/*
* Mesa 3-D graphics library
- * Version: 6.0
+ * Version: 6.5
*
- * Copyright (C) 1999-2004 Brian Paul All Rights Reserved.
+ * Copyright (C) 1999-2005 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"),
#include "glheader.h"
#include "colormac.h"
#include "context.h"
-#include "nvfragprog.h"
-#include "macros.h"
+#include "program_instruction.h"
#include "program.h"
#include "s_nvfragprog.h"
#include "s_span.h"
-#include "s_texture.h"
/* if 1, print some debugging info */
SWcontext *swrast = SWRAST_CONTEXT(ctx);
/* XXX use a float-valued TextureSample routine here!!! */
- swrast->TextureSample[unit](ctx, unit, ctx->Texture.Unit[unit]._Current,
+ swrast->TextureSample[unit](ctx, ctx->Texture.Unit[unit]._Current,
1, (const GLfloat (*)[4]) texcoord,
&lambda, &rgba);
color[0] = CHAN_TO_FLOAT(rgba[0]);
texcoord[0], texcoord[1], texcoord[3],
1.0F / texcoord[3]);
- swrast->TextureSample[unit](ctx, unit, ctx->Texture.Unit[unit]._Current,
+ swrast->TextureSample[unit](ctx, ctx->Texture.Unit[unit]._Current,
1, (const GLfloat (*)[4]) texcoord,
&lambda, &rgba);
color[0] = CHAN_TO_FLOAT(rgba[0]);
*/
static INLINE const GLfloat *
get_register_pointer( GLcontext *ctx,
- const struct fp_src_register *source,
+ const struct prog_src_register *source,
const struct fp_machine *machine,
const struct fragment_program *program )
{
ASSERT(source->Index < MAX_NV_FRAGMENT_PROGRAM_INPUTS);
src = machine->Inputs[source->Index];
break;
+ case PROGRAM_OUTPUT:
+ /* This is only for PRINT */
+ ASSERT(source->Index < MAX_NV_FRAGMENT_PROGRAM_OUTPUTS);
+ src = machine->Outputs[source->Index];
+ break;
case PROGRAM_LOCAL_PARAM:
ASSERT(source->Index < MAX_PROGRAM_LOCAL_PARAMS);
src = program->Base.LocalParams[source->Index];
ASSERT(source->Index < MAX_NV_FRAGMENT_PROGRAM_PARAMS);
src = ctx->FragmentProgram.Parameters[source->Index];
break;
-
case PROGRAM_STATE_VAR:
- /* Fallthrough */
-
+ /* Fallthrough */
case PROGRAM_NAMED_PARAM:
- ASSERT(source->Index < (GLint) program->Parameters->NumParameters);
- src = program->Parameters->Parameters[source->Index].Values;
+ ASSERT(source->Index < (GLint) program->Base.Parameters->NumParameters);
+ src = program->Base.Parameters->ParameterValues[source->Index];
break;
default:
- _mesa_problem(ctx, "Invalid input register file in fetch_vector4");
+ _mesa_problem(ctx, "Invalid input register file %d in fetch_vector4", source->File);
src = NULL;
}
return src;
*/
static void
fetch_vector4( GLcontext *ctx,
- const struct fp_src_register *source,
+ const struct prog_src_register *source,
const struct fp_machine *machine,
const struct fragment_program *program,
GLfloat result[4] )
const GLfloat *src = get_register_pointer(ctx, source, machine, program);
ASSERT(src);
- result[0] = src[source->Swizzle[0]];
- result[1] = src[source->Swizzle[1]];
- result[2] = src[source->Swizzle[2]];
- result[3] = src[source->Swizzle[3]];
+ result[0] = src[GET_SWZ(source->Swizzle, 0)];
+ result[1] = src[GET_SWZ(source->Swizzle, 1)];
+ result[2] = src[GET_SWZ(source->Swizzle, 2)];
+ result[3] = src[GET_SWZ(source->Swizzle, 3)];
if (source->NegateBase) {
result[0] = -result[0];
*/
static GLboolean
fetch_vector4_deriv( GLcontext *ctx,
- const struct fp_src_register *source,
+ const struct prog_src_register *source,
const struct sw_span *span,
char xOrY, GLint column, GLfloat result[4] )
{
if (xOrY == 'X') {
src[0] = 1.0;
src[1] = 0.0;
- src[2] = span->dzdx / ctx->DepthMaxF;
+ src[2] = span->dzdx / ctx->DrawBuffer->_DepthMaxF;
src[3] = span->dwdx;
}
else {
src[0] = 0.0;
src[1] = 1.0;
- src[2] = span->dzdy / ctx->DepthMaxF;
+ src[2] = span->dzdy / ctx->DrawBuffer->_DepthMaxF;
src[3] = span->dwdy;
}
break;
return GL_FALSE;
}
- result[0] = src[source->Swizzle[0]];
- result[1] = src[source->Swizzle[1]];
- result[2] = src[source->Swizzle[2]];
- result[3] = src[source->Swizzle[3]];
+ result[0] = src[GET_SWZ(source->Swizzle, 0)];
+ result[1] = src[GET_SWZ(source->Swizzle, 1)];
+ result[2] = src[GET_SWZ(source->Swizzle, 2)];
+ result[3] = src[GET_SWZ(source->Swizzle, 3)];
if (source->NegateBase) {
result[0] = -result[0];
*/
static void
fetch_vector1( GLcontext *ctx,
- const struct fp_src_register *source,
+ const struct prog_src_register *source,
const struct fp_machine *machine,
const struct fragment_program *program,
GLfloat result[4] )
const GLfloat *src = get_register_pointer(ctx, source, machine, program);
ASSERT(src);
- result[0] = src[source->Swizzle[0]];
+ result[0] = src[GET_SWZ(source->Swizzle, 0)];
if (source->NegateBase) {
result[0] = -result[0];
}
-/*
+/**
* Test value against zero and return GT, LT, EQ or UN if NaN.
*/
static INLINE GLuint
return COND_EQ;
}
-/*
+
+/**
* Test if the ccMaskRule is satisfied by the given condition code.
* Used to mask destination writes according to the current condition codee.
*/
* set-condition-code flags.
*/
static void
-store_vector4( const struct fp_instruction *inst,
+store_vector4( const struct prog_instruction *inst,
struct fp_machine *machine,
const GLfloat value[4] )
{
- const struct fp_dst_register *dest = &(inst->DstReg);
- const GLboolean clamp = inst->Saturate;
- const GLboolean updateCC = inst->UpdateCondRegister;
+ const struct prog_dst_register *dest = &(inst->DstReg);
+ const GLboolean clamp = inst->SaturateMode == SATURATE_ZERO_ONE;
+ const GLboolean updateCC = inst->CondUpdate;
GLfloat *dstReg;
+ GLfloat dummyReg[4];
GLfloat clampedValue[4];
- const GLboolean *writeMask = dest->WriteMask;
GLboolean condWriteMask[4];
+ GLuint writeMask = dest->WriteMask;
switch (dest->File) {
case PROGRAM_OUTPUT:
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 (dest->CondMask != COND_TR) {
- condWriteMask[0] = writeMask[0]
- && test_cc(machine->CondCodes[dest->CondSwizzle[0]], dest->CondMask);
- condWriteMask[1] = writeMask[1]
- && test_cc(machine->CondCodes[dest->CondSwizzle[1]], dest->CondMask);
- condWriteMask[2] = writeMask[2]
- && test_cc(machine->CondCodes[dest->CondSwizzle[2]], dest->CondMask);
- condWriteMask[3] = writeMask[3]
- && test_cc(machine->CondCodes[dest->CondSwizzle[3]], dest->CondMask);
- writeMask = condWriteMask;
+ condWriteMask[0] = GET_BIT(writeMask, 0)
+ && test_cc(machine->CondCodes[GET_SWZ(dest->CondSwizzle, 0)], dest->CondMask);
+ condWriteMask[1] = GET_BIT(writeMask, 1)
+ && test_cc(machine->CondCodes[GET_SWZ(dest->CondSwizzle, 1)], dest->CondMask);
+ condWriteMask[2] = GET_BIT(writeMask, 2)
+ && test_cc(machine->CondCodes[GET_SWZ(dest->CondSwizzle, 2)], dest->CondMask);
+ condWriteMask[3] = GET_BIT(writeMask, 3)
+ && test_cc(machine->CondCodes[GET_SWZ(dest->CondSwizzle, 3)], dest->CondMask);
+
+ writeMask = ((condWriteMask[0] << 0) |
+ (condWriteMask[1] << 1) |
+ (condWriteMask[2] << 2) |
+ (condWriteMask[3] << 3));
}
- if (writeMask[0]) {
+ if (GET_BIT(writeMask, 0)) {
dstReg[0] = value[0];
if (updateCC)
machine->CondCodes[0] = generate_cc(value[0]);
}
- if (writeMask[1]) {
+ if (GET_BIT(writeMask, 1)) {
dstReg[1] = value[1];
if (updateCC)
machine->CondCodes[1] = generate_cc(value[1]);
}
- if (writeMask[2]) {
+ if (GET_BIT(writeMask, 2)) {
dstReg[2] = value[2];
if (updateCC)
machine->CondCodes[2] = generate_cc(value[2]);
}
- if (writeMask[3]) {
+ if (GET_BIT(writeMask, 3)) {
dstReg[3] = value[3];
if (updateCC)
machine->CondCodes[3] = generate_cc(value[3]);
/* copy existing machine */
_mesa_memcpy(dMachine, machine, sizeof(struct fp_machine));
- /* Clear temporary registers */
- _mesa_bzero( (void*) machine->Temporaries,
- 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)) {
+ if (program->Base.InputsRead & (1 << FRAG_ATTRIB_WPOS)) {
GLfloat *wpos = (GLfloat*) machine->Inputs[FRAG_ATTRIB_WPOS];
if (xOrY == 'X') {
wpos[0] += 1.0F;
wpos[3] += span->dwdy;
}
}
- if (program->InputsRead & (1 << FRAG_ATTRIB_COL0)) {
+ if (program->Base.InputsRead & (1 << FRAG_ATTRIB_COL0)) {
GLfloat *col0 = (GLfloat*) machine->Inputs[FRAG_ATTRIB_COL0];
if (xOrY == 'X') {
col0[0] += span->drdx * (1.0F / CHAN_MAXF);
col0[3] += span->dady * (1.0F / CHAN_MAXF);
}
}
- if (program->InputsRead & (1 << FRAG_ATTRIB_COL1)) {
+ if (program->Base.InputsRead & (1 << FRAG_ATTRIB_COL1)) {
GLfloat *col1 = (GLfloat*) machine->Inputs[FRAG_ATTRIB_COL1];
if (xOrY == 'X') {
col1[0] += span->dsrdx * (1.0F / CHAN_MAXF);
col1[3] += 0.0; /*XXX fix */
}
}
- if (program->InputsRead & (1 << FRAG_ATTRIB_FOGC)) {
+ if (program->Base.InputsRead & (1 << FRAG_ATTRIB_FOGC)) {
GLfloat *fogc = (GLfloat*) machine->Inputs[FRAG_ATTRIB_FOGC];
if (xOrY == 'X') {
fogc[0] += span->dfogdx;
}
}
for (u = 0; u < ctx->Const.MaxTextureCoordUnits; u++) {
- if (program->InputsRead & (1 << (FRAG_ATTRIB_TEX0 + u))) {
+ if (program->Base.InputsRead & (1 << (FRAG_ATTRIB_TEX0 + u))) {
GLfloat *tex = (GLfloat*) machine->Inputs[FRAG_ATTRIB_TEX0 + u];
/* XXX perspective-correct interpolation */
if (xOrY == 'X') {
printf("execute fragment program --------------------\n");
#endif
- /* XXX: This should go someplace else, but it is safe here (and slow!)
- * - karl
- */
- _mesa_load_state_parameters(ctx, program->Parameters);
-
for (pc = 0; pc < maxInst; pc++) {
- const struct fp_instruction *inst = program->Instructions + pc;
+ const struct prog_instruction *inst = program->Base.Instructions + pc;
if (ctx->FragmentProgram.CallbackEnabled &&
ctx->FragmentProgram.Callback) {
}
switch (inst->Opcode) {
- case FP_OPCODE_ABS:
+ case OPCODE_ABS:
{
GLfloat a[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_ADD:
+ case OPCODE_ADD:
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_CMP:
+ case OPCODE_CMP:
{
GLfloat a[4], b[4], c[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_COS:
+ case OPCODE_COS:
{
GLfloat a[4], result[4];
fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_DDX: /* Partial derivative with respect to X */
+ case OPCODE_DDX: /* Partial derivative with respect to X */
{
GLfloat a[4], aNext[4], result[4];
struct fp_machine dMachine;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_DDY: /* Partial derivative with respect to Y */
+ case OPCODE_DDY: /* Partial derivative with respect to Y */
{
GLfloat a[4], aNext[4], result[4];
struct fp_machine dMachine;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_DP3:
+ case OPCODE_DP3:
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
#endif
}
break;
- case FP_OPCODE_DP4:
+ case OPCODE_DP4:
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
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:
+ case OPCODE_DPH:
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_DST: /* Distance vector */
+ case OPCODE_DST: /* Distance vector */
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_EX2: /* Exponential base 2 */
+ case OPCODE_EX2: /* Exponential base 2 */
{
GLfloat a[4], result[4];
fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_FLR:
+ case OPCODE_FLR:
{
GLfloat a[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_FRC:
+ case OPCODE_FRC:
{
GLfloat a[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_KIL:
+ case OPCODE_KIL_NV: /* NV_f_p only */
{
- const GLuint *swizzle = inst->DstReg.CondSwizzle;
+ 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)) {
+ if (test_cc(machine->CondCodes[GET_SWZ(swizzle, 0)], condMask) ||
+ test_cc(machine->CondCodes[GET_SWZ(swizzle, 1)], condMask) ||
+ test_cc(machine->CondCodes[GET_SWZ(swizzle, 2)], condMask) ||
+ test_cc(machine->CondCodes[GET_SWZ(swizzle, 3)], condMask)) {
+ return GL_FALSE;
+ }
+ }
+ break;
+ case 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 */
+ case OPCODE_LG2: /* log base 2 */
{
GLfloat a[4], result[4];
fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_LIT:
+ case OPCODE_LIT:
{
+ const GLfloat epsilon = 1.0F / 256.0F; /* from NV VP spec */
GLfloat a[4], result[4];
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;
+ a[0] = MAX2(a[0], 0.0F);
+ a[1] = MAX2(a[1], 0.0F);
+ /* XXX ARB version clamps a[3], NV version doesn't */
+ a[3] = CLAMP(a[3], -(128.0F - epsilon), (128.0F - epsilon));
result[0] = 1.0F;
result[1] = a[0];
- result[2] = (a[0] > 0.0F) ? (GLfloat) exp(a[3] * log(a[1])) : 0.0F;
+ /* XXX we could probably just use pow() here */
+ if (a[0] > 0.0F) {
+ if (a[1] == 0.0 && a[3] == 0.0)
+ result[2] = 1.0;
+ else
+ result[2] = EXPF(a[3] * LOGF(a[1]));
+ }
+ else {
+ result[2] = 0.0;
+ }
result[3] = 1.0F;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_LRP:
+ case OPCODE_LRP:
{
GLfloat a[4], b[4], c[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_MAD:
+ case OPCODE_MAD:
{
GLfloat a[4], b[4], c[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_MAX:
+ case OPCODE_MAX:
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
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:
+ case OPCODE_MIN:
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_MOV:
+ case OPCODE_MOV:
{
GLfloat result[4];
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:
+ case OPCODE_MUL:
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
#endif
}
break;
- case FP_OPCODE_PK2H: /* pack two 16-bit floats in one 32-bit float */
+ case OPCODE_PK2H: /* pack two 16-bit floats in one 32-bit float */
{
GLfloat a[4], result[4];
GLhalfNV hx, hy;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_PK2US: /* pack two GLushorts into one 32-bit float */
+ case OPCODE_PK2US: /* pack two GLushorts into one 32-bit float */
{
GLfloat a[4], result[4];
GLuint usx, usy, *rawResult = (GLuint *) result;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_PK4B: /* pack four GLbytes into one 32-bit float */
+ case OPCODE_PK4B: /* pack four GLbytes into one 32-bit float */
{
GLfloat a[4], result[4];
GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_PK4UB: /* pack four GLubytes into one 32-bit float */
+ case OPCODE_PK4UB: /* pack four GLubytes into one 32-bit float */
{
GLfloat a[4], result[4];
GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_POW:
+ case OPCODE_POW:
{
GLfloat a[4], b[4], result[4];
fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_RCP:
+ case OPCODE_RCP:
{
GLfloat a[4], result[4];
fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_RFL:
+ case OPCODE_RFL:
{
GLfloat axis[4], dir[4], result[4], tmp[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, axis );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_RSQ: /* 1 / sqrt() */
+ case OPCODE_RSQ: /* 1 / sqrt() */
{
GLfloat a[4], result[4];
fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
+ a[0] = FABSF(a[0]);
result[0] = result[1] = result[2] = result[3] = INV_SQRTF(a[0]);
store_vector4( inst, machine, result );
#if DEBUG_FRAG
- printf("RSQ %g = 1/sqrt(%g)\n", result[0], a[0]);
+ printf("RSQ %g = 1/sqrt(|%g|)\n", result[0], a[0]);
#endif
}
break;
- case FP_OPCODE_SCS: /* sine and cos */
+ case 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[0] = (GLfloat)_mesa_cos(a[0]);
+ result[1] = (GLfloat)_mesa_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 */
+ case OPCODE_SEQ: /* set on equal */
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SFL: /* set false, operands ignored */
+ case OPCODE_SFL: /* set false, operands ignored */
{
static const GLfloat result[4] = { 0.0F, 0.0F, 0.0F, 0.0F };
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SGE: /* set on greater or equal */
+ case OPCODE_SGE: /* set on greater or equal */
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SGT: /* set on greater */
+ case OPCODE_SGT: /* set on greater */
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SIN:
+ case OPCODE_SIN:
{
GLfloat a[4], result[4];
fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SLE: /* set on less or equal */
+ case OPCODE_SLE: /* set on less or equal */
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SLT: /* set on less */
+ case OPCODE_SLT: /* set on less */
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SNE: /* set on not equal */
+ case OPCODE_SNE: /* set on not equal */
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_STR: /* set true, operands ignored */
+ case OPCODE_STR: /* set true, operands ignored */
{
static const GLfloat result[4] = { 1.0F, 1.0F, 1.0F, 1.0F };
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SUB:
+ case OPCODE_SUB:
{
GLfloat a[4], b[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_SWZ:
+ case OPCODE_SWZ:
{
- const struct fp_src_register *source = &inst->SrcReg[0];
+ const struct prog_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)
+ for (i = 0; i < 4; i++) {
+ if (GET_SWZ(source->Swizzle, i) == SWIZZLE_ZERO)
result[i] = 0.0;
- else if (source->Swizzle[i] == SWIZZLE_ONE)
- result[i] = -1.0;
+ else if (GET_SWZ(source->Swizzle, i) == SWIZZLE_ONE)
+ result[i] = 1.0;
else
- result[i] = -src[source->Swizzle[i]];
- if (source->NegateBase)
+ result[i] = src[GET_SWZ(source->Swizzle, i)];
+
+ if (source->NegateBase & (1 << i))
result[i] = -result[i];
}
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_TEX:
+ case OPCODE_TEX: /* Both ARB and NV frag prog */
/* Texel lookup */
{
GLfloat texcoord[4], color[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
- /* XXX: Undo perspective divide from interpolate_texcoords() */
- fetch_texel( ctx, texcoord,
- span->array->lambda[inst->TexSrcUnit][column],
- inst->TexSrcUnit, color );
+ /* Note: we pass 0 for LOD. The ARB extension requires it
+ * while the NV extension says it's implementation dependant.
+ */
+ /* KW: Previously lambda was passed as zero, but I
+ * believe this is incorrect, the spec seems to
+ * indicate rather that lambda should not be
+ * changed/biased, unlike TXB where texcoord[3] is
+ * added to the lambda calculations. The lambda should
+ * still be calculated normally for TEX & TXP though,
+ * not set to zero. Otherwise it's very difficult to
+ * implement normal GL semantics through the fragment
+ * shader.
+ */
+ fetch_texel( ctx, texcoord,
+ span->array->lambda[inst->TexSrcUnit][column],
+ inst->TexSrcUnit, color );
+#if DEBUG_FRAG
+ if (color[3])
+ printf("color[3] = %f\n", color[3]);
+#endif
store_vector4( inst, machine, color );
}
break;
- case FP_OPCODE_TXB:
+ case OPCODE_TXB: /* GL_ARB_fragment_program only */
/* Texel lookup with LOD bias */
{
GLfloat texcoord[4], color[4], bias, lambda;
store_vector4( inst, machine, color );
}
break;
- case FP_OPCODE_TXD:
+ case OPCODE_TXD: /* GL_NV_fragment_program only */
/* Texture lookup w/ partial derivatives for LOD */
{
GLfloat texcoord[4], dtdx[4], dtdy[4], color[4];
store_vector4( inst, machine, color );
}
break;
- case FP_OPCODE_TXP:
- /* Texture lookup w/ perspective divide */
+ case 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 );
+ /* Not so sure about this test - if texcoord[3] is
+ * zero, we'd probably be fine except for an ASSERT in
+ * IROUND_POS() which gets triggered by the inf values created.
+ */
+ if (texcoord[3] != 0.0) {
+ texcoord[0] /= texcoord[3];
+ texcoord[1] /= texcoord[3];
+ texcoord[2] /= texcoord[3];
+ }
+ /* KW: Previously lambda was passed as zero, but I
+ * believe this is incorrect, the spec seems to
+ * indicate rather that lambda should not be
+ * changed/biased, unlike TXB where texcoord[3] is
+ * added to the lambda calculations. The lambda should
+ * still be calculated normally for TEX & TXP though,
+ * not set to zero.
+ */
+ fetch_texel( ctx, texcoord,
+ span->array->lambda[inst->TexSrcUnit][column],
+ inst->TexSrcUnit, color );
+ store_vector4( inst, machine, color );
+ }
+ break;
+ case OPCODE_TXP_NV: /* GL_NV_fragment_program only */
+ /* Texture lookup w/ projective divide */
{
GLfloat texcoord[4], color[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
- /* Already did perspective divide in interpolate_texcoords() */
+ if (inst->TexSrcTarget != TEXTURE_CUBE_INDEX &&
+ texcoord[3] != 0.0) {
+ 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 );
store_vector4( inst, machine, color );
}
break;
- case FP_OPCODE_UP2H: /* unpack two 16-bit floats */
+ case OPCODE_UP2H: /* unpack two 16-bit floats */
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_UP2US: /* unpack two GLushorts */
+ case OPCODE_UP2US: /* unpack two GLushorts */
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_UP4B: /* unpack four GLbytes */
+ case OPCODE_UP4B: /* unpack four GLbytes */
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) 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;
- case FP_OPCODE_UP4UB: /* unpack four GLubytes */
+ case OPCODE_UP4UB: /* unpack four GLubytes */
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) 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 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 */
+ case OPCODE_X2D: /* 2-D matrix transform */
{
GLfloat a[4], b[4], c[4], result[4];
fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
store_vector4( inst, machine, result );
}
break;
- case FP_OPCODE_END:
+ case OPCODE_PRINT:
+ {
+ if (inst->SrcReg[0].File != -1) {
+ GLfloat a[4];
+ fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a);
+ _mesa_printf("%s%g, %g, %g, %g\n", (const char *) inst->Data,
+ a[0], a[1], a[2], a[3]);
+ }
+ else {
+ _mesa_printf("%s\n", (const char *) inst->Data);
+ }
+ }
+ break;
+ case OPCODE_END:
return GL_TRUE;
default:
_mesa_problem(ctx, "Bad opcode %d in _mesa_exec_fragment_program",
const struct fragment_program *program,
const struct sw_span *span, GLuint col )
{
- GLuint inputsRead = program->InputsRead;
+ GLuint inputsRead = program->Base.InputsRead;
GLuint u;
if (ctx->FragmentProgram.CallbackEnabled)
inputsRead = ~0;
- /* Clear temporary registers */
- _mesa_bzero(machine->Temporaries,
- 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(machine->Temporaries,
+ MAX_NV_FRAGMENT_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
+ }
/* Load input registers */
if (inputsRead & (1 << FRAG_ATTRIB_WPOS)) {
GLfloat *wpos = machine->Inputs[FRAG_ATTRIB_WPOS];
+ ASSERT(span->arrayMask & SPAN_Z);
wpos[0] = (GLfloat) span->x + col;
wpos[1] = (GLfloat) span->y;
- wpos[2] = (GLfloat) span->array->z[col] / ctx->DepthMaxF;
+ wpos[2] = (GLfloat) span->array->z[col] / ctx->DrawBuffer->_DepthMaxF;
wpos[3] = span->w + col * span->dwdx;
}
if (inputsRead & (1 << FRAG_ATTRIB_COL0)) {
GLfloat *col0 = machine->Inputs[FRAG_ATTRIB_COL0];
+ ASSERT(span->arrayMask & SPAN_RGBA);
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]);
}
if (inputsRead & (1 << FRAG_ATTRIB_FOGC)) {
GLfloat *fogc = machine->Inputs[FRAG_ATTRIB_FOGC];
+ ASSERT(span->arrayMask & SPAN_FOG);
fogc[0] = span->array->fog[col];
fogc[1] = 0.0F;
fogc[2] = 0.0F;
}
+
+/**
+ * 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;
+ const struct fragment_program *program = ctx->FragmentProgram._Current;
GLuint i;
ctx->_CurrentProgram = GL_FRAGMENT_PROGRAM_ARB; /* or NV, doesn't matter */
+ if (program->Base.Parameters) {
+ _mesa_load_state_parameters(ctx, program->Base.Parameters);
+ }
+
for (i = 0; i < span->end; i++) {
if (span->array->mask[i]) {
init_machine(ctx, &ctx->FragmentProgram.Machine,
- ctx->FragmentProgram.Current, span, i);
+ ctx->FragmentProgram._Current, span, i);
if (!execute_program(ctx, program, ~0,
&ctx->FragmentProgram.Machine, span, i)) {
span->array->mask[i] = GL_FALSE; /* killed fragment */
+ span->writeAll = GL_FALSE;
}
/* Store output registers */
{
const GLfloat *colOut
- = ctx->FragmentProgram.Machine.Outputs[FRAG_OUTPUT_COLR];
+ = ctx->FragmentProgram.Machine.Outputs[FRAG_RESULT_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 (program->OutputsWritten & (1 << FRAG_OUTPUT_DEPR))
- span->array->z[i] = IROUND(ctx->FragmentProgram.Machine.Outputs[FRAG_OUTPUT_DEPR][0] * ctx->DepthMaxF);
+ if (program->Base.OutputsWritten & (1 << FRAG_RESULT_DEPR)) {
+ const GLfloat depth
+ = ctx->FragmentProgram.Machine.Outputs[FRAG_RESULT_DEPR][2];
+ span->array->z[i] = IROUND(depth * ctx->DrawBuffer->_DepthMaxF);
+ }
}
}
+ if (program->Base.OutputsWritten & (1 << FRAG_RESULT_DEPR)) {
+ span->interpMask &= ~SPAN_Z;
+ span->arrayMask |= SPAN_Z;
+ }
+
ctx->_CurrentProgram = 0;
}