-/* $Id: s_nvfragprog.c,v 1.10 2003/03/19 05:34:25 brianp Exp $ */
-
/*
* Mesa 3-D graphics library
* Version: 5.1
*/
-
#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"
#include "s_texture.h"
+/* if 1, print some debugging info */
+#define DEBUG_FRAG 0
+
+
/**
* Fetch a texel.
*/
static void
-fetch_texel( GLcontext *ctx, const GLfloat texcoord[4], GLuint unit,
- GLfloat color[4] )
+fetch_texel( GLcontext *ctx, const GLfloat texcoord[4], GLfloat lambda,
+ GLuint unit, GLfloat color[4] )
{
- const GLfloat *lambda = NULL;
GLchan rgba[4];
SWcontext *swrast = SWRAST_CONTEXT(ctx);
swrast->TextureSample[unit](ctx, unit, ctx->Texture.Unit[unit]._Current,
1, (const GLfloat (*)[4]) texcoord,
- lambda, &rgba);
+ &lambda, &rgba);
color[0] = CHAN_TO_FLOAT(rgba[0]);
color[1] = CHAN_TO_FLOAT(rgba[1]);
color[2] = CHAN_TO_FLOAT(rgba[2]);
const GLfloat texH = (GLfloat) texImg->HeightScale;
GLchan rgba[4];
- GLfloat lambda = _mesa_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
+ GLfloat lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
texdx[1], texdy[1], /* dt/dx, dt/dy */
texdx[3], texdy[2], /* dq/dx, dq/dy */
texW, texH,
}
+/**
+ * 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_NAMED_PARAM:
+ ASSERT(source->Index < (GLint) program->Parameters->NumParameters);
+ src = program->Parameters->Parameters[source->Index].Values;
+ break;
+ case PROGRAM_STATE_VAR:
+ src = NULL;
+ 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 = 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]];
ASSERT(xOrY == 'X' || xOrY == 'Y');
- switch (source->Register) {
+ assert(source->File == PROGRAM_INPUT);
+
+ switch (source->Index) {
case FRAG_ATTRIB_WPOS:
if (xOrY == 'X') {
src[0] = 1.0;
case FRAG_ATTRIB_TEX6:
case FRAG_ATTRIB_TEX7:
if (xOrY == 'X') {
- const GLuint u = source->Register - FRAG_ATTRIB_TEX0;
+ const GLuint u = source->Index - 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);
}
else {
- const GLuint u = source->Register - FRAG_ATTRIB_TEX0;
+ const GLuint u = source->Index - 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);
* 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 = machine->Registers[source->Register];
+ const GLfloat *src = get_register_pointer(ctx, source, machine, program);
+ ASSERT(src);
result[0] = src[source->Swizzle[0]];
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 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;
+ 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]) ||
+ IS_INF_OR_NAN(value[1]) ||
+ IS_INF_OR_NAN(value[2]) ||
+ IS_INF_OR_NAN(value[3]) )
+ printf("store %g %g %g %g\n", value[0], value[1], value[2], value[3]);
+#endif
+
if (clamp) {
clampedValue[0] = CLAMP(value[0], 0.0F, 1.0F);
clampedValue[1] = CLAMP(value[1], 0.0F, 1.0F);
_mesa_memcpy(dMachine, machine, sizeof(struct fp_machine));
/* Clear temporary registers */
- _mesa_bzero(machine->Registers + FP_TEMP_REG_START,
+ _mesa_bzero( (void*) machine->Temporaries,
MAX_NV_FRAGMENT_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
/* Add derivatives */
if (program->InputsRead & (1 << FRAG_ATTRIB_WPOS)) {
- GLfloat *wpos = 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;
}
}
if (program->InputsRead & (1 << FRAG_ATTRIB_COL0)) {
- GLfloat *col0 = 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);
}
}
if (program->InputsRead & (1 << FRAG_ATTRIB_COL1)) {
- GLfloat *col1 = 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);
}
}
if (program->InputsRead & (1 << FRAG_ATTRIB_FOGC)) {
- GLfloat *fogc = machine->Registers[FP_INPUT_REG_START+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))) {
- GLfloat *tex = machine->Registers[FP_INPUT_REG_START+FRAG_ATTRIB_TEX0+u];
+ GLfloat *tex = (GLfloat*) machine->Inputs[FRAG_ATTRIB_TEX0 + u];
if (xOrY == 'X') {
tex[0] += span->texStepX[u][0];
tex[1] += span->texStepX[u][1];
}
}
}
+
+ /* init condition codes */
+ dMachine->CondCodes[0] = COND_EQ;
+ dMachine->CondCodes[1] = COND_EQ;
+ dMachine->CondCodes[2] = COND_EQ;
+ dMachine->CondCodes[3] = COND_EQ;
}
static GLboolean
execute_program( GLcontext *ctx,
const struct fragment_program *program, GLuint maxInst,
- struct fp_machine *machine, const struct sw_span *span )
+ struct fp_machine *machine, const struct sw_span *span,
+ GLuint column )
{
- GLuint pc = 0;
+ GLuint pc;
+
+#if DEBUG_FRAG
+ printf("execute fragment program --------------------\n");
+#endif
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, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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];
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, 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;
* Finally, find the difference in the register values for
* the original and derivative runs.
*/
+ 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);
- fetch_vector4( &inst->SrcReg[0], &dMachine, aNext );
+ execute_program(ctx, program, pc, &dMachine, span, column);
+ 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];
if (!fetch_vector4_deriv(&inst->SrcReg[0], span, 'Y', result)) {
init_machine_deriv(ctx, machine, program, span,
'Y', &dMachine);
- execute_program(ctx, program, pc, &dMachine, span);
- fetch_vector4( &inst->SrcReg[0], &dMachine, aNext );
+ fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a);
+ execute_program(ctx, program, pc, &dMachine, span, column);
+ 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];
case FP_OPCODE_DP3:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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[0] * b[0] + a[1] * b[1] + a[2] * b[2];
store_vector4( inst, machine, result );
+#if DEBUG_FRAG
+ printf("DP3 %g = (%g %g %g) . (%g %g %g)\n",
+ result[0], a[0], a[1], a[2], b[0], b[1], b[2]);
+#endif
}
break;
case FP_OPCODE_DP4:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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 );
+ }
+ 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] + a[3] * b[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, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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];
case FP_OPCODE_EX2: /* Exponential base 2 */
{
GLfloat a[4], result[4];
- fetch_vector1( &inst->SrcReg[0], machine, 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 );
case FP_OPCODE_FLR:
{
GLfloat a[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, 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]);
case FP_OPCODE_FRC:
{
GLfloat a[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, 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]);
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_LG2: /* log base 2 */
{
GLfloat a[4], result[4];
- fetch_vector1( &inst->SrcReg[0], machine, 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 );
case FP_OPCODE_LIT:
{
GLfloat a[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, 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)_mesa_pow(2.0, a[3]) : 0.0F;
result[3] = 1.0F;
store_vector4( inst, machine, result );
}
case FP_OPCODE_LRP:
{
GLfloat a[4], b[4], c[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, b );
- fetch_vector4( &inst->SrcReg[2], machine, 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];
case FP_OPCODE_MAD:
{
GLfloat a[4], b[4], c[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, b );
- fetch_vector4( &inst->SrcReg[2], machine, 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];
case FP_OPCODE_MAX:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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]);
case FP_OPCODE_MIN:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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]);
case FP_OPCODE_MOV:
{
GLfloat result[4];
- fetch_vector4( &inst->SrcReg[0], machine, result );
+ fetch_vector4( ctx, &inst->SrcReg[0], machine, program, result );
store_vector4( inst, machine, result );
}
break;
case FP_OPCODE_MUL:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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];
result[3] = a[3] * b[3];
store_vector4( inst, machine, result );
+#if DEBUG_FRAG
+ printf("MUL (%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_PK2H: /* pack two 16-bit floats */
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
GLuint *rawResult = (GLuint *) result;
- fetch_vector4( &inst->SrcReg[0], machine, a );
+ fetch_vector4( ctx, &inst->SrcReg[0], machine, program, a );
rawResult[0] = rawResult[1] = rawResult[2] = rawResult[3]
= rawBits[0] | (rawBits[1] << 16);
store_vector4( inst, machine, result );
{
GLfloat a[4], result[4];
GLuint usx, usy, *rawResult = (GLuint *) result;
- fetch_vector4( &inst->SrcReg[0], machine, 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);
usx = IROUND(a[0] * 65535.0F);
{
GLfloat a[4], result[4];
GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result;
- fetch_vector4( &inst->SrcReg[0], machine, 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);
{
GLfloat a[4], result[4];
GLuint ubx, uby, ubz, ubw, *rawResult = (GLuint *) result;
- fetch_vector4( &inst->SrcReg[0], machine, 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);
case FP_OPCODE_POW:
{
GLfloat a[4], b[4], result[4];
- fetch_vector1( &inst->SrcReg[0], machine, a );
- fetch_vector1( &inst->SrcReg[1], machine, 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, a );
+ fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
+#if DEBUG_FRAG
+ if (a[0] == 0)
+ printf("RCP(0)\n");
+ else if (IS_INF_OR_NAN(a[0]))
+ printf("RCP(inf)\n");
+#endif
result[0] = result[1] = result[2] = result[3]
= 1.0F / a[0];
store_vector4( inst, machine, result );
case FP_OPCODE_RFL:
{
GLfloat axis[4], dir[4], result[4], tmp[4];
- fetch_vector4( &inst->SrcReg[0], machine, axis );
- fetch_vector4( &inst->SrcReg[1], machine, 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];
case FP_OPCODE_RSQ: /* 1 / sqrt() */
{
GLfloat a[4], result[4];
- fetch_vector1( &inst->SrcReg[0], machine, 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
+ printf("RSQ %g = 1/sqrt(%g)\n", result[0], a[0]);
+#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, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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;
case FP_OPCODE_SGE: /* set on greater or equal */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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;
case FP_OPCODE_SGT: /* set on greater */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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;
case FP_OPCODE_SIN:
{
GLfloat a[4], result[4];
- fetch_vector1( &inst->SrcReg[0], machine, 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, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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;
case FP_OPCODE_SLT: /* set on less */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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;
case FP_OPCODE_SNE: /* set on not equal */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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;
case FP_OPCODE_SUB:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, 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];
store_vector4( inst, machine, result );
}
break;
+ 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:
/* Texel lookup */
{
GLfloat texcoord[4], color[4];
- fetch_vector4( &inst->SrcReg[0], machine, texcoord );
+ fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
/* XXX: Undo perspective divide from interpolate_texcoords() */
- fetch_texel( ctx, texcoord, inst->TexSrcUnit, color );
+ fetch_texel( ctx, texcoord,
+ span->array->lambda[inst->TexSrcUnit][column],
+ inst->TexSrcUnit, color );
+ store_vector4( inst, machine, color );
+ }
+ break;
+ case FP_OPCODE_TXB:
+ /* 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;
/* Texture lookup w/ partial derivatives for LOD */
{
GLfloat texcoord[4], dtdx[4], dtdy[4], color[4];
- fetch_vector4( &inst->SrcReg[0], machine, texcoord );
- fetch_vector4( &inst->SrcReg[1], machine, dtdx );
- fetch_vector4( &inst->SrcReg[2], machine, 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 );
/* Texture lookup w/ perspective divide */
{
GLfloat texcoord[4], color[4];
- fetch_vector4( &inst->SrcReg[0], machine, texcoord );
+ fetch_vector4( ctx, &inst->SrcReg[0], machine, program, texcoord );
/* Already did perspective divide in interpolate_texcoords() */
- fetch_texel( ctx, texcoord, inst->TexSrcUnit, color );
+ fetch_texel( ctx, texcoord,
+ span->array->lambda[inst->TexSrcUnit][column],
+ inst->TexSrcUnit, color );
store_vector4( inst, machine, color );
}
break;
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
GLuint *rawResult = (GLuint *) result;
- fetch_vector1( &inst->SrcReg[0], machine, a );
+ fetch_vector1( ctx, &inst->SrcReg[0], machine, program, a );
rawResult[0] = rawBits[0] & 0xffff;
rawResult[1] = (rawBits[0] >> 16) & 0xffff;
rawResult[2] = rawBits[0] & 0xffff;
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
- fetch_vector1( &inst->SrcReg[0], machine, a );
+ fetch_vector1( ctx, &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];
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
- fetch_vector1( &inst->SrcReg[0], machine, 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;
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
- fetch_vector1( &inst->SrcReg[0], machine, 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;
case FP_OPCODE_X2D: /* 2-D matrix transform */
{
GLfloat a[4], b[4], c[4], result[4];
- fetch_vector4( &inst->SrcReg[0], machine, a );
- fetch_vector4( &inst->SrcReg[1], machine, b );
- fetch_vector4( &inst->SrcReg[2], machine, 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];
const struct fragment_program *program,
const struct sw_span *span, GLuint col )
{
- GLuint j, u;
+ GLuint inputsRead = program->InputsRead;
+ GLuint u;
+
+ if (ctx->FragmentProgram.CallbackEnabled)
+ inputsRead = ~0;
/* Clear temporary registers */
- _mesa_bzero(machine->Registers + FP_TEMP_REG_START,
+ _mesa_bzero(machine->Temporaries,
MAX_NV_FRAGMENT_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
- /* Load program local parameters */
- for (j = 0; j < MAX_NV_FRAGMENT_PROGRAM_PARAMS; j++) {
- COPY_4V(machine->Registers[FP_PROG_REG_START + j],
- program->LocalParams[j]);
- }
-
/* 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];
- if (ctx->Texture.Unit[u]._ReallyEnabled) {
- COPY_4V(tex, span->array->texcoords[u][col]);
- }
- else {
- COPY_4V(tex, ctx->Current.Attrib[VERT_ATTRIB_TEX0 + 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);*/
}
}
+
+ /* init condition codes */
+ machine->CondCodes[0] = COND_EQ;
+ machine->CondCodes[1] = COND_EQ;
+ machine->CondCodes[2] = COND_EQ;
+ machine->CondCodes[3] = COND_EQ;
}
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);
if (!execute_program(ctx, program, ~0,
- &ctx->FragmentProgram.Machine, span))
+ &ctx->FragmentProgram.Machine, span, i)) {
span->array->mask[i] = GL_FALSE; /* killed fragment */
+ }
/* 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;
}