X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fmesa%2Fswrast%2Fs_nvfragprog.c;h=88a6593aff45fd5dd8f5304f3cd8990fc25a991d;hb=89353febc0fdb1db6ff14633da604a59eee3913a;hp=59d7c0f826295cf6de3fff7a32d1371de80234b8;hpb=7bb832f9f7f417616c949e7152b955572b9d0212;p=mesa.git diff --git a/src/mesa/swrast/s_nvfragprog.c b/src/mesa/swrast/s_nvfragprog.c index 59d7c0f8262..88a6593aff4 100644 --- a/src/mesa/swrast/s_nvfragprog.c +++ b/src/mesa/swrast/s_nvfragprog.c @@ -1,8 +1,8 @@ /* * Mesa 3-D graphics library - * Version: 5.1 + * Version: 6.1 * - * Copyright (C) 1999-2003 Brian Paul All Rights Reserved. + * Copyright (C) 1999-2004 Brian Paul 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"), @@ -22,12 +22,21 @@ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ +/* + * Regarding GL_NV_fragment_program: + * + * Portions of this software may use or implement intellectual + * property owned and licensed by NVIDIA Corporation. NVIDIA disclaims + * any and all warranties with respect to such intellectual property, + * including any use thereof or modifications thereto. + */ #include "glheader.h" #include "colormac.h" #include "context.h" #include "nvfragprog.h" #include "macros.h" +#include "program.h" #include "s_nvfragprog.h" #include "s_span.h" @@ -37,7 +46,6 @@ /* if 1, print some debugging info */ #define DEBUG_FRAG 0 - /** * Fetch a texel. */ @@ -48,6 +56,7 @@ fetch_texel( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda, GLchan rgba[4]; SWcontext *swrast = SWRAST_CONTEXT(ctx); + /* XXX use a float-valued TextureSample routine here!!! */ swrast->TextureSample[unit](ctx, unit, ctx->Texture.Unit[unit]._Current, 1, (const GLfloat (*)[4]) texcoord, &lambda, &rgba); @@ -69,7 +78,7 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], { SWcontext *swrast = SWRAST_CONTEXT(ctx); const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current; - const struct gl_texture_image *texImg = texObj->Image[texObj->BaseLevel]; + const struct gl_texture_image *texImg = texObj->Image[0][texObj->BaseLevel]; const GLfloat texW = (GLfloat) texImg->WidthScale; const GLfloat texH = (GLfloat) texImg->HeightScale; GLchan rgba[4]; @@ -91,25 +100,63 @@ fetch_texel_deriv( GLcontext *ctx, const GLfloat texcoord[4], } +/** + * Return a pointer to the 4-element float vector specified by the given + * source register. + */ +static INLINE const GLfloat * +get_register_pointer( GLcontext *ctx, + const struct fp_src_register *source, + const struct fp_machine *machine, + const struct fragment_program *program ) +{ + const GLfloat *src; + switch (source->File) { + case PROGRAM_TEMPORARY: + ASSERT(source->Index < MAX_NV_FRAGMENT_PROGRAM_TEMPS); + src = machine->Temporaries[source->Index]; + break; + case PROGRAM_INPUT: + ASSERT(source->Index < MAX_NV_FRAGMENT_PROGRAM_INPUTS); + src = machine->Inputs[source->Index]; + break; + case PROGRAM_LOCAL_PARAM: + ASSERT(source->Index < MAX_PROGRAM_LOCAL_PARAMS); + src = program->Base.LocalParams[source->Index]; + break; + case PROGRAM_ENV_PARAM: + ASSERT(source->Index < MAX_NV_FRAGMENT_PROGRAM_PARAMS); + src = ctx->FragmentProgram.Parameters[source->Index]; + break; + + case PROGRAM_STATE_VAR: + /* Fallthrough */ + + case PROGRAM_NAMED_PARAM: + ASSERT(source->Index < (GLint) program->Parameters->NumParameters); + src = program->Parameters->Parameters[source->Index].Values; + break; + default: + _mesa_problem(ctx, "Invalid input register file in fetch_vector4"); + src = NULL; + } + return src; +} + /** * Fetch a 4-element float vector from the given source register. * Apply swizzling and negating as needed. */ static void -fetch_vector4( const struct fp_src_register *source, +fetch_vector4( GLcontext *ctx, + const struct fp_src_register *source, const struct fp_machine *machine, const struct fragment_program *program, GLfloat result[4] ) { - const GLfloat *src; - - if (source->IsParameter) { - src = program->Parameters[source->Register].Values; - } - else { - src = machine->Registers[source->Register]; - } + const GLfloat *src = get_register_pointer(ctx, source, machine, program); + ASSERT(src); result[0] = src[source->Swizzle[0]]; result[1] = src[source->Swizzle[1]]; @@ -143,26 +190,27 @@ fetch_vector4( const struct fp_src_register *source, * need to execute another instance of the program (ugh)! */ static GLboolean -fetch_vector4_deriv( const struct fp_src_register *source, +fetch_vector4_deriv( GLcontext *ctx, + const struct fp_src_register *source, const struct sw_span *span, - char xOrY, GLfloat result[4] ) + char xOrY, GLint column, GLfloat result[4] ) { GLfloat src[4]; ASSERT(xOrY == 'X' || xOrY == 'Y'); - switch (source->Register) { + switch (source->Index) { case FRAG_ATTRIB_WPOS: if (xOrY == 'X') { src[0] = 1.0; src[1] = 0.0; - src[2] = span->dzdx; + src[2] = span->dzdx / ctx->DepthMaxF; src[3] = span->dwdx; } else { src[0] = 0.0; src[1] = 1.0; - src[2] = span->dzdy; + src[2] = span->dzdy / ctx->DepthMaxF; src[3] = span->dwdy; } break; @@ -217,18 +265,23 @@ fetch_vector4_deriv( const struct fp_src_register *source, case FRAG_ATTRIB_TEX6: case FRAG_ATTRIB_TEX7: if (xOrY == 'X') { - const GLuint u = source->Register - FRAG_ATTRIB_TEX0; - src[0] = span->texStepX[u][0] * (1.0F / CHAN_MAXF); - src[1] = span->texStepX[u][1] * (1.0F / CHAN_MAXF); - src[2] = span->texStepX[u][2] * (1.0F / CHAN_MAXF); - src[3] = span->texStepX[u][3] * (1.0F / CHAN_MAXF); + const GLuint u = source->Index - FRAG_ATTRIB_TEX0; + /* this is a little tricky - I think I've got it right */ + const GLfloat invQ = 1.0f / (span->tex[u][3] + + span->texStepX[u][3] * column); + src[0] = span->texStepX[u][0] * invQ; + src[1] = span->texStepX[u][1] * invQ; + src[2] = span->texStepX[u][2] * invQ; + src[3] = span->texStepX[u][3] * invQ; } else { - const GLuint u = source->Register - FRAG_ATTRIB_TEX0; - src[0] = span->texStepY[u][0] * (1.0F / CHAN_MAXF); - src[1] = span->texStepY[u][1] * (1.0F / CHAN_MAXF); - src[2] = span->texStepY[u][2] * (1.0F / CHAN_MAXF); - src[3] = span->texStepY[u][3] * (1.0F / CHAN_MAXF); + const GLuint u = source->Index - FRAG_ATTRIB_TEX0; + /* Tricky, as above, but in Y direction */ + const GLfloat invQ = 1.0f / (span->tex[u][3] + span->texStepY[u][3]); + src[0] = span->texStepY[u][0] * invQ; + src[1] = span->texStepY[u][1] * invQ; + src[2] = span->texStepY[u][2] * invQ; + src[3] = span->texStepY[u][3] * invQ; } break; default: @@ -266,19 +319,14 @@ fetch_vector4_deriv( const struct fp_src_register *source, * As above, but only return result[0] element. */ static void -fetch_vector1( const struct fp_src_register *source, +fetch_vector1( GLcontext *ctx, + const struct fp_src_register *source, const struct fp_machine *machine, const struct fragment_program *program, GLfloat result[4] ) { - const GLfloat *src; - - if (source->IsParameter) { - src = program->Parameters[source->Register].Values; - } - else { - src = machine->Registers[source->Register]; - } + const GLfloat *src = get_register_pointer(ctx, source, machine, program); + ASSERT(src); result[0] = src[source->Swizzle[0]]; @@ -342,11 +390,27 @@ store_vector4( const struct fp_instruction *inst, const struct fp_dst_register *dest = &(inst->DstReg); const GLboolean clamp = inst->Saturate; const GLboolean updateCC = inst->UpdateCondRegister; - GLfloat *dstReg = machine->Registers[dest->Register]; + GLfloat *dstReg; + GLfloat dummyReg[4]; GLfloat clampedValue[4]; const GLboolean *writeMask = dest->WriteMask; GLboolean condWriteMask[4]; + switch (dest->File) { + case PROGRAM_OUTPUT: + dstReg = machine->Outputs[dest->Index]; + break; + case PROGRAM_TEMPORARY: + dstReg = machine->Temporaries[dest->Index]; + break; + case PROGRAM_WRITE_ONLY: + dstReg = dummyReg; + return; + default: + _mesa_problem(NULL, "bad register file in store_vector4(fp)"); + return; + } + #if DEBUG_FRAG if (value[0] > 1.0e10 || IS_INF_OR_NAN(value[0]) || @@ -418,13 +482,15 @@ init_machine_deriv( GLcontext *ctx, /* copy existing machine */ _mesa_memcpy(dMachine, machine, sizeof(struct fp_machine)); - /* Clear temporary registers */ - _mesa_bzero((GLfloat*) (machine->Registers + FP_TEMP_REG_START) , - MAX_NV_FRAGMENT_PROGRAM_TEMPS * 4 * sizeof(GLfloat)); + if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) { + /* Clear temporary registers (undefined for ARB_f_p) */ + _mesa_bzero( (void*) machine->Temporaries, + MAX_NV_FRAGMENT_PROGRAM_TEMPS * 4 * sizeof(GLfloat)); + } /* Add derivatives */ if (program->InputsRead & (1 << FRAG_ATTRIB_WPOS)) { - GLfloat *wpos = (GLfloat*) machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_WPOS]; + GLfloat *wpos = (GLfloat*) machine->Inputs[FRAG_ATTRIB_WPOS]; if (xOrY == 'X') { wpos[0] += 1.0F; wpos[1] += 0.0F; @@ -439,7 +505,7 @@ init_machine_deriv( GLcontext *ctx, } } if (program->InputsRead & (1 << FRAG_ATTRIB_COL0)) { - GLfloat *col0 = (GLfloat*) machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_COL0]; + GLfloat *col0 = (GLfloat*) machine->Inputs[FRAG_ATTRIB_COL0]; if (xOrY == 'X') { col0[0] += span->drdx * (1.0F / CHAN_MAXF); col0[1] += span->dgdx * (1.0F / CHAN_MAXF); @@ -454,7 +520,7 @@ init_machine_deriv( GLcontext *ctx, } } if (program->InputsRead & (1 << FRAG_ATTRIB_COL1)) { - GLfloat *col1 = (GLfloat*) machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_COL1]; + GLfloat *col1 = (GLfloat*) machine->Inputs[FRAG_ATTRIB_COL1]; if (xOrY == 'X') { col1[0] += span->dsrdx * (1.0F / CHAN_MAXF); col1[1] += span->dsgdx * (1.0F / CHAN_MAXF); @@ -469,7 +535,7 @@ init_machine_deriv( GLcontext *ctx, } } if (program->InputsRead & (1 << FRAG_ATTRIB_FOGC)) { - GLfloat *fogc = (GLfloat*) machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_FOGC]; + GLfloat *fogc = (GLfloat*) machine->Inputs[FRAG_ATTRIB_FOGC]; if (xOrY == 'X') { fogc[0] += span->dfogdx; } @@ -479,7 +545,8 @@ init_machine_deriv( GLcontext *ctx, } for (u = 0; u < ctx->Const.MaxTextureCoordUnits; u++) { if (program->InputsRead & (1 << (FRAG_ATTRIB_TEX0 + u))) { - GLfloat *tex = (GLfloat*) machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_TEX0+u]; + GLfloat *tex = (GLfloat*) machine->Inputs[FRAG_ATTRIB_TEX0 + u]; + /* XXX perspective-correct interpolation */ if (xOrY == 'X') { tex[0] += span->texStepX[u][0]; tex[1] += span->texStepX[u][1]; @@ -494,6 +561,12 @@ init_machine_deriv( GLcontext *ctx, } } } + + /* init condition codes */ + dMachine->CondCodes[0] = COND_EQ; + dMachine->CondCodes[1] = COND_EQ; + dMachine->CondCodes[2] = COND_EQ; + dMachine->CondCodes[3] = COND_EQ; } @@ -521,12 +594,31 @@ execute_program( GLcontext *ctx, for (pc = 0; pc < maxInst; pc++) { const struct fp_instruction *inst = program->Instructions + pc; + + if (ctx->FragmentProgram.CallbackEnabled && + ctx->FragmentProgram.Callback) { + ctx->FragmentProgram.CurrentPosition = inst->StringPos; + ctx->FragmentProgram.Callback(program->Base.Target, + ctx->FragmentProgram.CallbackData); + } + switch (inst->Opcode) { + case FP_OPCODE_ABS: + { + GLfloat a[4], result[4]; + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + result[0] = FABSF(a[0]); + result[1] = FABSF(a[1]); + result[2] = FABSF(a[2]); + result[3] = FABSF(a[3]); + store_vector4( inst, machine, result ); + } + break; case FP_OPCODE_ADD: { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = a[0] + b[0]; result[1] = a[1] + b[1]; result[2] = a[2] + b[2]; @@ -534,11 +626,24 @@ execute_program( GLcontext *ctx, store_vector4( inst, machine, result ); } break; + case FP_OPCODE_CMP: + { + GLfloat a[4], b[4], c[4], result[4]; + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[2], machine, program, c ); + result[0] = a[0] < 0.0F ? b[0] : c[0]; + result[1] = a[1] < 0.0F ? b[1] : c[1]; + result[2] = a[2] < 0.0F ? b[2] : c[2]; + result[3] = a[3] < 0.0F ? b[3] : c[3]; + store_vector4( inst, machine, result ); + } + break; case FP_OPCODE_COS: { GLfloat a[4], result[4]; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); - result[0] = result[1] = result[2] = result[3] = _mesa_cos(a[0]); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); + result[0] = result[1] = result[2] = result[3] = (GLfloat)_mesa_cos(a[0]); store_vector4( inst, machine, result ); } break; @@ -546,7 +651,8 @@ execute_program( GLcontext *ctx, { GLfloat a[4], aNext[4], result[4]; struct fp_machine dMachine; - if (!fetch_vector4_deriv(&inst->SrcReg[0], span, 'X', result)) { + if (!fetch_vector4_deriv(ctx, &inst->SrcReg[0], span, 'X', + column, result)) { /* This is tricky. Make a copy of the current machine state, * increment the input registers by the dx or dy partial * derivatives, then re-execute the program up to the @@ -554,11 +660,11 @@ execute_program( GLcontext *ctx, * Finally, find the difference in the register values for * the original and derivative runs. */ - fetch_vector4( &inst->SrcReg[0], machine, program, a); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a); init_machine_deriv(ctx, machine, program, span, 'X', &dMachine); execute_program(ctx, program, pc, &dMachine, span, column); - fetch_vector4( &inst->SrcReg[0], &dMachine, program, aNext ); + fetch_vector4( ctx, &inst->SrcReg[0], &dMachine, program, aNext ); result[0] = aNext[0] - a[0]; result[1] = aNext[1] - a[1]; result[2] = aNext[2] - a[2]; @@ -571,12 +677,13 @@ execute_program( GLcontext *ctx, { GLfloat a[4], aNext[4], result[4]; struct fp_machine dMachine; - if (!fetch_vector4_deriv(&inst->SrcReg[0], span, 'Y', result)) { + if (!fetch_vector4_deriv(ctx, &inst->SrcReg[0], span, 'Y', + column, result)) { init_machine_deriv(ctx, machine, program, span, 'Y', &dMachine); - fetch_vector4( &inst->SrcReg[0], machine, program, a); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a); execute_program(ctx, program, pc, &dMachine, span, column); - fetch_vector4( &inst->SrcReg[0], &dMachine, program, aNext ); + fetch_vector4( ctx, &inst->SrcReg[0], &dMachine, program, aNext ); result[0] = aNext[0] - a[0]; result[1] = aNext[1] - a[1]; result[2] = aNext[2] - a[2]; @@ -588,8 +695,8 @@ execute_program( GLcontext *ctx, case FP_OPCODE_DP3: { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = result[1] = result[2] = result[3] = a[0] * b[0] + a[1] * b[1] + a[2] * b[2]; store_vector4( inst, machine, result ); @@ -602,18 +709,32 @@ execute_program( GLcontext *ctx, case FP_OPCODE_DP4: { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = result[1] = result[2] = result[3] = a[0] * b[0] + a[1] * b[1] + a[2] * b[2] + a[3] * b[3]; store_vector4( inst, machine, result ); +#if DEBUG_FRAG + printf("DP4 %g = (%g, %g %g %g) . (%g, %g %g %g)\n", + result[0], a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]); +#endif + } + break; + case FP_OPCODE_DPH: + { + GLfloat a[4], b[4], result[4]; + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); + result[0] = result[1] = result[2] = result[3] = + a[0] * b[0] + a[1] * b[1] + a[2] * b[2] + b[3]; + store_vector4( inst, machine, result ); } break; case FP_OPCODE_DST: /* Distance vector */ { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = 1.0F; result[1] = a[1] * b[1]; result[2] = a[2]; @@ -624,7 +745,7 @@ execute_program( GLcontext *ctx, case FP_OPCODE_EX2: /* Exponential base 2 */ { GLfloat a[4], result[4]; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); result[0] = result[1] = result[2] = result[3] = (GLfloat) _mesa_pow(2.0, a[0]); store_vector4( inst, machine, result ); @@ -633,7 +754,7 @@ execute_program( GLcontext *ctx, case FP_OPCODE_FLR: { GLfloat a[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); result[0] = FLOORF(a[0]); result[1] = FLOORF(a[1]); result[2] = FLOORF(a[2]); @@ -644,7 +765,7 @@ execute_program( GLcontext *ctx, case FP_OPCODE_FRC: { GLfloat a[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); result[0] = a[0] - FLOORF(a[0]); result[1] = a[1] - FLOORF(a[1]); result[2] = a[2] - FLOORF(a[2]); @@ -652,21 +773,31 @@ execute_program( GLcontext *ctx, store_vector4( inst, machine, result ); } break; - case FP_OPCODE_KIL: + case FP_OPCODE_KIL_NV: /* NV_f_p only */ { const GLuint *swizzle = inst->DstReg.CondSwizzle; const GLuint condMask = inst->DstReg.CondMask; if (test_cc(machine->CondCodes[swizzle[0]], condMask) || test_cc(machine->CondCodes[swizzle[1]], condMask) || test_cc(machine->CondCodes[swizzle[2]], condMask) || - test_cc(machine->CondCodes[swizzle[3]], condMask)) + test_cc(machine->CondCodes[swizzle[3]], condMask)) { return GL_FALSE; + } + } + break; + case FP_OPCODE_KIL: /* ARB_f_p only */ + { + GLfloat a[4]; + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + if (a[0] < 0.0F || a[1] < 0.0F || a[2] < 0.0F || a[3] < 0.0F) { + return GL_FALSE; + } } break; case FP_OPCODE_LG2: /* log base 2 */ { GLfloat a[4], result[4]; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); result[0] = result[1] = result[2] = result[3] = LOG2(a[0]); store_vector4( inst, machine, result ); @@ -675,14 +806,14 @@ execute_program( GLcontext *ctx, case FP_OPCODE_LIT: { GLfloat a[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); if (a[0] < 0.0F) a[0] = 0.0F; if (a[1] < 0.0F) a[1] = 0.0F; result[0] = 1.0F; result[1] = a[0]; - result[2] = (a[0] > 0.0) ? _mesa_pow(2.0, a[3]) : 0.0F; + result[2] = (a[0] > 0.0F) ? (GLfloat) exp(a[3] * log(a[1])) : 0.0F; result[3] = 1.0F; store_vector4( inst, machine, result ); } @@ -690,9 +821,9 @@ execute_program( GLcontext *ctx, case FP_OPCODE_LRP: { GLfloat a[4], b[4], c[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); - fetch_vector4( &inst->SrcReg[2], machine, program, c ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[2], machine, program, c ); result[0] = a[0] * b[0] + (1.0F - a[0]) * c[0]; result[1] = a[1] * b[1] + (1.0F - a[1]) * c[1]; result[2] = a[2] * b[2] + (1.0F - a[2]) * c[2]; @@ -703,9 +834,9 @@ execute_program( GLcontext *ctx, case FP_OPCODE_MAD: { GLfloat a[4], b[4], c[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); - fetch_vector4( &inst->SrcReg[2], machine, program, c ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[2], machine, program, c ); result[0] = a[0] * b[0] + c[0]; result[1] = a[1] * b[1] + c[1]; result[2] = a[2] * b[2] + c[2]; @@ -716,20 +847,26 @@ execute_program( GLcontext *ctx, case FP_OPCODE_MAX: { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = MAX2(a[0], b[0]); result[1] = MAX2(a[1], b[1]); result[2] = MAX2(a[2], b[2]); result[3] = MAX2(a[3], b[3]); store_vector4( inst, machine, result ); +#if DEBUG_FRAG + printf("MAX (%g %g %g %g) = (%g %g %g %g), (%g %g %g %g)\n", + result[0], result[1], result[2], result[3], + a[0], a[1], a[2], a[3], + b[0], b[1], b[2], b[3]); +#endif } break; case FP_OPCODE_MIN: { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = MIN2(a[0], b[0]); result[1] = MIN2(a[1], b[1]); result[2] = MIN2(a[2], b[2]); @@ -740,15 +877,19 @@ execute_program( GLcontext *ctx, case FP_OPCODE_MOV: { GLfloat result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, result ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, result ); store_vector4( inst, machine, result ); +#if DEBUG_FRAG + printf("MOV (%g %g %g %g)\n", + result[0], result[1], result[2], result[3]); +#endif } break; case FP_OPCODE_MUL: { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = a[0] * b[0]; result[1] = a[1] * b[1]; result[2] = a[2] * b[2]; @@ -762,25 +903,28 @@ execute_program( GLcontext *ctx, #endif } break; - case FP_OPCODE_PK2H: /* pack two 16-bit floats */ - /* XXX this is probably wrong */ + case FP_OPCODE_PK2H: /* pack two 16-bit floats in one 32-bit float */ { GLfloat a[4], result[4]; - const GLuint *rawBits = (const GLuint *) a; + GLhalfNV hx, hy; GLuint *rawResult = (GLuint *) result; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); + GLuint twoHalves; + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + hx = _mesa_float_to_half(a[0]); + hy = _mesa_float_to_half(a[1]); + twoHalves = hx | (hy << 16); rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3] - = rawBits[0] | (rawBits[1] << 16); + = twoHalves; store_vector4( inst, machine, result ); } break; - case FP_OPCODE_PK2US: /* pack two GLushorts */ + case FP_OPCODE_PK2US: /* pack two GLushorts into one 32-bit float */ { GLfloat a[4], result[4]; GLuint usx, usy, *rawResult = (GLuint *) result; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); a[0] = CLAMP(a[0], 0.0F, 1.0F); - a[1] = CLAMP(a[0], 0.0F, 1.0F); + a[1] = CLAMP(a[1], 0.0F, 1.0F); usx = IROUND(a[0] * 65535.0F); usy = IROUND(a[1] * 65535.0F); rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3] @@ -788,11 +932,11 @@ execute_program( GLcontext *ctx, store_vector4( inst, machine, result ); } break; - case FP_OPCODE_PK4B: /* pack four GLbytes */ + case FP_OPCODE_PK4B: /* pack four GLbytes into one 32-bit float */ { GLfloat a[4], result[4]; GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); a[0] = CLAMP(a[0], -128.0F / 127.0F, 1.0F); a[1] = CLAMP(a[1], -128.0F / 127.0F, 1.0F); a[2] = CLAMP(a[2], -128.0F / 127.0F, 1.0F); @@ -806,11 +950,11 @@ execute_program( GLcontext *ctx, store_vector4( inst, machine, result ); } break; - case FP_OPCODE_PK4UB: /* pack four GLubytes */ + case FP_OPCODE_PK4UB: /* pack four GLubytes into one 32-bit float */ { GLfloat a[4], result[4]; GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); a[0] = CLAMP(a[0], 0.0F, 1.0F); a[1] = CLAMP(a[1], 0.0F, 1.0F); a[2] = CLAMP(a[2], 0.0F, 1.0F); @@ -827,17 +971,17 @@ execute_program( GLcontext *ctx, case FP_OPCODE_POW: { GLfloat a[4], b[4], result[4]; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); - fetch_vector1( &inst->SrcReg[1], machine, program, b ); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector1( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = result[1] = result[2] = result[3] - = _mesa_pow(a[0], b[0]); + = (GLfloat)_mesa_pow(a[0], b[0]); store_vector4( inst, machine, result ); } break; case FP_OPCODE_RCP: { GLfloat a[4], result[4]; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); #if DEBUG_FRAG if (a[0] == 0) printf("RCP(0)\n"); @@ -852,8 +996,8 @@ execute_program( GLcontext *ctx, case FP_OPCODE_RFL: { GLfloat axis[4], dir[4], result[4], tmp[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, axis ); - fetch_vector4( &inst->SrcReg[1], machine, program, dir ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, axis ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, dir ); tmp[3] = axis[0] * axis[0] + axis[1] * axis[1] + axis[2] * axis[2]; @@ -870,7 +1014,7 @@ execute_program( GLcontext *ctx, case FP_OPCODE_RSQ: /* 1 / sqrt() */ { GLfloat a[4], result[4]; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); result[0] = result[1] = result[2] = result[3] = INV_SQRTF(a[0]); store_vector4( inst, machine, result ); #if DEBUG_FRAG @@ -878,11 +1022,22 @@ execute_program( GLcontext *ctx, #endif } break; + case FP_OPCODE_SCS: /* sine and cos */ + { + GLfloat a[4], result[4]; + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); + result[0] = (GLfloat)cos(a[0]); + result[1] = (GLfloat)sin(a[0]); + result[2] = 0.0; /* undefined! */ + result[3] = 0.0; /* undefined! */ + store_vector4( inst, machine, result ); + } + break; case FP_OPCODE_SEQ: /* set on equal */ { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = (a[0] == b[0]) ? 1.0F : 0.0F; result[1] = (a[1] == b[1]) ? 1.0F : 0.0F; result[2] = (a[2] == b[2]) ? 1.0F : 0.0F; @@ -899,8 +1054,8 @@ execute_program( GLcontext *ctx, case FP_OPCODE_SGE: /* set on greater or equal */ { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = (a[0] >= b[0]) ? 1.0F : 0.0F; result[1] = (a[1] >= b[1]) ? 1.0F : 0.0F; result[2] = (a[2] >= b[2]) ? 1.0F : 0.0F; @@ -911,8 +1066,8 @@ execute_program( GLcontext *ctx, case FP_OPCODE_SGT: /* set on greater */ { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = (a[0] > b[0]) ? 1.0F : 0.0F; result[1] = (a[1] > b[1]) ? 1.0F : 0.0F; result[2] = (a[2] > b[2]) ? 1.0F : 0.0F; @@ -923,16 +1078,17 @@ execute_program( GLcontext *ctx, case FP_OPCODE_SIN: { GLfloat a[4], result[4]; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); - result[0] = result[1] = result[2] = result[3] = _mesa_sin(a[0]); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); + result[0] = result[1] = result[2] = + result[3] = (GLfloat)_mesa_sin(a[0]); store_vector4( inst, machine, result ); } break; case FP_OPCODE_SLE: /* set on less or equal */ { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = (a[0] <= b[0]) ? 1.0F : 0.0F; result[1] = (a[1] <= b[1]) ? 1.0F : 0.0F; result[2] = (a[2] <= b[2]) ? 1.0F : 0.0F; @@ -943,8 +1099,8 @@ execute_program( GLcontext *ctx, case FP_OPCODE_SLT: /* set on less */ { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = (a[0] < b[0]) ? 1.0F : 0.0F; result[1] = (a[1] < b[1]) ? 1.0F : 0.0F; result[2] = (a[2] < b[2]) ? 1.0F : 0.0F; @@ -955,8 +1111,8 @@ execute_program( GLcontext *ctx, case FP_OPCODE_SNE: /* set on not equal */ { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = (a[0] != b[0]) ? 1.0F : 0.0F; result[1] = (a[1] != b[1]) ? 1.0F : 0.0F; result[2] = (a[2] != b[2]) ? 1.0F : 0.0F; @@ -973,8 +1129,8 @@ execute_program( GLcontext *ctx, case FP_OPCODE_SUB: { GLfloat a[4], b[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); result[0] = a[0] - b[0]; result[1] = a[1] - b[1]; result[2] = a[2] - b[2]; @@ -982,36 +1138,91 @@ execute_program( GLcontext *ctx, store_vector4( inst, machine, result ); } break; - case FP_OPCODE_TEX: + case FP_OPCODE_SWZ: + { + const struct fp_src_register *source = &inst->SrcReg[0]; + const GLfloat *src = get_register_pointer(ctx, source, + machine, program); + GLfloat result[4]; + GLuint i; + + /* do extended swizzling here */ + for (i = 0; i < 3; i++) { + if (source->Swizzle[i] == SWIZZLE_ZERO) + result[i] = 0.0; + else if (source->Swizzle[i] == SWIZZLE_ONE) + result[i] = -1.0; + else + result[i] = -src[source->Swizzle[i]]; + if (source->NegateBase) + result[i] = -result[i]; + } + store_vector4( inst, machine, result ); + } + break; + case FP_OPCODE_TEX: /* Both ARB and NV frag prog */ /* Texel lookup */ { GLfloat texcoord[4], color[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, texcoord ); - /* XXX: Undo perspective divide from interpolate_texcoords() */ - fetch_texel( ctx, texcoord, - span->array->lambda[inst->TexSrcUnit][column], + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord ); + /* Note: we pass 0 for LOD. The ARB extension requires it + * while the NV extension says it's implementation dependant. + */ + fetch_texel( ctx, texcoord, 0.0F, inst->TexSrcUnit, color ); + store_vector4( inst, machine, color ); + } + break; + case FP_OPCODE_TXB: /* GL_ARB_fragment_program only */ + /* Texel lookup with LOD bias */ + { + GLfloat texcoord[4], color[4], bias, lambda; + + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord ); + /* texcoord[3] is the bias to add to lambda */ + bias = ctx->Texture.Unit[inst->TexSrcUnit].LodBias + + ctx->Texture.Unit[inst->TexSrcUnit]._Current->LodBias + + texcoord[3]; + lambda = span->array->lambda[inst->TexSrcUnit][column] + bias; + fetch_texel( ctx, texcoord, lambda, inst->TexSrcUnit, color ); store_vector4( inst, machine, color ); } break; - case FP_OPCODE_TXD: + case FP_OPCODE_TXD: /* GL_NV_fragment_program only */ /* Texture lookup w/ partial derivatives for LOD */ { GLfloat texcoord[4], dtdx[4], dtdy[4], color[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, texcoord ); - fetch_vector4( &inst->SrcReg[1], machine, program, dtdx ); - fetch_vector4( &inst->SrcReg[2], machine, program, dtdy ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, dtdx ); + fetch_vector4( ctx, &inst->SrcReg[2], machine, program, dtdy ); fetch_texel_deriv( ctx, texcoord, dtdx, dtdy, inst->TexSrcUnit, color ); store_vector4( inst, machine, color ); } break; - case FP_OPCODE_TXP: - /* Texture lookup w/ perspective divide */ + case FP_OPCODE_TXP: /* GL_ARB_fragment_program only */ + /* Texture lookup w/ projective divide */ + { + GLfloat texcoord[4], color[4]; + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord ); + texcoord[0] /= texcoord[3]; + texcoord[1] /= texcoord[3]; + texcoord[2] /= texcoord[3]; + /* Note: LOD=0 */ + fetch_texel( ctx, texcoord, 0.0F, inst->TexSrcUnit, color ); + store_vector4( inst, machine, color ); + } + break; + case FP_OPCODE_TXP_NV: /* GL_NV_fragment_program only */ + /* Texture lookup w/ projective divide */ { GLfloat texcoord[4], color[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, texcoord ); - /* Already did perspective divide in interpolate_texcoords() */ + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord ); + if (inst->TexSrcBit != TEXTURE_CUBE_BIT) { + texcoord[0] /= texcoord[3]; + texcoord[1] /= texcoord[3]; + texcoord[2] /= texcoord[3]; + } fetch_texel( ctx, texcoord, span->array->lambda[inst->TexSrcUnit][column], inst->TexSrcUnit, color ); @@ -1019,16 +1230,15 @@ execute_program( GLcontext *ctx, } break; case FP_OPCODE_UP2H: /* unpack two 16-bit floats */ - /* XXX this is probably wrong */ { GLfloat a[4], result[4]; const GLuint *rawBits = (const GLuint *) a; - GLuint *rawResult = (GLuint *) result; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); - rawResult[0] = rawBits[0] & 0xffff; - rawResult[1] = (rawBits[0] >> 16) & 0xffff; - rawResult[2] = rawBits[0] & 0xffff; - rawResult[3] = (rawBits[0] >> 16) & 0xffff; + GLhalfNV hx, hy; + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); + hx = rawBits[0] & 0xffff; + hy = rawBits[0] >> 16; + result[0] = result[2] = _mesa_half_to_float(hx); + result[1] = result[3] = _mesa_half_to_float(hy); store_vector4( inst, machine, result ); } break; @@ -1036,11 +1246,12 @@ execute_program( GLcontext *ctx, { GLfloat a[4], result[4]; const GLuint *rawBits = (const GLuint *) a; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); - result[0] = (GLfloat) ((rawBits[0] >> 0) & 0xffff) / 65535.0F; - result[1] = (GLfloat) ((rawBits[0] >> 16) & 0xffff) / 65535.0F; - result[2] = result[0]; - result[3] = result[1]; + GLushort usx, usy; + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); + usx = rawBits[0] & 0xffff; + usy = rawBits[0] >> 16; + result[0] = result[2] = usx * (1.0f / 65535.0f); + result[1] = result[3] = usy * (1.0f / 65535.0f); store_vector4( inst, machine, result ); } break; @@ -1048,11 +1259,11 @@ execute_program( GLcontext *ctx, { GLfloat a[4], result[4]; const GLuint *rawBits = (const GLuint *) a; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); result[0] = (((rawBits[0] >> 0) & 0xff) - 128) / 127.0F; - result[0] = (((rawBits[0] >> 8) & 0xff) - 128) / 127.0F; - result[0] = (((rawBits[0] >> 16) & 0xff) - 128) / 127.0F; - result[0] = (((rawBits[0] >> 24) & 0xff) - 128) / 127.0F; + result[1] = (((rawBits[0] >> 8) & 0xff) - 128) / 127.0F; + result[2] = (((rawBits[0] >> 16) & 0xff) - 128) / 127.0F; + result[3] = (((rawBits[0] >> 24) & 0xff) - 128) / 127.0F; store_vector4( inst, machine, result ); } break; @@ -1060,20 +1271,32 @@ execute_program( GLcontext *ctx, { GLfloat a[4], result[4]; const GLuint *rawBits = (const GLuint *) a; - fetch_vector1( &inst->SrcReg[0], machine, program, a ); + fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a ); result[0] = ((rawBits[0] >> 0) & 0xff) / 255.0F; - result[0] = ((rawBits[0] >> 8) & 0xff) / 255.0F; - result[0] = ((rawBits[0] >> 16) & 0xff) / 255.0F; - result[0] = ((rawBits[0] >> 24) & 0xff) / 255.0F; + result[1] = ((rawBits[0] >> 8) & 0xff) / 255.0F; + result[2] = ((rawBits[0] >> 16) & 0xff) / 255.0F; + result[3] = ((rawBits[0] >> 24) & 0xff) / 255.0F; + store_vector4( inst, machine, result ); + } + break; + case FP_OPCODE_XPD: /* cross product */ + { + GLfloat a[4], b[4], result[4]; + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); + result[0] = a[1] * b[2] - a[2] * b[1]; + result[1] = a[2] * b[0] - a[0] * b[2]; + result[2] = a[0] * b[1] - a[1] * b[0]; + result[3] = 1.0; store_vector4( inst, machine, result ); } break; case FP_OPCODE_X2D: /* 2-D matrix transform */ { GLfloat a[4], b[4], c[4], result[4]; - fetch_vector4( &inst->SrcReg[0], machine, program, a ); - fetch_vector4( &inst->SrcReg[1], machine, program, b ); - fetch_vector4( &inst->SrcReg[2], machine, program, c ); + fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a ); + fetch_vector4( ctx, &inst->SrcReg[1], machine, program, b ); + fetch_vector4( ctx, &inst->SrcReg[2], machine, program, c ); result[0] = a[0] + b[0] * c[0] + b[1] * c[1]; result[1] = a[1] + b[0] * c[2] + b[1] * c[3]; result[2] = a[2] + b[0] * c[0] + b[1] * c[1]; @@ -1098,86 +1321,109 @@ init_machine( GLcontext *ctx, struct fp_machine *machine, const struct fragment_program *program, const struct sw_span *span, GLuint col ) { - GLuint j, u; + GLuint inputsRead = program->InputsRead; + GLuint u; - /* Clear temporary registers */ - _mesa_bzero(machine->Registers + FP_TEMP_REG_START, - MAX_NV_FRAGMENT_PROGRAM_TEMPS * 4 * sizeof(GLfloat)); + if (ctx->FragmentProgram.CallbackEnabled) + inputsRead = ~0; - /* Load program local parameters */ - for (j = 0; j < MAX_NV_FRAGMENT_PROGRAM_PARAMS; j++) { - COPY_4V(machine->Registers[FP_PROG_REG_START + j], - program->Base.LocalParams[j]); + if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) { + /* Clear temporary registers (undefined for ARB_f_p) */ + _mesa_bzero(machine->Temporaries, + MAX_NV_FRAGMENT_PROGRAM_TEMPS * 4 * sizeof(GLfloat)); } /* Load input registers */ - if (program->InputsRead & (1 << FRAG_ATTRIB_WPOS)) { - GLfloat *wpos = machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_WPOS]; - wpos[0] = span->x + col; - wpos[1] = span->y; + if (inputsRead & (1 << FRAG_ATTRIB_WPOS)) { + GLfloat *wpos = machine->Inputs[FRAG_ATTRIB_WPOS]; + wpos[0] = (GLfloat) span->x + col; + wpos[1] = (GLfloat) span->y; wpos[2] = (GLfloat) span->array->z[col] / ctx->DepthMaxF; wpos[3] = span->w + col * span->dwdx; } - if (program->InputsRead & (1 << FRAG_ATTRIB_COL0)) { - GLfloat *col0 = machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_COL0]; + if (inputsRead & (1 << FRAG_ATTRIB_COL0)) { + GLfloat *col0 = machine->Inputs[FRAG_ATTRIB_COL0]; col0[0] = CHAN_TO_FLOAT(span->array->rgba[col][RCOMP]); col0[1] = CHAN_TO_FLOAT(span->array->rgba[col][GCOMP]); col0[2] = CHAN_TO_FLOAT(span->array->rgba[col][BCOMP]); col0[3] = CHAN_TO_FLOAT(span->array->rgba[col][ACOMP]); } - if (program->InputsRead & (1 << FRAG_ATTRIB_COL1)) { - GLfloat *col1 = machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_COL1]; + if (inputsRead & (1 << FRAG_ATTRIB_COL1)) { + GLfloat *col1 = machine->Inputs[FRAG_ATTRIB_COL1]; col1[0] = CHAN_TO_FLOAT(span->array->spec[col][RCOMP]); col1[1] = CHAN_TO_FLOAT(span->array->spec[col][GCOMP]); col1[2] = CHAN_TO_FLOAT(span->array->spec[col][BCOMP]); col1[3] = CHAN_TO_FLOAT(span->array->spec[col][ACOMP]); } - if (program->InputsRead & (1 << FRAG_ATTRIB_FOGC)) { - GLfloat *fogc = machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_FOGC]; + if (inputsRead & (1 << FRAG_ATTRIB_FOGC)) { + GLfloat *fogc = machine->Inputs[FRAG_ATTRIB_FOGC]; fogc[0] = span->array->fog[col]; fogc[1] = 0.0F; fogc[2] = 0.0F; fogc[3] = 0.0F; } for (u = 0; u < ctx->Const.MaxTextureCoordUnits; u++) { - if (program->InputsRead & (1 << (FRAG_ATTRIB_TEX0 + u))) { - GLfloat *tex = machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_TEX0+u]; - ASSERT(ctx->Texture._EnabledCoordUnits & (1 << u)); + if (inputsRead & (1 << (FRAG_ATTRIB_TEX0 + u))) { + GLfloat *tex = machine->Inputs[FRAG_ATTRIB_TEX0 + u]; + /*ASSERT(ctx->Texture._EnabledCoordUnits & (1 << u));*/ COPY_4V(tex, span->array->texcoords[u][col]); - ASSERT(tex[0] != 0 || tex[1] != 0 || tex[2] != 0); + /*ASSERT(tex[0] != 0 || tex[1] != 0 || tex[2] != 0);*/ } } + + /* init condition codes */ + machine->CondCodes[0] = COND_EQ; + machine->CondCodes[1] = COND_EQ; + machine->CondCodes[2] = COND_EQ; + machine->CondCodes[3] = COND_EQ; } + +/** + * Execute the current fragment program, operating on the given span. + */ void -_swrast_exec_nv_fragment_program( GLcontext *ctx, struct sw_span *span ) +_swrast_exec_fragment_program( GLcontext *ctx, struct sw_span *span ) { const struct fragment_program *program = ctx->FragmentProgram.Current; GLuint i; + ctx->_CurrentProgram = GL_FRAGMENT_PROGRAM_ARB; /* or NV, doesn't matter */ + for (i = 0; i < span->end; i++) { if (span->array->mask[i]) { init_machine(ctx, &ctx->FragmentProgram.Machine, ctx->FragmentProgram.Current, span, i); +#ifdef USE_TCC + if (!_swrast_execute_codegen_program(ctx, program, ~0, + &ctx->FragmentProgram.Machine, + span, i)) { + span->array->mask[i] = GL_FALSE; /* killed fragment */ + } +#else if (!execute_program(ctx, program, ~0, - &ctx->FragmentProgram.Machine, span, i)) + &ctx->FragmentProgram.Machine, span, i)) { span->array->mask[i] = GL_FALSE; /* killed fragment */ + } +#endif /* Store output registers */ { const GLfloat *colOut - = ctx->FragmentProgram.Machine.Registers[FP_OUTPUT_REG_START]; + = ctx->FragmentProgram.Machine.Outputs[FRAG_OUTPUT_COLR]; UNCLAMPED_FLOAT_TO_CHAN(span->array->rgba[i][RCOMP], colOut[0]); UNCLAMPED_FLOAT_TO_CHAN(span->array->rgba[i][GCOMP], colOut[1]); UNCLAMPED_FLOAT_TO_CHAN(span->array->rgba[i][BCOMP], colOut[2]); UNCLAMPED_FLOAT_TO_CHAN(span->array->rgba[i][ACOMP], colOut[3]); } /* depth value */ - if (ctx->FragmentProgram.Current->OutputsWritten & 2) - span->array->z[i] = IROUND(ctx->FragmentProgram.Machine.Registers[FP_OUTPUT_REG_START + 2][0] * ctx->DepthMaxF); + if (program->OutputsWritten & (1 << FRAG_OUTPUT_DEPR)) + span->array->z[i] = IROUND(ctx->FragmentProgram.Machine.Outputs[FRAG_OUTPUT_DEPR][0] * ctx->DepthMaxF); } } + + ctx->_CurrentProgram = 0; }