projects / mesa.git / blob
? search:


c1207cacd11e93247c82e96dd953d7ac7348bb6e
[mesa.git] / src / mesa / drivers / dri / mach64 / mach64_vbtmp.h
1 /* $XFree86$ */ /* -*- mode: c; c-basic-offset: 3 -*- */
2 /*
3 * Mesa 3-D graphics library
4 * Version: 3.5
5 *
6 * Copyright (C) 1999-2001 Brian Paul All Rights Reserved.
7 *
8 * Permission is hereby granted, free of charge, to any person obtaining a
9 * copy of this software and associated documentation files (the "Software"),
10 * to deal in the Software without restriction, including without limitation
11 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
12 * and/or sell copies of the Software, and to permit persons to whom the
13 * Software is furnished to do so, subject to the following conditions:
14 *
15 * The above copyright notice and this permission notice shall be included
16 * in all copies or substantial portions of the Software.
17 *
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
19 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
20 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
21 * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
22 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
23 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
24 *
25 * Authors:
26 * Keith Whitwell <keithw@valinux.com>
27 *
28 * Modified for mach64 by:
29 * Leif Delgass <ldelgass@retinalburn.net>
30 * José Fonseca <j_r_fonseca@yahoo.co.uk>
31 */
32
33
34 /* Unlike the other templates here, this assumes quite a bit about the
35 * underlying hardware. Specifically it assumes a d3d-like vertex
36 * format, with a layout more or less constrained to look like the
37 * following:
38 *
39 * union {
40 * struct {
41 * float x, y, z, w;
42 * struct { char r, g, b, a; } color;
43 * struct { char r, g, b, fog; } spec;
44 * float u0, v0;
45 * float u1, v1;
46 * float u2, v2;
47 * float u3, v3;
48 * } v;
49 * struct {
50 * float x, y, z, w;
51 * struct { char r, g, b, a; } color;
52 * struct { char r, g, b, fog; } spec;
53 * float u0, v0, q0;
54 * float u1, v1, q1;
55 * float u2, v2, q2;
56 * float u3, v3, q3;
57 * } pv;
58 * struct {
59 * float x, y, z;
60 * struct { char r, g, b, a; } color;
61 * } tv;
62 * float f[16];
63 * unsigned int ui[16];
64 * unsigned char ub4[4][16];
65 * }
66 *
67
68 * DO_XYZW: Emit xyz and maybe w coordinates.
69 * DO_RGBA: Emit color.
70 * DO_SPEC: Emit specular color.
71 * DO_FOG: Emit fog coordinate in specular alpha.
72 * DO_TEX0: Emit tex0 u,v coordinates.
73 * DO_TEX1: Emit tex1 u,v coordinates.
74 * DO_TEX2: Emit tex2 u,v coordinates.
75 * DO_TEX3: Emit tex3 u,v coordinates.
76 * DO_PTEX: Emit tex0,1,2,3 q coordinates where possible.
77 *
78 * HAVE_RGBA_COLOR: Hardware takes color in rgba order (else bgra).
79 *
80 * HAVE_HW_VIEWPORT: Hardware performs viewport transform.
81 * HAVE_HW_DIVIDE: Hardware performs perspective divide.
82 *
83 * HAVE_TINY_VERTICES: Hardware understands v.tv format.
84 * HAVE_PTEX_VERTICES: Hardware understands v.pv format.
85 * HAVE_NOTEX_VERTICES: Hardware understands v.v format with texcount 0.
86 *
87 * Additionally, this template assumes it is emitting *transformed*
88 * vertices; the modifications to emit untransformed vertices (ie. to
89 * t&l hardware) are probably too great to cooexist with the code
90 * already in this file.
91 *
92 * NOTE: The PTEX vertex format always includes TEX0 and TEX1, even if
93 * only TEX0 is enabled, in order to maintain a vertex size which is
94 * an exact number of quadwords.
95 */
96
97 #if (HAVE_HW_VIEWPORT)
98 #define VIEWPORT_X(dst,x) dst = x
99 #define VIEWPORT_Y(dst,y) dst = y
100 #define VIEWPORT_Z(dst,z) dst = z
101 #else
102 #define VIEWPORT_X(dst,x) dst = s[0] * x + s[12]
103 #define VIEWPORT_Y(dst,y) dst = s[5] * y + s[13]
104 #define VIEWPORT_Z(dst,z) dst = s[10] * z + s[14]
105 #endif
106
107 #if (HAVE_HW_DIVIDE && !HAVE_PTEX_VERTICES)
108 #error "can't cope with this combination"
109 #endif
110
111 #ifndef LOCALVARS
112 #define LOCALVARS
113 #endif
114
115 #ifndef CHECK_HW_DIVIDE
116 #define CHECK_HW_DIVIDE 1
117 #endif
118
119 #if (HAVE_HW_DIVIDE || DO_SPEC || DO_TEX0 || DO_FOG || !HAVE_TINY_VERTICES)
120
121 static void TAG(emit)( GLcontext *ctx,
122 GLuint start, GLuint end,
123 void *dest,
124 GLuint stride )
125 {
126 LOCALVARS
127 struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
128 GLfloat (*tc0)[4], (*tc1)[4], (*fog)[4];
129 GLfloat (*tc2)[4], (*tc3)[4];
130 GLfloat (*spec)[4];
131 GLfloat (*col)[4];
132 GLuint col_stride;
133 GLuint tc0_stride, tc1_stride, spec_stride, fog_stride;
134 GLuint tc2_stride, tc3_stride;
135 GLuint tc0_size, tc1_size;
136 GLuint tc2_size, tc3_size;
137 GLfloat (*coord)[4];
138 GLuint coord_stride;
139 VERTEX *v = (VERTEX *)dest;
140 const GLfloat *s = GET_VIEWPORT_MAT();
141 const GLubyte *mask = VB->ClipMask;
142 int i;
143
144 /* fprintf(stderr, "%s(big) importable %d %d..%d\n", */
145 /* __FUNCTION__, VB->importable_data, start, end); */
146
147 if (HAVE_HW_VIEWPORT && HAVE_HW_DIVIDE && CHECK_HW_DIVIDE) {
148 (void) s;
149 coord = VB->ClipPtr->data;
150 coord_stride = VB->ClipPtr->stride;
151 }
152 else {
153 coord = VB->NdcPtr->data;
154 coord_stride = VB->NdcPtr->stride;
155 }
156
157 if (DO_TEX3) {
158 const GLuint t3 = GET_TEXSOURCE(3);
159 tc3 = VB->TexCoordPtr[t3]->data;
160 tc3_stride = VB->TexCoordPtr[t3]->stride;
161 if (DO_PTEX)
162 tc3_size = VB->TexCoordPtr[t3]->size;
163 }
164
165 if (DO_TEX2) {
166 const GLuint t2 = GET_TEXSOURCE(2);
167 tc2 = VB->TexCoordPtr[t2]->data;
168 tc2_stride = VB->TexCoordPtr[t2]->stride;
169 if (DO_PTEX)
170 tc2_size = VB->TexCoordPtr[t2]->size;
171 }
172
173 if (DO_TEX1) {
174 const GLuint t1 = GET_TEXSOURCE(1);
175 tc1 = VB->TexCoordPtr[t1]->data;
176 tc1_stride = VB->TexCoordPtr[t1]->stride;
177 if (DO_PTEX)
178 tc1_size = VB->TexCoordPtr[t1]->size;
179 }
180
181 if (DO_TEX0) {
182 const GLuint t0 = GET_TEXSOURCE(0);
183 tc0_stride = VB->TexCoordPtr[t0]->stride;
184 tc0 = VB->TexCoordPtr[t0]->data;
185 if (DO_PTEX)
186 tc0_size = VB->TexCoordPtr[t0]->size;
187 }
188
189 if (DO_RGBA) {
190 col = VB->ColorPtr[0]->data;
191 col_stride = VB->ColorPtr[0]->stride;
192 }
193
194 if (DO_SPEC) {
195 spec = VB->SecondaryColorPtr[0]->data;
196 spec_stride = VB->SecondaryColorPtr[0]->stride;
197 } else {
198 spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
199 spec_stride = 0;
200 }
201
202 if (DO_FOG) {
203 if (VB->FogCoordPtr) {
204 fog = VB->FogCoordPtr->data;
205 fog_stride = VB->FogCoordPtr->stride;
206 } else {
207 static GLfloat tmp[4] = {0, 0, 0, 0};
208 fog = &tmp;
209 fog_stride = 0;
210 }
211 }
212
213 /* May have nonstandard strides:
214 */
215 if (start) {
216 coord = (GLfloat (*)[4])((GLubyte *)coord + start * coord_stride);
217 if (DO_TEX0)
218 tc0 = (GLfloat (*)[4])((GLubyte *)tc0 + start * tc0_stride);
219 if (DO_TEX1)
220 tc1 = (GLfloat (*)[4])((GLubyte *)tc1 + start * tc1_stride);
221 if (DO_TEX2)
222 tc2 = (GLfloat (*)[4])((GLubyte *)tc2 + start * tc2_stride);
223 if (DO_TEX3)
224 tc3 = (GLfloat (*)[4])((GLubyte *)tc3 + start * tc3_stride);
225 if (DO_RGBA)
226 STRIDE_4F(col, start * col_stride);
227 if (DO_SPEC)
228 STRIDE_4F(spec, start * spec_stride);
229 if (DO_FOG)
230 STRIDE_4F(fog, start * fog_stride);
231 // fog = (GLfloat (*)[4])((GLubyte *)fog + start * fog_stride);
232 /* STRIDE_F(fog, start * fog_stride); */
233 }
234
235 for (i=start; i < end; i++, v = (VERTEX *)((GLubyte *)v + stride)) {
236 if (DO_XYZW) {
237 if (HAVE_HW_VIEWPORT || mask[i] == 0) {
238 /* unclipped */
239 VIEWPORT_X(v->v.x, coord[0][0]);
240 VIEWPORT_Y(v->v.y, coord[0][1]);
241 VIEWPORT_Z(v->v.z, coord[0][2]);
242 v->v.w = coord[0][3];
243 } else {
244 /* clipped */
245 v->v.w = 1.0;
246 }
247 if (MACH64_DEBUG & DEBUG_VERBOSE_PRIMS) {
248 fprintf(stderr, "%s: vert (importable) %d: %.2f %.2f %.2f %f\n",
249 __FUNCTION__, i, v->v.x, v->v.y, v->v.z, v->v.w);
250 }
251 coord = (GLfloat (*)[4])((GLubyte *)coord + coord_stride);
252 }
253 if (DO_RGBA) {
254 if (HAVE_RGBA_COLOR) {
255 *(GLuint *)&v->v.color = *(GLuint *)&col[0];
256 STRIDE_4F(col, col_stride);
257 } else {
258 v->v.color.blue = col[0][2];
259 v->v.color.green = col[0][1];
260 v->v.color.red = col[0][0];
261 v->v.color.alpha = col[0][3];
262 STRIDE_4F(col, col_stride);
263 }
264 }
265 if (DO_SPEC) {
266 v->v.specular.red = spec[0][0];
267 v->v.specular.green = spec[0][1];
268 v->v.specular.blue = spec[0][2];
269 STRIDE_4F(spec, spec_stride);
270 }
271 if (DO_FOG) {
272 v->v.specular.alpha = fog[0][0] * 255.0;
273 /* STRIDE_F(fog, fog_stride); */
274 fog = (GLfloat (*)[4])((GLubyte *)fog + fog_stride);
275 }
276 if (DO_TEX0) {
277 v->v.u0 = tc0[0][0];
278 v->v.v0 = tc0[0][1];
279 if (MACH64_DEBUG & DEBUG_VERBOSE_PRIMS) {
280 fprintf(stderr, "%s: vert (importable) %d: u0: %.2f, v0: %.2f, w: %f\n",
281 __FUNCTION__, i, v->v.u0, v->v.v0, v->v.w);
282 }
283 #ifdef MACH64_PREMULT_TEXCOORDS
284 v->v.u0 *= v->v.w;
285 v->v.v0 *= v->v.w;
286 #endif
287 if (DO_PTEX) {
288 if (HAVE_PTEX_VERTICES) {
289 if (tc0_size == 4)
290 v->pv.q0 = tc0[0][3];
291 else
292 v->pv.q0 = 1.0;
293 }
294 else if (tc0_size == 4) {
295 #ifdef MACH64_PREMULT_TEXCOORDS
296 v->v.w *= tc0[0][3];
297 #else
298 float rhw = 1.0 / tc0[0][3];
299 v->v.w *= tc0[0][3];
300 v->v.u0 *= rhw;
301 v->v.v0 *= rhw;
302 #endif
303 }
304 }
305 tc0 = (GLfloat (*)[4])((GLubyte *)tc0 + tc0_stride);
306 }
307 if (DO_TEX1) {
308 if (DO_PTEX) {
309 v->pv.u1 = tc1[0][0];
310 v->pv.v1 = tc1[0][1];
311 if (tc1_size == 4)
312 v->pv.q1 = tc1[0][3];
313 else
314 v->pv.q1 = 1.0;
315 }
316 else {
317 v->v.u1 = tc1[0][0];
318 v->v.v1 = tc1[0][1];
319 }
320 #ifdef MACH64_PREMULT_TEXCOORDS
321 v->v.u1 *= v->v.w;
322 v->v.v1 *= v->v.w;
323 #endif
324 tc1 = (GLfloat (*)[4])((GLubyte *)tc1 + tc1_stride);
325 }
326 else if (DO_PTEX) {
327 *(GLuint *)&v->pv.q1 = 0; /* avoid culling on radeon */
328 }
329 if (DO_TEX2) {
330 if (DO_PTEX) {
331 v->pv.u2 = tc2[0][0];
332 v->pv.v2 = tc2[0][1];
333 if (tc2_size == 4)
334 v->pv.q2 = tc2[0][3];
335 else
336 v->pv.q2 = 1.0;
337 }
338 else {
339 v->v.u2 = tc2[0][0];
340 v->v.v2 = tc2[0][1];
341 }
342 tc2 = (GLfloat (*)[4])((GLubyte *)tc2 + tc2_stride);
343 }
344 if (DO_TEX3) {
345 if (DO_PTEX) {
346 v->pv.u3 = tc3[0][0];
347 v->pv.v3 = tc3[0][1];
348 if (tc3_size == 4)
349 v->pv.q3 = tc3[0][3];
350 else
351 v->pv.q3 = 1.0;
352 }
353 else {
354 v->v.u3 = tc3[0][0];
355 v->v.v3 = tc3[0][1];
356 }
357 tc3 = (GLfloat (*)[4])((GLubyte *)tc3 + tc3_stride);
358 }
359 }
360 }
361
362 #else
363 #if DO_XYZW
364
365 #if HAVE_HW_DIVIDE
366 #error "cannot use tiny vertices with hw perspective divide"
367 #endif
368
369 static void TAG(emit)( GLcontext *ctx, GLuint start, GLuint end,
370 void *dest, GLuint stride )
371 {
372 LOCALVARS
373 struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
374 GLfloat (*col)[4];
375 GLuint col_stride;
376 GLfloat (*coord)[4] = VB->NdcPtr->data;
377 GLuint coord_stride = VB->NdcPtr->stride;
378 GLfloat *v = (GLfloat *)dest;
379 const GLubyte *mask = VB->ClipMask;
380 const GLfloat *s = GET_VIEWPORT_MAT();
381 int i;
382
383 (void) s;
384
385 ASSERT(stride == 4);
386
387 col = VB->ColorPtr[0]->data;
388 col_stride = VB->ColorPtr[0]->stride;
389
390 /* Pack what's left into a 4-dword vertex. Color is in a different
391 * place, and there is no 'w' coordinate.
392 */
393 if (start) {
394 coord = (GLfloat (*)[4])((GLubyte *)coord + start * coord_stride);
395 STRIDE_4F(col, start * col_stride);
396 }
397
398 for (i=start; i < end; i++, v+=4) {
399 if (HAVE_HW_VIEWPORT || mask[i] == 0) {
400 VIEWPORT_X(v[0], coord[0][0]);
401 VIEWPORT_Y(v[1], coord[0][1]);
402 VIEWPORT_Z(v[2], coord[0][2]);
403 }
404 coord = (GLfloat (*)[4])((GLubyte *)coord + coord_stride);
405 if (DO_RGBA) {
406 if (HAVE_RGBA_COLOR) {
407 *(GLuint *)&v[3] = *(GLuint *)col;
408 }
409 else {
410 GLubyte *b = (GLubyte *)&v[3];
411 UNCLAMPED_FLOAT_TO_UBYTE(b[0], col[0][2]);
412 UNCLAMPED_FLOAT_TO_UBYTE(b[1], col[0][1]);
413 UNCLAMPED_FLOAT_TO_UBYTE(b[2], col[0][0]);
414 UNCLAMPED_FLOAT_TO_UBYTE(b[3], col[0][3]);
415 }
416 STRIDE_4F( col, col_stride );
417 }
418 if (MACH64_DEBUG & DEBUG_VERBOSE_PRIMS) {
419 fprintf(stderr, "vert (importable) %d: %.2f %.2f %.2f %x\n",
420 i, v[0], v[1], v[2], *(int *)&v[3]);
421 }
422 }
423 }
424 #else
425 static void TAG(emit)( GLcontext *ctx, GLuint start, GLuint end,
426 void *dest, GLuint stride )
427 {
428 LOCALVARS
429 struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
430 GLfloat (*col)[4];
431 GLuint col_stride;
432 GLfloat *v = (GLfloat *)dest;
433 int i;
434
435 col = VB->ColorPtr[0]->data;
436 col_stride = VB->ColorPtr[0]->stride;
437
438 if (start)
439 STRIDE_4F(col, col_stride * start);
440
441 /* Need to figure out where color is:
442 */
443 if (GET_VERTEX_FORMAT() == TINY_VERTEX_FORMAT)
444 v += 3;
445 else
446 v += 4;
447
448 for (i=start; i < end; i++, STRIDE_F(v, stride)) {
449 if (HAVE_RGBA_COLOR) {
450 *(GLuint *)v = *(GLuint *)col[0];
451 }
452 else {
453 GLubyte *b = (GLubyte *)v;
454 UNCLAMPED_FLOAT_TO_UBYTE(b[0], col[0][2]);
455 UNCLAMPED_FLOAT_TO_UBYTE(b[1], col[0][1]);
456 UNCLAMPED_FLOAT_TO_UBYTE(b[2], col[0][0]);
457 UNCLAMPED_FLOAT_TO_UBYTE(b[3], col[0][3]);
458 }
459 STRIDE_4F( col, col_stride );
460 }
461 }
462 #endif /* emit */
463 #endif /* emit */
464
465 #if (DO_XYZW) && (DO_RGBA)
466
467
468 #if (HAVE_PTEX_VERTICES)
469 static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
470 {
471 LOCALVARS
472 struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
473
474 /* Force 'missing' texcoords to something valid.
475 */
476 if (DO_TEX3 && VB->TexCoordPtr[2] == 0)
477 VB->TexCoordPtr[2] = VB->TexCoordPtr[3];
478
479 if (DO_TEX2 && VB->TexCoordPtr[1] == 0)
480 VB->TexCoordPtr[1] = VB->TexCoordPtr[2];
481
482 if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
483 VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
484
485 if (DO_PTEX)
486 return GL_TRUE;
487
488 if ((DO_TEX3 && VB->TexCoordPtr[GET_TEXSOURCE(3)]->size == 4) ||
489 (DO_TEX2 && VB->TexCoordPtr[GET_TEXSOURCE(2)]->size == 4) ||
490 (DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4) ||
491 (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4))
492 return GL_FALSE;
493
494 return GL_TRUE;
495 }
496 #else
497 static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
498 {
499 LOCALVARS
500 struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
501
502 /* Force 'missing' texcoords to something valid.
503 */
504 if (DO_TEX3 && VB->TexCoordPtr[2] == 0)
505 VB->TexCoordPtr[2] = VB->TexCoordPtr[3];
506
507 if (DO_TEX2 && VB->TexCoordPtr[1] == 0)
508 VB->TexCoordPtr[1] = VB->TexCoordPtr[2];
509
510 if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
511 VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
512
513 if (DO_PTEX)
514 return GL_TRUE;
515
516 /* No hardware support for projective texture. Can fake it for
517 * TEX0 only.
518 */
519 if ((DO_TEX3 && VB->TexCoordPtr[GET_TEXSOURCE(3)]->size == 4) ||
520 (DO_TEX2 && VB->TexCoordPtr[GET_TEXSOURCE(2)]->size == 4) ||
521 (DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4)) {
522 PTEX_FALLBACK();
523 return GL_FALSE;
524 }
525
526 if (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4) {
527 if (DO_TEX1 || DO_TEX2 || DO_TEX3) {
528 PTEX_FALLBACK();
529 }
530 return GL_FALSE;
531 }
532
533 return GL_TRUE;
534 }
535 #endif /* ptex */
536
537
538 static void TAG(interp)( GLcontext *ctx,
539 GLfloat t,
540 GLuint edst, GLuint eout, GLuint ein,
541 GLboolean force_boundary )
542 {
543 LOCALVARS
544 struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
545 GLubyte *ddverts = GET_VERTEX_STORE();
546 GLuint size = GET_VERTEX_SIZE();
547 const GLfloat *dstclip = VB->ClipPtr->data[edst];
548 GLfloat w;
549 const GLfloat *s = GET_VIEWPORT_MAT();
550
551 VERTEX *dst = (VERTEX *)(ddverts + (edst * size));
552 VERTEX *in = (VERTEX *)(ddverts + (ein * size));
553 VERTEX *out = (VERTEX *)(ddverts + (eout * size));
554
555 (void)s;
556
557 if (HAVE_HW_DIVIDE && CHECK_HW_DIVIDE) {
558 VIEWPORT_X( dst->v.x, dstclip[0] );
559 VIEWPORT_Y( dst->v.y, dstclip[1] );
560 VIEWPORT_Z( dst->v.z, dstclip[2] );
561 w = dstclip[3];
562 }
563 else {
564 w = (dstclip[3] == 0.0F) ? 1.0 : (1.0 / dstclip[3]);
565 VIEWPORT_X( dst->v.x, dstclip[0] * w );
566 VIEWPORT_Y( dst->v.y, dstclip[1] * w );
567 VIEWPORT_Z( dst->v.z, dstclip[2] * w );
568 }
569
570 if (MACH64_DEBUG & DEBUG_VERBOSE_PRIMS) {
571 fprintf( stderr, "%s: dst vert: %.2f %.2f %.2f %f\n",
572 __FUNCTION__,
573 dst->v.x,
574 dst->v.y,
575 dst->v.z,
576 w );
577 }
578
579 if ((HAVE_HW_DIVIDE && CHECK_HW_DIVIDE) ||
580 DO_FOG || DO_SPEC || DO_TEX0 || DO_TEX1 ||
581 DO_TEX2 || DO_TEX3 || !HAVE_TINY_VERTICES) {
582
583 dst->v.w = w;
584
585 INTERP_UB( t, dst->ub4[4][0], out->ub4[4][0], in->ub4[4][0] );
586 INTERP_UB( t, dst->ub4[4][1], out->ub4[4][1], in->ub4[4][1] );
587 INTERP_UB( t, dst->ub4[4][2], out->ub4[4][2], in->ub4[4][2] );
588 INTERP_UB( t, dst->ub4[4][3], out->ub4[4][3], in->ub4[4][3] );
589
590 if (DO_SPEC) {
591 INTERP_UB( t, dst->ub4[5][0], out->ub4[5][0], in->ub4[5][0] );
592 INTERP_UB( t, dst->ub4[5][1], out->ub4[5][1], in->ub4[5][1] );
593 INTERP_UB( t, dst->ub4[5][2], out->ub4[5][2], in->ub4[5][2] );
594 }
595 if (DO_FOG) {
596 INTERP_UB( t, dst->ub4[5][3], out->ub4[5][3], in->ub4[5][3] );
597 }
598 if (DO_TEX0) {
599 if (DO_PTEX) {
600 if (HAVE_PTEX_VERTICES) {
601 INTERP_F( t, dst->pv.u0, out->pv.u0, in->pv.u0 );
602 INTERP_F( t, dst->pv.v0, out->pv.v0, in->pv.v0 );
603 INTERP_F( t, dst->pv.q0, out->pv.q0, in->pv.q0 );
604 } else {
605 GLfloat wout = VB->NdcPtr->data[eout][3];
606 GLfloat win = VB->NdcPtr->data[ein][3];
607 GLfloat qout = out->pv.w / wout;
608 GLfloat qin = in->pv.w / win;
609 GLfloat qdst, rqdst;
610
611 ASSERT( !HAVE_HW_DIVIDE );
612
613 INTERP_F( t, dst->v.u0, out->v.u0 * qout, in->v.u0 * qin );
614 INTERP_F( t, dst->v.v0, out->v.v0 * qout, in->v.v0 * qin );
615 INTERP_F( t, qdst, qout, qin );
616
617 rqdst = 1.0 / qdst;
618 dst->v.u0 *= rqdst;
619 dst->v.v0 *= rqdst;
620 dst->v.w *= rqdst;
621 }
622 }
623 else {
624 #ifdef MACH64_PREMULT_TEXCOORDS
625 GLfloat qout = 1 / out->v.w;
626 GLfloat qin = 1 / in->v.w;
627
628 INTERP_F( t, dst->v.u0, out->v.u0 * qout, in->v.u0 * qin);
629 INTERP_F( t, dst->v.v0, out->v.v0 * qout, in->v.v0 * qin);
630
631 dst->v.u0 *= w;
632 dst->v.v0 *= w;
633 #else
634 INTERP_F( t, dst->v.u0, out->v.u0, in->v.u0 );
635 INTERP_F( t, dst->v.v0, out->v.v0, in->v.v0 );
636 #endif
637 }
638 }
639 if (DO_TEX1) {
640 if (DO_PTEX) {
641 INTERP_F( t, dst->pv.u1, out->pv.u1, in->pv.u1 );
642 INTERP_F( t, dst->pv.v1, out->pv.v1, in->pv.v1 );
643 INTERP_F( t, dst->pv.q1, out->pv.q1, in->pv.q1 );
644 } else {
645 #ifdef MACH64_PREMULT_TEXCOORDS
646 GLfloat qout = 1 / out->v.w;
647 GLfloat qin = 1 / in->v.w;
648
649 INTERP_F( t, dst->v.u1, out->v.u1 * qout, in->v.u1 * qin );
650 INTERP_F( t, dst->v.v1, out->v.v1 * qout, in->v.v1 * qin );
651
652 dst->v.u1 *= w;
653 dst->v.v1 *= w;
654 #else
655 INTERP_F( t, dst->v.u1, out->v.u1, in->v.u1 );
656 INTERP_F( t, dst->v.v1, out->v.v1, in->v.v1 );
657 #endif
658 }
659 }
660 else if (DO_PTEX) {
661 dst->pv.q0 = 0.0; /* must be a valid float on radeon */
662 }
663 if (DO_TEX2) {
664 if (DO_PTEX) {
665 INTERP_F( t, dst->pv.u2, out->pv.u2, in->pv.u2 );
666 INTERP_F( t, dst->pv.v2, out->pv.v2, in->pv.v2 );
667 INTERP_F( t, dst->pv.q2, out->pv.q2, in->pv.q2 );
668 } else {
669 INTERP_F( t, dst->v.u2, out->v.u2, in->v.u2 );
670 INTERP_F( t, dst->v.v2, out->v.v2, in->v.v2 );
671 }
672 }
673 if (DO_TEX3) {
674 if (DO_PTEX) {
675 INTERP_F( t, dst->pv.u3, out->pv.u3, in->pv.u3 );
676 INTERP_F( t, dst->pv.v3, out->pv.v3, in->pv.v3 );
677 INTERP_F( t, dst->pv.q3, out->pv.q3, in->pv.q3 );
678 } else {
679 INTERP_F( t, dst->v.u3, out->v.u3, in->v.u3 );
680 INTERP_F( t, dst->v.v3, out->v.v3, in->v.v3 );
681 }
682 }
683 } else {
684 /* 4-dword vertex. Color is in v[3] and there is no oow coordinate.
685 */
686 INTERP_UB( t, dst->ub4[3][0], out->ub4[3][0], in->ub4[3][0] );
687 INTERP_UB( t, dst->ub4[3][1], out->ub4[3][1], in->ub4[3][1] );
688 INTERP_UB( t, dst->ub4[3][2], out->ub4[3][2], in->ub4[3][2] );
689 INTERP_UB( t, dst->ub4[3][3], out->ub4[3][3], in->ub4[3][3] );
690 }
691 }
692
693 #endif /* rgba && xyzw */
694
695
696 static void TAG(init)( void )
697 {
698 setup_tab[IND].emit = TAG(emit);
699
700 #if (DO_XYZW && DO_RGBA)
701 setup_tab[IND].check_tex_sizes = TAG(check_tex_sizes);
702 setup_tab[IND].interp = TAG(interp);
703 #endif
704
705 if (DO_SPEC)
706 setup_tab[IND].copy_pv = copy_pv_rgba4_spec5;
707 else if (HAVE_HW_DIVIDE || DO_SPEC || DO_FOG || DO_TEX0 || DO_TEX1 ||
708 DO_TEX2 || DO_TEX3 || !HAVE_TINY_VERTICES)
709 setup_tab[IND].copy_pv = copy_pv_rgba4;
710 else
711 setup_tab[IND].copy_pv = copy_pv_rgba3;
712
713 if (DO_TEX3) {
714 if (DO_PTEX) {
715 ASSERT(HAVE_PTEX_VERTICES);
716 setup_tab[IND].vertex_format = PROJ_TEX3_VERTEX_FORMAT;
717 setup_tab[IND].vertex_size = 18;
718 }
719 else {
720 setup_tab[IND].vertex_format = TEX3_VERTEX_FORMAT;
721 setup_tab[IND].vertex_size = 14;
722 }
723 }
724 else if (DO_TEX2) {
725 if (DO_PTEX) {
726 ASSERT(HAVE_PTEX_VERTICES);
727 setup_tab[IND].vertex_format = PROJ_TEX3_VERTEX_FORMAT;
728 setup_tab[IND].vertex_size = 18;
729 }
730 else {
731 setup_tab[IND].vertex_format = TEX2_VERTEX_FORMAT;
732 setup_tab[IND].vertex_size = 12;
733 }
734 }
735 else if (DO_TEX1) {
736 if (DO_PTEX) {
737 ASSERT(HAVE_PTEX_VERTICES);
738 setup_tab[IND].vertex_format = PROJ_TEX1_VERTEX_FORMAT;
739 setup_tab[IND].vertex_size = 12;
740 }
741 else {
742 setup_tab[IND].vertex_format = TEX1_VERTEX_FORMAT;
743 setup_tab[IND].vertex_size = 10;
744 }
745 }
746 else if (DO_TEX0) {
747 if (DO_PTEX && HAVE_PTEX_VERTICES) {
748 setup_tab[IND].vertex_format = PROJ_TEX1_VERTEX_FORMAT;
749 setup_tab[IND].vertex_size = 12;
750 } else {
751 setup_tab[IND].vertex_format = TEX0_VERTEX_FORMAT;
752 setup_tab[IND].vertex_size = 8;
753 }
754 }
755 else if (!HAVE_HW_DIVIDE && !DO_SPEC && !DO_FOG && HAVE_TINY_VERTICES) {
756 setup_tab[IND].vertex_format = TINY_VERTEX_FORMAT;
757 setup_tab[IND].vertex_size = 4;
758 } else if (HAVE_NOTEX_VERTICES) {
759 setup_tab[IND].vertex_format = NOTEX_VERTEX_FORMAT;
760 setup_tab[IND].vertex_size = 6;
761 } else {
762 setup_tab[IND].vertex_format = TEX0_VERTEX_FORMAT;
763 setup_tab[IND].vertex_size = 8;
764 }
765
766 }
767
768
769 #undef IND
770 #undef TAG