/*
* Mesa 3-D graphics library
- * Version: 7.3
*
* Copyright (C) 1999-2008 Brian Paul All Rights Reserved.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
*/
/**
*/
+#include "c99_math.h"
#include "main/glheader.h"
-#include "main/colormac.h"
#include "main/macros.h"
#include "prog_execute.h"
#include "prog_instruction.h"
/**
* Set x to positive or negative infinity.
*/
-#if defined(USE_IEEE) || defined(_WIN32)
#define SET_POS_INFINITY(x) \
do { \
fi_type fi; \
fi.i = 0xFF800000; \
x = fi.f; \
} while (0)
-#elif defined(VMS)
-#define SET_POS_INFINITY(x) x = __MAXFLOAT
-#define SET_NEG_INFINITY(x) x = -__MAXFLOAT
-#else
-#define SET_POS_INFINITY(x) x = (GLfloat) HUGE_VAL
-#define SET_NEG_INFINITY(x) x = (GLfloat) -HUGE_VAL
-#endif
#define SET_FLOAT_BITS(x, bits) ((fi_type *) (void *) &(x))->i = bits
static const GLfloat ZeroVec[4] = { 0.0F, 0.0F, 0.0F, 0.0F };
-
-/**
- * Return TRUE for +0 and other positive values, FALSE otherwise.
- * Used for RCC opcode.
- */
-static INLINE GLboolean
-positive(float x)
-{
- fi_type fi;
- fi.f = x;
- if (fi.i & 0x80000000)
- return GL_FALSE;
- return GL_TRUE;
-}
-
-
-
/**
* Return a pointer to the 4-element float vector specified by the given
* source register.
*/
-static INLINE const GLfloat *
+static inline const GLfloat *
get_src_register_pointer(const struct prog_src_register *source,
const struct gl_program_machine *machine)
{
return machine->VertAttribs[reg];
}
else {
- if (reg >= FRAG_ATTRIB_MAX)
+ if (reg >= VARYING_SLOT_MAX)
return ZeroVec;
return machine->Attribs[reg][machine->CurElement];
}
return ZeroVec;
return machine->Outputs[reg];
- case PROGRAM_LOCAL_PARAM:
- if (reg >= MAX_PROGRAM_LOCAL_PARAMS)
- return ZeroVec;
- return machine->CurProgram->LocalParams[reg];
-
- case PROGRAM_ENV_PARAM:
- if (reg >= MAX_PROGRAM_ENV_PARAMS)
- return ZeroVec;
- return machine->EnvParams[reg];
-
case PROGRAM_STATE_VAR:
/* Fallthrough */
case PROGRAM_CONSTANT:
/* Fallthrough */
case PROGRAM_UNIFORM:
- /* Fallthrough */
- case PROGRAM_NAMED_PARAM:
if (reg >= (GLint) prog->Parameters->NumParameters)
return ZeroVec;
- return prog->Parameters->ParameterValues[reg];
+ return (GLfloat *) prog->Parameters->ParameterValues[reg];
+
+ case PROGRAM_SYSTEM_VALUE:
+ assert(reg < (GLint) ARRAY_SIZE(machine->SystemValues));
+ return machine->SystemValues[reg];
default:
_mesa_problem(NULL,
"Invalid src register file %d in get_src_register_pointer()",
source->File);
- return NULL;
+ return ZeroVec;
}
}
* Return a pointer to the 4-element float vector specified by the given
* destination register.
*/
-static INLINE GLfloat *
+static inline GLfloat *
get_dst_register_pointer(const struct prog_dst_register *dest,
struct gl_program_machine *machine)
{
return dummyReg;
return machine->Outputs[reg];
- case PROGRAM_WRITE_ONLY:
- return dummyReg;
-
default:
_mesa_problem(NULL,
"Invalid dest register file %d in get_dst_register_pointer()",
dest->File);
- return NULL;
+ return dummyReg;
}
}
const struct gl_program_machine *machine, GLfloat result[4])
{
const GLfloat *src = get_src_register_pointer(source, machine);
- ASSERT(src);
if (source->Swizzle == SWIZZLE_NOOP) {
/* no swizzling */
COPY_4V(result, src);
}
else {
- ASSERT(GET_SWZ(source->Swizzle, 0) <= 3);
- ASSERT(GET_SWZ(source->Swizzle, 1) <= 3);
- ASSERT(GET_SWZ(source->Swizzle, 2) <= 3);
- ASSERT(GET_SWZ(source->Swizzle, 3) <= 3);
+ assert(GET_SWZ(source->Swizzle, 0) <= 3);
+ assert(GET_SWZ(source->Swizzle, 1) <= 3);
+ assert(GET_SWZ(source->Swizzle, 2) <= 3);
+ assert(GET_SWZ(source->Swizzle, 3) <= 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)];
}
if (source->Abs) {
- result[0] = FABSF(result[0]);
- result[1] = FABSF(result[1]);
- result[2] = FABSF(result[2]);
- result[3] = FABSF(result[3]);
+ result[0] = fabsf(result[0]);
+ result[1] = fabsf(result[1]);
+ result[2] = fabsf(result[2]);
+ result[3] = fabsf(result[3]);
}
if (source->Negate) {
- ASSERT(source->Negate == NEGATE_XYZW);
+ assert(source->Negate == NEGATE_XYZW);
result[0] = -result[0];
result[1] = -result[1];
result[2] = -result[2];
}
-/**
- * Fetch a 4-element uint vector from the given source register.
- * Apply swizzling but not negation/abs.
- */
-static void
-fetch_vector4ui(const struct prog_src_register *source,
- const struct gl_program_machine *machine, GLuint result[4])
-{
- const GLuint *src = (GLuint *) get_src_register_pointer(source, machine);
- ASSERT(src);
-
- if (source->Swizzle == SWIZZLE_NOOP) {
- /* no swizzling */
- COPY_4V(result, src);
- }
- else {
- ASSERT(GET_SWZ(source->Swizzle, 0) <= 3);
- ASSERT(GET_SWZ(source->Swizzle, 1) <= 3);
- ASSERT(GET_SWZ(source->Swizzle, 2) <= 3);
- ASSERT(GET_SWZ(source->Swizzle, 3) <= 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)];
- }
-
- /* Note: no Negate or Abs here */
-}
-
-
-
/**
* Fetch the derivative with respect to X or Y for the given register.
* XXX this currently only works for fragment program input attribs.
*/
static void
-fetch_vector4_deriv(GLcontext * ctx,
+fetch_vector4_deriv(struct gl_context * ctx,
const struct prog_src_register *source,
const struct gl_program_machine *machine,
char xOrY, GLfloat result[4])
if (source->File == PROGRAM_INPUT &&
source->Index < (GLint) machine->NumDeriv) {
const GLint col = machine->CurElement;
- const GLfloat w = machine->Attribs[FRAG_ATTRIB_WPOS][col][3];
+ const GLfloat w = machine->Attribs[VARYING_SLOT_POS][col][3];
const GLfloat invQ = 1.0f / w;
GLfloat deriv[4];
result[3] = deriv[GET_SWZ(source->Swizzle, 3)];
if (source->Abs) {
- result[0] = FABSF(result[0]);
- result[1] = FABSF(result[1]);
- result[2] = FABSF(result[2]);
- result[3] = FABSF(result[3]);
+ result[0] = fabsf(result[0]);
+ result[1] = fabsf(result[1]);
+ result[2] = fabsf(result[2]);
+ result[3] = fabsf(result[3]);
}
if (source->Negate) {
- ASSERT(source->Negate == NEGATE_XYZW);
+ assert(source->Negate == NEGATE_XYZW);
result[0] = -result[0];
result[1] = -result[1];
result[2] = -result[2];
const struct gl_program_machine *machine, GLfloat result[4])
{
const GLfloat *src = get_src_register_pointer(source, machine);
- ASSERT(src);
result[0] = src[GET_SWZ(source->Swizzle, 0)];
if (source->Abs) {
- result[0] = FABSF(result[0]);
+ result[0] = fabsf(result[0]);
}
if (source->Negate) {
result[0] = -result[0];
}
-static GLuint
-fetch_vector1ui(const struct prog_src_register *source,
- const struct gl_program_machine *machine)
-{
- const GLuint *src = (GLuint *) get_src_register_pointer(source, machine);
- return src[GET_SWZ(source->Swizzle, 0)];
-}
-
-
/**
* Fetch texel from texture. Use partial derivatives when possible.
*/
-static INLINE void
-fetch_texel(GLcontext *ctx,
+static inline void
+fetch_texel(struct gl_context *ctx,
const struct gl_program_machine *machine,
const struct prog_instruction *inst,
const GLfloat texcoord[4], GLfloat lodBias,
*/
if (machine->NumDeriv > 0 &&
inst->SrcReg[0].File == PROGRAM_INPUT &&
- inst->SrcReg[0].Index == FRAG_ATTRIB_TEX0 + inst->TexSrcUnit) {
+ inst->SrcReg[0].Index == VARYING_SLOT_TEX0 + inst->TexSrcUnit) {
/* simple texture fetch for which we should have derivatives */
GLuint attr = inst->SrcReg[0].Index;
machine->FetchTexelDeriv(ctx, texcoord,
}
-/**
- * Test value against zero and return GT, LT, EQ or UN if NaN.
- */
-static INLINE GLuint
-generate_cc(float value)
-{
- if (value != value)
- return COND_UN; /* NaN */
- if (value > 0.0F)
- return COND_GT;
- if (value < 0.0F)
- return COND_LT;
- return COND_EQ;
-}
-
-
-/**
- * Test if the ccMaskRule is satisfied by the given condition code.
- * Used to mask destination writes according to the current condition code.
- */
-static INLINE GLboolean
-test_cc(GLuint condCode, GLuint ccMaskRule)
-{
- switch (ccMaskRule) {
- case COND_EQ: return (condCode == COND_EQ);
- case COND_NE: return (condCode != COND_EQ);
- case COND_LT: return (condCode == COND_LT);
- case COND_GE: return (condCode == COND_GT || condCode == COND_EQ);
- case COND_LE: return (condCode == COND_LT || condCode == COND_EQ);
- case COND_GT: return (condCode == COND_GT);
- case COND_TR: return GL_TRUE;
- case COND_FL: return GL_FALSE;
- default: return GL_TRUE;
- }
-}
-
-
-/**
- * Evaluate the 4 condition codes against a predicate and return GL_TRUE
- * or GL_FALSE to indicate result.
- */
-static INLINE GLboolean
-eval_condition(const struct gl_program_machine *machine,
- const struct prog_instruction *inst)
-{
- const GLuint swizzle = inst->DstReg.CondSwizzle;
- const GLuint condMask = inst->DstReg.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_TRUE;
- }
- else {
- return GL_FALSE;
- }
-}
-
-
-
/**
* Store 4 floats into a register. Observe the instructions saturate and
* set-condition-code flags.
struct gl_program_machine *machine, const GLfloat value[4])
{
const struct prog_dst_register *dstReg = &(inst->DstReg);
- const GLboolean clamp = inst->SaturateMode == SATURATE_ZERO_ONE;
+ const GLboolean clamp = inst->Saturate;
GLuint writeMask = dstReg->WriteMask;
GLfloat clampedValue[4];
GLfloat *dst = get_dst_register_pointer(dstReg, machine);
value = clampedValue;
}
- if (dstReg->CondMask != COND_TR) {
- /* condition codes may turn off some writes */
- if (writeMask & WRITEMASK_X) {
- if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 0)],
- dstReg->CondMask))
- writeMask &= ~WRITEMASK_X;
- }
- if (writeMask & WRITEMASK_Y) {
- if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 1)],
- dstReg->CondMask))
- writeMask &= ~WRITEMASK_Y;
- }
- if (writeMask & WRITEMASK_Z) {
- if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 2)],
- dstReg->CondMask))
- writeMask &= ~WRITEMASK_Z;
- }
- if (writeMask & WRITEMASK_W) {
- if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 3)],
- dstReg->CondMask))
- writeMask &= ~WRITEMASK_W;
- }
- }
-
#ifdef NAN_CHECK
assert(!IS_INF_OR_NAN(value[0]));
assert(!IS_INF_OR_NAN(value[0]));
dst[2] = value[2];
if (writeMask & WRITEMASK_W)
dst[3] = value[3];
-
- if (inst->CondUpdate) {
- if (writeMask & WRITEMASK_X)
- machine->CondCodes[0] = generate_cc(value[0]);
- if (writeMask & WRITEMASK_Y)
- machine->CondCodes[1] = generate_cc(value[1]);
- if (writeMask & WRITEMASK_Z)
- 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
- }
}
-/**
- * Store 4 uints into a register. Observe the set-condition-code flags.
- */
-static void
-store_vector4ui(const struct prog_instruction *inst,
- struct gl_program_machine *machine, const GLuint value[4])
-{
- const struct prog_dst_register *dstReg = &(inst->DstReg);
- GLuint writeMask = dstReg->WriteMask;
- GLuint *dst = (GLuint *) get_dst_register_pointer(dstReg, machine);
-
- if (dstReg->CondMask != COND_TR) {
- /* condition codes may turn off some writes */
- if (writeMask & WRITEMASK_X) {
- if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 0)],
- dstReg->CondMask))
- writeMask &= ~WRITEMASK_X;
- }
- if (writeMask & WRITEMASK_Y) {
- if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 1)],
- dstReg->CondMask))
- writeMask &= ~WRITEMASK_Y;
- }
- if (writeMask & WRITEMASK_Z) {
- if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 2)],
- dstReg->CondMask))
- writeMask &= ~WRITEMASK_Z;
- }
- if (writeMask & WRITEMASK_W) {
- if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 3)],
- dstReg->CondMask))
- writeMask &= ~WRITEMASK_W;
- }
- }
-
- if (writeMask & WRITEMASK_X)
- dst[0] = value[0];
- if (writeMask & WRITEMASK_Y)
- dst[1] = value[1];
- if (writeMask & WRITEMASK_Z)
- dst[2] = value[2];
- if (writeMask & WRITEMASK_W)
- dst[3] = value[3];
-
- if (inst->CondUpdate) {
- if (writeMask & WRITEMASK_X)
- machine->CondCodes[0] = generate_cc((float)value[0]);
- if (writeMask & WRITEMASK_Y)
- machine->CondCodes[1] = generate_cc((float)value[1]);
- if (writeMask & WRITEMASK_Z)
- machine->CondCodes[2] = generate_cc((float)value[2]);
- if (writeMask & WRITEMASK_W)
- machine->CondCodes[3] = generate_cc((float)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
- }
-}
-
-
-
/**
* Execute the given vertex/fragment program.
*
* \return GL_TRUE if program completed or GL_FALSE if program executed KIL.
*/
GLboolean
-_mesa_execute_program(GLcontext * ctx,
+_mesa_execute_program(struct gl_context * ctx,
const struct gl_program *program,
struct gl_program_machine *machine)
{
const GLuint numInst = program->NumInstructions;
- const GLuint maxExec = 10000;
+ const GLuint maxExec = 65536;
GLuint pc, numExec = 0;
machine->CurProgram = program;
{
GLfloat a[4], result[4];
fetch_vector4(&inst->SrcReg[0], machine, a);
- result[0] = FABSF(a[0]);
- result[1] = FABSF(a[1]);
- result[2] = FABSF(a[2]);
- result[3] = FABSF(a[3]);
+ 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;
}
}
break;
- case OPCODE_AND: /* bitwise AND */
- {
- GLuint a[4], b[4], result[4];
- fetch_vector4ui(&inst->SrcReg[0], machine, a);
- fetch_vector4ui(&inst->SrcReg[1], machine, 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_vector4ui(inst, machine, result);
- }
- break;
case OPCODE_ARL:
{
GLfloat t[4];
break;
case OPCODE_BGNLOOP:
/* no-op */
- ASSERT(program->Instructions[inst->BranchTarget].Opcode
+ assert(program->Instructions[inst->BranchTarget].Opcode
== OPCODE_ENDLOOP);
break;
case OPCODE_ENDLOOP:
/* subtract 1 here since pc is incremented by for(pc) loop */
- ASSERT(program->Instructions[inst->BranchTarget].Opcode
+ assert(program->Instructions[inst->BranchTarget].Opcode
== OPCODE_BGNLOOP);
pc = inst->BranchTarget - 1; /* go to matching BNGLOOP */
break;
break;
case OPCODE_ENDSUB: /* end subroutine */
break;
- case OPCODE_BRA: /* branch (conditional) */
- if (eval_condition(machine, inst)) {
- /* take branch */
- /* Subtract 1 here since we'll do pc++ below */
- pc = inst->BranchTarget - 1;
- }
- break;
case OPCODE_BRK: /* break out of loop (conditional) */
- ASSERT(program->Instructions[inst->BranchTarget].Opcode
+ assert(program->Instructions[inst->BranchTarget].Opcode
== OPCODE_ENDLOOP);
- if (eval_condition(machine, inst)) {
- /* break out of loop */
- /* pc++ at end of for-loop will put us after the ENDLOOP inst */
- pc = inst->BranchTarget;
- }
+ /* break out of loop */
+ /* pc++ at end of for-loop will put us after the ENDLOOP inst */
+ pc = inst->BranchTarget;
break;
case OPCODE_CONT: /* continue loop (conditional) */
- ASSERT(program->Instructions[inst->BranchTarget].Opcode
+ assert(program->Instructions[inst->BranchTarget].Opcode
== OPCODE_ENDLOOP);
- if (eval_condition(machine, inst)) {
- /* continue at ENDLOOP */
- /* Subtract 1 here since we'll do pc++ at end of for-loop */
- pc = inst->BranchTarget - 1;
- }
+ /* continue at ENDLOOP */
+ /* Subtract 1 here since we'll do pc++ at end of for-loop */
+ pc = inst->BranchTarget - 1;
break;
case OPCODE_CAL: /* Call subroutine (conditional) */
- if (eval_condition(machine, inst)) {
- /* call the subroutine */
- if (machine->StackDepth >= MAX_PROGRAM_CALL_DEPTH) {
- return GL_TRUE; /* Per GL_NV_vertex_program2 spec */
- }
- 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;
+ /* call the subroutine */
+ if (machine->StackDepth >= MAX_PROGRAM_CALL_DEPTH) {
+ return GL_TRUE; /* Per GL_NV_vertex_program2 spec */
}
+ 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], result[4];
fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = result[1] = result[2] = result[3]
- = (GLfloat) cos(a[0]);
+ = cosf(a[0]);
store_vector4(inst, machine, result);
}
break;
}
}
break;
- case OPCODE_DP2A:
- {
- GLfloat a[4], b[4], c, result[4];
- fetch_vector4(&inst->SrcReg[0], machine, a);
- fetch_vector4(&inst->SrcReg[1], machine, b);
- fetch_vector1(&inst->SrcReg[1], machine, &c);
- result[0] = result[1] = result[2] = result[3] = DOT2(a, b) + c;
- store_vector4(inst, machine, result);
- if (DEBUG_PROG) {
- printf("DP2A %g = (%g %g) . (%g %g) + %g\n",
- result[0], a[0], a[1], b[0], b[1], c);
- }
- }
- break;
case OPCODE_DP3:
{
GLfloat a[4], b[4], result[4];
{
GLfloat t[4], q[4], floor_t0;
fetch_vector1(&inst->SrcReg[0], machine, t);
- floor_t0 = FLOORF(t[0]);
+ floor_t0 = floorf(t[0]);
if (floor_t0 > FLT_MAX_EXP) {
SET_POS_INFINITY(q[0]);
SET_POS_INFINITY(q[2]);
q[2] = 0.0F;
}
else {
- q[0] = LDEXPF(1.0, (int) floor_t0);
+ q[0] = ldexpf(1.0, (int) floor_t0);
/* Note: GL_NV_vertex_program expects
* result.z = result.x * APPX(result.y)
* We do what the ARB extension says.
*/
- q[2] = (GLfloat) pow(2.0, t[0]);
+ q[2] = exp2f(t[0]);
}
q[1] = t[0] - floor_t0;
q[3] = 1.0F;
{
GLfloat a[4], result[4], val;
fetch_vector1(&inst->SrcReg[0], machine, a);
- val = (GLfloat) pow(2.0, a[0]);
+ val = exp2f(a[0]);
/*
if (IS_INF_OR_NAN(val))
val = 1.0e10;
{
GLfloat a[4], result[4];
fetch_vector4(&inst->SrcReg[0], machine, a);
- result[0] = FLOORF(a[0]);
- result[1] = FLOORF(a[1]);
- result[2] = FLOORF(a[2]);
- result[3] = FLOORF(a[3]);
+ result[0] = floorf(a[0]);
+ result[1] = floorf(a[1]);
+ result[2] = floorf(a[2]);
+ result[3] = floorf(a[3]);
store_vector4(inst, machine, result);
}
break;
{
GLfloat a[4], result[4];
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]);
- result[3] = a[3] - FLOORF(a[3]);
+ result[0] = a[0] - floorf(a[0]);
+ result[1] = a[1] - floorf(a[1]);
+ result[2] = a[2] - floorf(a[2]);
+ result[3] = a[3] - floorf(a[3]);
store_vector4(inst, machine, result);
}
break;
case OPCODE_IF:
{
GLboolean cond;
- ASSERT(program->Instructions[inst->BranchTarget].Opcode
+ assert(program->Instructions[inst->BranchTarget].Opcode
== OPCODE_ELSE ||
program->Instructions[inst->BranchTarget].Opcode
== OPCODE_ENDIF);
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);
+ cond = (a[0] != 0.0F);
}
if (DEBUG_PROG) {
printf("IF: %d\n", cond);
break;
case OPCODE_ELSE:
/* goto ENDIF */
- ASSERT(program->Instructions[inst->BranchTarget].Opcode
+ assert(program->Instructions[inst->BranchTarget].Opcode
== OPCODE_ENDIF);
assert(inst->BranchTarget >= 0);
pc = inst->BranchTarget;
case OPCODE_ENDIF:
/* nothing */
break;
- case OPCODE_KIL_NV: /* NV_f_p only (conditional) */
- if (eval_condition(machine, inst)) {
- return GL_FALSE;
- }
- break;
case OPCODE_KIL: /* ARB_f_p only */
{
GLfloat a[4];
val = -FLT_MAX;
}
else {
- val = (float)(log(a[0]) * 1.442695F);
+ val = logf(a[0]) * 1.442695F;
}
result[0] = result[1] = result[2] = result[3] = val;
store_vector4(inst, machine, result);
result[1] = a[0];
/* XXX we could probably just use pow() here */
if (a[0] > 0.0F) {
- if (a[1] == 0.0 && a[3] == 0.0)
+ if (a[1] == 0.0F && a[3] == 0.0F)
result[2] = 1.0F;
else
- result[2] = (GLfloat) pow(a[1], a[3]);
+ result[2] = powf(a[1], a[3]);
}
else {
result[2] = 0.0F;
{
GLfloat t[4], q[4], abs_t0;
fetch_vector1(&inst->SrcReg[0], machine, t);
- abs_t0 = FABSF(t[0]);
+ abs_t0 = fabsf(t[0]);
if (abs_t0 != 0.0F) {
- /* Since we really can't handle infinite values on VMS
- * like other OSes we'll use __MAXFLOAT to represent
- * infinity. This may need some tweaking.
- */
-#ifdef VMS
- if (abs_t0 == __MAXFLOAT)
-#else
if (IS_INF_OR_NAN(abs_t0))
-#endif
{
SET_POS_INFINITY(q[0]);
q[1] = 1.0F;
}
else {
int exponent;
- GLfloat mantissa = FREXPF(t[0], &exponent);
+ GLfloat mantissa = frexpf(t[0], &exponent);
q[0] = (GLfloat) (exponent - 1);
- q[1] = (GLfloat) (2.0 * mantissa); /* map [.5, 1) -> [1, 2) */
+ q[1] = 2.0F * mantissa; /* map [.5, 1) -> [1, 2) */
/* The fast LOG2 macro doesn't meet the precision
* requirements.
*/
- q[2] = (float)(log(t[0]) * 1.442695F);
+ q[2] = logf(t[0]) * 1.442695F;
}
}
else {
break;
case OPCODE_NOP:
break;
- case OPCODE_NOT: /* bitwise NOT */
- {
- GLuint a[4], result[4];
- fetch_vector4ui(&inst->SrcReg[0], machine, a);
- result[0] = ~a[0];
- result[1] = ~a[1];
- result[2] = ~a[2];
- result[3] = ~a[3];
- store_vector4ui(inst, machine, result);
- }
- break;
- case OPCODE_NRM3: /* 3-component normalization */
- {
- GLfloat a[4], result[4];
- GLfloat tmp;
- fetch_vector4(&inst->SrcReg[0], machine, a);
- tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2];
- if (tmp != 0.0F)
- tmp = INV_SQRTF(tmp);
- result[0] = tmp * a[0];
- result[1] = tmp * a[1];
- result[2] = tmp * a[2];
- result[3] = 0.0; /* undefined, but prevent valgrind warnings */
- store_vector4(inst, machine, result);
- }
- break;
- case OPCODE_NRM4: /* 4-component normalization */
- {
- GLfloat a[4], result[4];
- GLfloat tmp;
- fetch_vector4(&inst->SrcReg[0], machine, a);
- tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2] + a[3] * a[3];
- if (tmp != 0.0F)
- tmp = INV_SQRTF(tmp);
- result[0] = tmp * a[0];
- result[1] = tmp * a[1];
- result[2] = tmp * a[2];
- result[3] = tmp * a[3];
- store_vector4(inst, machine, result);
- }
- break;
- case OPCODE_OR: /* bitwise OR */
- {
- GLuint a[4], b[4], result[4];
- fetch_vector4ui(&inst->SrcReg[0], machine, a);
- fetch_vector4ui(&inst->SrcReg[1], machine, 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_vector4ui(inst, machine, result);
- }
- break;
- case OPCODE_PK2H: /* pack two 16-bit floats in one 32-bit float */
- {
- GLfloat a[4];
- GLuint result[4];
- GLhalfNV hx, hy;
- fetch_vector4(&inst->SrcReg[0], machine, a);
- hx = _mesa_float_to_half(a[0]);
- hy = _mesa_float_to_half(a[1]);
- result[0] =
- result[1] =
- result[2] =
- result[3] = hx | (hy << 16);
- store_vector4ui(inst, machine, result);
- }
- break;
- case OPCODE_PK2US: /* pack two GLushorts into one 32-bit float */
- {
- GLfloat a[4];
- GLuint result[4], usx, usy;
- 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);
- usy = IROUND(a[1] * 65535.0F);
- result[0] =
- result[1] =
- result[2] =
- result[3] = usx | (usy << 16);
- store_vector4ui(inst, machine, result);
- }
- break;
- case OPCODE_PK4B: /* pack four GLbytes into one 32-bit float */
- {
- GLfloat a[4];
- GLuint result[4], ubx, uby, ubz, ubw;
- 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);
- a[3] = CLAMP(a[3], -128.0F / 127.0F, 1.0F);
- ubx = IROUND(127.0F * a[0] + 128.0F);
- uby = IROUND(127.0F * a[1] + 128.0F);
- ubz = IROUND(127.0F * a[2] + 128.0F);
- ubw = IROUND(127.0F * a[3] + 128.0F);
- result[0] =
- result[1] =
- result[2] =
- result[3] = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
- store_vector4ui(inst, machine, result);
- }
- break;
- case OPCODE_PK4UB: /* pack four GLubytes into one 32-bit float */
- {
- GLfloat a[4];
- GLuint result[4], ubx, uby, ubz, ubw;
- 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);
- a[3] = CLAMP(a[3], 0.0F, 1.0F);
- ubx = IROUND(255.0F * a[0]);
- uby = IROUND(255.0F * a[1]);
- ubz = IROUND(255.0F * a[2]);
- ubw = IROUND(255.0F * a[3]);
- result[0] =
- result[1] =
- result[2] =
- result[3] = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
- store_vector4ui(inst, machine, result);
- }
- break;
case OPCODE_POW:
{
GLfloat a[4], b[4], result[4];
fetch_vector1(&inst->SrcReg[0], machine, a);
fetch_vector1(&inst->SrcReg[1], machine, b);
result[0] = result[1] = result[2] = result[3]
- = (GLfloat) pow(a[0], b[0]);
- store_vector4(inst, machine, result);
- }
- break;
- case OPCODE_RCC: /* clamped riciprocal */
- {
- const float largest = 1.884467e+19, smallest = 5.42101e-20;
- GLfloat a[4], r, result[4];
- fetch_vector1(&inst->SrcReg[0], machine, a);
- if (DEBUG_PROG) {
- if (a[0] == 0)
- printf("RCC(0)\n");
- else if (IS_INF_OR_NAN(a[0]))
- printf("RCC(inf)\n");
- }
- if (a[0] == 1.0F) {
- r = 1.0F;
- }
- else {
- r = 1.0F / a[0];
- }
- if (positive(r)) {
- if (r > largest) {
- r = largest;
- }
- else if (r < smallest) {
- r = smallest;
- }
- }
- else {
- if (r < -largest) {
- r = -largest;
- }
- else if (r > -smallest) {
- r = -smallest;
- }
- }
- result[0] = result[1] = result[2] = result[3] = r;
+ = powf(a[0], b[0]);
store_vector4(inst, machine, result);
}
break;
}
break;
case OPCODE_RET: /* return from subroutine (conditional) */
- if (eval_condition(machine, inst)) {
- if (machine->StackDepth == 0) {
- return GL_TRUE; /* Per GL_NV_vertex_program2 spec */
- }
- /* 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(&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];
- result[1] = tmpX * axis[1] - dir[1];
- result[2] = tmpX * axis[2] - dir[2];
- /* result[3] is never written! XXX enforce in parser! */
- store_vector4(inst, machine, result);
+ if (machine->StackDepth == 0) {
+ return GL_TRUE; /* Per GL_NV_vertex_program2 spec */
}
+ /* subtract one because of pc++ in the for loop */
+ pc = machine->CallStack[--machine->StackDepth] - 1;
break;
case OPCODE_RSQ: /* 1 / sqrt() */
{
GLfloat a[4], result[4];
fetch_vector1(&inst->SrcReg[0], machine, a);
- a[0] = FABSF(a[0]);
- result[0] = result[1] = result[2] = result[3] = INV_SQRTF(a[0]);
+ a[0] = fabsf(a[0]);
+ result[0] = result[1] = result[2] = result[3] = 1.0f / sqrtf(a[0]);
store_vector4(inst, machine, result);
if (DEBUG_PROG) {
printf("RSQ %g = 1/sqrt(|%g|)\n", result[0], a[0]);
{
GLfloat a[4], result[4];
fetch_vector1(&inst->SrcReg[0], machine, a);
- result[0] = (GLfloat) cos(a[0]);
- result[1] = (GLfloat) sin(a[0]);
- result[2] = 0.0; /* undefined! */
- result[3] = 0.0; /* undefined! */
+ result[0] = cosf(a[0]);
+ result[1] = sinf(a[0]);
+ result[2] = 0.0F; /* undefined! */
+ result[3] = 0.0F; /* undefined! */
store_vector4(inst, machine, result);
}
break;
}
}
break;
- 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 OPCODE_SGE: /* set on greater or equal */
{
GLfloat a[4], b[4], result[4];
GLfloat a[4], result[4];
fetch_vector1(&inst->SrcReg[0], machine, a);
result[0] = result[1] = result[2] = result[3]
- = (GLfloat) sin(a[0]);
+ = sinf(a[0]);
store_vector4(inst, machine, result);
}
break;
store_vector4(inst, machine, result);
}
break;
- 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 OPCODE_SUB:
{
GLfloat a[4], b[4], result[4];
else if (swz == SWIZZLE_ONE)
result[i] = 1.0;
else {
- ASSERT(swz >= 0);
- ASSERT(swz <= 3);
+ assert(swz <= 3);
result[i] = src[swz];
}
if (source->Negate & (1 << i))
GLfloat texcoord[4], color[4];
fetch_vector4(&inst->SrcReg[0], machine, texcoord);
+ /* For TEX, texcoord.Q should not be used and its value should not
+ * matter (at most, we pass coord.xyz to texture3D() in GLSL).
+ * Set Q=1 so that FetchTexelDeriv() doesn't get a garbage value
+ * which is effectively what happens when the texcoord swizzle
+ * is .xyzz
+ */
+ texcoord[3] = 1.0f;
+
fetch_texel(ctx, machine, inst, texcoord, 0.0, color);
if (DEBUG_PROG) {
fetch_texel(ctx, machine, inst, texcoord, lodBias, color);
+ if (DEBUG_PROG) {
+ printf("TXB (%g, %g, %g, %g) = texture[%d][%g %g %g %g]"
+ " bias %g\n",
+ color[0], color[1], color[2], color[3],
+ inst->TexSrcUnit,
+ texcoord[0],
+ texcoord[1],
+ texcoord[2],
+ texcoord[3],
+ lodBias);
+ }
+
store_vector4(inst, machine, color);
}
break;
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
+ * 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) {
+ if (texcoord[3] != 0.0F) {
texcoord[0] /= texcoord[3];
texcoord[1] /= texcoord[3];
texcoord[2] /= texcoord[3];
fetch_vector4(&inst->SrcReg[0], machine, texcoord);
if (inst->TexSrcTarget != TEXTURE_CUBE_INDEX &&
- texcoord[3] != 0.0) {
+ texcoord[3] != 0.0F) {
texcoord[0] /= texcoord[3];
texcoord[1] /= texcoord[3];
texcoord[2] /= texcoord[3];
store_vector4(inst, machine, result);
}
break;
- case OPCODE_UP2H: /* unpack two 16-bit floats */
- {
- const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine);
- GLfloat result[4];
- GLushort hx, hy;
- hx = raw & 0xffff;
- hy = raw >> 16;
- result[0] = result[2] = _mesa_half_to_float(hx);
- result[1] = result[3] = _mesa_half_to_float(hy);
- store_vector4(inst, machine, result);
- }
- break;
- case OPCODE_UP2US: /* unpack two GLushorts */
- {
- const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine);
- GLfloat result[4];
- GLushort usx, usy;
- usx = raw & 0xffff;
- usy = raw >> 16;
- result[0] = result[2] = usx * (1.0f / 65535.0f);
- result[1] = result[3] = usy * (1.0f / 65535.0f);
- store_vector4(inst, machine, result);
- }
- break;
- case OPCODE_UP4B: /* unpack four GLbytes */
- {
- const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine);
- GLfloat result[4];
- result[0] = (((raw >> 0) & 0xff) - 128) / 127.0F;
- result[1] = (((raw >> 8) & 0xff) - 128) / 127.0F;
- result[2] = (((raw >> 16) & 0xff) - 128) / 127.0F;
- result[3] = (((raw >> 24) & 0xff) - 128) / 127.0F;
- store_vector4(inst, machine, result);
- }
- break;
- case OPCODE_UP4UB: /* unpack four GLubytes */
- {
- const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine);
- GLfloat result[4];
- result[0] = ((raw >> 0) & 0xff) / 255.0F;
- result[1] = ((raw >> 8) & 0xff) / 255.0F;
- result[2] = ((raw >> 16) & 0xff) / 255.0F;
- result[3] = ((raw >> 24) & 0xff) / 255.0F;
- store_vector4(inst, machine, result);
- }
- break;
- case OPCODE_XOR: /* bitwise XOR */
- {
- GLuint a[4], b[4], result[4];
- fetch_vector4ui(&inst->SrcReg[0], machine, a);
- fetch_vector4ui(&inst->SrcReg[1], machine, 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_vector4ui(inst, machine, result);
- }
- break;
case OPCODE_XPD: /* cross product */
{
GLfloat a[4], b[4], result[4];
}
}
break;
- case 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);
- 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];
- result[3] = a[3] + b[0] * c[2] + b[1] * c[3];
- store_vector4(inst, machine, result);
- }
- break;
- case OPCODE_PRINT:
- {
- if (inst->SrcReg[0].File != PROGRAM_UNDEFINED) {
- GLfloat a[4];
- fetch_vector4(&inst->SrcReg[0], machine, a);
- printf("%s%g, %g, %g, %g\n", (const char *) inst->Data,
- a[0], a[1], a[2], a[3]);
- }
- else {
- printf("%s\n", (const char *) inst->Data);
- }
- }
- break;
case OPCODE_END:
return GL_TRUE;
default:
projects / mesa.git / blobdiff