lms.be 30/08/95) - * HP has on it's 8-bit 700-series computers, a feature called + * HP has on its 8-bit 700-series computers, a feature called * 'Color Recovery'. This allows near 24-bit output (so they say). * It is enabled by selecting the 8-bit TrueColor visual AND * corresponding colormap (see tkInitWindow) AND doing some special diff --git a/src/mesa/es/state_tracker/st_cb_drawtex.c b/src/mesa/es/state_tracker/st_cb_drawtex.c index 0a5cba9d927..f75f4861a2f 100644 --- a/src/mesa/es/state_tracker/st_cb_drawtex.c +++ b/src/mesa/es/state_tracker/st_cb_drawtex.c @@ -120,8 +120,11 @@ st_DrawTex(GLcontext *ctx, GLfloat x, GLfloat y, GLfloat z, GLboolean emitColor; uint semantic_names[2 + MAX_TEXTURE_UNITS]; uint semantic_indexes[2 + MAX_TEXTURE_UNITS]; + struct pipe_vertex_element velements[2 + MAX_TEXTURE_UNITS]; GLbitfield inputs = VERT_BIT_POS; + st_validate_state(st); + /* determine if we need vertex color */ if (ctx->FragmentProgram._Current->Base.InputsRead & FRAG_BIT_COL0) emitColor = GL_TRUE; @@ -232,6 +235,7 @@ st_DrawTex(GLcontext *ctx, GLfloat x, GLfloat y, GLfloat z, cso_save_viewport(cso); cso_save_vertex_shader(cso); + cso_save_vertex_elements(cso); { void *vs = lookup_shader(pipe, numAttribs, @@ -239,6 +243,14 @@ st_DrawTex(GLcontext *ctx, GLfloat x, GLfloat y, GLfloat z, cso_set_vertex_shader_handle(cso, vs); } + for (i = 0; i < numAttribs; i++) { + velements[i].src_offset = i * 4 * sizeof(float); + velements[i].instance_divisor = 0; + velements[i].vertex_buffer_index = 0; + velements[i].src_format = PIPE_FORMAT_R32G32B32A32_FLOAT; + } + cso_set_vertex_elements(cso, numAttribs, velements); + /* viewport state: viewport matching window dims */ { const struct gl_framebuffer *fb = st->ctx->DrawBuffer; @@ -270,6 +282,7 @@ st_DrawTex(GLcontext *ctx, GLfloat x, GLfloat y, GLfloat z, /* restore state */ cso_restore_viewport(cso); cso_restore_vertex_shader(cso); + cso_restore_vertex_elements(cso); } diff --git a/src/mesa/glapi/glapi_getproc.c b/src/mesa/glapi/glapi_getproc.c index 46b466920b6..295657875dc 100644 --- a/src/mesa/glapi/glapi_getproc.c +++ b/src/mesa/glapi/glapi_getproc.c @@ -44,11 +44,14 @@ #include "glapi/glapioffsets.h" +/********************************************************************** + * Static function management. + */ + + #if !defined(DISPATCH_FUNCTION_SIZE) && !defined(XFree86Server) # define NEED_FUNCTION_POINTER #endif - -/* The code in this file is auto-generated with Python */ #include "glapi/glprocs.h" @@ -57,7 +60,7 @@ * and return the corresponding glprocs_table_t entry. */ static const glprocs_table_t * -find_entry( const char * n ) +get_static_proc( const char * n ) { GLuint i; for (i = 0; static_functions[i].Name_offset >= 0; i++) { @@ -83,11 +86,12 @@ find_entry( const char * n ) static GLint get_static_proc_offset(const char *funcName) { - const glprocs_table_t * const f = find_entry( funcName ); - if (f) { - return f->Offset; + const glprocs_table_t * const f = get_static_proc( funcName ); + if (f == NULL) { + return -1; } - return -1; + + return f->Offset; } @@ -100,25 +104,32 @@ get_static_proc_offset(const char *funcName) static _glapi_proc get_static_proc_address(const char *funcName) { - const glprocs_table_t * const f = find_entry( funcName ); - if (f) { + const glprocs_table_t * const f = get_static_proc( funcName ); + if (f == NULL) { + return NULL; + } + #if defined(DISPATCH_FUNCTION_SIZE) && defined(GLX_INDIRECT_RENDERING) - return (f->Address == NULL) - ? get_entrypoint_address(f->Offset) - : f->Address; + return (f->Address == NULL) + ? get_entrypoint_address(f->Offset) + : f->Address; #elif defined(DISPATCH_FUNCTION_SIZE) - return get_entrypoint_address(f->Offset); + return get_entrypoint_address(f->Offset); #else - return f->Address; + return f->Address; #endif - } - else { - return NULL; - } } -#endif /* !defined(XFree86Server) */ +#else +static _glapi_proc +get_static_proc_address(const char *funcName) +{ + (void) funcName; + return NULL; +} + +#endif /* !defined(XFree86Server) */ /** @@ -199,6 +210,56 @@ static struct _glapi_function ExtEntryTable[MAX_EXTENSION_FUNCS]; static GLuint NumExtEntryPoints = 0; +static struct _glapi_function * +get_extension_proc(const char *funcName) +{ + GLuint i; + for (i = 0; i < NumExtEntryPoints; i++) { + if (strcmp(ExtEntryTable[i].name, funcName) == 0) { + return & ExtEntryTable[i]; + } + } + return NULL; +} + + +static GLint +get_extension_proc_offset(const char *funcName) +{ + const struct _glapi_function * const f = get_extension_proc( funcName ); + if (f == NULL) { + return -1; + } + + return f->dispatch_offset; +} + + +static _glapi_proc +get_extension_proc_address(const char *funcName) +{ + const struct _glapi_function * const f = get_extension_proc( funcName ); + if (f == NULL) { + return NULL; + } + + return f->dispatch_stub; +} + + +static const char * +get_extension_proc_name(GLuint offset) +{ + GLuint i; + for (i = 0; i < NumExtEntryPoints; i++) { + if (ExtEntryTable[i].dispatch_offset == offset) { + return ExtEntryTable[i].name; + } + } + return NULL; +} + + /** * strdup() is actually not a standard ANSI C or POSIX routine. * Irix will not define it if ANSI mode is in effect. @@ -232,20 +293,55 @@ static struct _glapi_function * add_function_name( const char * funcName ) { struct _glapi_function * entry = NULL; - - if (NumExtEntryPoints < MAX_EXTENSION_FUNCS) { - _glapi_proc entrypoint = generate_entrypoint(~0); - if (entrypoint != NULL) { - entry = & ExtEntryTable[NumExtEntryPoints]; - - ExtEntryTable[NumExtEntryPoints].name = str_dup(funcName); - ExtEntryTable[NumExtEntryPoints].parameter_signature = NULL; - ExtEntryTable[NumExtEntryPoints].dispatch_offset = ~0; - ExtEntryTable[NumExtEntryPoints].dispatch_stub = entrypoint; - NumExtEntryPoints++; - } + _glapi_proc entrypoint = NULL; + char * name_dup = NULL; + + if (NumExtEntryPoints >= MAX_EXTENSION_FUNCS) + return NULL; + + if (funcName == NULL) + return NULL; + + name_dup = str_dup(funcName); + if (name_dup == NULL) + return NULL; + + entrypoint = generate_entrypoint(~0); + + if (entrypoint == NULL) { + free(name_dup); + return NULL; } + entry = & ExtEntryTable[NumExtEntryPoints]; + NumExtEntryPoints++; + + entry->name = name_dup; + entry->parameter_signature = NULL; + entry->dispatch_offset = ~0; + entry->dispatch_stub = entrypoint; + + return entry; +} + + +static struct _glapi_function * +set_entry_info( struct _glapi_function * entry, const char * signature, unsigned offset ) +{ + char * sig_dup = NULL; + + if (signature == NULL) + return NULL; + + sig_dup = str_dup(signature); + if (sig_dup == NULL) + return NULL; + + fill_in_entrypoint_offset(entry->dispatch_stub, offset); + + entry->parameter_signature = sig_dup; + entry->dispatch_offset = offset; + return entry; } @@ -307,88 +403,103 @@ _glapi_add_dispatch( const char * const * function_names, struct _glapi_function * entry[8]; GLboolean is_static[8]; unsigned i; - unsigned j; int offset = ~0; - int new_offset; + init_glapi_relocs_once(); (void) memset( is_static, 0, sizeof( is_static ) ); (void) memset( entry, 0, sizeof( entry ) ); + /* Find the _single_ dispatch offset for all function names that already + * exist (and have a dispatch offset). + */ + for ( i = 0 ; function_names[i] != NULL ; i++ ) { - /* Do some trivial validation on the name of the function. - */ + const char * funcName = function_names[i]; + int static_offset; + int extension_offset; - if (!function_names[i] || function_names[i][0] != 'g' || function_names[i][1] != 'l') + if (funcName[0] != 'g' || funcName[1] != 'l') return -1; - - /* Determine if the named function already exists. If the function does - * exist, it must have the same parameter signature as the function - * being added. - */ - - new_offset = get_static_proc_offset(function_names[i]); - if (new_offset >= 0) { + + /* search built-in functions */ + static_offset = get_static_proc_offset(funcName); + + if (static_offset >= 0) { + + is_static[i] = GL_TRUE; + /* FIXME: Make sure the parameter signatures match! How do we get * FIXME: the parameter signature for static functions? */ - if ( (offset != ~0) && (new_offset != offset) ) { + if ( (offset != ~0) && (static_offset != offset) ) { return -1; } - is_static[i] = GL_TRUE; - offset = new_offset; + offset = static_offset; + + continue; } - - - for ( j = 0 ; j < NumExtEntryPoints ; j++ ) { - if (strcmp(ExtEntryTable[j].name, function_names[i]) == 0) { - /* The offset may be ~0 if the function name was added by - * glXGetProcAddress but never filled in by the driver. - */ - - if (ExtEntryTable[j].dispatch_offset != ~0) { - if (strcmp(real_sig, ExtEntryTable[j].parameter_signature) - != 0) { - return -1; - } - - if ( (offset != ~0) && (ExtEntryTable[j].dispatch_offset != offset) ) { - return -1; - } - - offset = ExtEntryTable[j].dispatch_offset; - } - - entry[i] = & ExtEntryTable[j]; - break; + + /* search added extension functions */ + entry[i] = get_extension_proc(funcName); + + if (entry[i] != NULL) { + extension_offset = entry[i]->dispatch_offset; + + /* The offset may be ~0 if the function name was added by + * glXGetProcAddress but never filled in by the driver. + */ + + if (extension_offset == ~0) { + continue; + } + + if (strcmp(real_sig, entry[i]->parameter_signature) != 0) { + return -1; + } + + if ( (offset != ~0) && (extension_offset != offset) ) { + return -1; } + + offset = extension_offset; } } + /* If all function names are either new (or with no dispatch offset), + * allocate a new dispatch offset. + */ + if (offset == ~0) { offset = next_dynamic_offset; next_dynamic_offset++; } + /* Fill in the dispatch offset for the new function names (and those with + * no dispatch offset). + */ + for ( i = 0 ; function_names[i] != NULL ; i++ ) { - if (! is_static[i] ) { + if (is_static[i]) { + continue; + } + + /* generate entrypoints for new function names */ + if (entry[i] == NULL) { + entry[i] = add_function_name( function_names[i] ); if (entry[i] == NULL) { - entry[i] = add_function_name( function_names[i] ); - if (entry[i] == NULL) { - /* FIXME: Possible memory leak here. - */ - return -1; - } + /* FIXME: Possible memory leak here. */ + return -1; } + } - entry[i]->parameter_signature = str_dup(real_sig); - fill_in_entrypoint_offset(entry[i]->dispatch_stub, offset); - entry[i]->dispatch_offset = offset; + if (entry[i]->dispatch_offset == ~0) { + set_entry_info( entry[i], real_sig, offset ); } } - + return offset; } @@ -399,13 +510,13 @@ _glapi_add_dispatch( const char * const * function_names, PUBLIC GLint _glapi_get_proc_offset(const char *funcName) { + GLint offset; + /* search extension functions first */ - GLuint i; - for (i = 0; i < NumExtEntryPoints; i++) { - if (strcmp(ExtEntryTable[i].name, funcName) == 0) { - return ExtEntryTable[i].dispatch_offset; - } - } + offset = get_extension_proc_offset(funcName); + if (offset >= 0) + return offset; + /* search static functions */ return get_static_proc_offset(funcName); } @@ -420,8 +531,10 @@ _glapi_get_proc_offset(const char *funcName) PUBLIC _glapi_proc _glapi_get_proc_address(const char *funcName) { + _glapi_proc func; struct _glapi_function * entry; - GLuint i; + + init_glapi_relocs_once(); #ifdef MANGLE /* skip the prefix on the name */ @@ -433,23 +546,21 @@ _glapi_get_proc_address(const char *funcName) #endif /* search extension functions first */ - for (i = 0; i < NumExtEntryPoints; i++) { - if (strcmp(ExtEntryTable[i].name, funcName) == 0) { - return ExtEntryTable[i].dispatch_stub; - } - } + func = get_extension_proc_address(funcName); + if (func) + return func; -#if !defined( XFree86Server ) /* search static functions */ - { - const _glapi_proc func = get_static_proc_address(funcName); - if (func) - return func; - } -#endif /* !defined( XFree86Server ) */ + func = get_static_proc_address(funcName); + if (func) + return func; + /* generate entrypoint, dispatch offset must be filled in by the driver */ entry = add_function_name(funcName); - return (entry == NULL) ? NULL : entry->dispatch_stub; + if (entry == NULL) + return NULL; + + return entry->dispatch_stub; } @@ -461,7 +572,6 @@ _glapi_get_proc_address(const char *funcName) const char * _glapi_get_proc_name(GLuint offset) { - GLuint i; const char * n; /* search built-in functions */ @@ -471,12 +581,7 @@ _glapi_get_proc_name(GLuint offset) } /* search added extension functions */ - for (i = 0; i < NumExtEntryPoints; i++) { - if (ExtEntryTable[i].dispatch_offset == offset) { - return ExtEntryTable[i].name; - } - } - return NULL; + return get_extension_proc_name(offset); } diff --git a/src/mesa/main/dd.h b/src/mesa/main/dd.h index 197de09b22a..7c02faaa535 100644 --- a/src/mesa/main/dd.h +++ b/src/mesa/main/dd.h @@ -1072,7 +1072,7 @@ struct dd_function_table { * These are the initial values to be installed into dispatch by * mesa. If the T&L driver wants to modify the dispatch table * while installed, it must do so itself. It would be possible for - * the vertexformat to install it's own initial values for these + * the vertexformat to install its own initial values for these * functions, but this way there is an obvious list of what is * expected of the driver. * diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c index 14c533e0d43..7c442e390c2 100644 --- a/src/mesa/main/fbobject.c +++ b/src/mesa/main/fbobject.c @@ -1549,6 +1549,7 @@ framebuffer_texture(GLcontext *ctx, const char *caller, GLenum target, texObj = _mesa_lookup_texture(ctx, texture); if (texObj != NULL) { if (textarget == 0) { + /* XXX what's the purpose of this? */ err = (texObj->Target != GL_TEXTURE_3D) && (texObj->Target != GL_TEXTURE_1D_ARRAY_EXT) && (texObj->Target != GL_TEXTURE_2D_ARRAY_EXT); @@ -1559,6 +1560,13 @@ framebuffer_texture(GLcontext *ctx, const char *caller, GLenum target, : (texObj->Target != textarget); } } + else { + /* can't render to a non-existant texture */ + _mesa_error(ctx, GL_INVALID_OPERATION, + "glFramebufferTexture%sEXT(non existant texture)", + caller); + return; + } if (err) { _mesa_error(ctx, GL_INVALID_OPERATION, diff --git a/src/mesa/main/framebuffer.c b/src/mesa/main/framebuffer.c index 6a85162d5da..5a654e5c2a3 100644 --- a/src/mesa/main/framebuffer.c +++ b/src/mesa/main/framebuffer.c @@ -986,6 +986,10 @@ _mesa_dest_buffer_exists(GLcontext *ctx, GLenum format) return GL_TRUE; } + +/** + * Used to answer the GL_IMPLEMENTATION_COLOR_READ_FORMAT_OES query. + */ GLenum _mesa_get_color_read_format(GLcontext *ctx) { @@ -999,6 +1003,10 @@ _mesa_get_color_read_format(GLcontext *ctx) } } + +/** + * Used to answer the GL_IMPLEMENTATION_COLOR_READ_TYPE_OES query. + */ GLenum _mesa_get_color_read_type(GLcontext *ctx) { diff --git a/src/mesa/main/texcompress_fxt1.c b/src/mesa/main/texcompress_fxt1.c index 149853f7acd..04acf05e528 100644 --- a/src/mesa/main/texcompress_fxt1.c +++ b/src/mesa/main/texcompress_fxt1.c @@ -476,7 +476,7 @@ fxt1_lloyd (GLfloat vec[][MAX_COMP], GLint nv, * for each sample color * sort to nearest vector. * - * replace each vector with the centroid of it's matching colors. + * replace each vector with the centroid of its matching colors. * * repeat until RMS doesn't improve. * diff --git a/src/mesa/math/m_debug_util.h b/src/mesa/math/m_debug_util.h index 2e67db8e55d..ed11c849ece 100644 --- a/src/mesa/math/m_debug_util.h +++ b/src/mesa/math/m_debug_util.h @@ -61,7 +61,7 @@ extern long counter_overhead; */ extern char *mesa_profile; -/* Modify the the number of tests if you like. +/* Modify the number of tests if you like. * We take the minimum of all results, because every error should be * positive (time used by other processes, task switches etc). * It is assumed that all calculations are done in the cache. diff --git a/src/mesa/math/m_matrix.c b/src/mesa/math/m_matrix.c index ef8a40fbecb..4b33d0bbb37 100644 --- a/src/mesa/math/m_matrix.c +++ b/src/mesa/math/m_matrix.c @@ -889,7 +889,7 @@ _math_matrix_rotate( GLmatrix *mat, * Y-axis to bring the axis vector parallel with the X-axis. The * rotation about the X-axis is then performed. Ry and Rz are * simply the respective inverse transforms to bring the arbitrary - * axis back to it's original orientation. The first transforms + * axis back to its original orientation. The first transforms * Rz' and Ry' are considered inverses, since the data from the * arbitrary axis gives you info on how to get to it, not how * to get away from it, and an inverse must be applied. diff --git a/src/mesa/shader/lex.yy.c b/src/mesa/shader/lex.yy.c index d1af35fedb6..a08617ff8d4 100644 --- a/src/mesa/shader/lex.yy.c +++ b/src/mesa/shader/lex.yy.c @@ -53,6 +53,7 @@ typedef int flex_int32_t; typedef unsigned char flex_uint8_t; typedef unsigned short int flex_uint16_t; typedef unsigned int flex_uint32_t; +#endif /* ! C99 */ /* Limits of integral types. */ #ifndef INT8_MIN @@ -83,8 +84,6 @@ typedef unsigned int flex_uint32_t; #define UINT32_MAX (4294967295U) #endif -#endif /* ! C99 */ - #endif /* ! FLEXINT_H */ #ifdef __cplusplus @@ -158,15 +157,7 @@ typedef void* yyscan_t; /* Size of default input buffer. */ #ifndef YY_BUF_SIZE -#ifdef __ia64__ -/* On IA-64, the buffer size is 16k, not 8k. - * Moreover, YY_BUF_SIZE is 2*YY_READ_BUF_SIZE in the general case. - * Ditto for the __ia64__ case accordingly. - */ -#define YY_BUF_SIZE 32768 -#else #define YY_BUF_SIZE 16384 -#endif /* __ia64__ */ #endif /* The state buf must be large enough to hold one state per character in the main buffer. @@ -1161,7 +1152,7 @@ handle_ident(struct asm_parser_state *state, const char *text, YYSTYPE *lval) } while(0); #define YY_EXTRA_TYPE struct asm_parser_state * -#line 1165 "lex.yy.c" +#line 1156 "lex.yy.c" #define INITIAL 0 @@ -1298,12 +1289,7 @@ static int input (yyscan_t yyscanner ); /* Amount of stuff to slurp up with each read. */ #ifndef YY_READ_BUF_SIZE -#ifdef __ia64__ -/* On IA-64, the buffer size is 16k, not 8k */ -#define YY_READ_BUF_SIZE 16384 -#else #define YY_READ_BUF_SIZE 8192 -#endif /* __ia64__ */ #endif /* Copy whatever the last rule matched to the standard output. */ @@ -1311,7 +1297,7 @@ static int input (yyscan_t yyscanner ); /* This used to be an fputs(), but since the string might contain NUL's, * we now use fwrite(). */ -#define ECHO do { if (fwrite( yytext, yyleng, 1, yyout )) {} } while (0) +#define ECHO fwrite( yytext, yyleng, 1, yyout ) #endif /* Gets input and stuffs it into "buf". number of characters read, or YY_NULL, @@ -1322,7 +1308,7 @@ static int input (yyscan_t yyscanner ); if ( YY_CURRENT_BUFFER_LVALUE->yy_is_interactive ) \ { \ int c = '*'; \ - size_t n; \ + unsigned n; \ for ( n = 0; n < max_size && \ (c = getc( yyin )) != EOF && c != '\n'; ++n ) \ buf[n] = (char) c; \ @@ -1410,7 +1396,7 @@ YY_DECL #line 157 "program_lexer.l" -#line 1414 "lex.yy.c" +#line 1400 "lex.yy.c" yylval = yylval_param; @@ -2212,7 +2198,7 @@ case 142: YY_RULE_SETUP #line 326 "program_lexer.l" { - yylval->real = _mesa_strtod(yytext, NULL); + yylval->real = (float) _mesa_strtod(yytext, NULL); return REAL; } YY_BREAK @@ -2224,7 +2210,7 @@ YY_DO_BEFORE_ACTION; /* set up yytext again */ YY_RULE_SETUP #line 330 "program_lexer.l" { - yylval->real = _mesa_strtod(yytext, NULL); + yylval->real = (float) _mesa_strtod(yytext, NULL); return REAL; } YY_BREAK @@ -2232,7 +2218,7 @@ case 144: YY_RULE_SETUP #line 334 "program_lexer.l" { - yylval->real = _mesa_strtod(yytext, NULL); + yylval->real = (float) _mesa_strtod(yytext, NULL); return REAL; } YY_BREAK @@ -2240,7 +2226,7 @@ case 145: YY_RULE_SETUP #line 338 "program_lexer.l" { - yylval->real = _mesa_strtod(yytext, NULL); + yylval->real = (float) _mesa_strtod(yytext, NULL); return REAL; } YY_BREAK @@ -2474,7 +2460,7 @@ YY_RULE_SETUP #line 481 "program_lexer.l" ECHO; YY_BREAK -#line 2478 "lex.yy.c" +#line 2464 "lex.yy.c" case YY_STATE_EOF(INITIAL): yyterminate(); @@ -3242,8 +3228,8 @@ YY_BUFFER_STATE yy_scan_string (yyconst char * yystr , yyscan_t yyscanner) /** Setup the input buffer state to scan the given bytes. The next call to yylex() will * scan from a @e copy of @a bytes. - * @param yybytes the byte buffer to scan - * @param _yybytes_len the number of bytes in the buffer pointed to by @a bytes. + * @param bytes the byte buffer to scan + * @param len the number of bytes in the buffer pointed to by @a bytes. * @param yyscanner The scanner object. * @return the newly allocated buffer state object. */ diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h index 224350caac6..28c797a4ba8 100644 --- a/src/mesa/shader/prog_instruction.h +++ b/src/mesa/shader/prog_instruction.h @@ -97,8 +97,8 @@ #define COND_EQ 2 /**< equal to zero */ #define COND_LT 3 /**< less than zero */ #define COND_UN 4 /**< unordered (NaN) */ -#define COND_GE 5 /**< greater then or equal to zero */ -#define COND_LE 6 /**< less then or equal to zero */ +#define COND_GE 5 /**< greater than or equal to zero */ +#define COND_LE 6 /**< less than or equal to zero */ #define COND_NE 7 /**< not equal to zero */ #define COND_TR 8 /**< always true */ #define COND_FL 9 /**< always false */ diff --git a/src/mesa/shader/program_lexer.l b/src/mesa/shader/program_lexer.l index 83bc5089d9e..b00765793dc 100644 --- a/src/mesa/shader/program_lexer.l +++ b/src/mesa/shader/program_lexer.l @@ -324,19 +324,19 @@ ARRAYSHADOW2D { return_token_or_IDENTIFIER(require_ARB_fp && require return INTEGER; } {num}?{frac}{exp}? { - yylval->real = _mesa_strtod(yytext, NULL); + yylval->real = (float) _mesa_strtod(yytext, NULL); return REAL; } {num}"."/[^.] { - yylval->real = _mesa_strtod(yytext, NULL); + yylval->real = (float) _mesa_strtod(yytext, NULL); return REAL; } {num}{exp} { - yylval->real = _mesa_strtod(yytext, NULL); + yylval->real = (float) _mesa_strtod(yytext, NULL); return REAL; } {num}"."{exp} { - yylval->real = _mesa_strtod(yytext, NULL); + yylval->real = (float) _mesa_strtod(yytext, NULL); return REAL; } diff --git a/src/mesa/shader/program_parser.h b/src/mesa/shader/program_parser.h index 730466c30f5..be952d4b9c8 100644 --- a/src/mesa/shader/program_parser.h +++ b/src/mesa/shader/program_parser.h @@ -62,7 +62,7 @@ struct asm_symbol { */ unsigned param_binding_swizzle; - /* This is how many entries in the the program_parameter_list we take up + /* This is how many entries in the program_parameter_list we take up * with our state tokens or constants. Note that this is _not_ the same as * the number of param registers we eventually use. */ diff --git a/src/mesa/shader/slang/library/slang_common_builtin.gc b/src/mesa/shader/slang/library/slang_common_builtin.gc index 8b7771c2846..a25ca55bc42 100644 --- a/src/mesa/shader/slang/library/slang_common_builtin.gc +++ b/src/mesa/shader/slang/library/slang_common_builtin.gc @@ -695,7 +695,7 @@ vec3 normalize(const vec3 v) { // const float s = inversesqrt(dot(v, v)); // __retVal = v * s; -// XXX note, we _could_ use __retVal.w instead of tmp and and save a +// XXX note, we _could_ use __retVal.w instead of tmp and save a // register, but that's actually a compilation error because v is a vec3 // and the .w suffix is illegal. Oh well. float tmp; diff --git a/src/mesa/state_tracker/st_atom_pixeltransfer.c b/src/mesa/state_tracker/st_atom_pixeltransfer.c index e766b3a9038..03e33361448 100644 --- a/src/mesa/state_tracker/st_atom_pixeltransfer.c +++ b/src/mesa/state_tracker/st_atom_pixeltransfer.c @@ -138,7 +138,6 @@ static void load_color_map_texture(GLcontext *ctx, struct pipe_texture *pt) { struct pipe_context *pipe = ctx->st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *transfer; const GLuint rSize = ctx->PixelMaps.RtoR.Size; const GLuint gSize = ctx->PixelMaps.GtoG.Size; @@ -151,7 +150,7 @@ load_color_map_texture(GLcontext *ctx, struct pipe_texture *pt) transfer = st_cond_flush_get_tex_transfer(st_context(ctx), pt, 0, 0, 0, PIPE_TRANSFER_WRITE, 0, 0, texSize, texSize); - dest = (uint *) screen->transfer_map(screen, transfer); + dest = (uint *) pipe->transfer_map(pipe, transfer); /* Pack four 1D maps into a 2D texture: * R map is placed horizontally, indexed by S, in channel 0 @@ -172,8 +171,8 @@ load_color_map_texture(GLcontext *ctx, struct pipe_texture *pt) } } - screen->transfer_unmap(screen, transfer); - screen->tex_transfer_destroy(transfer); + pipe->transfer_unmap(pipe, transfer); + pipe->tex_transfer_destroy(pipe, transfer); } diff --git a/src/mesa/state_tracker/st_cb_accum.c b/src/mesa/state_tracker/st_cb_accum.c index 33e43ddcc4c..01aba3e3dd4 100644 --- a/src/mesa/state_tracker/st_cb_accum.c +++ b/src/mesa/state_tracker/st_cb_accum.c @@ -129,7 +129,6 @@ accum_accum(struct st_context *st, GLfloat value, struct st_renderbuffer *color_strb) { struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *color_trans; size_t stride = acc_strb->stride; GLubyte *data = acc_strb->data; @@ -145,7 +144,7 @@ accum_accum(struct st_context *st, GLfloat value, buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - pipe_get_tile_rgba(color_trans, 0, 0, width, height, buf); + pipe_get_tile_rgba(pipe, color_trans, 0, 0, width, height, buf); switch (acc_strb->format) { case PIPE_FORMAT_R16G16B16A16_SNORM: @@ -166,7 +165,7 @@ accum_accum(struct st_context *st, GLfloat value, } free(buf); - screen->tex_transfer_destroy(color_trans); + pipe->tex_transfer_destroy(pipe, color_trans); } @@ -177,7 +176,6 @@ accum_load(struct st_context *st, GLfloat value, struct st_renderbuffer *color_strb) { struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *color_trans; size_t stride = acc_strb->stride; GLubyte *data = acc_strb->data; @@ -194,7 +192,7 @@ accum_load(struct st_context *st, GLfloat value, buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - pipe_get_tile_rgba(color_trans, 0, 0, width, height, buf); + pipe_get_tile_rgba(pipe, color_trans, 0, 0, width, height, buf); switch (acc_strb->format) { case PIPE_FORMAT_R16G16B16A16_SNORM: @@ -215,7 +213,7 @@ accum_load(struct st_context *st, GLfloat value, } free(buf); - screen->tex_transfer_destroy(color_trans); + pipe->tex_transfer_destroy(pipe, color_trans); } @@ -226,7 +224,6 @@ accum_return(GLcontext *ctx, GLfloat value, struct st_renderbuffer *color_strb) { struct pipe_context *pipe = ctx->st->pipe; - struct pipe_screen *screen = pipe->screen; const GLubyte *colormask = ctx->Color.ColorMask[0]; enum pipe_transfer_usage usage; struct pipe_transfer *color_trans; @@ -251,7 +248,7 @@ accum_return(GLcontext *ctx, GLfloat value, width, height); if (usage & PIPE_TRANSFER_READ) - pipe_get_tile_rgba(color_trans, 0, 0, width, height, buf); + pipe_get_tile_rgba(pipe, color_trans, 0, 0, width, height, buf); switch (acc_strb->format) { case PIPE_FORMAT_R16G16B16A16_SNORM: @@ -280,10 +277,10 @@ accum_return(GLcontext *ctx, GLfloat value, _mesa_problem(NULL, "unexpected format in st_clear_accum_buffer()"); } - pipe_put_tile_rgba(color_trans, 0, 0, width, height, buf); + pipe_put_tile_rgba(pipe, color_trans, 0, 0, width, height, buf); free(buf); - screen->tex_transfer_destroy(color_trans); + pipe->tex_transfer_destroy(pipe, color_trans); } diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c index 25d33b933e3..9a0446bb710 100644 --- a/src/mesa/state_tracker/st_cb_bitmap.c +++ b/src/mesa/state_tracker/st_cb_bitmap.c @@ -259,7 +259,6 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, const GLubyte *bitmap) { struct pipe_context *pipe = ctx->st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *transfer; ubyte *dest; struct pipe_texture *pt; @@ -285,7 +284,7 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, PIPE_TRANSFER_WRITE, 0, 0, width, height); - dest = screen->transfer_map(screen, transfer); + dest = pipe->transfer_map(pipe, transfer); /* Put image into texture transfer */ memset(dest, 0xff, height * transfer->stride); @@ -295,8 +294,8 @@ make_bitmap_texture(GLcontext *ctx, GLsizei width, GLsizei height, _mesa_unmap_pbo_source(ctx, unpack); /* Release transfer */ - screen->transfer_unmap(screen, transfer); - screen->tex_transfer_destroy(transfer); + pipe->transfer_unmap(pipe, transfer); + pipe->tex_transfer_destroy(pipe, transfer); return pt; } @@ -520,7 +519,6 @@ static void reset_cache(struct st_context *st) { struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; struct bitmap_cache *cache = st->bitmap.cache; /*memset(cache->buffer, 0xff, sizeof(cache->buffer));*/ @@ -532,7 +530,7 @@ reset_cache(struct st_context *st) cache->ymax = -1000000; if (cache->trans) { - screen->tex_transfer_destroy(cache->trans); + pipe->tex_transfer_destroy(pipe, cache->trans); cache->trans = NULL; } @@ -570,7 +568,6 @@ static void create_cache_trans(struct st_context *st) { struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; struct bitmap_cache *cache = st->bitmap.cache; if (cache->trans) @@ -583,7 +580,7 @@ create_cache_trans(struct st_context *st) PIPE_TRANSFER_WRITE, 0, 0, BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT); - cache->buffer = screen->transfer_map(screen, cache->trans); + cache->buffer = pipe->transfer_map(pipe, cache->trans); /* init image to all 0xff */ memset(cache->buffer, 0xff, cache->trans->stride * BITMAP_CACHE_HEIGHT); @@ -601,7 +598,6 @@ st_flush_bitmap_cache(struct st_context *st) if (st->ctx->DrawBuffer) { struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_sampler_view *sv; assert(cache->xmin <= cache->xmax); @@ -618,10 +614,10 @@ st_flush_bitmap_cache(struct st_context *st) if (cache->trans) { if (0) print_cache(cache); - screen->transfer_unmap(screen, cache->trans); + pipe->transfer_unmap(pipe, cache->trans); cache->buffer = NULL; - screen->tex_transfer_destroy(cache->trans); + pipe->tex_transfer_destroy(pipe, cache->trans); cache->trans = NULL; } @@ -837,7 +833,6 @@ void st_destroy_bitmap(struct st_context *st) { struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; struct bitmap_cache *cache = st->bitmap.cache; @@ -854,8 +849,8 @@ st_destroy_bitmap(struct st_context *st) if (cache) { if (cache->trans) { - screen->transfer_unmap(screen, cache->trans); - screen->tex_transfer_destroy(cache->trans); + pipe->transfer_unmap(pipe, cache->trans); + pipe->tex_transfer_destroy(pipe, cache->trans); } pipe_texture_reference(&st->bitmap.cache->texture, NULL); free(st->bitmap.cache); diff --git a/src/mesa/state_tracker/st_cb_drawpixels.c b/src/mesa/state_tracker/st_cb_drawpixels.c index 236010c5b5d..75be79fd4b0 100644 --- a/src/mesa/state_tracker/st_cb_drawpixels.c +++ b/src/mesa/state_tracker/st_cb_drawpixels.c @@ -293,6 +293,51 @@ base_format(GLenum format) /** + * Create a temporary texture to hold an image of the given size. + * If width, height are not POT and the driver only handles POT textures, + * allocate the next larger size of texture that is POT. + */ +static struct pipe_texture * +alloc_texture(struct st_context *st, GLsizei width, GLsizei height, + enum pipe_format texFormat) +{ + struct pipe_context *pipe = st->pipe; + struct pipe_screen *screen = pipe->screen; + struct pipe_texture *pt; + int ptw, pth; + + ptw = width; + pth = height; + + /* Need to use POT texture? */ + if (!screen->get_param(screen, PIPE_CAP_NPOT_TEXTURES)) { + int l2pt, maxSize; + + l2pt = util_logbase2(width); + if (1 << l2pt != width) { + ptw = 1 << (l2pt + 1); + } + + l2pt = util_logbase2(height); + if (1 << l2pt != height) { + pth = 1 << (l2pt + 1); + } + + /* Check against maximum texture size */ + maxSize = 1 << (pipe->screen->get_param(pipe->screen, + PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1); + assert(ptw <= maxSize); + assert(pth <= maxSize); + } + + pt = st_texture_create(st, PIPE_TEXTURE_2D, texFormat, 0, + ptw, pth, 1, PIPE_TEXTURE_USAGE_SAMPLER); + + return pt; +} + + +/** * Make texture containing an image for glDrawPixels image. * If 'pixels' is NULL, leave the texture image data undefined. */ @@ -304,13 +349,11 @@ make_texture(struct st_context *st, { GLcontext *ctx = st->ctx; struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; gl_format mformat; struct pipe_texture *pt; enum pipe_format pipeFormat; GLuint cpp; GLenum baseFormat; - int ptw, pth; baseFormat = base_format(format); @@ -325,29 +368,8 @@ make_texture(struct st_context *st, if (!pixels) return NULL; - /* Need to use POT texture? */ - ptw = width; - pth = height; - if (!screen->get_param(screen, PIPE_CAP_NPOT_TEXTURES)) { - int l2pt, maxSize; - - l2pt = util_logbase2(width); - if (1<<l2pt != width) { - ptw = 1<<(l2pt+1); - } - l2pt = util_logbase2(height); - if (1<<l2pt != height) { - pth = 1<<(l2pt+1); - } - - /* Check against maximum texture size */ - maxSize = 1 << (pipe->screen->get_param(pipe->screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1); - assert(ptw <= maxSize); - assert(pth <= maxSize); - } - - pt = st_texture_create(st, PIPE_TEXTURE_2D, pipeFormat, 0, ptw, pth, 1, - PIPE_TEXTURE_USAGE_SAMPLER); + /* alloc temporary texture */ + pt = alloc_texture(st, width, height, pipeFormat); if (!pt) { _mesa_unmap_pbo_source(ctx, unpack); return NULL; @@ -368,7 +390,7 @@ make_texture(struct st_context *st, width, height); /* map texture transfer */ - dest = screen->transfer_map(screen, transfer); + dest = pipe->transfer_map(pipe, transfer); /* Put image into texture transfer. @@ -388,8 +410,8 @@ make_texture(struct st_context *st, unpack); /* unmap */ - screen->transfer_unmap(screen, transfer); - screen->tex_transfer_destroy(transfer); + pipe->transfer_unmap(pipe, transfer); + pipe->tex_transfer_destroy(pipe, transfer); assert(success); @@ -405,7 +427,7 @@ make_texture(struct st_context *st, /** * Draw quad with texcoords and optional color. - * Coords are window coords with y=0=bottom. + * Coords are gallium window coords with y=0=top. * \param color may be null * \param invertTex if true, flip texcoords vertically */ @@ -595,10 +617,15 @@ draw_textured_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z, cso_set_fragment_sampler_views(cso, 1, &sv); } - /* Compute window coords (y=0=bottom) with pixel zoom. + /* Compute Gallium window coords (y=0=top) with pixel zoom. * Recall that these coords are transformed by the current * vertex shader and viewport transformation. */ + if (st_fb_orientation(ctx->DrawBuffer) == Y_0_BOTTOM) { + y = ctx->DrawBuffer->Height - (int) (y + height * ctx->Pixel.ZoomY); + invertTex = !invertTex; + } + x0 = (GLfloat) x; x1 = x + width * ctx->Pixel.ZoomX; y0 = (GLfloat) y; @@ -630,7 +657,6 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, { struct st_context *st = st_context(ctx); struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; struct st_renderbuffer *strb; enum pipe_transfer_usage usage; struct pipe_transfer *pt; @@ -664,7 +690,7 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, usage, x, y, width, height); - stmap = screen->transfer_map(screen, pt); + stmap = pipe->transfer_map(pipe, pt); pixels = _mesa_map_pbo_source(ctx, &clippedUnpack, pixels); assert(pixels); @@ -764,8 +790,8 @@ draw_stencil_pixels(GLcontext *ctx, GLint x, GLint y, _mesa_unmap_pbo_source(ctx, &clippedUnpack); /* unmap the stencil buffer */ - screen->transfer_unmap(screen, pt); - screen->tex_transfer_destroy(pt); + pipe->transfer_unmap(pipe, pt); + pipe->tex_transfer_destroy(pipe, pt); } @@ -833,7 +859,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, GLint dstx, GLint dsty) { struct st_renderbuffer *rbDraw = st_renderbuffer(ctx->DrawBuffer->_StencilBuffer); - struct pipe_screen *screen = ctx->st->pipe->screen; + struct pipe_context *pipe = ctx->st->pipe; enum pipe_transfer_usage usage; struct pipe_transfer *ptDraw; ubyte *drawMap; @@ -869,7 +895,7 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, assert(util_format_get_blockheight(ptDraw->texture->format) == 1); /* map the stencil buffer */ - drawMap = screen->transfer_map(screen, ptDraw); + drawMap = pipe->transfer_map(pipe, ptDraw); /* draw */ /* XXX PixelZoom not handled yet */ @@ -922,8 +948,8 @@ copy_stencil_pixels(GLcontext *ctx, GLint srcx, GLint srcy, free(buffer); /* unmap the stencil buffer */ - screen->transfer_unmap(screen, ptDraw); - screen->tex_transfer_destroy(ptDraw); + pipe->transfer_unmap(pipe, ptDraw); + pipe->tex_transfer_destroy(pipe, ptDraw); } @@ -941,7 +967,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, struct pipe_sampler_view *sv; GLfloat *color; enum pipe_format srcFormat, texFormat; - int ptw, pth; + GLboolean invertTex = GL_FALSE; pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE, NULL); @@ -1017,8 +1043,8 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, } } - if (st_fb_orientation(ctx->DrawBuffer) == Y_0_TOP) { - srcy = ctx->DrawBuffer->Height - srcy - height; + if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { + srcy = ctx->ReadBuffer->Height - srcy - height; if (srcy < 0) { height -= -srcy; @@ -1027,32 +1053,12 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, if (height < 0) return; - } - /* Need to use POT texture? */ - ptw = width; - pth = height; - if (!screen->get_param(screen, PIPE_CAP_NPOT_TEXTURES)) { - int l2pt, maxSize; - - l2pt = util_logbase2(width); - if (1<<l2pt != width) { - ptw = 1<<(l2pt+1); - } - l2pt = util_logbase2(height); - if (1<<l2pt != height) { - pth = 1<<(l2pt+1); - } - - /* Check against maximum texture size */ - maxSize = 1 << (pipe->screen->get_param(pipe->screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS) - 1); - assert(ptw <= maxSize); - assert(pth <= maxSize); + invertTex = !invertTex; } - pt = st_texture_create(st, PIPE_TEXTURE_2D, texFormat, 0, - ptw, pth, 1, - PIPE_TEXTURE_USAGE_SAMPLER); + /* alloc temporary texture */ + pt = alloc_texture(st, width, height, texFormat); if (!pt) return; @@ -1062,6 +1068,9 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, return; } + /* Make temporary texture which is a copy of the src region. + * We'll draw a quad with this texture to draw the dest image. + */ if (srcFormat == texFormat) { /* copy source framebuffer surface into mipmap/texture */ struct pipe_surface *psRead = screen->get_tex_surface(screen, @@ -1082,6 +1091,13 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, psRead, srcx, srcy, width, height); } + + if (0) { + /* debug */ + debug_dump_surface(pipe, "copypixsrcsurf", psRead); + debug_dump_surface(pipe, "copypixtemptex", psTex); + } + pipe_surface_reference(&psRead, NULL); pipe_surface_reference(&psTex, NULL); } @@ -1109,21 +1125,21 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, /* alternate path using get/put_tile() */ GLfloat *buf = (GLfloat *) malloc(width * height * 4 * sizeof(GLfloat)); - pipe_get_tile_rgba(ptRead, 0, 0, width, height, buf); - pipe_put_tile_rgba(ptTex, 0, 0, width, height, buf); + pipe_get_tile_rgba(pipe, ptRead, 0, 0, width, height, buf); + pipe_put_tile_rgba(pipe, ptTex, 0, 0, width, height, buf); free(buf); } else { /* GL_DEPTH */ GLuint *buf = (GLuint *) malloc(width * height * sizeof(GLuint)); - pipe_get_tile_z(ptRead, 0, 0, width, height, buf); - pipe_put_tile_z(ptTex, 0, 0, width, height, buf); + pipe_get_tile_z(pipe, ptRead, 0, 0, width, height, buf); + pipe_put_tile_z(pipe, ptTex, 0, 0, width, height, buf); free(buf); } - screen->tex_transfer_destroy(ptRead); - screen->tex_transfer_destroy(ptTex); + pipe->tex_transfer_destroy(pipe, ptRead); + pipe->tex_transfer_destroy(pipe, ptTex); } /* draw textured quad */ @@ -1132,7 +1148,7 @@ st_CopyPixels(GLcontext *ctx, GLint srcx, GLint srcy, sv, driver_vp, driver_fp, - color, GL_TRUE); + color, invertTex); pipe_texture_reference(&pt, NULL); pipe_sampler_view_reference(&sv, NULL); diff --git a/src/mesa/state_tracker/st_cb_fbo.c b/src/mesa/state_tracker/st_cb_fbo.c index 00e9d1dccbd..abf0c8d6cb1 100644 --- a/src/mesa/state_tracker/st_cb_fbo.c +++ b/src/mesa/state_tracker/st_cb_fbo.c @@ -379,6 +379,8 @@ st_render_texture(GLcontext *ctx, PIPE_BUFFER_USAGE_GPU_READ | PIPE_BUFFER_USAGE_GPU_WRITE); + strb->format = pt->format; + strb->Base.Format = st_pipe_format_to_mesa_format(pt->format); strb->Base.DataType = st_format_datatype(pt->format); diff --git a/src/mesa/state_tracker/st_cb_flush.c b/src/mesa/state_tracker/st_cb_flush.c index 1329f807bc9..28a384ba49b 100644 --- a/src/mesa/state_tracker/st_cb_flush.c +++ b/src/mesa/state_tracker/st_cb_flush.c @@ -79,7 +79,7 @@ display_front_buffer(struct st_context *st) /* Hook for copying "fake" frontbuffer if necessary: */ st->pipe->screen->flush_frontbuffer( st->pipe->screen, front_surf, - st->pipe->priv ); + st->winsys_drawable_handle ); /* st->frontbuffer_status = FRONT_STATUS_UNDEFINED; diff --git a/src/mesa/state_tracker/st_cb_readpixels.c b/src/mesa/state_tracker/st_cb_readpixels.c index 952d9ce9156..080a5f9bfb8 100644 --- a/src/mesa/state_tracker/st_cb_readpixels.c +++ b/src/mesa/state_tracker/st_cb_readpixels.c @@ -63,7 +63,7 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, GLvoid *pixels) { struct gl_framebuffer *fb = ctx->ReadBuffer; - struct pipe_screen *screen = ctx->st->pipe->screen; + struct pipe_context *pipe = ctx->st->pipe; struct st_renderbuffer *strb = st_renderbuffer(fb->_StencilBuffer); struct pipe_transfer *pt; ubyte *stmap; @@ -81,7 +81,7 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, width, height); /* map the stencil buffer */ - stmap = screen->transfer_map(screen, pt); + stmap = pipe->transfer_map(pipe, pt); /* width should never be > MAX_WIDTH since we did clipping earlier */ ASSERT(width <= MAX_WIDTH); @@ -161,8 +161,8 @@ st_read_stencil_pixels(GLcontext *ctx, GLint x, GLint y, } /* unmap the stencil buffer */ - screen->transfer_unmap(screen, pt); - screen->tex_transfer_destroy(pt); + pipe->transfer_unmap(pipe, pt); + pipe->tex_transfer_destroy(pipe, pt); } @@ -234,13 +234,13 @@ st_fast_readpixels(GLcontext *ctx, struct st_renderbuffer *strb, { struct pipe_context *pipe = ctx->st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *trans; const GLubyte *map; GLubyte *dst; GLint row, col, dy, dstStride; if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { + /* convert GL Y to Gallium Y */ y = strb->texture->height0 - y - height; } @@ -252,17 +252,22 @@ st_fast_readpixels(GLcontext *ctx, struct st_renderbuffer *strb, return GL_FALSE; } - map = screen->transfer_map(screen, trans); + map = pipe->transfer_map(pipe, trans); if (!map) { - screen->tex_transfer_destroy(trans); + pipe->tex_transfer_destroy(pipe, trans); return GL_FALSE; } + /* We always write to the user/dest buffer from low addr to high addr + * but the read order depends on renderbuffer orientation + */ if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { + /* read source rows from bottom to top */ y = height - 1; dy = -1; } else { + /* read source rows from top to bottom */ y = 0; dy = 1; } @@ -311,8 +316,8 @@ st_fast_readpixels(GLcontext *ctx, struct st_renderbuffer *strb, ; /* nothing */ } - screen->transfer_unmap(screen, trans); - screen->tex_transfer_destroy(trans); + pipe->transfer_unmap(pipe, trans); + pipe->tex_transfer_destroy(pipe, trans); } return GL_TRUE; @@ -331,7 +336,6 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, GLvoid *dest) { struct pipe_context *pipe = ctx->st->pipe; - struct pipe_screen *screen = pipe->screen; GLfloat temp[MAX_WIDTH][4]; const GLbitfield transferOps = ctx->_ImageTransferState; GLsizei i, j; @@ -396,6 +400,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, } if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) { + /* convert GL Y to Gallium Y */ y = strb->Base.Height - y - height; } @@ -436,7 +441,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, GLuint ztemp[MAX_WIDTH]; GLfloat zfloat[MAX_WIDTH]; const double scale = 1.0 / ((1 << 24) - 1); - pipe_get_tile_raw(trans, 0, y, width, 1, ztemp, 0); + pipe_get_tile_raw(pipe, trans, 0, y, width, 1, ztemp, 0); y += yStep; for (j = 0; j < width; j++) { zfloat[j] = (float) (scale * (ztemp[j] & 0xffffff)); @@ -451,7 +456,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, assert(format == GL_DEPTH_STENCIL_EXT); for (i = 0; i < height; i++) { GLuint *zshort = (GLuint *)dst; - pipe_get_tile_raw(trans, 0, y, width, 1, dst, 0); + pipe_get_tile_raw(pipe, trans, 0, y, width, 1, dst, 0); y += yStep; /* Reverse into 24/8 */ for (j = 0; j < width; j++) { @@ -468,7 +473,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, GLuint ztemp[MAX_WIDTH]; GLfloat zfloat[MAX_WIDTH]; const double scale = 1.0 / ((1 << 24) - 1); - pipe_get_tile_raw(trans, 0, y, width, 1, ztemp, 0); + pipe_get_tile_raw(pipe, trans, 0, y, width, 1, ztemp, 0); y += yStep; for (j = 0; j < width; j++) { zfloat[j] = (float) (scale * ((ztemp[j] >> 8) & 0xffffff)); @@ -482,7 +487,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, /* XXX: unreachable code -- should be before st_read_stencil_pixels */ assert(format == GL_DEPTH_STENCIL_EXT); for (i = 0; i < height; i++) { - pipe_get_tile_raw(trans, 0, y, width, 1, dst, 0); + pipe_get_tile_raw(pipe, trans, 0, y, width, 1, dst, 0); y += yStep; dst += dstStride; } @@ -493,7 +498,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, GLushort ztemp[MAX_WIDTH]; GLfloat zfloat[MAX_WIDTH]; const double scale = 1.0 / 0xffff; - pipe_get_tile_raw(trans, 0, y, width, 1, ztemp, 0); + pipe_get_tile_raw(pipe, trans, 0, y, width, 1, ztemp, 0); y += yStep; for (j = 0; j < width; j++) { zfloat[j] = (float) (scale * ztemp[j]); @@ -508,7 +513,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, GLuint ztemp[MAX_WIDTH]; GLfloat zfloat[MAX_WIDTH]; const double scale = 1.0 / 0xffffffff; - pipe_get_tile_raw(trans, 0, y, width, 1, ztemp, 0); + pipe_get_tile_raw(pipe, trans, 0, y, width, 1, ztemp, 0); y += yStep; for (j = 0; j < width; j++) { zfloat[j] = (float) (scale * ztemp[j]); @@ -522,7 +527,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, /* RGBA format */ /* Do a row at a time to flip image data vertically */ for (i = 0; i < height; i++) { - pipe_get_tile_rgba(trans, 0, y, width, 1, df); + pipe_get_tile_rgba(pipe, trans, 0, y, width, 1, df); y += yStep; df += dfStride; if (!dfStride) { @@ -534,7 +539,7 @@ st_readpixels(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei height, } } - screen->tex_transfer_destroy(trans); + pipe->tex_transfer_destroy(pipe, trans); _mesa_unmap_pbo_dest(ctx, &clippedPacking); } diff --git a/src/mesa/state_tracker/st_cb_texture.c b/src/mesa/state_tracker/st_cb_texture.c index 76846560f9f..04c0ef8c84c 100644 --- a/src/mesa/state_tracker/st_cb_texture.c +++ b/src/mesa/state_tracker/st_cb_texture.c @@ -375,7 +375,8 @@ compress_with_blit(GLcontext * ctx, { const GLuint dstImageOffsets[1] = {0}; struct st_texture_image *stImage = st_texture_image(texImage); - struct pipe_screen *screen = ctx->st->pipe->screen; + struct pipe_context *pipe = ctx->st->pipe; + struct pipe_screen *screen = pipe->screen; gl_format mesa_format; struct pipe_texture templ; struct pipe_texture *src_tex; @@ -425,7 +426,7 @@ compress_with_blit(GLcontext * ctx, 0, 0, 0, /* face, level are zero */ PIPE_TRANSFER_WRITE, 0, 0, width, height); /* x, y, w, h */ - map = screen->transfer_map(screen, tex_xfer); + map = pipe->transfer_map(pipe, tex_xfer); _mesa_texstore(ctx, 2, GL_RGBA, mesa_format, map, /* dest ptr */ @@ -437,8 +438,8 @@ compress_with_blit(GLcontext * ctx, pixels, /* source data */ unpack); /* source data packing */ - screen->transfer_unmap(screen, tex_xfer); - screen->tex_transfer_destroy(tex_xfer); + pipe->transfer_unmap(pipe, tex_xfer); + pipe->tex_transfer_destroy(pipe, tex_xfer); /* copy / compress image */ util_blit_pixels_tex(ctx->st->blit, @@ -813,7 +814,8 @@ decompress_with_blit(GLcontext * ctx, GLenum target, GLint level, struct gl_texture_object *texObj, struct gl_texture_image *texImage) { - struct pipe_screen *screen = ctx->st->pipe->screen; + struct pipe_context *pipe = ctx->st->pipe; + struct pipe_screen *screen = pipe->screen; struct st_texture_image *stImage = st_texture_image(texImage); const GLuint width = texImage->Width; const GLuint height = texImage->Height; @@ -852,7 +854,7 @@ decompress_with_blit(GLcontext * ctx, GLenum target, GLint level, if (st_equal_formats(stImage->pt->format, format, type)) { /* memcpy */ const uint bytesPerRow = width * util_format_get_blocksize(stImage->pt->format); - ubyte *map = screen->transfer_map(screen, tex_xfer); + ubyte *map = pipe->transfer_map(pipe, tex_xfer); GLuint row; for (row = 0; row < height; row++) { GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width, @@ -860,7 +862,7 @@ decompress_with_blit(GLcontext * ctx, GLenum target, GLint level, memcpy(dest, map, bytesPerRow); map += tex_xfer->stride; } - screen->transfer_unmap(screen, tex_xfer); + pipe->transfer_unmap(pipe, tex_xfer); } else { /* format translation via floats */ @@ -875,7 +877,7 @@ decompress_with_blit(GLcontext * ctx, GLenum target, GLint level, debug_printf("%s: fallback format translation\n", __FUNCTION__); /* get float[4] rgba row from surface */ - pipe_get_tile_rgba(tex_xfer, 0, row, width, 1, rgba); + pipe_get_tile_rgba(pipe, tex_xfer, 0, row, width, 1, rgba); _mesa_pack_rgba_span_float(ctx, width, (GLfloat (*)[4]) rgba, format, type, dest, &ctx->Pack, transferOps); @@ -1260,7 +1262,6 @@ fallback_copy_texsubimage(GLcontext *ctx, GLenum target, GLint level, GLsizei width, GLsizei height) { struct pipe_context *pipe = ctx->st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_transfer *src_trans; GLvoid *texDest; enum pipe_transfer_usage transfer_usage; @@ -1313,11 +1314,11 @@ fallback_copy_texsubimage(GLcontext *ctx, GLenum target, GLint level, /* To avoid a large temp memory allocation, do copy row by row */ for (row = 0; row < height; row++, srcY += yStep) { uint data[MAX_WIDTH]; - pipe_get_tile_z(src_trans, 0, srcY, width, 1, data); + pipe_get_tile_z(pipe, src_trans, 0, srcY, width, 1, data); if (scaleOrBias) { _mesa_scale_and_bias_depth_uint(ctx, width, data); } - pipe_put_tile_z(stImage->transfer, 0, row, width, 1, data); + pipe_put_tile_z(pipe, stImage->transfer, 0, row, width, 1, data); } } else { @@ -1339,7 +1340,7 @@ fallback_copy_texsubimage(GLcontext *ctx, GLenum target, GLint level, /* XXX this usually involves a lot of int/float conversion. * try to avoid that someday. */ - pipe_get_tile_rgba(src_trans, 0, 0, width, height, tempSrc); + pipe_get_tile_rgba(pipe, src_trans, 0, 0, width, height, tempSrc); /* Store into texture memory. * Note that this does some special things such as pixel transfer @@ -1367,7 +1368,7 @@ fallback_copy_texsubimage(GLcontext *ctx, GLenum target, GLint level, } st_texture_image_unmap(ctx->st, stImage); - screen->tex_transfer_destroy(src_trans); + pipe->tex_transfer_destroy(pipe, src_trans); } diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c index e978415c200..ce1a2ea33cd 100644 --- a/src/mesa/state_tracker/st_context.c +++ b/src/mesa/state_tracker/st_context.c @@ -287,7 +287,7 @@ st_make_current(struct st_context *st, } _mesa_check_init_viewport(st->ctx, draw->InitWidth, draw->InitHeight); - st->pipe->priv = winsys_drawable_handle; + st->winsys_drawable_handle = winsys_drawable_handle; return GL_TRUE; } diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h index ae73817c0e6..4f3a67f41b2 100644 --- a/src/mesa/state_tracker/st_context.h +++ b/src/mesa/state_tracker/st_context.h @@ -186,6 +186,7 @@ struct st_context struct cso_context *cso_context; int force_msaa; + void *winsys_drawable_handle; }; diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index 8a6e1ed4662..7f45e3f5484 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -184,7 +184,7 @@ st_pipe_vertex_format(GLenum type, GLuint size, GLenum format, /* this is an odd-ball case */ assert(type == GL_UNSIGNED_BYTE); assert(normalized); - return PIPE_FORMAT_A8R8G8B8_UNORM; + return PIPE_FORMAT_B8G8R8A8_UNORM; } if (normalized) { @@ -273,7 +273,8 @@ is_interleaved_arrays(const struct st_vertex_program *vp, } *userSpace = (num_client_arrays == vpv->num_inputs); - /* printf("user space: %d (%d %d)\n", (int) *userSpace,num_client_arrays,vp->num_inputs); */ + /* debug_printf("user space: %s (%d arrays, %d inputs)\n", + (int)*userSpace ? "Yes" : "No", num_client_arrays, vp->num_inputs); */ return GL_TRUE; } @@ -293,6 +294,8 @@ get_arrays_bounds(const struct st_vertex_program *vp, const GLubyte *high_addr = NULL; GLuint attr; + /* debug_printf("get_arrays_bounds: Handling %u attrs\n", vpv->num_inputs); */ + for (attr = 0; attr < vpv->num_inputs; attr++) { const GLuint mesaAttr = vp->index_to_input[attr]; const GLint stride = arrays[mesaAttr]->StrideB; @@ -301,6 +304,9 @@ get_arrays_bounds(const struct st_vertex_program *vp, _mesa_sizeof_type(arrays[mesaAttr]->Type)); const GLubyte *end = start + (max_index * stride) + sz; + /* debug_printf("attr %u: stride %d size %u start %p end %p\n", + attr, stride, sz, start, end); */ + if (attr == 0) { low_addr = start; high_addr = end; @@ -348,7 +354,8 @@ setup_interleaved_attribs(GLcontext *ctx, const GLubyte *low, *high; get_arrays_bounds(vp, vpv, arrays, max_index, &low, &high); - /*printf("buffer range: %p %p %d\n", low, high, high-low);*/ + /* debug_printf("buffer range: %p %p range %d max index %u\n", + low, high, high - low, max_index); */ offset0 = low; if (userSpace) { diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c index e03dd30f0f5..290ee36b0fe 100644 --- a/src/mesa/state_tracker/st_extensions.c +++ b/src/mesa/state_tracker/st_extensions.c @@ -172,6 +172,7 @@ void st_init_extensions(struct st_context *st) ctx->Extensions.ARB_vertex_array_object = GL_TRUE; ctx->Extensions.ARB_vertex_buffer_object = GL_TRUE; ctx->Extensions.ARB_vertex_program = GL_TRUE; + ctx->Extensions.ARB_window_pos = GL_TRUE; ctx->Extensions.EXT_blend_color = GL_TRUE; ctx->Extensions.EXT_blend_func_separate = GL_TRUE; @@ -196,9 +197,17 @@ void st_init_extensions(struct st_context *st) ctx->Extensions.APPLE_vertex_array_object = GL_TRUE; + ctx->Extensions.MESA_pack_invert = GL_TRUE; + ctx->Extensions.NV_blend_square = GL_TRUE; ctx->Extensions.NV_texgen_reflection = GL_TRUE; ctx->Extensions.NV_texture_env_combine4 = GL_TRUE; + ctx->Extensions.NV_texture_rectangle = GL_TRUE; +#if 0 + /* possibly could support the following two */ + ctx->Extensions.NV_vertex_program = GL_TRUE; + ctx->Extensions.NV_vertex_program1_1 = GL_TRUE; +#endif #if FEATURE_OES_draw_texture ctx->Extensions.OES_draw_texture = GL_TRUE; @@ -236,7 +245,6 @@ void st_init_extensions(struct st_context *st) if (screen->get_param(screen, PIPE_CAP_NPOT_TEXTURES)) { ctx->Extensions.ARB_texture_non_power_of_two = GL_TRUE; - ctx->Extensions.NV_texture_rectangle = GL_TRUE; } if (screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS) > 1) { diff --git a/src/mesa/state_tracker/st_framebuffer.c b/src/mesa/state_tracker/st_framebuffer.c index 1d35e8d6574..0a91183f89d 100644 --- a/src/mesa/state_tracker/st_framebuffer.c +++ b/src/mesa/state_tracker/st_framebuffer.c @@ -167,9 +167,7 @@ st_set_framebuffer_surface(struct st_framebuffer *stfb, uint surfIndex, struct pipe_surface *surf) { GET_CURRENT_CONTEXT(ctx); - static const GLuint invalid_size = 9999999; struct st_renderbuffer *strb; - GLuint width, height, i; /* sanity checks */ assert(ST_SURFACE_FRONT_LEFT == BUFFER_FRONT_LEFT); @@ -183,18 +181,17 @@ st_set_framebuffer_surface(struct st_framebuffer *stfb, strb = st_renderbuffer(stfb->Base.Attachment[surfIndex].Renderbuffer); if (!strb) { - if (surfIndex == ST_SURFACE_FRONT_LEFT) { - /* Delayed creation when the window system supplies a fake front buffer */ - struct st_renderbuffer *strb_back - = st_renderbuffer(stfb->Base.Attachment[ST_SURFACE_BACK_LEFT].Renderbuffer); - struct gl_renderbuffer *rb - = st_new_renderbuffer_fb(surf->format, strb_back->Base.NumSamples, FALSE); - _mesa_add_renderbuffer(&stfb->Base, BUFFER_FRONT_LEFT, rb); - strb = st_renderbuffer(rb); - } else { - /* fail */ + /* create new renderbuffer for this surface now */ + const GLuint numSamples = stfb->Base.Visual.samples; + struct gl_renderbuffer *rb = + st_new_renderbuffer_fb(surf->format, numSamples, FALSE); + if (!rb) { + /* out of memory */ + _mesa_warning(ctx, "Out of memory allocating renderbuffer"); return; } + _mesa_add_renderbuffer(&stfb->Base, surfIndex, rb); + strb = st_renderbuffer(rb); } /* replace the renderbuffer's surface/texture pointers */ @@ -206,39 +203,16 @@ st_set_framebuffer_surface(struct st_framebuffer *stfb, * But when we do, we need to start setting this dirty bit * to ensure the renderbuffer attachements are up-to-date * via update_framebuffer. + * Core Mesa's state validation will update the parent framebuffer's + * size info, etc. */ ctx->st->dirty.st |= ST_NEW_FRAMEBUFFER; + ctx->NewState |= _NEW_BUFFERS; } /* update renderbuffer's width/height */ strb->Base.Width = surf->width; strb->Base.Height = surf->height; - - /* Try to update the framebuffer's width/height from the renderbuffer - * sizes. Before we start drawing, all the rbs _should_ be the same size. - */ - width = height = invalid_size; - for (i = 0; i < BUFFER_COUNT; i++) { - if (stfb->Base.Attachment[i].Renderbuffer) { - if (width == invalid_size) { - width = stfb->Base.Attachment[i].Renderbuffer->Width; - height = stfb->Base.Attachment[i].Renderbuffer->Height; - } - else if (width != stfb->Base.Attachment[i].Renderbuffer->Width || - height != stfb->Base.Attachment[i].Renderbuffer->Height) { - /* inconsistant renderbuffer sizes, bail out */ - return; - } - } - } - - if (width != invalid_size) { - /* OK, the renderbuffers are of a consistant size, so update the - * parent framebuffer's size. - */ - stfb->Base.Width = width; - stfb->Base.Height = height; - } } diff --git a/src/mesa/state_tracker/st_gen_mipmap.c b/src/mesa/state_tracker/st_gen_mipmap.c index f67d7b4cb5c..b2521433c87 100644 --- a/src/mesa/state_tracker/st_gen_mipmap.c +++ b/src/mesa/state_tracker/st_gen_mipmap.c @@ -106,7 +106,6 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, struct gl_texture_object *texObj) { struct pipe_context *pipe = ctx->st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_texture *pt = st_get_texobj_texture(texObj); const uint baseLevel = texObj->BaseLevel; const uint lastLevel = pt->last_level; @@ -142,8 +141,8 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, u_minify(pt->width0, dstLevel), u_minify(pt->height0, dstLevel)); - srcData = (ubyte *) screen->transfer_map(screen, srcTrans); - dstData = (ubyte *) screen->transfer_map(screen, dstTrans); + srcData = (ubyte *) pipe->transfer_map(pipe, srcTrans); + dstData = (ubyte *) pipe->transfer_map(pipe, dstTrans); srcStride = srcTrans->stride / util_format_get_blocksize(srcTrans->texture->format); dstStride = dstTrans->stride / util_format_get_blocksize(dstTrans->texture->format); @@ -161,11 +160,11 @@ fallback_generate_mipmap(GLcontext *ctx, GLenum target, dstData, dstStride); /* stride in texels */ - screen->transfer_unmap(screen, srcTrans); - screen->transfer_unmap(screen, dstTrans); + pipe->transfer_unmap(pipe, srcTrans); + pipe->transfer_unmap(pipe, dstTrans); - screen->tex_transfer_destroy(srcTrans); - screen->tex_transfer_destroy(dstTrans); + pipe->tex_transfer_destroy(pipe, srcTrans); + pipe->tex_transfer_destroy(pipe, dstTrans); } } diff --git a/src/mesa/state_tracker/st_inlines.h b/src/mesa/state_tracker/st_inlines.h index e105870bc75..7fcde7b1a96 100644 --- a/src/mesa/state_tracker/st_inlines.h +++ b/src/mesa/state_tracker/st_inlines.h @@ -53,11 +53,11 @@ st_cond_flush_get_tex_transfer(struct st_context *st, unsigned int x, unsigned int y, unsigned int w, unsigned int h) { - struct pipe_screen *screen = st->pipe->screen; + struct pipe_context *context = st->pipe; st_teximage_flush_before_map(st, pt, face, level, usage); - return screen->get_tex_transfer(screen, pt, face, level, zslice, usage, - x, y, w, h); + return context->get_tex_transfer(context, pt, face, level, zslice, usage, + x, y, w, h); } static INLINE struct pipe_transfer * @@ -70,9 +70,9 @@ st_no_flush_get_tex_transfer(struct st_context *st, unsigned int x, unsigned int y, unsigned int w, unsigned int h) { - struct pipe_screen *screen = st->pipe->screen; + struct pipe_context *context = st->pipe; - return screen->get_tex_transfer(screen, pt, face, level, + return context->get_tex_transfer(context, pt, face, level, zslice, usage, x, y, w, h); } diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c index 5a45c4358a9..10a38befb41 100644 --- a/src/mesa/state_tracker/st_texture.c +++ b/src/mesa/state_tracker/st_texture.c @@ -192,7 +192,6 @@ st_texture_image_map(struct st_context *st, struct st_texture_image *stImage, GLuint x, GLuint y, GLuint w, GLuint h) { struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; struct pipe_texture *pt = stImage->pt; DBG("%s \n", __FUNCTION__); @@ -202,7 +201,7 @@ st_texture_image_map(struct st_context *st, struct st_texture_image *stImage, usage, x, y, w, h); if (stImage->transfer) - return screen->transfer_map(screen, stImage->transfer); + return pipe->transfer_map(pipe, stImage->transfer); else return NULL; } @@ -212,13 +211,13 @@ void st_texture_image_unmap(struct st_context *st, struct st_texture_image *stImage) { - struct pipe_screen *screen = st->pipe->screen; + struct pipe_context *pipe = st->pipe; DBG("%s\n", __FUNCTION__); - screen->transfer_unmap(screen, stImage->transfer); + pipe->transfer_unmap(pipe, stImage->transfer); - screen->tex_transfer_destroy(stImage->transfer); + pipe->tex_transfer_destroy(pipe, stImage->transfer); } @@ -238,8 +237,7 @@ st_surface_data(struct pipe_context *pipe, const void *src, unsigned src_stride, unsigned srcx, unsigned srcy, unsigned width, unsigned height) { - struct pipe_screen *screen = pipe->screen; - void *map = screen->transfer_map(screen, dst); + void *map = pipe->transfer_map(pipe, dst); assert(dst->texture); util_copy_rect(map, @@ -250,7 +248,7 @@ st_surface_data(struct pipe_context *pipe, src, src_stride, srcx, srcy); - screen->transfer_unmap(screen, dst); + pipe->transfer_unmap(pipe, dst); } @@ -265,7 +263,6 @@ st_texture_image_data(struct st_context *st, GLuint src_row_stride, GLuint src_image_stride) { struct pipe_context *pipe = st->pipe; - struct pipe_screen *screen = pipe->screen; GLuint depth = u_minify(dst->depth0, level); GLuint i; const GLubyte *srcUB = src; @@ -287,7 +284,7 @@ st_texture_image_data(struct st_context *st, u_minify(dst->width0, level), u_minify(dst->height0, level)); /* width, height */ - screen->tex_transfer_destroy(dst_transfer); + pipe->tex_transfer_destroy(pipe, dst_transfer); srcUB += src_image_stride; } diff --git a/src/mesa/swrast/s_depth.c b/src/mesa/swrast/s_depth.c index 3e36cf9a7e5..ed637cac124 100644 --- a/src/mesa/swrast/s_depth.c +++ b/src/mesa/swrast/s_depth.c @@ -526,7 +526,7 @@ _swrast_depth_clamp_span( GLcontext *ctx, SWspan *span ) /* Convert floating point values in [0,1] to device Z coordinates in * [0, DepthMax]. - * ex: If the the Z buffer has 24 bits, DepthMax = 0xffffff. + * ex: If the Z buffer has 24 bits, DepthMax = 0xffffff. * * XXX this all falls apart if we have 31 or more bits of Z because * the triangle rasterization code produces unsigned Z values. Negative diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c index 29f070686fe..687c8eb0bf8 100644 --- a/src/mesa/swrast/s_span.c +++ b/src/mesa/swrast/s_span.c @@ -60,6 +60,7 @@ void _swrast_span_default_attribs(GLcontext *ctx, SWspan *span) { + GLchan r, g, b, a; /* Z*/ { const GLfloat depthMax = ctx->DrawBuffer->_DepthMaxF; @@ -80,7 +81,6 @@ _swrast_span_default_attribs(GLcontext *ctx, SWspan *span) span->attrStepY[FRAG_ATTRIB_WPOS][3] = 0.0; /* primary color, or color index */ - GLchan r, g, b, a; UNCLAMPED_FLOAT_TO_CHAN(r, ctx->Current.RasterColor[0]); UNCLAMPED_FLOAT_TO_CHAN(g, ctx->Current.RasterColor[1]); UNCLAMPED_FLOAT_TO_CHAN(b, ctx->Current.RasterColor[2]); diff --git a/src/mesa/vbo/vbo_save_loopback.c b/src/mesa/vbo/vbo_save_loopback.c index f253c854d2b..3f581ea02da 100644 --- a/src/mesa/vbo/vbo_save_loopback.c +++ b/src/mesa/vbo/vbo_save_loopback.c @@ -78,7 +78,7 @@ struct loopback_attr { }; /* Don't emit ends and begins on wrapped primitives. Don't replay - * wrapped vertices. If we get here, it's probably because the the + * wrapped vertices. If we get here, it's probably because the * precalculated wrapping is wrong. */ static void loopback_prim( GLcontext *ctx, |