projects / mesa.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Replace gl_frag_attrib enum with gl_varying_slot.
[mesa.git] / src / mesa / program / prog_execute.c
1 /*
2 * Mesa 3-D graphics library
3 * Version: 7.3
4 *
5 * Copyright (C) 1999-2008 Brian Paul All Rights Reserved.
6 *
7 * Permission is hereby granted, free of charge, to any person obtaining a
8 * copy of this software and associated documentation files (the "Software"),
9 * to deal in the Software without restriction, including without limitation
10 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
11 * and/or sell copies of the Software, and to permit persons to whom the
12 * Software is furnished to do so, subject to the following conditions:
13 *
14 * The above copyright notice and this permission notice shall be included
15 * in all copies or substantial portions of the Software.
16 *
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
18 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
19 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
20 * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
21 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
22 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
23 */
24
25 /**
26 * \file prog_execute.c
27 * Software interpreter for vertex/fragment programs.
28 * \author Brian Paul
29 */
30
31 /*
32 * NOTE: we do everything in single-precision floating point; we don't
33 * currently observe the single/half/fixed-precision qualifiers.
34 *
35 */
36
37
38 #include "main/glheader.h"
39 #include "main/colormac.h"
40 #include "main/macros.h"
41 #include "prog_execute.h"
42 #include "prog_instruction.h"
43 #include "prog_parameter.h"
44 #include "prog_print.h"
45 #include "prog_noise.h"
46
47
48 /* debug predicate */
49 #define DEBUG_PROG 0
50
51
52 /**
53 * Set x to positive or negative infinity.
54 */
55 #if defined(USE_IEEE) || defined(_WIN32)
56 #define SET_POS_INFINITY(x) \
57 do { \
58 fi_type fi; \
59 fi.i = 0x7F800000; \
60 x = fi.f; \
61 } while (0)
62 #define SET_NEG_INFINITY(x) \
63 do { \
64 fi_type fi; \
65 fi.i = 0xFF800000; \
66 x = fi.f; \
67 } while (0)
68 #else
69 #define SET_POS_INFINITY(x) x = (GLfloat) HUGE_VAL
70 #define SET_NEG_INFINITY(x) x = (GLfloat) -HUGE_VAL
71 #endif
72
73 #define SET_FLOAT_BITS(x, bits) ((fi_type *) (void *) &(x))->i = bits
74
75
76 static const GLfloat ZeroVec[4] = { 0.0F, 0.0F, 0.0F, 0.0F };
77
78
79 /**
80 * Return a pointer to the 4-element float vector specified by the given
81 * source register.
82 */
83 static inline const GLfloat *
84 get_src_register_pointer(const struct prog_src_register *source,
85 const struct gl_program_machine *machine)
86 {
87 const struct gl_program *prog = machine->CurProgram;
88 GLint reg = source->Index;
89
90 if (source->RelAddr) {
91 /* add address register value to src index/offset */
92 reg += machine->AddressReg[0][0];
93 if (reg < 0) {
94 return ZeroVec;
95 }
96 }
97
98 switch (source->File) {
99 case PROGRAM_TEMPORARY:
100 if (reg >= MAX_PROGRAM_TEMPS)
101 return ZeroVec;
102 return machine->Temporaries[reg];
103
104 case PROGRAM_INPUT:
105 if (prog->Target == GL_VERTEX_PROGRAM_ARB) {
106 if (reg >= VERT_ATTRIB_MAX)
107 return ZeroVec;
108 return machine->VertAttribs[reg];
109 }
110 else {
111 if (reg >= VARYING_SLOT_MAX)
112 return ZeroVec;
113 return machine->Attribs[reg][machine->CurElement];
114 }
115
116 case PROGRAM_OUTPUT:
117 if (reg >= MAX_PROGRAM_OUTPUTS)
118 return ZeroVec;
119 return machine->Outputs[reg];
120
121 case PROGRAM_LOCAL_PARAM:
122 if (reg >= MAX_PROGRAM_LOCAL_PARAMS)
123 return ZeroVec;
124 return machine->CurProgram->LocalParams[reg];
125
126 case PROGRAM_ENV_PARAM:
127 if (reg >= MAX_PROGRAM_ENV_PARAMS)
128 return ZeroVec;
129 return machine->EnvParams[reg];
130
131 case PROGRAM_STATE_VAR:
132 /* Fallthrough */
133 case PROGRAM_CONSTANT:
134 /* Fallthrough */
135 case PROGRAM_UNIFORM:
136 if (reg >= (GLint) prog->Parameters->NumParameters)
137 return ZeroVec;
138 return (GLfloat *) prog->Parameters->ParameterValues[reg];
139
140 case PROGRAM_SYSTEM_VALUE:
141 assert(reg < Elements(machine->SystemValues));
142 return machine->SystemValues[reg];
143
144 default:
145 _mesa_problem(NULL,
146 "Invalid src register file %d in get_src_register_pointer()",
147 source->File);
148 return NULL;
149 }
150 }
151
152
153 /**
154 * Return a pointer to the 4-element float vector specified by the given
155 * destination register.
156 */
157 static inline GLfloat *
158 get_dst_register_pointer(const struct prog_dst_register *dest,
159 struct gl_program_machine *machine)
160 {
161 static GLfloat dummyReg[4];
162 GLint reg = dest->Index;
163
164 if (dest->RelAddr) {
165 /* add address register value to src index/offset */
166 reg += machine->AddressReg[0][0];
167 if (reg < 0) {
168 return dummyReg;
169 }
170 }
171
172 switch (dest->File) {
173 case PROGRAM_TEMPORARY:
174 if (reg >= MAX_PROGRAM_TEMPS)
175 return dummyReg;
176 return machine->Temporaries[reg];
177
178 case PROGRAM_OUTPUT:
179 if (reg >= MAX_PROGRAM_OUTPUTS)
180 return dummyReg;
181 return machine->Outputs[reg];
182
183 default:
184 _mesa_problem(NULL,
185 "Invalid dest register file %d in get_dst_register_pointer()",
186 dest->File);
187 return NULL;
188 }
189 }
190
191
192
193 /**
194 * Fetch a 4-element float vector from the given source register.
195 * Apply swizzling and negating as needed.
196 */
197 static void
198 fetch_vector4(const struct prog_src_register *source,
199 const struct gl_program_machine *machine, GLfloat result[4])
200 {
201 const GLfloat *src = get_src_register_pointer(source, machine);
202 ASSERT(src);
203
204 if (source->Swizzle == SWIZZLE_NOOP) {
205 /* no swizzling */
206 COPY_4V(result, src);
207 }
208 else {
209 ASSERT(GET_SWZ(source->Swizzle, 0) <= 3);
210 ASSERT(GET_SWZ(source->Swizzle, 1) <= 3);
211 ASSERT(GET_SWZ(source->Swizzle, 2) <= 3);
212 ASSERT(GET_SWZ(source->Swizzle, 3) <= 3);
213 result[0] = src[GET_SWZ(source->Swizzle, 0)];
214 result[1] = src[GET_SWZ(source->Swizzle, 1)];
215 result[2] = src[GET_SWZ(source->Swizzle, 2)];
216 result[3] = src[GET_SWZ(source->Swizzle, 3)];
217 }
218
219 if (source->Abs) {
220 result[0] = FABSF(result[0]);
221 result[1] = FABSF(result[1]);
222 result[2] = FABSF(result[2]);
223 result[3] = FABSF(result[3]);
224 }
225 if (source->Negate) {
226 ASSERT(source->Negate == NEGATE_XYZW);
227 result[0] = -result[0];
228 result[1] = -result[1];
229 result[2] = -result[2];
230 result[3] = -result[3];
231 }
232
233 #ifdef NAN_CHECK
234 assert(!IS_INF_OR_NAN(result[0]));
235 assert(!IS_INF_OR_NAN(result[0]));
236 assert(!IS_INF_OR_NAN(result[0]));
237 assert(!IS_INF_OR_NAN(result[0]));
238 #endif
239 }
240
241
242 /**
243 * Fetch a 4-element uint vector from the given source register.
244 * Apply swizzling but not negation/abs.
245 */
246 static void
247 fetch_vector4ui(const struct prog_src_register *source,
248 const struct gl_program_machine *machine, GLuint result[4])
249 {
250 const GLuint *src = (GLuint *) get_src_register_pointer(source, machine);
251 ASSERT(src);
252
253 if (source->Swizzle == SWIZZLE_NOOP) {
254 /* no swizzling */
255 COPY_4V(result, src);
256 }
257 else {
258 ASSERT(GET_SWZ(source->Swizzle, 0) <= 3);
259 ASSERT(GET_SWZ(source->Swizzle, 1) <= 3);
260 ASSERT(GET_SWZ(source->Swizzle, 2) <= 3);
261 ASSERT(GET_SWZ(source->Swizzle, 3) <= 3);
262 result[0] = src[GET_SWZ(source->Swizzle, 0)];
263 result[1] = src[GET_SWZ(source->Swizzle, 1)];
264 result[2] = src[GET_SWZ(source->Swizzle, 2)];
265 result[3] = src[GET_SWZ(source->Swizzle, 3)];
266 }
267
268 /* Note: no Negate or Abs here */
269 }
270
271
272
273 /**
274 * Fetch the derivative with respect to X or Y for the given register.
275 * XXX this currently only works for fragment program input attribs.
276 */
277 static void
278 fetch_vector4_deriv(struct gl_context * ctx,
279 const struct prog_src_register *source,
280 const struct gl_program_machine *machine,
281 char xOrY, GLfloat result[4])
282 {
283 if (source->File == PROGRAM_INPUT &&
284 source->Index < (GLint) machine->NumDeriv) {
285 const GLint col = machine->CurElement;
286 const GLfloat w = machine->Attribs[VARYING_SLOT_POS][col][3];
287 const GLfloat invQ = 1.0f / w;
288 GLfloat deriv[4];
289
290 if (xOrY == 'X') {
291 deriv[0] = machine->DerivX[source->Index][0] * invQ;
292 deriv[1] = machine->DerivX[source->Index][1] * invQ;
293 deriv[2] = machine->DerivX[source->Index][2] * invQ;
294 deriv[3] = machine->DerivX[source->Index][3] * invQ;
295 }
296 else {
297 deriv[0] = machine->DerivY[source->Index][0] * invQ;
298 deriv[1] = machine->DerivY[source->Index][1] * invQ;
299 deriv[2] = machine->DerivY[source->Index][2] * invQ;
300 deriv[3] = machine->DerivY[source->Index][3] * invQ;
301 }
302
303 result[0] = deriv[GET_SWZ(source->Swizzle, 0)];
304 result[1] = deriv[GET_SWZ(source->Swizzle, 1)];
305 result[2] = deriv[GET_SWZ(source->Swizzle, 2)];
306 result[3] = deriv[GET_SWZ(source->Swizzle, 3)];
307
308 if (source->Abs) {
309 result[0] = FABSF(result[0]);
310 result[1] = FABSF(result[1]);
311 result[2] = FABSF(result[2]);
312 result[3] = FABSF(result[3]);
313 }
314 if (source->Negate) {
315 ASSERT(source->Negate == NEGATE_XYZW);
316 result[0] = -result[0];
317 result[1] = -result[1];
318 result[2] = -result[2];
319 result[3] = -result[3];
320 }
321 }
322 else {
323 ASSIGN_4V(result, 0.0, 0.0, 0.0, 0.0);
324 }
325 }
326
327
328 /**
329 * As above, but only return result[0] element.
330 */
331 static void
332 fetch_vector1(const struct prog_src_register *source,
333 const struct gl_program_machine *machine, GLfloat result[4])
334 {
335 const GLfloat *src = get_src_register_pointer(source, machine);
336 ASSERT(src);
337
338 result[0] = src[GET_SWZ(source->Swizzle, 0)];
339
340 if (source->Abs) {
341 result[0] = FABSF(result[0]);
342 }
343 if (source->Negate) {
344 result[0] = -result[0];
345 }
346 }
347
348
349 static GLuint
350 fetch_vector1ui(const struct prog_src_register *source,
351 const struct gl_program_machine *machine)
352 {
353 const GLuint *src = (GLuint *) get_src_register_pointer(source, machine);
354 return src[GET_SWZ(source->Swizzle, 0)];
355 }
356
357
358 /**
359 * Fetch texel from texture. Use partial derivatives when possible.
360 */
361 static inline void
362 fetch_texel(struct gl_context *ctx,
363 const struct gl_program_machine *machine,
364 const struct prog_instruction *inst,
365 const GLfloat texcoord[4], GLfloat lodBias,
366 GLfloat color[4])
367 {
368 const GLuint unit = machine->Samplers[inst->TexSrcUnit];
369
370 /* Note: we only have the right derivatives for fragment input attribs.
371 */
372 if (machine->NumDeriv > 0 &&
373 inst->SrcReg[0].File == PROGRAM_INPUT &&
374 inst->SrcReg[0].Index == VARYING_SLOT_TEX0 + inst->TexSrcUnit) {
375 /* simple texture fetch for which we should have derivatives */
376 GLuint attr = inst->SrcReg[0].Index;
377 machine->FetchTexelDeriv(ctx, texcoord,
378 machine->DerivX[attr],
379 machine->DerivY[attr],
380 lodBias, unit, color);
381 }
382 else {
383 machine->FetchTexelLod(ctx, texcoord, lodBias, unit, color);
384 }
385 }
386
387
388 /**
389 * Test value against zero and return GT, LT, EQ or UN if NaN.
390 */
391 static inline GLuint
392 generate_cc(float value)
393 {
394 if (value != value)
395 return COND_UN; /* NaN */
396 if (value > 0.0F)
397 return COND_GT;
398 if (value < 0.0F)
399 return COND_LT;
400 return COND_EQ;
401 }
402
403
404 /**
405 * Test if the ccMaskRule is satisfied by the given condition code.
406 * Used to mask destination writes according to the current condition code.
407 */
408 static inline GLboolean
409 test_cc(GLuint condCode, GLuint ccMaskRule)
410 {
411 switch (ccMaskRule) {
412 case COND_EQ: return (condCode == COND_EQ);
413 case COND_NE: return (condCode != COND_EQ);
414 case COND_LT: return (condCode == COND_LT);
415 case COND_GE: return (condCode == COND_GT || condCode == COND_EQ);
416 case COND_LE: return (condCode == COND_LT || condCode == COND_EQ);
417 case COND_GT: return (condCode == COND_GT);
418 case COND_TR: return GL_TRUE;
419 case COND_FL: return GL_FALSE;
420 default: return GL_TRUE;
421 }
422 }
423
424
425 /**
426 * Evaluate the 4 condition codes against a predicate and return GL_TRUE
427 * or GL_FALSE to indicate result.
428 */
429 static inline GLboolean
430 eval_condition(const struct gl_program_machine *machine,
431 const struct prog_instruction *inst)
432 {
433 const GLuint swizzle = inst->DstReg.CondSwizzle;
434 const GLuint condMask = inst->DstReg.CondMask;
435 if (test_cc(machine->CondCodes[GET_SWZ(swizzle, 0)], condMask) ||
436 test_cc(machine->CondCodes[GET_SWZ(swizzle, 1)], condMask) ||
437 test_cc(machine->CondCodes[GET_SWZ(swizzle, 2)], condMask) ||
438 test_cc(machine->CondCodes[GET_SWZ(swizzle, 3)], condMask)) {
439 return GL_TRUE;
440 }
441 else {
442 return GL_FALSE;
443 }
444 }
445
446
447
448 /**
449 * Store 4 floats into a register. Observe the instructions saturate and
450 * set-condition-code flags.
451 */
452 static void
453 store_vector4(const struct prog_instruction *inst,
454 struct gl_program_machine *machine, const GLfloat value[4])
455 {
456 const struct prog_dst_register *dstReg = &(inst->DstReg);
457 const GLboolean clamp = inst->SaturateMode == SATURATE_ZERO_ONE;
458 GLuint writeMask = dstReg->WriteMask;
459 GLfloat clampedValue[4];
460 GLfloat *dst = get_dst_register_pointer(dstReg, machine);
461
462 #if 0
463 if (value[0] > 1.0e10 ||
464 IS_INF_OR_NAN(value[0]) ||
465 IS_INF_OR_NAN(value[1]) ||
466 IS_INF_OR_NAN(value[2]) || IS_INF_OR_NAN(value[3]))
467 printf("store %g %g %g %g\n", value[0], value[1], value[2], value[3]);
468 #endif
469
470 if (clamp) {
471 clampedValue[0] = CLAMP(value[0], 0.0F, 1.0F);
472 clampedValue[1] = CLAMP(value[1], 0.0F, 1.0F);
473 clampedValue[2] = CLAMP(value[2], 0.0F, 1.0F);
474 clampedValue[3] = CLAMP(value[3], 0.0F, 1.0F);
475 value = clampedValue;
476 }
477
478 if (dstReg->CondMask != COND_TR) {
479 /* condition codes may turn off some writes */
480 if (writeMask & WRITEMASK_X) {
481 if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 0)],
482 dstReg->CondMask))
483 writeMask &= ~WRITEMASK_X;
484 }
485 if (writeMask & WRITEMASK_Y) {
486 if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 1)],
487 dstReg->CondMask))
488 writeMask &= ~WRITEMASK_Y;
489 }
490 if (writeMask & WRITEMASK_Z) {
491 if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 2)],
492 dstReg->CondMask))
493 writeMask &= ~WRITEMASK_Z;
494 }
495 if (writeMask & WRITEMASK_W) {
496 if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 3)],
497 dstReg->CondMask))
498 writeMask &= ~WRITEMASK_W;
499 }
500 }
501
502 #ifdef NAN_CHECK
503 assert(!IS_INF_OR_NAN(value[0]));
504 assert(!IS_INF_OR_NAN(value[0]));
505 assert(!IS_INF_OR_NAN(value[0]));
506 assert(!IS_INF_OR_NAN(value[0]));
507 #endif
508
509 if (writeMask & WRITEMASK_X)
510 dst[0] = value[0];
511 if (writeMask & WRITEMASK_Y)
512 dst[1] = value[1];
513 if (writeMask & WRITEMASK_Z)
514 dst[2] = value[2];
515 if (writeMask & WRITEMASK_W)
516 dst[3] = value[3];
517
518 if (inst->CondUpdate) {
519 if (writeMask & WRITEMASK_X)
520 machine->CondCodes[0] = generate_cc(value[0]);
521 if (writeMask & WRITEMASK_Y)
522 machine->CondCodes[1] = generate_cc(value[1]);
523 if (writeMask & WRITEMASK_Z)
524 machine->CondCodes[2] = generate_cc(value[2]);
525 if (writeMask & WRITEMASK_W)
526 machine->CondCodes[3] = generate_cc(value[3]);
527 #if DEBUG_PROG
528 printf("CondCodes=(%s,%s,%s,%s) for:\n",
529 _mesa_condcode_string(machine->CondCodes[0]),
530 _mesa_condcode_string(machine->CondCodes[1]),
531 _mesa_condcode_string(machine->CondCodes[2]),
532 _mesa_condcode_string(machine->CondCodes[3]));
533 #endif
534 }
535 }
536
537
538 /**
539 * Store 4 uints into a register. Observe the set-condition-code flags.
540 */
541 static void
542 store_vector4ui(const struct prog_instruction *inst,
543 struct gl_program_machine *machine, const GLuint value[4])
544 {
545 const struct prog_dst_register *dstReg = &(inst->DstReg);
546 GLuint writeMask = dstReg->WriteMask;
547 GLuint *dst = (GLuint *) get_dst_register_pointer(dstReg, machine);
548
549 if (dstReg->CondMask != COND_TR) {
550 /* condition codes may turn off some writes */
551 if (writeMask & WRITEMASK_X) {
552 if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 0)],
553 dstReg->CondMask))
554 writeMask &= ~WRITEMASK_X;
555 }
556 if (writeMask & WRITEMASK_Y) {
557 if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 1)],
558 dstReg->CondMask))
559 writeMask &= ~WRITEMASK_Y;
560 }
561 if (writeMask & WRITEMASK_Z) {
562 if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 2)],
563 dstReg->CondMask))
564 writeMask &= ~WRITEMASK_Z;
565 }
566 if (writeMask & WRITEMASK_W) {
567 if (!test_cc(machine->CondCodes[GET_SWZ(dstReg->CondSwizzle, 3)],
568 dstReg->CondMask))
569 writeMask &= ~WRITEMASK_W;
570 }
571 }
572
573 if (writeMask & WRITEMASK_X)
574 dst[0] = value[0];
575 if (writeMask & WRITEMASK_Y)
576 dst[1] = value[1];
577 if (writeMask & WRITEMASK_Z)
578 dst[2] = value[2];
579 if (writeMask & WRITEMASK_W)
580 dst[3] = value[3];
581
582 if (inst->CondUpdate) {
583 if (writeMask & WRITEMASK_X)
584 machine->CondCodes[0] = generate_cc((float)value[0]);
585 if (writeMask & WRITEMASK_Y)
586 machine->CondCodes[1] = generate_cc((float)value[1]);
587 if (writeMask & WRITEMASK_Z)
588 machine->CondCodes[2] = generate_cc((float)value[2]);
589 if (writeMask & WRITEMASK_W)
590 machine->CondCodes[3] = generate_cc((float)value[3]);
591 #if DEBUG_PROG
592 printf("CondCodes=(%s,%s,%s,%s) for:\n",
593 _mesa_condcode_string(machine->CondCodes[0]),
594 _mesa_condcode_string(machine->CondCodes[1]),
595 _mesa_condcode_string(machine->CondCodes[2]),
596 _mesa_condcode_string(machine->CondCodes[3]));
597 #endif
598 }
599 }
600
601
602
603 /**
604 * Execute the given vertex/fragment program.
605 *
606 * \param ctx rendering context
607 * \param program the program to execute
608 * \param machine machine state (must be initialized)
609 * \return GL_TRUE if program completed or GL_FALSE if program executed KIL.
610 */
611 GLboolean
612 _mesa_execute_program(struct gl_context * ctx,
613 const struct gl_program *program,
614 struct gl_program_machine *machine)
615 {
616 const GLuint numInst = program->NumInstructions;
617 const GLuint maxExec = 65536;
618 GLuint pc, numExec = 0;
619
620 machine->CurProgram = program;
621
622 if (DEBUG_PROG) {
623 printf("execute program %u --------------------\n", program->Id);
624 }
625
626 if (program->Target == GL_VERTEX_PROGRAM_ARB) {
627 machine->EnvParams = ctx->VertexProgram.Parameters;
628 }
629 else {
630 machine->EnvParams = ctx->FragmentProgram.Parameters;
631 }
632
633 for (pc = 0; pc < numInst; pc++) {
634 const struct prog_instruction *inst = program->Instructions + pc;
635
636 if (DEBUG_PROG) {
637 _mesa_print_instruction(inst);
638 }
639
640 switch (inst->Opcode) {
641 case OPCODE_ABS:
642 {
643 GLfloat a[4], result[4];
644 fetch_vector4(&inst->SrcReg[0], machine, a);
645 result[0] = FABSF(a[0]);
646 result[1] = FABSF(a[1]);
647 result[2] = FABSF(a[2]);
648 result[3] = FABSF(a[3]);
649 store_vector4(inst, machine, result);
650 }
651 break;
652 case OPCODE_ADD:
653 {
654 GLfloat a[4], b[4], result[4];
655 fetch_vector4(&inst->SrcReg[0], machine, a);
656 fetch_vector4(&inst->SrcReg[1], machine, b);
657 result[0] = a[0] + b[0];
658 result[1] = a[1] + b[1];
659 result[2] = a[2] + b[2];
660 result[3] = a[3] + b[3];
661 store_vector4(inst, machine, result);
662 if (DEBUG_PROG) {
663 printf("ADD (%g %g %g %g) = (%g %g %g %g) + (%g %g %g %g)\n",
664 result[0], result[1], result[2], result[3],
665 a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]);
666 }
667 }
668 break;
669 case OPCODE_AND: /* bitwise AND */
670 {
671 GLuint a[4], b[4], result[4];
672 fetch_vector4ui(&inst->SrcReg[0], machine, a);
673 fetch_vector4ui(&inst->SrcReg[1], machine, b);
674 result[0] = a[0] & b[0];
675 result[1] = a[1] & b[1];
676 result[2] = a[2] & b[2];
677 result[3] = a[3] & b[3];
678 store_vector4ui(inst, machine, result);
679 }
680 break;
681 case OPCODE_ARL:
682 {
683 GLfloat t[4];
684 fetch_vector4(&inst->SrcReg[0], machine, t);
685 machine->AddressReg[0][0] = IFLOOR(t[0]);
686 if (DEBUG_PROG) {
687 printf("ARL %d\n", machine->AddressReg[0][0]);
688 }
689 }
690 break;
691 case OPCODE_BGNLOOP:
692 /* no-op */
693 ASSERT(program->Instructions[inst->BranchTarget].Opcode
694 == OPCODE_ENDLOOP);
695 break;
696 case OPCODE_ENDLOOP:
697 /* subtract 1 here since pc is incremented by for(pc) loop */
698 ASSERT(program->Instructions[inst->BranchTarget].Opcode
699 == OPCODE_BGNLOOP);
700 pc = inst->BranchTarget - 1; /* go to matching BNGLOOP */
701 break;
702 case OPCODE_BGNSUB: /* begin subroutine */
703 break;
704 case OPCODE_ENDSUB: /* end subroutine */
705 break;
706 case OPCODE_BRK: /* break out of loop (conditional) */
707 ASSERT(program->Instructions[inst->BranchTarget].Opcode
708 == OPCODE_ENDLOOP);
709 if (eval_condition(machine, inst)) {
710 /* break out of loop */
711 /* pc++ at end of for-loop will put us after the ENDLOOP inst */
712 pc = inst->BranchTarget;
713 }
714 break;
715 case OPCODE_CONT: /* continue loop (conditional) */
716 ASSERT(program->Instructions[inst->BranchTarget].Opcode
717 == OPCODE_ENDLOOP);
718 if (eval_condition(machine, inst)) {
719 /* continue at ENDLOOP */
720 /* Subtract 1 here since we'll do pc++ at end of for-loop */
721 pc = inst->BranchTarget - 1;
722 }
723 break;
724 case OPCODE_CAL: /* Call subroutine (conditional) */
725 if (eval_condition(machine, inst)) {
726 /* call the subroutine */
727 if (machine->StackDepth >= MAX_PROGRAM_CALL_DEPTH) {
728 return GL_TRUE; /* Per GL_NV_vertex_program2 spec */
729 }
730 machine->CallStack[machine->StackDepth++] = pc + 1; /* next inst */
731 /* Subtract 1 here since we'll do pc++ at end of for-loop */
732 pc = inst->BranchTarget - 1;
733 }
734 break;
735 case OPCODE_CMP:
736 {
737 GLfloat a[4], b[4], c[4], result[4];
738 fetch_vector4(&inst->SrcReg[0], machine, a);
739 fetch_vector4(&inst->SrcReg[1], machine, b);
740 fetch_vector4(&inst->SrcReg[2], machine, c);
741 result[0] = a[0] < 0.0F ? b[0] : c[0];
742 result[1] = a[1] < 0.0F ? b[1] : c[1];
743 result[2] = a[2] < 0.0F ? b[2] : c[2];
744 result[3] = a[3] < 0.0F ? b[3] : c[3];
745 store_vector4(inst, machine, result);
746 if (DEBUG_PROG) {
747 printf("CMP (%g %g %g %g) = (%g %g %g %g) < 0 ? (%g %g %g %g) : (%g %g %g %g)\n",
748 result[0], result[1], result[2], result[3],
749 a[0], a[1], a[2], a[3],
750 b[0], b[1], b[2], b[3],
751 c[0], c[1], c[2], c[3]);
752 }
753 }
754 break;
755 case OPCODE_COS:
756 {
757 GLfloat a[4], result[4];
758 fetch_vector1(&inst->SrcReg[0], machine, a);
759 result[0] = result[1] = result[2] = result[3]
760 = (GLfloat) cos(a[0]);
761 store_vector4(inst, machine, result);
762 }
763 break;
764 case OPCODE_DDX: /* Partial derivative with respect to X */
765 {
766 GLfloat result[4];
767 fetch_vector4_deriv(ctx, &inst->SrcReg[0], machine,
768 'X', result);
769 store_vector4(inst, machine, result);
770 }
771 break;
772 case OPCODE_DDY: /* Partial derivative with respect to Y */
773 {
774 GLfloat result[4];
775 fetch_vector4_deriv(ctx, &inst->SrcReg[0], machine,
776 'Y', result);
777 store_vector4(inst, machine, result);
778 }
779 break;
780 case OPCODE_DP2:
781 {
782 GLfloat a[4], b[4], result[4];
783 fetch_vector4(&inst->SrcReg[0], machine, a);
784 fetch_vector4(&inst->SrcReg[1], machine, b);
785 result[0] = result[1] = result[2] = result[3] = DOT2(a, b);
786 store_vector4(inst, machine, result);
787 if (DEBUG_PROG) {
788 printf("DP2 %g = (%g %g) . (%g %g)\n",
789 result[0], a[0], a[1], b[0], b[1]);
790 }
791 }
792 break;
793 case OPCODE_DP2A:
794 {
795 GLfloat a[4], b[4], c, result[4];
796 fetch_vector4(&inst->SrcReg[0], machine, a);
797 fetch_vector4(&inst->SrcReg[1], machine, b);
798 fetch_vector1(&inst->SrcReg[1], machine, &c);
799 result[0] = result[1] = result[2] = result[3] = DOT2(a, b) + c;
800 store_vector4(inst, machine, result);
801 if (DEBUG_PROG) {
802 printf("DP2A %g = (%g %g) . (%g %g) + %g\n",
803 result[0], a[0], a[1], b[0], b[1], c);
804 }
805 }
806 break;
807 case OPCODE_DP3:
808 {
809 GLfloat a[4], b[4], result[4];
810 fetch_vector4(&inst->SrcReg[0], machine, a);
811 fetch_vector4(&inst->SrcReg[1], machine, b);
812 result[0] = result[1] = result[2] = result[3] = DOT3(a, b);
813 store_vector4(inst, machine, result);
814 if (DEBUG_PROG) {
815 printf("DP3 %g = (%g %g %g) . (%g %g %g)\n",
816 result[0], a[0], a[1], a[2], b[0], b[1], b[2]);
817 }
818 }
819 break;
820 case OPCODE_DP4:
821 {
822 GLfloat a[4], b[4], result[4];
823 fetch_vector4(&inst->SrcReg[0], machine, a);
824 fetch_vector4(&inst->SrcReg[1], machine, b);
825 result[0] = result[1] = result[2] = result[3] = DOT4(a, b);
826 store_vector4(inst, machine, result);
827 if (DEBUG_PROG) {
828 printf("DP4 %g = (%g, %g %g %g) . (%g, %g %g %g)\n",
829 result[0], a[0], a[1], a[2], a[3],
830 b[0], b[1], b[2], b[3]);
831 }
832 }
833 break;
834 case OPCODE_DPH:
835 {
836 GLfloat a[4], b[4], result[4];
837 fetch_vector4(&inst->SrcReg[0], machine, a);
838 fetch_vector4(&inst->SrcReg[1], machine, b);
839 result[0] = result[1] = result[2] = result[3] = DOT3(a, b) + b[3];
840 store_vector4(inst, machine, result);
841 }
842 break;
843 case OPCODE_DST: /* Distance vector */
844 {
845 GLfloat a[4], b[4], result[4];
846 fetch_vector4(&inst->SrcReg[0], machine, a);
847 fetch_vector4(&inst->SrcReg[1], machine, b);
848 result[0] = 1.0F;
849 result[1] = a[1] * b[1];
850 result[2] = a[2];
851 result[3] = b[3];
852 store_vector4(inst, machine, result);
853 }
854 break;
855 case OPCODE_EXP:
856 {
857 GLfloat t[4], q[4], floor_t0;
858 fetch_vector1(&inst->SrcReg[0], machine, t);
859 floor_t0 = FLOORF(t[0]);
860 if (floor_t0 > FLT_MAX_EXP) {
861 SET_POS_INFINITY(q[0]);
862 SET_POS_INFINITY(q[2]);
863 }
864 else if (floor_t0 < FLT_MIN_EXP) {
865 q[0] = 0.0F;
866 q[2] = 0.0F;
867 }
868 else {
869 q[0] = LDEXPF(1.0, (int) floor_t0);
870 /* Note: GL_NV_vertex_program expects
871 * result.z = result.x * APPX(result.y)
872 * We do what the ARB extension says.
873 */
874 q[2] = (GLfloat) pow(2.0, t[0]);
875 }
876 q[1] = t[0] - floor_t0;
877 q[3] = 1.0F;
878 store_vector4( inst, machine, q );
879 }
880 break;
881 case OPCODE_EX2: /* Exponential base 2 */
882 {
883 GLfloat a[4], result[4], val;
884 fetch_vector1(&inst->SrcReg[0], machine, a);
885 val = (GLfloat) pow(2.0, a[0]);
886 /*
887 if (IS_INF_OR_NAN(val))
888 val = 1.0e10;
889 */
890 result[0] = result[1] = result[2] = result[3] = val;
891 store_vector4(inst, machine, result);
892 }
893 break;
894 case OPCODE_FLR:
895 {
896 GLfloat a[4], result[4];
897 fetch_vector4(&inst->SrcReg[0], machine, a);
898 result[0] = FLOORF(a[0]);
899 result[1] = FLOORF(a[1]);
900 result[2] = FLOORF(a[2]);
901 result[3] = FLOORF(a[3]);
902 store_vector4(inst, machine, result);
903 }
904 break;
905 case OPCODE_FRC:
906 {
907 GLfloat a[4], result[4];
908 fetch_vector4(&inst->SrcReg[0], machine, a);
909 result[0] = a[0] - FLOORF(a[0]);
910 result[1] = a[1] - FLOORF(a[1]);
911 result[2] = a[2] - FLOORF(a[2]);
912 result[3] = a[3] - FLOORF(a[3]);
913 store_vector4(inst, machine, result);
914 }
915 break;
916 case OPCODE_IF:
917 {
918 GLboolean cond;
919 ASSERT(program->Instructions[inst->BranchTarget].Opcode
920 == OPCODE_ELSE ||
921 program->Instructions[inst->BranchTarget].Opcode
922 == OPCODE_ENDIF);
923 /* eval condition */
924 if (inst->SrcReg[0].File != PROGRAM_UNDEFINED) {
925 GLfloat a[4];
926 fetch_vector1(&inst->SrcReg[0], machine, a);
927 cond = (a[0] != 0.0);
928 }
929 else {
930 cond = eval_condition(machine, inst);
931 }
932 if (DEBUG_PROG) {
933 printf("IF: %d\n", cond);
934 }
935 /* do if/else */
936 if (cond) {
937 /* do if-clause (just continue execution) */
938 }
939 else {
940 /* go to the instruction after ELSE or ENDIF */
941 assert(inst->BranchTarget >= 0);
942 pc = inst->BranchTarget;
943 }
944 }
945 break;
946 case OPCODE_ELSE:
947 /* goto ENDIF */
948 ASSERT(program->Instructions[inst->BranchTarget].Opcode
949 == OPCODE_ENDIF);
950 assert(inst->BranchTarget >= 0);
951 pc = inst->BranchTarget;
952 break;
953 case OPCODE_ENDIF:
954 /* nothing */
955 break;
956 case OPCODE_KIL_NV: /* NV_f_p only (conditional) */
957 if (eval_condition(machine, inst)) {
958 return GL_FALSE;
959 }
960 break;
961 case OPCODE_KIL: /* ARB_f_p only */
962 {
963 GLfloat a[4];
964 fetch_vector4(&inst->SrcReg[0], machine, a);
965 if (DEBUG_PROG) {
966 printf("KIL if (%g %g %g %g) <= 0.0\n",
967 a[0], a[1], a[2], a[3]);
968 }
969
970 if (a[0] < 0.0F || a[1] < 0.0F || a[2] < 0.0F || a[3] < 0.0F) {
971 return GL_FALSE;
972 }
973 }
974 break;
975 case OPCODE_LG2: /* log base 2 */
976 {
977 GLfloat a[4], result[4], val;
978 fetch_vector1(&inst->SrcReg[0], machine, a);
979 /* The fast LOG2 macro doesn't meet the precision requirements.
980 */
981 if (a[0] == 0.0F) {
982 val = -FLT_MAX;
983 }
984 else {
985 val = (float)(log(a[0]) * 1.442695F);
986 }
987 result[0] = result[1] = result[2] = result[3] = val;
988 store_vector4(inst, machine, result);
989 }
990 break;
991 case OPCODE_LIT:
992 {
993 const GLfloat epsilon = 1.0F / 256.0F; /* from NV VP spec */
994 GLfloat a[4], result[4];
995 fetch_vector4(&inst->SrcReg[0], machine, a);
996 a[0] = MAX2(a[0], 0.0F);
997 a[1] = MAX2(a[1], 0.0F);
998 /* XXX ARB version clamps a[3], NV version doesn't */
999 a[3] = CLAMP(a[3], -(128.0F - epsilon), (128.0F - epsilon));
1000 result[0] = 1.0F;
1001 result[1] = a[0];
1002 /* XXX we could probably just use pow() here */
1003 if (a[0] > 0.0F) {
1004 if (a[1] == 0.0 && a[3] == 0.0)
1005 result[2] = 1.0F;
1006 else
1007 result[2] = (GLfloat) pow(a[1], a[3]);
1008 }
1009 else {
1010 result[2] = 0.0F;
1011 }
1012 result[3] = 1.0F;
1013 store_vector4(inst, machine, result);
1014 if (DEBUG_PROG) {
1015 printf("LIT (%g %g %g %g) : (%g %g %g %g)\n",
1016 result[0], result[1], result[2], result[3],
1017 a[0], a[1], a[2], a[3]);
1018 }
1019 }
1020 break;
1021 case OPCODE_LOG:
1022 {
1023 GLfloat t[4], q[4], abs_t0;
1024 fetch_vector1(&inst->SrcReg[0], machine, t);
1025 abs_t0 = FABSF(t[0]);
1026 if (abs_t0 != 0.0F) {
1027 if (IS_INF_OR_NAN(abs_t0))
1028 {
1029 SET_POS_INFINITY(q[0]);
1030 q[1] = 1.0F;
1031 SET_POS_INFINITY(q[2]);
1032 }
1033 else {
1034 int exponent;
1035 GLfloat mantissa = FREXPF(t[0], &exponent);
1036 q[0] = (GLfloat) (exponent - 1);
1037 q[1] = (GLfloat) (2.0 * mantissa); /* map [.5, 1) -> [1, 2) */
1038
1039 /* The fast LOG2 macro doesn't meet the precision
1040 * requirements.
1041 */
1042 q[2] = (float)(log(t[0]) * 1.442695F);
1043 }
1044 }
1045 else {
1046 SET_NEG_INFINITY(q[0]);
1047 q[1] = 1.0F;
1048 SET_NEG_INFINITY(q[2]);
1049 }
1050 q[3] = 1.0;
1051 store_vector4(inst, machine, q);
1052 }
1053 break;
1054 case OPCODE_LRP:
1055 {
1056 GLfloat a[4], b[4], c[4], result[4];
1057 fetch_vector4(&inst->SrcReg[0], machine, a);
1058 fetch_vector4(&inst->SrcReg[1], machine, b);
1059 fetch_vector4(&inst->SrcReg[2], machine, c);
1060 result[0] = a[0] * b[0] + (1.0F - a[0]) * c[0];
1061 result[1] = a[1] * b[1] + (1.0F - a[1]) * c[1];
1062 result[2] = a[2] * b[2] + (1.0F - a[2]) * c[2];
1063 result[3] = a[3] * b[3] + (1.0F - a[3]) * c[3];
1064 store_vector4(inst, machine, result);
1065 if (DEBUG_PROG) {
1066 printf("LRP (%g %g %g %g) = (%g %g %g %g), "
1067 "(%g %g %g %g), (%g %g %g %g)\n",
1068 result[0], result[1], result[2], result[3],
1069 a[0], a[1], a[2], a[3],
1070 b[0], b[1], b[2], b[3], c[0], c[1], c[2], c[3]);
1071 }
1072 }
1073 break;
1074 case OPCODE_MAD:
1075 {
1076 GLfloat a[4], b[4], c[4], result[4];
1077 fetch_vector4(&inst->SrcReg[0], machine, a);
1078 fetch_vector4(&inst->SrcReg[1], machine, b);
1079 fetch_vector4(&inst->SrcReg[2], machine, c);
1080 result[0] = a[0] * b[0] + c[0];
1081 result[1] = a[1] * b[1] + c[1];
1082 result[2] = a[2] * b[2] + c[2];
1083 result[3] = a[3] * b[3] + c[3];
1084 store_vector4(inst, machine, result);
1085 if (DEBUG_PROG) {
1086 printf("MAD (%g %g %g %g) = (%g %g %g %g) * "
1087 "(%g %g %g %g) + (%g %g %g %g)\n",
1088 result[0], result[1], result[2], result[3],
1089 a[0], a[1], a[2], a[3],
1090 b[0], b[1], b[2], b[3], c[0], c[1], c[2], c[3]);
1091 }
1092 }
1093 break;
1094 case OPCODE_MAX:
1095 {
1096 GLfloat a[4], b[4], result[4];
1097 fetch_vector4(&inst->SrcReg[0], machine, a);
1098 fetch_vector4(&inst->SrcReg[1], machine, b);
1099 result[0] = MAX2(a[0], b[0]);
1100 result[1] = MAX2(a[1], b[1]);
1101 result[2] = MAX2(a[2], b[2]);
1102 result[3] = MAX2(a[3], b[3]);
1103 store_vector4(inst, machine, result);
1104 if (DEBUG_PROG) {
1105 printf("MAX (%g %g %g %g) = (%g %g %g %g), (%g %g %g %g)\n",
1106 result[0], result[1], result[2], result[3],
1107 a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]);
1108 }
1109 }
1110 break;
1111 case OPCODE_MIN:
1112 {
1113 GLfloat a[4], b[4], result[4];
1114 fetch_vector4(&inst->SrcReg[0], machine, a);
1115 fetch_vector4(&inst->SrcReg[1], machine, b);
1116 result[0] = MIN2(a[0], b[0]);
1117 result[1] = MIN2(a[1], b[1]);
1118 result[2] = MIN2(a[2], b[2]);
1119 result[3] = MIN2(a[3], b[3]);
1120 store_vector4(inst, machine, result);
1121 }
1122 break;
1123 case OPCODE_MOV:
1124 {
1125 GLfloat result[4];
1126 fetch_vector4(&inst->SrcReg[0], machine, result);
1127 store_vector4(inst, machine, result);
1128 if (DEBUG_PROG) {
1129 printf("MOV (%g %g %g %g)\n",
1130 result[0], result[1], result[2], result[3]);
1131 }
1132 }
1133 break;
1134 case OPCODE_MUL:
1135 {
1136 GLfloat a[4], b[4], result[4];
1137 fetch_vector4(&inst->SrcReg[0], machine, a);
1138 fetch_vector4(&inst->SrcReg[1], machine, b);
1139 result[0] = a[0] * b[0];
1140 result[1] = a[1] * b[1];
1141 result[2] = a[2] * b[2];
1142 result[3] = a[3] * b[3];
1143 store_vector4(inst, machine, result);
1144 if (DEBUG_PROG) {
1145 printf("MUL (%g %g %g %g) = (%g %g %g %g) * (%g %g %g %g)\n",
1146 result[0], result[1], result[2], result[3],
1147 a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]);
1148 }
1149 }
1150 break;
1151 case OPCODE_NOISE1:
1152 {
1153 GLfloat a[4], result[4];
1154 fetch_vector1(&inst->SrcReg[0], machine, a);
1155 result[0] =
1156 result[1] =
1157 result[2] =
1158 result[3] = _mesa_noise1(a[0]);
1159 store_vector4(inst, machine, result);
1160 }
1161 break;
1162 case OPCODE_NOISE2:
1163 {
1164 GLfloat a[4], result[4];
1165 fetch_vector4(&inst->SrcReg[0], machine, a);
1166 result[0] =
1167 result[1] =
1168 result[2] = result[3] = _mesa_noise2(a[0], a[1]);
1169 store_vector4(inst, machine, result);
1170 }
1171 break;
1172 case OPCODE_NOISE3:
1173 {
1174 GLfloat a[4], result[4];
1175 fetch_vector4(&inst->SrcReg[0], machine, a);
1176 result[0] =
1177 result[1] =
1178 result[2] =
1179 result[3] = _mesa_noise3(a[0], a[1], a[2]);
1180 store_vector4(inst, machine, result);
1181 }
1182 break;
1183 case OPCODE_NOISE4:
1184 {
1185 GLfloat a[4], result[4];
1186 fetch_vector4(&inst->SrcReg[0], machine, a);
1187 result[0] =
1188 result[1] =
1189 result[2] =
1190 result[3] = _mesa_noise4(a[0], a[1], a[2], a[3]);
1191 store_vector4(inst, machine, result);
1192 }
1193 break;
1194 case OPCODE_NOP:
1195 break;
1196 case OPCODE_NOT: /* bitwise NOT */
1197 {
1198 GLuint a[4], result[4];
1199 fetch_vector4ui(&inst->SrcReg[0], machine, a);
1200 result[0] = ~a[0];
1201 result[1] = ~a[1];
1202 result[2] = ~a[2];
1203 result[3] = ~a[3];
1204 store_vector4ui(inst, machine, result);
1205 }
1206 break;
1207 case OPCODE_NRM3: /* 3-component normalization */
1208 {
1209 GLfloat a[4], result[4];
1210 GLfloat tmp;
1211 fetch_vector4(&inst->SrcReg[0], machine, a);
1212 tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2];
1213 if (tmp != 0.0F)
1214 tmp = INV_SQRTF(tmp);
1215 result[0] = tmp * a[0];
1216 result[1] = tmp * a[1];
1217 result[2] = tmp * a[2];
1218 result[3] = 0.0; /* undefined, but prevent valgrind warnings */
1219 store_vector4(inst, machine, result);
1220 }
1221 break;
1222 case OPCODE_NRM4: /* 4-component normalization */
1223 {
1224 GLfloat a[4], result[4];
1225 GLfloat tmp;
1226 fetch_vector4(&inst->SrcReg[0], machine, a);
1227 tmp = a[0] * a[0] + a[1] * a[1] + a[2] * a[2] + a[3] * a[3];
1228 if (tmp != 0.0F)
1229 tmp = INV_SQRTF(tmp);
1230 result[0] = tmp * a[0];
1231 result[1] = tmp * a[1];
1232 result[2] = tmp * a[2];
1233 result[3] = tmp * a[3];
1234 store_vector4(inst, machine, result);
1235 }
1236 break;
1237 case OPCODE_OR: /* bitwise OR */
1238 {
1239 GLuint a[4], b[4], result[4];
1240 fetch_vector4ui(&inst->SrcReg[0], machine, a);
1241 fetch_vector4ui(&inst->SrcReg[1], machine, b);
1242 result[0] = a[0] | b[0];
1243 result[1] = a[1] | b[1];
1244 result[2] = a[2] | b[2];
1245 result[3] = a[3] | b[3];
1246 store_vector4ui(inst, machine, result);
1247 }
1248 break;
1249 case OPCODE_PK2H: /* pack two 16-bit floats in one 32-bit float */
1250 {
1251 GLfloat a[4];
1252 GLuint result[4];
1253 GLhalfNV hx, hy;
1254 fetch_vector4(&inst->SrcReg[0], machine, a);
1255 hx = _mesa_float_to_half(a[0]);
1256 hy = _mesa_float_to_half(a[1]);
1257 result[0] =
1258 result[1] =
1259 result[2] =
1260 result[3] = hx | (hy << 16);
1261 store_vector4ui(inst, machine, result);
1262 }
1263 break;
1264 case OPCODE_PK2US: /* pack two GLushorts into one 32-bit float */
1265 {
1266 GLfloat a[4];
1267 GLuint result[4], usx, usy;
1268 fetch_vector4(&inst->SrcReg[0], machine, a);
1269 a[0] = CLAMP(a[0], 0.0F, 1.0F);
1270 a[1] = CLAMP(a[1], 0.0F, 1.0F);
1271 usx = F_TO_I(a[0] * 65535.0F);
1272 usy = F_TO_I(a[1] * 65535.0F);
1273 result[0] =
1274 result[1] =
1275 result[2] =
1276 result[3] = usx | (usy << 16);
1277 store_vector4ui(inst, machine, result);
1278 }
1279 break;
1280 case OPCODE_PK4B: /* pack four GLbytes into one 32-bit float */
1281 {
1282 GLfloat a[4];
1283 GLuint result[4], ubx, uby, ubz, ubw;
1284 fetch_vector4(&inst->SrcReg[0], machine, a);
1285 a[0] = CLAMP(a[0], -128.0F / 127.0F, 1.0F);
1286 a[1] = CLAMP(a[1], -128.0F / 127.0F, 1.0F);
1287 a[2] = CLAMP(a[2], -128.0F / 127.0F, 1.0F);
1288 a[3] = CLAMP(a[3], -128.0F / 127.0F, 1.0F);
1289 ubx = F_TO_I(127.0F * a[0] + 128.0F);
1290 uby = F_TO_I(127.0F * a[1] + 128.0F);
1291 ubz = F_TO_I(127.0F * a[2] + 128.0F);
1292 ubw = F_TO_I(127.0F * a[3] + 128.0F);
1293 result[0] =
1294 result[1] =
1295 result[2] =
1296 result[3] = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
1297 store_vector4ui(inst, machine, result);
1298 }
1299 break;
1300 case OPCODE_PK4UB: /* pack four GLubytes into one 32-bit float */
1301 {
1302 GLfloat a[4];
1303 GLuint result[4], ubx, uby, ubz, ubw;
1304 fetch_vector4(&inst->SrcReg[0], machine, a);
1305 a[0] = CLAMP(a[0], 0.0F, 1.0F);
1306 a[1] = CLAMP(a[1], 0.0F, 1.0F);
1307 a[2] = CLAMP(a[2], 0.0F, 1.0F);
1308 a[3] = CLAMP(a[3], 0.0F, 1.0F);
1309 ubx = F_TO_I(255.0F * a[0]);
1310 uby = F_TO_I(255.0F * a[1]);
1311 ubz = F_TO_I(255.0F * a[2]);
1312 ubw = F_TO_I(255.0F * a[3]);
1313 result[0] =
1314 result[1] =
1315 result[2] =
1316 result[3] = ubx | (uby << 8) | (ubz << 16) | (ubw << 24);
1317 store_vector4ui(inst, machine, result);
1318 }
1319 break;
1320 case OPCODE_POW:
1321 {
1322 GLfloat a[4], b[4], result[4];
1323 fetch_vector1(&inst->SrcReg[0], machine, a);
1324 fetch_vector1(&inst->SrcReg[1], machine, b);
1325 result[0] = result[1] = result[2] = result[3]
1326 = (GLfloat) pow(a[0], b[0]);
1327 store_vector4(inst, machine, result);
1328 }
1329 break;
1330
1331 case OPCODE_RCP:
1332 {
1333 GLfloat a[4], result[4];
1334 fetch_vector1(&inst->SrcReg[0], machine, a);
1335 if (DEBUG_PROG) {
1336 if (a[0] == 0)
1337 printf("RCP(0)\n");
1338 else if (IS_INF_OR_NAN(a[0]))
1339 printf("RCP(inf)\n");
1340 }
1341 result[0] = result[1] = result[2] = result[3] = 1.0F / a[0];
1342 store_vector4(inst, machine, result);
1343 }
1344 break;
1345 case OPCODE_RET: /* return from subroutine (conditional) */
1346 if (eval_condition(machine, inst)) {
1347 if (machine->StackDepth == 0) {
1348 return GL_TRUE; /* Per GL_NV_vertex_program2 spec */
1349 }
1350 /* subtract one because of pc++ in the for loop */
1351 pc = machine->CallStack[--machine->StackDepth] - 1;
1352 }
1353 break;
1354 case OPCODE_RFL: /* reflection vector */
1355 {
1356 GLfloat axis[4], dir[4], result[4], tmpX, tmpW;
1357 fetch_vector4(&inst->SrcReg[0], machine, axis);
1358 fetch_vector4(&inst->SrcReg[1], machine, dir);
1359 tmpW = DOT3(axis, axis);
1360 tmpX = (2.0F * DOT3(axis, dir)) / tmpW;
1361 result[0] = tmpX * axis[0] - dir[0];
1362 result[1] = tmpX * axis[1] - dir[1];
1363 result[2] = tmpX * axis[2] - dir[2];
1364 /* result[3] is never written! XXX enforce in parser! */
1365 store_vector4(inst, machine, result);
1366 }
1367 break;
1368 case OPCODE_RSQ: /* 1 / sqrt() */
1369 {
1370 GLfloat a[4], result[4];
1371 fetch_vector1(&inst->SrcReg[0], machine, a);
1372 a[0] = FABSF(a[0]);
1373 result[0] = result[1] = result[2] = result[3] = INV_SQRTF(a[0]);
1374 store_vector4(inst, machine, result);
1375 if (DEBUG_PROG) {
1376 printf("RSQ %g = 1/sqrt(|%g|)\n", result[0], a[0]);
1377 }
1378 }
1379 break;
1380 case OPCODE_SCS: /* sine and cos */
1381 {
1382 GLfloat a[4], result[4];
1383 fetch_vector1(&inst->SrcReg[0], machine, a);
1384 result[0] = (GLfloat) cos(a[0]);
1385 result[1] = (GLfloat) sin(a[0]);
1386 result[2] = 0.0; /* undefined! */
1387 result[3] = 0.0; /* undefined! */
1388 store_vector4(inst, machine, result);
1389 }
1390 break;
1391 case OPCODE_SEQ: /* set on equal */
1392 {
1393 GLfloat a[4], b[4], result[4];
1394 fetch_vector4(&inst->SrcReg[0], machine, a);
1395 fetch_vector4(&inst->SrcReg[1], machine, b);
1396 result[0] = (a[0] == b[0]) ? 1.0F : 0.0F;
1397 result[1] = (a[1] == b[1]) ? 1.0F : 0.0F;
1398 result[2] = (a[2] == b[2]) ? 1.0F : 0.0F;
1399 result[3] = (a[3] == b[3]) ? 1.0F : 0.0F;
1400 store_vector4(inst, machine, result);
1401 if (DEBUG_PROG) {
1402 printf("SEQ (%g %g %g %g) = (%g %g %g %g) == (%g %g %g %g)\n",
1403 result[0], result[1], result[2], result[3],
1404 a[0], a[1], a[2], a[3],
1405 b[0], b[1], b[2], b[3]);
1406 }
1407 }
1408 break;
1409 case OPCODE_SFL: /* set false, operands ignored */
1410 {
1411 static const GLfloat result[4] = { 0.0F, 0.0F, 0.0F, 0.0F };
1412 store_vector4(inst, machine, result);
1413 }
1414 break;
1415 case OPCODE_SGE: /* set on greater or equal */
1416 {
1417 GLfloat a[4], b[4], result[4];
1418 fetch_vector4(&inst->SrcReg[0], machine, a);
1419 fetch_vector4(&inst->SrcReg[1], machine, b);
1420 result[0] = (a[0] >= b[0]) ? 1.0F : 0.0F;
1421 result[1] = (a[1] >= b[1]) ? 1.0F : 0.0F;
1422 result[2] = (a[2] >= b[2]) ? 1.0F : 0.0F;
1423 result[3] = (a[3] >= b[3]) ? 1.0F : 0.0F;
1424 store_vector4(inst, machine, result);
1425 if (DEBUG_PROG) {
1426 printf("SGE (%g %g %g %g) = (%g %g %g %g) >= (%g %g %g %g)\n",
1427 result[0], result[1], result[2], result[3],
1428 a[0], a[1], a[2], a[3],
1429 b[0], b[1], b[2], b[3]);
1430 }
1431 }
1432 break;
1433 case OPCODE_SGT: /* set on greater */
1434 {
1435 GLfloat a[4], b[4], result[4];
1436 fetch_vector4(&inst->SrcReg[0], machine, a);
1437 fetch_vector4(&inst->SrcReg[1], machine, b);
1438 result[0] = (a[0] > b[0]) ? 1.0F : 0.0F;
1439 result[1] = (a[1] > b[1]) ? 1.0F : 0.0F;
1440 result[2] = (a[2] > b[2]) ? 1.0F : 0.0F;
1441 result[3] = (a[3] > b[3]) ? 1.0F : 0.0F;
1442 store_vector4(inst, machine, result);
1443 if (DEBUG_PROG) {
1444 printf("SGT (%g %g %g %g) = (%g %g %g %g) > (%g %g %g %g)\n",
1445 result[0], result[1], result[2], result[3],
1446 a[0], a[1], a[2], a[3],
1447 b[0], b[1], b[2], b[3]);
1448 }
1449 }
1450 break;
1451 case OPCODE_SIN:
1452 {
1453 GLfloat a[4], result[4];
1454 fetch_vector1(&inst->SrcReg[0], machine, a);
1455 result[0] = result[1] = result[2] = result[3]
1456 = (GLfloat) sin(a[0]);
1457 store_vector4(inst, machine, result);
1458 }
1459 break;
1460 case OPCODE_SLE: /* set on less or equal */
1461 {
1462 GLfloat a[4], b[4], result[4];
1463 fetch_vector4(&inst->SrcReg[0], machine, a);
1464 fetch_vector4(&inst->SrcReg[1], machine, b);
1465 result[0] = (a[0] <= b[0]) ? 1.0F : 0.0F;
1466 result[1] = (a[1] <= b[1]) ? 1.0F : 0.0F;
1467 result[2] = (a[2] <= b[2]) ? 1.0F : 0.0F;
1468 result[3] = (a[3] <= b[3]) ? 1.0F : 0.0F;
1469 store_vector4(inst, machine, result);
1470 if (DEBUG_PROG) {
1471 printf("SLE (%g %g %g %g) = (%g %g %g %g) <= (%g %g %g %g)\n",
1472 result[0], result[1], result[2], result[3],
1473 a[0], a[1], a[2], a[3],
1474 b[0], b[1], b[2], b[3]);
1475 }
1476 }
1477 break;
1478 case OPCODE_SLT: /* set on less */
1479 {
1480 GLfloat a[4], b[4], result[4];
1481 fetch_vector4(&inst->SrcReg[0], machine, a);
1482 fetch_vector4(&inst->SrcReg[1], machine, b);
1483 result[0] = (a[0] < b[0]) ? 1.0F : 0.0F;
1484 result[1] = (a[1] < b[1]) ? 1.0F : 0.0F;
1485 result[2] = (a[2] < b[2]) ? 1.0F : 0.0F;
1486 result[3] = (a[3] < b[3]) ? 1.0F : 0.0F;
1487 store_vector4(inst, machine, result);
1488 if (DEBUG_PROG) {
1489 printf("SLT (%g %g %g %g) = (%g %g %g %g) < (%g %g %g %g)\n",
1490 result[0], result[1], result[2], result[3],
1491 a[0], a[1], a[2], a[3],
1492 b[0], b[1], b[2], b[3]);
1493 }
1494 }
1495 break;
1496 case OPCODE_SNE: /* set on not equal */
1497 {
1498 GLfloat a[4], b[4], result[4];
1499 fetch_vector4(&inst->SrcReg[0], machine, a);
1500 fetch_vector4(&inst->SrcReg[1], machine, b);
1501 result[0] = (a[0] != b[0]) ? 1.0F : 0.0F;
1502 result[1] = (a[1] != b[1]) ? 1.0F : 0.0F;
1503 result[2] = (a[2] != b[2]) ? 1.0F : 0.0F;
1504 result[3] = (a[3] != b[3]) ? 1.0F : 0.0F;
1505 store_vector4(inst, machine, result);
1506 if (DEBUG_PROG) {
1507 printf("SNE (%g %g %g %g) = (%g %g %g %g) != (%g %g %g %g)\n",
1508 result[0], result[1], result[2], result[3],
1509 a[0], a[1], a[2], a[3],
1510 b[0], b[1], b[2], b[3]);
1511 }
1512 }
1513 break;
1514 case OPCODE_SSG: /* set sign (-1, 0 or +1) */
1515 {
1516 GLfloat a[4], result[4];
1517 fetch_vector4(&inst->SrcReg[0], machine, a);
1518 result[0] = (GLfloat) ((a[0] > 0.0F) - (a[0] < 0.0F));
1519 result[1] = (GLfloat) ((a[1] > 0.0F) - (a[1] < 0.0F));
1520 result[2] = (GLfloat) ((a[2] > 0.0F) - (a[2] < 0.0F));
1521 result[3] = (GLfloat) ((a[3] > 0.0F) - (a[3] < 0.0F));
1522 store_vector4(inst, machine, result);
1523 }
1524 break;
1525 case OPCODE_STR: /* set true, operands ignored */
1526 {
1527 static const GLfloat result[4] = { 1.0F, 1.0F, 1.0F, 1.0F };
1528 store_vector4(inst, machine, result);
1529 }
1530 break;
1531 case OPCODE_SUB:
1532 {
1533 GLfloat a[4], b[4], result[4];
1534 fetch_vector4(&inst->SrcReg[0], machine, a);
1535 fetch_vector4(&inst->SrcReg[1], machine, b);
1536 result[0] = a[0] - b[0];
1537 result[1] = a[1] - b[1];
1538 result[2] = a[2] - b[2];
1539 result[3] = a[3] - b[3];
1540 store_vector4(inst, machine, result);
1541 if (DEBUG_PROG) {
1542 printf("SUB (%g %g %g %g) = (%g %g %g %g) - (%g %g %g %g)\n",
1543 result[0], result[1], result[2], result[3],
1544 a[0], a[1], a[2], a[3], b[0], b[1], b[2], b[3]);
1545 }
1546 }
1547 break;
1548 case OPCODE_SWZ: /* extended swizzle */
1549 {
1550 const struct prog_src_register *source = &inst->SrcReg[0];
1551 const GLfloat *src = get_src_register_pointer(source, machine);
1552 GLfloat result[4];
1553 GLuint i;
1554 for (i = 0; i < 4; i++) {
1555 const GLuint swz = GET_SWZ(source->Swizzle, i);
1556 if (swz == SWIZZLE_ZERO)
1557 result[i] = 0.0;
1558 else if (swz == SWIZZLE_ONE)
1559 result[i] = 1.0;
1560 else {
1561 ASSERT(swz >= 0);
1562 ASSERT(swz <= 3);
1563 result[i] = src[swz];
1564 }
1565 if (source->Negate & (1 << i))
1566 result[i] = -result[i];
1567 }
1568 store_vector4(inst, machine, result);
1569 }
1570 break;
1571 case OPCODE_TEX: /* Both ARB and NV frag prog */
1572 /* Simple texel lookup */
1573 {
1574 GLfloat texcoord[4], color[4];
1575 fetch_vector4(&inst->SrcReg[0], machine, texcoord);
1576
1577 /* For TEX, texcoord.Q should not be used and its value should not
1578 * matter (at most, we pass coord.xyz to texture3D() in GLSL).
1579 * Set Q=1 so that FetchTexelDeriv() doesn't get a garbage value
1580 * which is effectively what happens when the texcoord swizzle
1581 * is .xyzz
1582 */
1583 texcoord[3] = 1.0f;
1584
1585 fetch_texel(ctx, machine, inst, texcoord, 0.0, color);
1586
1587 if (DEBUG_PROG) {
1588 printf("TEX (%g, %g, %g, %g) = texture[%d][%g, %g, %g, %g]\n",
1589 color[0], color[1], color[2], color[3],
1590 inst->TexSrcUnit,
1591 texcoord[0], texcoord[1], texcoord[2], texcoord[3]);
1592 }
1593 store_vector4(inst, machine, color);
1594 }
1595 break;
1596 case OPCODE_TXB: /* GL_ARB_fragment_program only */
1597 /* Texel lookup with LOD bias */
1598 {
1599 GLfloat texcoord[4], color[4], lodBias;
1600
1601 fetch_vector4(&inst->SrcReg[0], machine, texcoord);
1602
1603 /* texcoord[3] is the bias to add to lambda */
1604 lodBias = texcoord[3];
1605
1606 fetch_texel(ctx, machine, inst, texcoord, lodBias, color);
1607
1608 if (DEBUG_PROG) {
1609 printf("TXB (%g, %g, %g, %g) = texture[%d][%g %g %g %g]"
1610 " bias %g\n",
1611 color[0], color[1], color[2], color[3],
1612 inst->TexSrcUnit,
1613 texcoord[0],
1614 texcoord[1],
1615 texcoord[2],
1616 texcoord[3],
1617 lodBias);
1618 }
1619
1620 store_vector4(inst, machine, color);
1621 }
1622 break;
1623 case OPCODE_TXD: /* GL_NV_fragment_program only */
1624 /* Texture lookup w/ partial derivatives for LOD */
1625 {
1626 GLfloat texcoord[4], dtdx[4], dtdy[4], color[4];
1627 fetch_vector4(&inst->SrcReg[0], machine, texcoord);
1628 fetch_vector4(&inst->SrcReg[1], machine, dtdx);
1629 fetch_vector4(&inst->SrcReg[2], machine, dtdy);
1630 machine->FetchTexelDeriv(ctx, texcoord, dtdx, dtdy,
1631 0.0, /* lodBias */
1632 inst->TexSrcUnit, color);
1633 store_vector4(inst, machine, color);
1634 }
1635 break;
1636 case OPCODE_TXL:
1637 /* Texel lookup with explicit LOD */
1638 {
1639 GLfloat texcoord[4], color[4], lod;
1640
1641 fetch_vector4(&inst->SrcReg[0], machine, texcoord);
1642
1643 /* texcoord[3] is the LOD */
1644 lod = texcoord[3];
1645
1646 machine->FetchTexelLod(ctx, texcoord, lod,
1647 machine->Samplers[inst->TexSrcUnit], color);
1648
1649 store_vector4(inst, machine, color);
1650 }
1651 break;
1652 case OPCODE_TXP: /* GL_ARB_fragment_program only */
1653 /* Texture lookup w/ projective divide */
1654 {
1655 GLfloat texcoord[4], color[4];
1656
1657 fetch_vector4(&inst->SrcReg[0], machine, texcoord);
1658 /* Not so sure about this test - if texcoord[3] is
1659 * zero, we'd probably be fine except for an ASSERT in
1660 * IROUND_POS() which gets triggered by the inf values created.
1661 */
1662 if (texcoord[3] != 0.0) {
1663 texcoord[0] /= texcoord[3];
1664 texcoord[1] /= texcoord[3];
1665 texcoord[2] /= texcoord[3];
1666 }
1667
1668 fetch_texel(ctx, machine, inst, texcoord, 0.0, color);
1669
1670 store_vector4(inst, machine, color);
1671 }
1672 break;
1673 case OPCODE_TXP_NV: /* GL_NV_fragment_program only */
1674 /* Texture lookup w/ projective divide, as above, but do not
1675 * do the divide by w if sampling from a cube map.
1676 */
1677 {
1678 GLfloat texcoord[4], color[4];
1679
1680 fetch_vector4(&inst->SrcReg[0], machine, texcoord);
1681 if (inst->TexSrcTarget != TEXTURE_CUBE_INDEX &&
1682 texcoord[3] != 0.0) {
1683 texcoord[0] /= texcoord[3];
1684 texcoord[1] /= texcoord[3];
1685 texcoord[2] /= texcoord[3];
1686 }
1687
1688 fetch_texel(ctx, machine, inst, texcoord, 0.0, color);
1689
1690 store_vector4(inst, machine, color);
1691 }
1692 break;
1693 case OPCODE_TRUNC: /* truncate toward zero */
1694 {
1695 GLfloat a[4], result[4];
1696 fetch_vector4(&inst->SrcReg[0], machine, a);
1697 result[0] = (GLfloat) (GLint) a[0];
1698 result[1] = (GLfloat) (GLint) a[1];
1699 result[2] = (GLfloat) (GLint) a[2];
1700 result[3] = (GLfloat) (GLint) a[3];
1701 store_vector4(inst, machine, result);
1702 }
1703 break;
1704 case OPCODE_UP2H: /* unpack two 16-bit floats */
1705 {
1706 const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine);
1707 GLfloat result[4];
1708 GLushort hx, hy;
1709 hx = raw & 0xffff;
1710 hy = raw >> 16;
1711 result[0] = result[2] = _mesa_half_to_float(hx);
1712 result[1] = result[3] = _mesa_half_to_float(hy);
1713 store_vector4(inst, machine, result);
1714 }
1715 break;
1716 case OPCODE_UP2US: /* unpack two GLushorts */
1717 {
1718 const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine);
1719 GLfloat result[4];
1720 GLushort usx, usy;
1721 usx = raw & 0xffff;
1722 usy = raw >> 16;
1723 result[0] = result[2] = usx * (1.0f / 65535.0f);
1724 result[1] = result[3] = usy * (1.0f / 65535.0f);
1725 store_vector4(inst, machine, result);
1726 }
1727 break;
1728 case OPCODE_UP4B: /* unpack four GLbytes */
1729 {
1730 const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine);
1731 GLfloat result[4];
1732 result[0] = (((raw >> 0) & 0xff) - 128) / 127.0F;
1733 result[1] = (((raw >> 8) & 0xff) - 128) / 127.0F;
1734 result[2] = (((raw >> 16) & 0xff) - 128) / 127.0F;
1735 result[3] = (((raw >> 24) & 0xff) - 128) / 127.0F;
1736 store_vector4(inst, machine, result);
1737 }
1738 break;
1739 case OPCODE_UP4UB: /* unpack four GLubytes */
1740 {
1741 const GLuint raw = fetch_vector1ui(&inst->SrcReg[0], machine);
1742 GLfloat result[4];
1743 result[0] = ((raw >> 0) & 0xff) / 255.0F;
1744 result[1] = ((raw >> 8) & 0xff) / 255.0F;
1745 result[2] = ((raw >> 16) & 0xff) / 255.0F;
1746 result[3] = ((raw >> 24) & 0xff) / 255.0F;
1747 store_vector4(inst, machine, result);
1748 }
1749 break;
1750 case OPCODE_XOR: /* bitwise XOR */
1751 {
1752 GLuint a[4], b[4], result[4];
1753 fetch_vector4ui(&inst->SrcReg[0], machine, a);
1754 fetch_vector4ui(&inst->SrcReg[1], machine, b);
1755 result[0] = a[0] ^ b[0];
1756 result[1] = a[1] ^ b[1];
1757 result[2] = a[2] ^ b[2];
1758 result[3] = a[3] ^ b[3];
1759 store_vector4ui(inst, machine, result);
1760 }
1761 break;
1762 case OPCODE_XPD: /* cross product */
1763 {
1764 GLfloat a[4], b[4], result[4];
1765 fetch_vector4(&inst->SrcReg[0], machine, a);
1766 fetch_vector4(&inst->SrcReg[1], machine, b);
1767 result[0] = a[1] * b[2] - a[2] * b[1];
1768 result[1] = a[2] * b[0] - a[0] * b[2];
1769 result[2] = a[0] * b[1] - a[1] * b[0];
1770 result[3] = 1.0;
1771 store_vector4(inst, machine, result);
1772 if (DEBUG_PROG) {
1773 printf("XPD (%g %g %g %g) = (%g %g %g) X (%g %g %g)\n",
1774 result[0], result[1], result[2], result[3],
1775 a[0], a[1], a[2], b[0], b[1], b[2]);
1776 }
1777 }
1778 break;
1779 case OPCODE_X2D: /* 2-D matrix transform */
1780 {
1781 GLfloat a[4], b[4], c[4], result[4];
1782 fetch_vector4(&inst->SrcReg[0], machine, a);
1783 fetch_vector4(&inst->SrcReg[1], machine, b);
1784 fetch_vector4(&inst->SrcReg[2], machine, c);
1785 result[0] = a[0] + b[0] * c[0] + b[1] * c[1];
1786 result[1] = a[1] + b[0] * c[2] + b[1] * c[3];
1787 result[2] = a[2] + b[0] * c[0] + b[1] * c[1];
1788 result[3] = a[3] + b[0] * c[2] + b[1] * c[3];
1789 store_vector4(inst, machine, result);
1790 }
1791 break;
1792 case OPCODE_PRINT:
1793 {
1794 if (inst->SrcReg[0].File != PROGRAM_UNDEFINED) {
1795 GLfloat a[4];
1796 fetch_vector4(&inst->SrcReg[0], machine, a);
1797 printf("%s%g, %g, %g, %g\n", (const char *) inst->Data,
1798 a[0], a[1], a[2], a[3]);
1799 }
1800 else {
1801 printf("%s\n", (const char *) inst->Data);
1802 }
1803 }
1804 break;
1805 case OPCODE_END:
1806 return GL_TRUE;
1807 default:
1808 _mesa_problem(ctx, "Bad opcode %d in _mesa_execute_program",
1809 inst->Opcode);
1810 return GL_TRUE; /* return value doesn't matter */
1811 }
1812
1813 numExec++;
1814 if (numExec > maxExec) {
1815 static GLboolean reported = GL_FALSE;
1816 if (!reported) {
1817 _mesa_problem(ctx, "Infinite loop detected in fragment program");
1818 reported = GL_TRUE;
1819 }
1820 return GL_TRUE;
1821 }
1822
1823 } /* for pc */
1824
1825 return GL_TRUE;
1826 }