/*
* Mesa 3-D graphics library
- * Version: 6.5.3
+ * Version: 7.0.3
*
* Copyright (C) 1999-2007 Brian Paul All Rights Reserved.
*
#include "prog_instruction.h"
#include "prog_parameter.h"
#include "prog_print.h"
-#include "slang_library_noise.h"
+#include "shader/slang/slang_library_noise.h"
-/* See comments below for info about this */
-#define LAMBDA_ZERO 1
-
/* debug predicate */
#define DEBUG_PROG 0
* source register.
*/
static INLINE const GLfloat *
-get_register_pointer(GLcontext * ctx,
- const struct prog_src_register *source,
+get_register_pointer(const struct prog_src_register *source,
const struct gl_program_machine *machine)
{
if (source->RelAddr) {
const GLint reg = source->Index + machine->AddressReg[0][0];
- ASSERT( (source->File == PROGRAM_ENV_PARAM) ||
- (source->File == PROGRAM_STATE_VAR) );
- if (reg < 0 || reg > MAX_NV_VERTEX_PROGRAM_PARAMS)
- return ZeroVec;
- else if (source->File == PROGRAM_ENV_PARAM)
- return ctx->VertexProgram.Parameters[reg];
+ if (source->File == PROGRAM_ENV_PARAM)
+ if (reg < 0 || reg >= MAX_PROGRAM_ENV_PARAMS)
+ return ZeroVec;
+ else
+ return machine->EnvParams[reg];
else {
+ const struct gl_program_parameter_list *params;
ASSERT(source->File == PROGRAM_LOCAL_PARAM ||
source->File == PROGRAM_STATE_VAR);
- return machine->CurProgram->Parameters->ParameterValues[reg];
+ params = machine->CurProgram->Parameters;
+ if (reg < 0 || reg >= params->NumParameters)
+ return ZeroVec;
+ else
+ return params->ParameterValues[reg];
}
}
case PROGRAM_ENV_PARAM:
ASSERT(source->Index < MAX_PROGRAM_ENV_PARAMS);
- if (machine->CurProgram->Target == GL_VERTEX_PROGRAM_ARB)
- return ctx->VertexProgram.Parameters[source->Index];
- else
- return ctx->FragmentProgram.Parameters[source->Index];
+ return machine->EnvParams[source->Index];
case PROGRAM_STATE_VAR:
/* Fallthrough */
return machine->CurProgram->Parameters->ParameterValues[source->Index];
default:
- _mesa_problem(ctx,
+ _mesa_problem(NULL,
"Invalid input register file %d in get_register_pointer()",
source->File);
return NULL;
const GLfloat *reg;
src.File = file;
src.Index = index;
- reg = get_register_pointer(ctx, &src, CurrentMachine);
+ reg = get_register_pointer(&src, CurrentMachine);
COPY_4V(val, reg);
}
}
* Apply swizzling and negating as needed.
*/
static void
-fetch_vector4(GLcontext * ctx,
- const struct prog_src_register *source,
+fetch_vector4(const struct prog_src_register *source,
const struct gl_program_machine *machine, GLfloat result[4])
{
- const GLfloat *src = get_register_pointer(ctx, source, machine);
+ const GLfloat *src = get_register_pointer(source, machine);
ASSERT(src);
if (source->Swizzle == SWIZZLE_NOOP) {
}
}
-#if 0
+
/**
- * Fetch the derivative with respect to X for the given register.
- * \return GL_TRUE if it was easily computed or GL_FALSE if we
- * need to execute another instance of the program (ugh)!
+ * Fetch the derivative with respect to X or Y for the given register.
+ * XXX this currently only works for fragment program input attribs.
*/
-static GLboolean
+static void
fetch_vector4_deriv(GLcontext * ctx,
const struct prog_src_register *source,
- const SWspan * span,
- char xOrY, GLint column, GLfloat result[4])
+ const struct gl_program_machine *machine,
+ char xOrY, GLfloat result[4])
{
- GLfloat src[4];
-
- ASSERT(xOrY == 'X' || xOrY == 'Y');
+ if (source->File == PROGRAM_INPUT && source->Index < machine->NumDeriv) {
+ const GLint col = machine->CurElement;
+ const GLfloat w = machine->Attribs[FRAG_ATTRIB_WPOS][col][3];
+ const GLfloat invQ = 1.0f / w;
+ GLfloat deriv[4];
- switch (source->Index) {
- case FRAG_ATTRIB_WPOS:
if (xOrY == 'X') {
- src[0] = 1.0;
- src[1] = 0.0;
- src[2] = span->attrStepX[FRAG_ATTRIB_WPOS][2]
- / ctx->DrawBuffer->_DepthMaxF;
- src[3] = span->attrStepX[FRAG_ATTRIB_WPOS][3];
+ deriv[0] = machine->DerivX[source->Index][0] * invQ;
+ deriv[1] = machine->DerivX[source->Index][1] * invQ;
+ deriv[2] = machine->DerivX[source->Index][2] * invQ;
+ deriv[3] = machine->DerivX[source->Index][3] * invQ;
}
else {
- src[0] = 0.0;
- src[1] = 1.0;
- src[2] = span->attrStepY[FRAG_ATTRIB_WPOS][2]
- / ctx->DrawBuffer->_DepthMaxF;
- src[3] = span->attrStepY[FRAG_ATTRIB_WPOS][3];
- }
- break;
- case FRAG_ATTRIB_COL0:
- case FRAG_ATTRIB_COL1:
- if (xOrY == 'X') {
- src[0] = span->attrStepX[source->Index][0] * (1.0F / CHAN_MAXF);
- src[1] = span->attrStepX[source->Index][1] * (1.0F / CHAN_MAXF);
- src[2] = span->attrStepX[source->Index][2] * (1.0F / CHAN_MAXF);
- src[3] = span->attrStepX[source->Index][3] * (1.0F / CHAN_MAXF);
- }
- else {
- src[0] = span->attrStepY[source->Index][0] * (1.0F / CHAN_MAXF);
- src[1] = span->attrStepY[source->Index][1] * (1.0F / CHAN_MAXF);
- src[2] = span->attrStepY[source->Index][2] * (1.0F / CHAN_MAXF);
- src[3] = span->attrStepY[source->Index][3] * (1.0F / CHAN_MAXF);
- }
- break;
- case FRAG_ATTRIB_FOGC:
- if (xOrY == 'X') {
- src[0] = span->attrStepX[FRAG_ATTRIB_FOGC][0] * (1.0F / CHAN_MAXF);
- src[1] = 0.0;
- src[2] = 0.0;
- src[3] = 0.0;
+ deriv[0] = machine->DerivY[source->Index][0] * invQ;
+ deriv[1] = machine->DerivY[source->Index][1] * invQ;
+ deriv[2] = machine->DerivY[source->Index][2] * invQ;
+ deriv[3] = machine->DerivY[source->Index][3] * invQ;
}
- else {
- src[0] = span->attrStepY[FRAG_ATTRIB_FOGC][0] * (1.0F / CHAN_MAXF);
- src[1] = 0.0;
- src[2] = 0.0;
- src[3] = 0.0;
+
+ result[0] = deriv[GET_SWZ(source->Swizzle, 0)];
+ result[1] = deriv[GET_SWZ(source->Swizzle, 1)];
+ result[2] = deriv[GET_SWZ(source->Swizzle, 2)];
+ result[3] = deriv[GET_SWZ(source->Swizzle, 3)];
+
+ if (source->NegateBase) {
+ result[0] = -result[0];
+ result[1] = -result[1];
+ result[2] = -result[2];
+ result[3] = -result[3];
}
- break;
- default:
- assert(source->Index < FRAG_ATTRIB_MAX);
- /* texcoord or varying */
- if (xOrY == 'X') {
- /* this is a little tricky - I think I've got it right */
- const GLfloat invQ = 1.0f / (span->attrStart[source->Index][3]
- +
- span->attrStepX[source->Index][3] *
- column);
- src[0] = span->attrStepX[source->Index][0] * invQ;
- src[1] = span->attrStepX[source->Index][1] * invQ;
- src[2] = span->attrStepX[source->Index][2] * invQ;
- src[3] = span->attrStepX[source->Index][3] * invQ;
+ if (source->Abs) {
+ result[0] = FABSF(result[0]);
+ result[1] = FABSF(result[1]);
+ result[2] = FABSF(result[2]);
+ result[3] = FABSF(result[3]);
}
- else {
- /* Tricky, as above, but in Y direction */
- const GLfloat invQ = 1.0f / (span->attrStart[source->Index][3]
- + span->attrStepY[source->Index][3]);
- src[0] = span->attrStepY[source->Index][0] * invQ;
- src[1] = span->attrStepY[source->Index][1] * invQ;
- src[2] = span->attrStepY[source->Index][2] * invQ;
- src[3] = span->attrStepY[source->Index][3] * invQ;
+ if (source->NegateAbs) {
+ result[0] = -result[0];
+ result[1] = -result[1];
+ result[2] = -result[2];
+ result[3] = -result[3];
}
- break;
}
-
- 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];
- result[1] = -result[1];
- result[2] = -result[2];
- result[3] = -result[3];
- }
- if (source->Abs) {
- result[0] = FABSF(result[0]);
- result[1] = FABSF(result[1]);
- result[2] = FABSF(result[2]);
- result[3] = FABSF(result[3]);
- }
- if (source->NegateAbs) {
- result[0] = -result[0];
- result[1] = -result[1];
- result[2] = -result[2];
- result[3] = -result[3];
+ else {
+ ASSIGN_4V(result, 0.0, 0.0, 0.0, 0.0);
}
- return GL_TRUE;
}
-#endif
/**
* As above, but only return result[0] element.
*/
static void
-fetch_vector1(GLcontext * ctx,
- const struct prog_src_register *source,
+fetch_vector1(const struct prog_src_register *source,
const struct gl_program_machine *machine, GLfloat result[4])
{
- const GLfloat *src = get_register_pointer(ctx, source, machine);
+ const GLfloat *src = get_register_pointer(source, machine);
ASSERT(src);
result[0] = src[GET_SWZ(source->Swizzle, 0)];
}
+/**
+ * Fetch texel from texture. Use partial derivatives when possible.
+ */
+static INLINE void
+fetch_texel(GLcontext *ctx,
+ const struct gl_program_machine *machine,
+ const struct prog_instruction *inst,
+ const GLfloat texcoord[4], GLfloat lodBias,
+ GLfloat color[4])
+{
+ /* Note: we only have the right derivatives for fragment input attribs.
+ */
+ if (machine->NumDeriv > 0 &&
+ inst->SrcReg[0].File == PROGRAM_INPUT &&
+ inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit) {
+ /* simple texture fetch for which we should have derivatives */
+ GLuint attr = inst->SrcReg[0].Index;
+ machine->FetchTexelDeriv(ctx, texcoord,
+ machine->DerivX[attr],
+ machine->DerivY[attr],
+ lodBias,
+ inst->TexSrcUnit, color);
+ }
+ else {
+ machine->FetchTexelLod(ctx, texcoord, lodBias,
+ inst->TexSrcUnit, color);
+ }
+}
+
+
/**
* Test value against zero and return GT, LT, EQ or UN if NaN.
*/
machine->CondCodes[2] = generate_cc(value[2]);
if (writeMask & WRITEMASK_W)
machine->CondCodes[3] = generate_cc(value[3]);
+#if DEBUG_PROG
+ printf("CondCodes=(%s,%s,%s,%s) for:\n",
+ _mesa_condcode_string(machine->CondCodes[0]),
+ _mesa_condcode_string(machine->CondCodes[1]),
+ _mesa_condcode_string(machine->CondCodes[2]),
+ _mesa_condcode_string(machine->CondCodes[3]));
+#endif
}
}
-#if 0
-/**
- * Initialize a new machine state instance from an existing one, adding
- * the partial derivatives onto the input registers.
- * Used to implement DDX and DDY instructions in non-trivial cases.
- */
-static void
-init_machine_deriv(GLcontext * ctx,
- const struct gl_program_machine *machine,
- const struct gl_fragment_program *program,
- const SWspan * span, char xOrY,
- struct gl_program_machine *dMachine)
-{
- GLuint attr;
-
- ASSERT(xOrY == 'X' || xOrY == 'Y');
-
- /* copy existing machine */
- _mesa_memcpy(dMachine, machine, sizeof(struct gl_program_machine));
-
- if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) {
- /* XXX also need to do this when using valgrind */
- /* Clear temporary registers (undefined for ARB_f_p) */
- _mesa_bzero((void *) machine->Temporaries,
- MAX_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
- }
-
- /* Add derivatives */
- if (program->Base.InputsRead & FRAG_BIT_WPOS) {
- GLfloat *wpos = machine->Attribs[FRAG_ATTRIB_WPOS][machine->CurElement];
- if (xOrY == 'X') {
- wpos[0] += 1.0F;
- wpos[1] += 0.0F;
- wpos[2] += span->attrStepX[FRAG_ATTRIB_WPOS][2];
- wpos[3] += span->attrStepX[FRAG_ATTRIB_WPOS][3];
- }
- else {
- wpos[0] += 0.0F;
- wpos[1] += 1.0F;
- wpos[2] += span->attrStepY[FRAG_ATTRIB_WPOS][2];
- wpos[3] += span->attrStepY[FRAG_ATTRIB_WPOS][3];
- }
- }
-
- /* primary, secondary colors */
- for (attr = FRAG_ATTRIB_COL0; attr <= FRAG_ATTRIB_COL1; attr++) {
- if (program->Base.InputsRead & (1 << attr)) {
- GLfloat *col = machine->Attribs[attr][machine->CurElement];
- if (xOrY == 'X') {
- col[0] += span->attrStepX[attr][0] * (1.0F / CHAN_MAXF);
- col[1] += span->attrStepX[attr][1] * (1.0F / CHAN_MAXF);
- col[2] += span->attrStepX[attr][2] * (1.0F / CHAN_MAXF);
- col[3] += span->attrStepX[attr][3] * (1.0F / CHAN_MAXF);
- }
- else {
- col[0] += span->attrStepY[attr][0] * (1.0F / CHAN_MAXF);
- col[1] += span->attrStepY[attr][1] * (1.0F / CHAN_MAXF);
- col[2] += span->attrStepY[attr][2] * (1.0F / CHAN_MAXF);
- col[3] += span->attrStepY[attr][3] * (1.0F / CHAN_MAXF);
- }
- }
- }
- if (program->Base.InputsRead & FRAG_BIT_FOGC) {
- GLfloat *fogc = machine->Attribs[FRAG_ATTRIB_FOGC][machine->CurElement];
- if (xOrY == 'X') {
- fogc[0] += span->attrStepX[FRAG_ATTRIB_FOGC][0];
- }
- else {
- fogc[0] += span->attrStepY[FRAG_ATTRIB_FOGC][0];
- }
- }
- /* texcoord and varying vars */
- for (attr = FRAG_ATTRIB_TEX0; attr < FRAG_ATTRIB_MAX; attr++) {
- if (program->Base.InputsRead & (1 << attr)) {
- GLfloat *val = machine->Attribs[attr][machine->CurElement];
- /* XXX perspective-correct interpolation */
- if (xOrY == 'X') {
- val[0] += span->attrStepX[attr][0];
- val[1] += span->attrStepX[attr][1];
- val[2] += span->attrStepX[attr][2];
- val[3] += span->attrStepX[attr][3];
- }
- else {
- val[0] += span->attrStepY[attr][0];
- val[1] += span->attrStepY[attr][1];
- val[2] += span->attrStepY[attr][2];
- val[3] += span->attrStepY[attr][3];
- }
- }
- }
-
- /* init condition codes */
- dMachine->CondCodes[0] = COND_EQ;
- dMachine->CondCodes[1] = COND_EQ;
- dMachine->CondCodes[2] = COND_EQ;
- dMachine->CondCodes[3] = COND_EQ;
-}
-#endif
-
-
/**
* Execute the given vertex/fragment program.
*
- * \param ctx - rendering context
- * \param program - the fragment program to execute
- * \param machine - machine state (register file)
+ * \param ctx rendering context
+ * \param program the program to execute
+ * \param machine machine state (must be initialized)
* \return GL_TRUE if program completed or GL_FALSE if program executed KIL.
*/
GLboolean
CurrentMachine = machine;
#endif
+ if (program->Target == GL_VERTEX_PROGRAM_ARB) {
+ machine->EnvParams = ctx->VertexProgram.Parameters;
+ }
+ else {
+ machine->EnvParams = ctx->FragmentProgram.Parameters;
+ }
+
for (pc = 0; pc < numInst; pc++) {
const struct prog_instruction *inst = program->Instructions + pc;
case OPCODE_ABS:
{
GLfloat a[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
result[0] = FABSF(a[0]);
result[1] = FABSF(a[1]);
result[2] = FABSF(a[2]);
case OPCODE_ADD:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
result[0] = a[0] + b[0];
result[1] = a[1] + b[1];
result[2] = a[2] + b[2];
case OPCODE_ARL:
{
GLfloat t[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, t);
+ fetch_vector4(&inst->SrcReg[0], machine, t);
machine->AddressReg[0][0] = (GLint) FLOORF(t[0]);
}
break;
if (machine->StackDepth >= MAX_PROGRAM_CALL_DEPTH) {
return GL_TRUE; /* Per GL_NV_vertex_program2 spec */
}
- machine->CallStack[machine->StackDepth++] = pc + 1;
- pc = inst->BranchTarget; /* XXX - 1 ??? */
+ machine->CallStack[machine->StackDepth++] = pc + 1; /* next inst */
+ /* Subtract 1 here since we'll do pc++ at end of for-loop */
+ pc = inst->BranchTarget - 1;
}
break;
case OPCODE_CMP:
{
GLfloat a[4], b[4], c[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
- fetch_vector4(ctx, &inst->SrcReg[2], machine, c);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[2], machine, 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];
case OPCODE_COS:
{
GLfloat a[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = result[1] = result[2] = result[3]
= (GLfloat) _mesa_cos(a[0]);
store_vector4(inst, machine, result);
break;
case OPCODE_DDX: /* Partial derivative with respect to X */
{
-#if 0
- GLfloat a[4], aNext[4], result[4];
- struct gl_program_machine dMachine;
- 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
- * preceeding instruction, then fetch the source register.
- * 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, 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];
- result[3] = aNext[3] - a[3];
- }
+ GLfloat result[4];
+ fetch_vector4_deriv(ctx, &inst->SrcReg[0], machine,
+ 'X', result);
store_vector4(inst, machine, result);
-#else
- store_vector4(inst, machine, ZeroVec);
-#endif
}
break;
case OPCODE_DDY: /* Partial derivative with respect to Y */
{
-#if 0
- GLfloat a[4], aNext[4], result[4];
- struct gl_program_machine dMachine;
- if (!fetch_vector4_deriv(ctx, &inst->SrcReg[0], span, 'Y',
- column, result)) {
- init_machine_deriv(ctx, machine, program, span,
- 'Y', &dMachine);
- 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];
- result[3] = aNext[3] - a[3];
- }
+ GLfloat result[4];
+ fetch_vector4_deriv(ctx, &inst->SrcReg[0], machine,
+ 'Y', result);
store_vector4(inst, machine, result);
-#else
- store_vector4(inst, machine, ZeroVec);
-#endif
}
break;
case OPCODE_DP3:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
result[0] = result[1] = result[2] = result[3] = DOT3(a, b);
store_vector4(inst, machine, result);
if (DEBUG_PROG) {
case OPCODE_DP4:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
result[0] = result[1] = result[2] = result[3] = DOT4(a, b);
store_vector4(inst, machine, result);
if (DEBUG_PROG) {
case OPCODE_DPH:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, 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);
case OPCODE_DST: /* Distance vector */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
result[0] = 1.0F;
result[1] = a[1] * b[1];
result[2] = a[2];
case OPCODE_EXP:
{
GLfloat t[4], q[4], floor_t0;
- fetch_vector1(ctx, &inst->SrcReg[0], machine, t);
+ fetch_vector1(&inst->SrcReg[0], machine, t);
floor_t0 = FLOORF(t[0]);
if (floor_t0 > FLT_MAX_EXP) {
SET_POS_INFINITY(q[0]);
case OPCODE_EX2: /* Exponential base 2 */
{
GLfloat a[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = result[1] = result[2] = result[3] =
(GLfloat) _mesa_pow(2.0, a[0]);
store_vector4(inst, machine, result);
case OPCODE_FLR:
{
GLfloat a[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
result[0] = FLOORF(a[0]);
result[1] = FLOORF(a[1]);
result[2] = FLOORF(a[2]);
case OPCODE_FRC:
{
GLfloat a[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
result[0] = a[0] - FLOORF(a[0]);
result[1] = a[1] - FLOORF(a[1]);
result[2] = a[2] - FLOORF(a[2]);
}
break;
case OPCODE_IF:
- if (eval_condition(machine, inst)) {
- /* do if-clause (just continue execution) */
- }
- else {
- /* go to the instruction after ELSE or ENDIF */
- assert(inst->BranchTarget >= 0);
- pc = inst->BranchTarget - 1;
+ {
+ GLboolean cond;
+ /* eval condition */
+ if (inst->SrcReg[0].File != PROGRAM_UNDEFINED) {
+ GLfloat a[4];
+ fetch_vector1(&inst->SrcReg[0], machine, a);
+ cond = (a[0] != 0.0);
+ }
+ else {
+ cond = eval_condition(machine, inst);
+ }
+ if (DEBUG_PROG) {
+ printf("IF: %d\n", cond);
+ }
+ /* do if/else */
+ if (cond) {
+ /* do if-clause (just continue execution) */
+ }
+ else {
+ /* go to the instruction after ELSE or ENDIF */
+ assert(inst->BranchTarget >= 0);
+ pc = inst->BranchTarget - 1;
+ }
}
break;
case OPCODE_ELSE:
case OPCODE_INT: /* float to int */
{
GLfloat a[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
result[0] = (GLfloat) (GLint) a[0];
result[1] = (GLfloat) (GLint) a[1];
result[2] = (GLfloat) (GLint) a[2];
case OPCODE_KIL: /* ARB_f_p only */
{
GLfloat a[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
if (a[0] < 0.0F || a[1] < 0.0F || a[2] < 0.0F || a[3] < 0.0F) {
return GL_FALSE;
}
case OPCODE_LG2: /* log base 2 */
{
GLfloat a[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = result[1] = result[2] = result[3] = LOG2(a[0]);
store_vector4(inst, machine, result);
}
{
const GLfloat epsilon = 1.0F / 256.0F; /* from NV VP spec */
GLfloat a[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
a[0] = MAX2(a[0], 0.0F);
a[1] = MAX2(a[1], 0.0F);
/* XXX ARB version clamps a[3], NV version doesn't */
case OPCODE_LOG:
{
GLfloat t[4], q[4], abs_t0;
- fetch_vector1(ctx, &inst->SrcReg[0], machine, t);
+ fetch_vector1(&inst->SrcReg[0], machine, t);
abs_t0 = FABSF(t[0]);
if (abs_t0 != 0.0F) {
/* Since we really can't handle infinite values on VMS
case OPCODE_LRP:
{
GLfloat a[4], b[4], c[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
- fetch_vector4(ctx, &inst->SrcReg[2], machine, c);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[2], machine, 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 OPCODE_MAD:
{
GLfloat a[4], b[4], c[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
- fetch_vector4(ctx, &inst->SrcReg[2], machine, c);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[2], machine, 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 OPCODE_MAX:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
result[0] = MAX2(a[0], b[0]);
result[1] = MAX2(a[1], b[1]);
result[2] = MAX2(a[2], b[2]);
case OPCODE_MIN:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
result[0] = MIN2(a[0], b[0]);
result[1] = MIN2(a[1], b[1]);
result[2] = MIN2(a[2], b[2]);
case OPCODE_MOV:
{
GLfloat result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, result);
+ fetch_vector4(&inst->SrcReg[0], machine, result);
store_vector4(inst, machine, result);
if (DEBUG_PROG) {
printf("MOV (%g %g %g %g)\n",
case OPCODE_MUL:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
result[0] = a[0] * b[0];
result[1] = a[1] * b[1];
result[2] = a[2] * b[2];
case OPCODE_NOISE1:
{
GLfloat a[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] =
result[1] =
result[2] = result[3] = _slang_library_noise1(a[0]);
case OPCODE_NOISE2:
{
GLfloat a[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
result[0] =
result[1] =
result[2] = result[3] = _slang_library_noise2(a[0], a[1]);
case OPCODE_NOISE3:
{
GLfloat a[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
result[0] =
result[1] =
result[2] =
case OPCODE_NOISE4:
{
GLfloat a[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
result[0] =
result[1] =
result[2] =
GLhalfNV hx, hy;
GLuint *rawResult = (GLuint *) result;
GLuint twoHalves;
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
hx = _mesa_float_to_half(a[0]);
hy = _mesa_float_to_half(a[1]);
twoHalves = hx | (hy << 16);
{
GLfloat a[4], result[4];
GLuint usx, usy, *rawResult = (GLuint *) result;
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
a[0] = CLAMP(a[0], 0.0F, 1.0F);
a[1] = CLAMP(a[1], 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(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, 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(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, 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 OPCODE_POW:
{
GLfloat a[4], b[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector1(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[1], machine, b);
result[0] = result[1] = result[2] = result[3]
= (GLfloat) _mesa_pow(a[0], b[0]);
store_vector4(inst, machine, result);
case OPCODE_RCP:
{
GLfloat a[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
if (DEBUG_PROG) {
if (a[0] == 0)
printf("RCP(0)\n");
if (machine->StackDepth == 0) {
return GL_TRUE; /* Per GL_NV_vertex_program2 spec */
}
- pc = machine->CallStack[--machine->StackDepth];
+ /* subtract one because of pc++ in the for loop */
+ pc = machine->CallStack[--machine->StackDepth] - 1;
}
break;
case OPCODE_RFL: /* reflection vector */
{
GLfloat axis[4], dir[4], result[4], tmpX, tmpW;
- fetch_vector4(ctx, &inst->SrcReg[0], machine, axis);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, dir);
+ fetch_vector4(&inst->SrcReg[0], machine, axis);
+ fetch_vector4(&inst->SrcReg[1], machine, dir);
tmpW = DOT3(axis, axis);
tmpX = (2.0F * DOT3(axis, dir)) / tmpW;
result[0] = tmpX * axis[0] - dir[0];
case OPCODE_RSQ: /* 1 / sqrt() */
{
GLfloat a[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
a[0] = FABSF(a[0]);
result[0] = result[1] = result[2] = result[3] = INV_SQRTF(a[0]);
store_vector4(inst, machine, result);
case OPCODE_SCS: /* sine and cos */
{
GLfloat a[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = (GLfloat) _mesa_cos(a[0]);
result[1] = (GLfloat) _mesa_sin(a[0]);
result[2] = 0.0; /* undefined! */
case OPCODE_SEQ: /* set on equal */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, 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;
result[3] = (a[3] == b[3]) ? 1.0F : 0.0F;
store_vector4(inst, machine, result);
+ if (DEBUG_PROG) {
+ printf("SEQ (%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]);
+ }
}
break;
case OPCODE_SFL: /* set false, operands ignored */
case OPCODE_SGE: /* set on greater or equal */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, 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;
result[3] = (a[3] >= b[3]) ? 1.0F : 0.0F;
store_vector4(inst, machine, result);
+ if (DEBUG_PROG) {
+ printf("SGE (%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]);
+ }
}
break;
case OPCODE_SGT: /* set on greater */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, 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;
result[3] = (a[3] > b[3]) ? 1.0F : 0.0F;
store_vector4(inst, machine, result);
if (DEBUG_PROG) {
- printf("SGT %g %g %g %g\n",
- result[0], result[1], result[2], result[3]);
+ printf("SGT (%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]);
}
}
break;
case OPCODE_SIN:
{
GLfloat a[4], result[4];
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = result[1] = result[2] = result[3]
= (GLfloat) _mesa_sin(a[0]);
store_vector4(inst, machine, result);
case OPCODE_SLE: /* set on less or equal */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, 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;
result[3] = (a[3] <= b[3]) ? 1.0F : 0.0F;
store_vector4(inst, machine, result);
+ if (DEBUG_PROG) {
+ printf("SLE (%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]);
+ }
}
break;
case OPCODE_SLT: /* set on less */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, 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;
result[3] = (a[3] < b[3]) ? 1.0F : 0.0F;
store_vector4(inst, machine, result);
+ if (DEBUG_PROG) {
+ printf("SLT (%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]);
+ }
}
break;
case OPCODE_SNE: /* set on not equal */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, 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;
result[3] = (a[3] != b[3]) ? 1.0F : 0.0F;
store_vector4(inst, machine, result);
+ if (DEBUG_PROG) {
+ printf("SNE (%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]);
+ }
}
break;
case OPCODE_STR: /* set true, operands ignored */
case OPCODE_SUB:
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
result[0] = a[0] - b[0];
result[1] = a[1] - b[1];
result[2] = a[2] - b[2];
case OPCODE_SWZ: /* extended swizzle */
{
const struct prog_src_register *source = &inst->SrcReg[0];
- const GLfloat *src = get_register_pointer(ctx, source, machine);
+ const GLfloat *src = get_register_pointer(source, machine);
GLfloat result[4];
GLuint i;
for (i = 0; i < 4; i++) {
}
break;
case OPCODE_TEX: /* Both ARB and NV frag prog */
- /* Texel lookup */
+ /* Simple texel lookup */
{
- /* Note: only use the precomputed lambda value when we're
- * sampling texture unit [K] with texcoord[K].
- * Otherwise, the lambda value may have no relation to the
- * instruction's texcoord or texture image. Using the wrong
- * lambda is usually bad news.
- * The rest of the time, just use zero (until we get a more
- * sophisticated way of computing lambda).
- */
- GLfloat coord[4], color[4], lambda;
-#if 0
- if (inst->SrcReg[0].File == PROGRAM_INPUT &&
- inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit)
- lambda = span->array->lambda[inst->TexSrcUnit][column];
- else
-#endif
- lambda = 0.0;
- fetch_vector4(ctx, &inst->SrcReg[0], machine, coord);
- machine->FetchTexelLod(ctx, coord, lambda, inst->TexSrcUnit,
- color);
+ GLfloat texcoord[4], color[4];
+ fetch_vector4(&inst->SrcReg[0], machine, texcoord);
+
+ fetch_texel(ctx, machine, inst, texcoord, 0.0, color);
+
if (DEBUG_PROG) {
- printf("TEX (%g, %g, %g, %g) = texture[%d][%g, %g, %g, %g], "
- "lod %f\n",
+ printf("TEX (%g, %g, %g, %g) = texture[%d][%g, %g, %g, %g]\n",
color[0], color[1], color[2], color[3],
inst->TexSrcUnit,
- coord[0], coord[1], coord[2], coord[3], lambda);
+ texcoord[0], texcoord[1], texcoord[2], texcoord[3]);
}
store_vector4(inst, machine, color);
}
{
const struct gl_texture_unit *texUnit
= &ctx->Texture.Unit[inst->TexSrcUnit];
- GLfloat coord[4], color[4], lambda, bias;
-#if 0
- if (inst->SrcReg[0].File == PROGRAM_INPUT &&
- inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit)
- lambda = span->array->lambda[inst->TexSrcUnit][column];
- else
-#endif
- lambda = 0.0;
- fetch_vector4(ctx, &inst->SrcReg[0], machine, coord);
- /* coord[3] is the bias to add to lambda */
- bias = texUnit->LodBias + coord[3];
- if (texUnit->_Current)
- bias += texUnit->_Current->LodBias;
- machine->FetchTexelLod(ctx, coord, lambda + bias,
- inst->TexSrcUnit, color);
+ GLfloat texcoord[4], color[4], lodBias;
+
+ fetch_vector4(&inst->SrcReg[0], machine, texcoord);
+
+ /* texcoord[3] is the bias to add to lambda */
+ lodBias = texUnit->LodBias + texcoord[3];
+ if (texUnit->_Current) {
+ lodBias += texUnit->_Current->LodBias;
+ }
+
+ fetch_texel(ctx, machine, inst, texcoord, lodBias, 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(ctx, &inst->SrcReg[0], machine, texcoord);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, dtdx);
- fetch_vector4(ctx, &inst->SrcReg[2], machine, dtdy);
+ fetch_vector4(&inst->SrcReg[0], machine, texcoord);
+ fetch_vector4(&inst->SrcReg[1], machine, dtdx);
+ fetch_vector4(&inst->SrcReg[2], machine, dtdy);
machine->FetchTexelDeriv(ctx, texcoord, dtdx, dtdy,
+ 0.0, /* lodBias */
inst->TexSrcUnit, color);
store_vector4(inst, machine, color);
}
case OPCODE_TXP: /* GL_ARB_fragment_program only */
/* Texture lookup w/ projective divide */
{
- GLfloat texcoord[4], color[4], lambda;
-#if 0
- if (inst->SrcReg[0].File == PROGRAM_INPUT &&
- inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit)
- lambda = span->array->lambda[inst->TexSrcUnit][column];
- else
-#endif
- lambda = 0.0;
- fetch_vector4(ctx, &inst->SrcReg[0], machine, texcoord);
+ GLfloat texcoord[4], color[4];
+
+ fetch_vector4(&inst->SrcReg[0], machine, 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.
texcoord[1] /= texcoord[3];
texcoord[2] /= texcoord[3];
}
- machine->FetchTexelLod(ctx, texcoord, lambda,
- inst->TexSrcUnit, color);
+
+ fetch_texel(ctx, machine, inst, texcoord, 0.0, color);
+
store_vector4(inst, machine, color);
}
break;
case OPCODE_TXP_NV: /* GL_NV_fragment_program only */
- /* Texture lookup w/ projective divide */
+ /* Texture lookup w/ projective divide, as above, but do not
+ * do the divide by w if sampling from a cube map.
+ */
{
- GLfloat texcoord[4], color[4], lambda;
-#if 0
- if (inst->SrcReg[0].File == PROGRAM_INPUT &&
- inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit)
- lambda = span->array->lambda[inst->TexSrcUnit][column];
- else
-#endif
- lambda = 0.0;
- fetch_vector4(ctx, &inst->SrcReg[0], machine, texcoord);
+ GLfloat texcoord[4], color[4];
+
+ fetch_vector4(&inst->SrcReg[0], machine, texcoord);
if (inst->TexSrcTarget != TEXTURE_CUBE_INDEX &&
texcoord[3] != 0.0) {
texcoord[0] /= texcoord[3];
texcoord[1] /= texcoord[3];
texcoord[2] /= texcoord[3];
}
- machine->FetchTexelLod(ctx, texcoord, lambda,
- inst->TexSrcUnit, color);
+
+ fetch_texel(ctx, machine, inst, texcoord, 0.0, color);
+
store_vector4(inst, machine, color);
}
break;
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
GLhalfNV hx, hy;
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
hx = rawBits[0] & 0xffff;
hy = rawBits[0] >> 16;
result[0] = result[2] = _mesa_half_to_float(hx);
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
GLushort usx, usy;
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
usx = rawBits[0] & 0xffff;
usy = rawBits[0] >> 16;
result[0] = result[2] = usx * (1.0f / 65535.0f);
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = (((rawBits[0] >> 0) & 0xff) - 128) / 127.0F;
result[1] = (((rawBits[0] >> 8) & 0xff) - 128) / 127.0F;
result[2] = (((rawBits[0] >> 16) & 0xff) - 128) / 127.0F;
{
GLfloat a[4], result[4];
const GLuint *rawBits = (const GLuint *) a;
- fetch_vector1(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = ((rawBits[0] >> 0) & 0xff) / 255.0F;
result[1] = ((rawBits[0] >> 8) & 0xff) / 255.0F;
result[2] = ((rawBits[0] >> 16) & 0xff) / 255.0F;
case OPCODE_XPD: /* cross product */
{
GLfloat a[4], b[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, 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);
+ if (DEBUG_PROG) {
+ printf("XPD (%g %g %g %g) = (%g %g %g) X (%g %g %g)\n",
+ result[0], result[1], result[2], result[3],
+ a[0], a[1], a[2], b[0], b[1], b[2]);
+ }
}
break;
case OPCODE_X2D: /* 2-D matrix transform */
{
GLfloat a[4], b[4], c[4], result[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
- fetch_vector4(ctx, &inst->SrcReg[1], machine, b);
- fetch_vector4(ctx, &inst->SrcReg[2], machine, c);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[1], machine, b);
+ fetch_vector4(&inst->SrcReg[2], machine, 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];
{
if (inst->SrcReg[0].File != -1) {
GLfloat a[4];
- fetch_vector4(ctx, &inst->SrcReg[0], machine, a);
+ fetch_vector4(&inst->SrcReg[0], machine, a);
_mesa_printf("%s%g, %g, %g, %g\n", (const char *) inst->Data,
a[0], a[1], a[2], a[3]);
}
projects / mesa.git / blobdiff