1 /**************************************************************************
2 *
3 * Copyright 2007 VMware, Inc.
4 * All Rights Reserved.
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a
7 * copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sub license, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
13 *
14 * The above copyright notice and this permission notice (including the
15 * next paragraph) shall be included in all copies or substantial portions
16 * 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
20 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
21 * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
22 * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
23 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
24 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
25 *
26 **************************************************************************/
27
28 /*
29 * Authors:
30 * Keith Whitwell <keithw@vmware.com>
31 */
32
33 #include "util/u_memory.h"
34 #include "util/u_format.h"
35 #include "util/u_half.h"
36 #include "util/u_math.h"
37 #include "pipe/p_state.h"
38 #include "translate.h"
39
40
41 #define DRAW_DBG 0
42
43 typedef void (*fetch_func)(void *dst,
44 const uint8_t *src,
45 unsigned i, unsigned j);
46 typedef void (*emit_func)(const void *attrib, void *ptr);
47
48
49
50 struct translate_generic {
51 struct translate translate;
52
53 struct {
54 enum translate_element_type type;
55
56 fetch_func fetch;
57 unsigned buffer;
58 unsigned input_offset;
59 unsigned instance_divisor;
60
61 emit_func emit;
62 unsigned output_offset;
63
64 const uint8_t *input_ptr;
65 unsigned input_stride;
66 unsigned max_index;
67
68 /* this value is set to -1 if this is a normal element with output_format != input_format:
69 * in this case, u_format is used to do a full conversion
70 *
71 * this value is set to the format size in bytes if output_format == input_format or for 32-bit instance ids:
72 * in this case, memcpy is used to copy this amount of bytes
73 */
74 int copy_size;
75
76 } attrib[TRANSLATE_MAX_ATTRIBS];
77
78 unsigned nr_attrib;
79 };
80
81
translate_generic(struct translate * translate)82 static struct translate_generic *translate_generic( struct translate *translate )
83 {
84 return (struct translate_generic *)translate;
85 }
86
87 /**
88 * Fetch a dword[4] vertex attribute from memory, doing format/type
89 * conversion as needed.
90 *
91 * This is probably needed/dupliocated elsewhere, eg format
92 * conversion, texture sampling etc.
93 */
94 #define ATTRIB( NAME, SZ, SRCTYPE, DSTTYPE, TO ) \
95 static void \
96 emit_##NAME(const void *attrib, void *ptr) \
97 { \
98 unsigned i; \
99 SRCTYPE *in = (SRCTYPE *)attrib; \
100 DSTTYPE *out = (DSTTYPE *)ptr; \
101 \
102 for (i = 0; i < SZ; i++) { \
103 out[i] = TO(in[i]); \
104 } \
105 }
106
107
108 #define TO_64_FLOAT(x) ((double) x)
109 #define TO_32_FLOAT(x) (x)
110 #define TO_16_FLOAT(x) util_float_to_half(x)
111
112 #define TO_8_USCALED(x) ((unsigned char) x)
113 #define TO_16_USCALED(x) ((unsigned short) x)
114 #define TO_32_USCALED(x) ((unsigned int) x)
115
116 #define TO_8_SSCALED(x) ((char) x)
117 #define TO_16_SSCALED(x) ((short) x)
118 #define TO_32_SSCALED(x) ((int) x)
119
120 #define TO_8_UNORM(x) ((unsigned char) (x * 255.0f))
121 #define TO_16_UNORM(x) ((unsigned short) (x * 65535.0f))
122 #define TO_32_UNORM(x) ((unsigned int) (x * 4294967295.0f))
123
124 #define TO_8_SNORM(x) ((char) (x * 127.0f))
125 #define TO_16_SNORM(x) ((short) (x * 32767.0f))
126 #define TO_32_SNORM(x) ((int) (x * 2147483647.0f))
127
128 #define TO_32_FIXED(x) ((int) (x * 65536.0f))
129
130 #define TO_INT(x) (x)
131
132
133 ATTRIB( R64G64B64A64_FLOAT, 4, float, double, TO_64_FLOAT )
134 ATTRIB( R64G64B64_FLOAT, 3, float, double, TO_64_FLOAT )
135 ATTRIB( R64G64_FLOAT, 2, float, double, TO_64_FLOAT )
136 ATTRIB( R64_FLOAT, 1, float, double, TO_64_FLOAT )
137
138 ATTRIB( R32G32B32A32_FLOAT, 4, float, float, TO_32_FLOAT )
139 ATTRIB( R32G32B32_FLOAT, 3, float, float, TO_32_FLOAT )
140 ATTRIB( R32G32_FLOAT, 2, float, float, TO_32_FLOAT )
141 ATTRIB( R32_FLOAT, 1, float, float, TO_32_FLOAT )
142
143 ATTRIB( R16G16B16A16_FLOAT, 4, float, ushort, TO_16_FLOAT )
144 ATTRIB( R16G16B16_FLOAT, 3, float, ushort, TO_16_FLOAT )
145 ATTRIB( R16G16_FLOAT, 2, float, ushort, TO_16_FLOAT )
146 ATTRIB( R16_FLOAT, 1, float, ushort, TO_16_FLOAT )
147
148 ATTRIB( R32G32B32A32_USCALED, 4, float, unsigned, TO_32_USCALED )
149 ATTRIB( R32G32B32_USCALED, 3, float, unsigned, TO_32_USCALED )
150 ATTRIB( R32G32_USCALED, 2, float, unsigned, TO_32_USCALED )
151 ATTRIB( R32_USCALED, 1, float, unsigned, TO_32_USCALED )
152
153 ATTRIB( R32G32B32A32_SSCALED, 4, float, int, TO_32_SSCALED )
154 ATTRIB( R32G32B32_SSCALED, 3, float, int, TO_32_SSCALED )
155 ATTRIB( R32G32_SSCALED, 2, float, int, TO_32_SSCALED )
156 ATTRIB( R32_SSCALED, 1, float, int, TO_32_SSCALED )
157
158 ATTRIB( R32G32B32A32_UNORM, 4, float, unsigned, TO_32_UNORM )
159 ATTRIB( R32G32B32_UNORM, 3, float, unsigned, TO_32_UNORM )
160 ATTRIB( R32G32_UNORM, 2, float, unsigned, TO_32_UNORM )
161 ATTRIB( R32_UNORM, 1, float, unsigned, TO_32_UNORM )
162
163 ATTRIB( R32G32B32A32_SNORM, 4, float, int, TO_32_SNORM )
164 ATTRIB( R32G32B32_SNORM, 3, float, int, TO_32_SNORM )
165 ATTRIB( R32G32_SNORM, 2, float, int, TO_32_SNORM )
166 ATTRIB( R32_SNORM, 1, float, int, TO_32_SNORM )
167
168 ATTRIB( R16G16B16A16_USCALED, 4, float, ushort, TO_16_USCALED )
169 ATTRIB( R16G16B16_USCALED, 3, float, ushort, TO_16_USCALED )
170 ATTRIB( R16G16_USCALED, 2, float, ushort, TO_16_USCALED )
171 ATTRIB( R16_USCALED, 1, float, ushort, TO_16_USCALED )
172
173 ATTRIB( R16G16B16A16_SSCALED, 4, float, short, TO_16_SSCALED )
174 ATTRIB( R16G16B16_SSCALED, 3, float, short, TO_16_SSCALED )
175 ATTRIB( R16G16_SSCALED, 2, float, short, TO_16_SSCALED )
176 ATTRIB( R16_SSCALED, 1, float, short, TO_16_SSCALED )
177
178 ATTRIB( R16G16B16A16_UNORM, 4, float, ushort, TO_16_UNORM )
179 ATTRIB( R16G16B16_UNORM, 3, float, ushort, TO_16_UNORM )
180 ATTRIB( R16G16_UNORM, 2, float, ushort, TO_16_UNORM )
181 ATTRIB( R16_UNORM, 1, float, ushort, TO_16_UNORM )
182
183 ATTRIB( R16G16B16A16_SNORM, 4, float, short, TO_16_SNORM )
184 ATTRIB( R16G16B16_SNORM, 3, float, short, TO_16_SNORM )
185 ATTRIB( R16G16_SNORM, 2, float, short, TO_16_SNORM )
186 ATTRIB( R16_SNORM, 1, float, short, TO_16_SNORM )
187
188 ATTRIB( R8G8B8A8_USCALED, 4, float, ubyte, TO_8_USCALED )
189 ATTRIB( R8G8B8_USCALED, 3, float, ubyte, TO_8_USCALED )
190 ATTRIB( R8G8_USCALED, 2, float, ubyte, TO_8_USCALED )
191 ATTRIB( R8_USCALED, 1, float, ubyte, TO_8_USCALED )
192
193 ATTRIB( R8G8B8A8_SSCALED, 4, float, char, TO_8_SSCALED )
194 ATTRIB( R8G8B8_SSCALED, 3, float, char, TO_8_SSCALED )
195 ATTRIB( R8G8_SSCALED, 2, float, char, TO_8_SSCALED )
196 ATTRIB( R8_SSCALED, 1, float, char, TO_8_SSCALED )
197
198 ATTRIB( R8G8B8A8_UNORM, 4, float, ubyte, TO_8_UNORM )
199 ATTRIB( R8G8B8_UNORM, 3, float, ubyte, TO_8_UNORM )
200 ATTRIB( R8G8_UNORM, 2, float, ubyte, TO_8_UNORM )
201 ATTRIB( R8_UNORM, 1, float, ubyte, TO_8_UNORM )
202
203 ATTRIB( R8G8B8A8_SNORM, 4, float, char, TO_8_SNORM )
204 ATTRIB( R8G8B8_SNORM, 3, float, char, TO_8_SNORM )
205 ATTRIB( R8G8_SNORM, 2, float, char, TO_8_SNORM )
206 ATTRIB( R8_SNORM, 1, float, char, TO_8_SNORM )
207
208 ATTRIB( R32G32B32A32_UINT, 4, uint32_t, unsigned, TO_INT )
209 ATTRIB( R32G32B32_UINT, 3, uint32_t, unsigned, TO_INT )
210 ATTRIB( R32G32_UINT, 2, uint32_t, unsigned, TO_INT )
211 ATTRIB( R32_UINT, 1, uint32_t, unsigned, TO_INT )
212
213 ATTRIB( R16G16B16A16_UINT, 4, uint32_t, ushort, TO_INT )
214 ATTRIB( R16G16B16_UINT, 3, uint32_t, ushort, TO_INT )
215 ATTRIB( R16G16_UINT, 2, uint32_t, ushort, TO_INT )
216 ATTRIB( R16_UINT, 1, uint32_t, ushort, TO_INT )
217
218 ATTRIB( R8G8B8A8_UINT, 4, uint32_t, ubyte, TO_INT )
219 ATTRIB( R8G8B8_UINT, 3, uint32_t, ubyte, TO_INT )
220 ATTRIB( R8G8_UINT, 2, uint32_t, ubyte, TO_INT )
221 ATTRIB( R8_UINT, 1, uint32_t, ubyte, TO_INT )
222
223 ATTRIB( R32G32B32A32_SINT, 4, int32_t, int, TO_INT )
224 ATTRIB( R32G32B32_SINT, 3, int32_t, int, TO_INT )
225 ATTRIB( R32G32_SINT, 2, int32_t, int, TO_INT )
226 ATTRIB( R32_SINT, 1, int32_t, int, TO_INT )
227
228 ATTRIB( R16G16B16A16_SINT, 4, int32_t, short, TO_INT )
229 ATTRIB( R16G16B16_SINT, 3, int32_t, short, TO_INT )
230 ATTRIB( R16G16_SINT, 2, int32_t, short, TO_INT )
231 ATTRIB( R16_SINT, 1, int32_t, short, TO_INT )
232
233 ATTRIB( R8G8B8A8_SINT, 4, int32_t, char, TO_INT )
234 ATTRIB( R8G8B8_SINT, 3, int32_t, char, TO_INT )
235 ATTRIB( R8G8_SINT, 2, int32_t, char, TO_INT )
236 ATTRIB( R8_SINT, 1, int32_t, char, TO_INT )
237
238 static void
emit_A8R8G8B8_UNORM(const void * attrib,void * ptr)239 emit_A8R8G8B8_UNORM( const void *attrib, void *ptr)
240 {
241 float *in = (float *)attrib;
242 ubyte *out = (ubyte *)ptr;
243 out[0] = TO_8_UNORM(in[3]);
244 out[1] = TO_8_UNORM(in[0]);
245 out[2] = TO_8_UNORM(in[1]);
246 out[3] = TO_8_UNORM(in[2]);
247 }
248
249 static void
emit_B8G8R8A8_UNORM(const void * attrib,void * ptr)250 emit_B8G8R8A8_UNORM( const void *attrib, void *ptr)
251 {
252 float *in = (float *)attrib;
253 ubyte *out = (ubyte *)ptr;
254 out[2] = TO_8_UNORM(in[0]);
255 out[1] = TO_8_UNORM(in[1]);
256 out[0] = TO_8_UNORM(in[2]);
257 out[3] = TO_8_UNORM(in[3]);
258 }
259
260 static void
emit_B10G10R10A2_UNORM(const void * attrib,void * ptr)261 emit_B10G10R10A2_UNORM( const void *attrib, void *ptr )
262 {
263 float *src = (float *)ptr;
264 uint32_t value = 0;
265 value |= ((uint32_t)(CLAMP(src[2], 0, 1) * 0x3ff)) & 0x3ff;
266 value |= (((uint32_t)(CLAMP(src[1], 0, 1) * 0x3ff)) & 0x3ff) << 10;
267 value |= (((uint32_t)(CLAMP(src[0], 0, 1) * 0x3ff)) & 0x3ff) << 20;
268 value |= ((uint32_t)(CLAMP(src[3], 0, 1) * 0x3)) << 30;
269 *(uint32_t *)attrib = util_le32_to_cpu(value);
270 }
271
272 static void
emit_B10G10R10A2_USCALED(const void * attrib,void * ptr)273 emit_B10G10R10A2_USCALED( const void *attrib, void *ptr )
274 {
275 float *src = (float *)ptr;
276 uint32_t value = 0;
277 value |= ((uint32_t)CLAMP(src[2], 0, 1023)) & 0x3ff;
278 value |= (((uint32_t)CLAMP(src[1], 0, 1023)) & 0x3ff) << 10;
279 value |= (((uint32_t)CLAMP(src[0], 0, 1023)) & 0x3ff) << 20;
280 value |= ((uint32_t)CLAMP(src[3], 0, 3)) << 30;
281 *(uint32_t *)attrib = util_le32_to_cpu(value);
282 }
283
284 static void
emit_B10G10R10A2_SNORM(const void * attrib,void * ptr)285 emit_B10G10R10A2_SNORM( const void *attrib, void *ptr )
286 {
287 float *src = (float *)ptr;
288 uint32_t value = 0;
289 value |= (uint32_t)(((uint32_t)(CLAMP(src[2], -1, 1) * 0x1ff)) & 0x3ff) ;
290 value |= (uint32_t)((((uint32_t)(CLAMP(src[1], -1, 1) * 0x1ff)) & 0x3ff) << 10) ;
291 value |= (uint32_t)((((uint32_t)(CLAMP(src[0], -1, 1) * 0x1ff)) & 0x3ff) << 20) ;
292 value |= (uint32_t)(((uint32_t)(CLAMP(src[3], -1, 1) * 0x1)) << 30) ;
293 *(uint32_t *)attrib = util_le32_to_cpu(value);
294 }
295
296 static void
emit_B10G10R10A2_SSCALED(const void * attrib,void * ptr)297 emit_B10G10R10A2_SSCALED( const void *attrib, void *ptr )
298 {
299 float *src = (float *)ptr;
300 uint32_t value = 0;
301 value |= (uint32_t)(((uint32_t)CLAMP(src[2], -512, 511)) & 0x3ff) ;
302 value |= (uint32_t)((((uint32_t)CLAMP(src[1], -512, 511)) & 0x3ff) << 10) ;
303 value |= (uint32_t)((((uint32_t)CLAMP(src[0], -512, 511)) & 0x3ff) << 20) ;
304 value |= (uint32_t)(((uint32_t)CLAMP(src[3], -2, 1)) << 30) ;
305 *(uint32_t *)attrib = util_le32_to_cpu(value);
306 }
307
308 static void
emit_R10G10B10A2_UNORM(const void * attrib,void * ptr)309 emit_R10G10B10A2_UNORM( const void *attrib, void *ptr )
310 {
311 float *src = (float *)ptr;
312 uint32_t value = 0;
313 value |= ((uint32_t)(CLAMP(src[0], 0, 1) * 0x3ff)) & 0x3ff;
314 value |= (((uint32_t)(CLAMP(src[1], 0, 1) * 0x3ff)) & 0x3ff) << 10;
315 value |= (((uint32_t)(CLAMP(src[2], 0, 1) * 0x3ff)) & 0x3ff) << 20;
316 value |= ((uint32_t)(CLAMP(src[3], 0, 1) * 0x3)) << 30;
317 *(uint32_t *)attrib = util_le32_to_cpu(value);
318 }
319
320 static void
emit_R10G10B10A2_USCALED(const void * attrib,void * ptr)321 emit_R10G10B10A2_USCALED( const void *attrib, void *ptr )
322 {
323 float *src = (float *)ptr;
324 uint32_t value = 0;
325 value |= ((uint32_t)CLAMP(src[0], 0, 1023)) & 0x3ff;
326 value |= (((uint32_t)CLAMP(src[1], 0, 1023)) & 0x3ff) << 10;
327 value |= (((uint32_t)CLAMP(src[2], 0, 1023)) & 0x3ff) << 20;
328 value |= ((uint32_t)CLAMP(src[3], 0, 3)) << 30;
329 *(uint32_t *)attrib = util_le32_to_cpu(value);
330 }
331
332 static void
emit_R10G10B10A2_SNORM(const void * attrib,void * ptr)333 emit_R10G10B10A2_SNORM( const void *attrib, void *ptr )
334 {
335 float *src = (float *)ptr;
336 uint32_t value = 0;
337 value |= (uint32_t)(((uint32_t)(CLAMP(src[0], -1, 1) * 0x1ff)) & 0x3ff) ;
338 value |= (uint32_t)((((uint32_t)(CLAMP(src[1], -1, 1) * 0x1ff)) & 0x3ff) << 10) ;
339 value |= (uint32_t)((((uint32_t)(CLAMP(src[2], -1, 1) * 0x1ff)) & 0x3ff) << 20) ;
340 value |= (uint32_t)(((uint32_t)(CLAMP(src[3], -1, 1) * 0x1)) << 30) ;
341 *(uint32_t *)attrib = util_le32_to_cpu(value);
342 }
343
344 static void
emit_R10G10B10A2_SSCALED(const void * attrib,void * ptr)345 emit_R10G10B10A2_SSCALED( const void *attrib, void *ptr)
346 {
347 float *src = (float *)ptr;
348 uint32_t value = 0;
349 value |= (uint32_t)(((uint32_t)CLAMP(src[0], -512, 511)) & 0x3ff) ;
350 value |= (uint32_t)((((uint32_t)CLAMP(src[1], -512, 511)) & 0x3ff) << 10) ;
351 value |= (uint32_t)((((uint32_t)CLAMP(src[2], -512, 511)) & 0x3ff) << 20) ;
352 value |= (uint32_t)(((uint32_t)CLAMP(src[3], -2, 1)) << 30) ;
353 *(uint32_t *)attrib = util_le32_to_cpu(value);
354 }
355
356 static void
emit_NULL(const void * attrib,void * ptr)357 emit_NULL( const void *attrib, void *ptr )
358 {
359 /* do nothing is the only sensible option */
360 }
361
get_emit_func(enum pipe_format format)362 static emit_func get_emit_func( enum pipe_format format )
363 {
364 switch (format) {
365 case PIPE_FORMAT_R64_FLOAT:
366 return &emit_R64_FLOAT;
367 case PIPE_FORMAT_R64G64_FLOAT:
368 return &emit_R64G64_FLOAT;
369 case PIPE_FORMAT_R64G64B64_FLOAT:
370 return &emit_R64G64B64_FLOAT;
371 case PIPE_FORMAT_R64G64B64A64_FLOAT:
372 return &emit_R64G64B64A64_FLOAT;
373
374 case PIPE_FORMAT_R32_FLOAT:
375 return &emit_R32_FLOAT;
376 case PIPE_FORMAT_R32G32_FLOAT:
377 return &emit_R32G32_FLOAT;
378 case PIPE_FORMAT_R32G32B32_FLOAT:
379 return &emit_R32G32B32_FLOAT;
380 case PIPE_FORMAT_R32G32B32A32_FLOAT:
381 return &emit_R32G32B32A32_FLOAT;
382
383 case PIPE_FORMAT_R16_FLOAT:
384 return &emit_R16_FLOAT;
385 case PIPE_FORMAT_R16G16_FLOAT:
386 return &emit_R16G16_FLOAT;
387 case PIPE_FORMAT_R16G16B16_FLOAT:
388 return &emit_R16G16B16_FLOAT;
389 case PIPE_FORMAT_R16G16B16A16_FLOAT:
390 return &emit_R16G16B16A16_FLOAT;
391
392 case PIPE_FORMAT_R32_UNORM:
393 return &emit_R32_UNORM;
394 case PIPE_FORMAT_R32G32_UNORM:
395 return &emit_R32G32_UNORM;
396 case PIPE_FORMAT_R32G32B32_UNORM:
397 return &emit_R32G32B32_UNORM;
398 case PIPE_FORMAT_R32G32B32A32_UNORM:
399 return &emit_R32G32B32A32_UNORM;
400
401 case PIPE_FORMAT_R32_USCALED:
402 return &emit_R32_USCALED;
403 case PIPE_FORMAT_R32G32_USCALED:
404 return &emit_R32G32_USCALED;
405 case PIPE_FORMAT_R32G32B32_USCALED:
406 return &emit_R32G32B32_USCALED;
407 case PIPE_FORMAT_R32G32B32A32_USCALED:
408 return &emit_R32G32B32A32_USCALED;
409
410 case PIPE_FORMAT_R32_SNORM:
411 return &emit_R32_SNORM;
412 case PIPE_FORMAT_R32G32_SNORM:
413 return &emit_R32G32_SNORM;
414 case PIPE_FORMAT_R32G32B32_SNORM:
415 return &emit_R32G32B32_SNORM;
416 case PIPE_FORMAT_R32G32B32A32_SNORM:
417 return &emit_R32G32B32A32_SNORM;
418
419 case PIPE_FORMAT_R32_SSCALED:
420 return &emit_R32_SSCALED;
421 case PIPE_FORMAT_R32G32_SSCALED:
422 return &emit_R32G32_SSCALED;
423 case PIPE_FORMAT_R32G32B32_SSCALED:
424 return &emit_R32G32B32_SSCALED;
425 case PIPE_FORMAT_R32G32B32A32_SSCALED:
426 return &emit_R32G32B32A32_SSCALED;
427
428 case PIPE_FORMAT_R16_UNORM:
429 return &emit_R16_UNORM;
430 case PIPE_FORMAT_R16G16_UNORM:
431 return &emit_R16G16_UNORM;
432 case PIPE_FORMAT_R16G16B16_UNORM:
433 return &emit_R16G16B16_UNORM;
434 case PIPE_FORMAT_R16G16B16A16_UNORM:
435 return &emit_R16G16B16A16_UNORM;
436
437 case PIPE_FORMAT_R16_USCALED:
438 return &emit_R16_USCALED;
439 case PIPE_FORMAT_R16G16_USCALED:
440 return &emit_R16G16_USCALED;
441 case PIPE_FORMAT_R16G16B16_USCALED:
442 return &emit_R16G16B16_USCALED;
443 case PIPE_FORMAT_R16G16B16A16_USCALED:
444 return &emit_R16G16B16A16_USCALED;
445
446 case PIPE_FORMAT_R16_SNORM:
447 return &emit_R16_SNORM;
448 case PIPE_FORMAT_R16G16_SNORM:
449 return &emit_R16G16_SNORM;
450 case PIPE_FORMAT_R16G16B16_SNORM:
451 return &emit_R16G16B16_SNORM;
452 case PIPE_FORMAT_R16G16B16A16_SNORM:
453 return &emit_R16G16B16A16_SNORM;
454
455 case PIPE_FORMAT_R16_SSCALED:
456 return &emit_R16_SSCALED;
457 case PIPE_FORMAT_R16G16_SSCALED:
458 return &emit_R16G16_SSCALED;
459 case PIPE_FORMAT_R16G16B16_SSCALED:
460 return &emit_R16G16B16_SSCALED;
461 case PIPE_FORMAT_R16G16B16A16_SSCALED:
462 return &emit_R16G16B16A16_SSCALED;
463
464 case PIPE_FORMAT_R8_UNORM:
465 return &emit_R8_UNORM;
466 case PIPE_FORMAT_R8G8_UNORM:
467 return &emit_R8G8_UNORM;
468 case PIPE_FORMAT_R8G8B8_UNORM:
469 return &emit_R8G8B8_UNORM;
470 case PIPE_FORMAT_R8G8B8A8_UNORM:
471 return &emit_R8G8B8A8_UNORM;
472
473 case PIPE_FORMAT_R8_USCALED:
474 return &emit_R8_USCALED;
475 case PIPE_FORMAT_R8G8_USCALED:
476 return &emit_R8G8_USCALED;
477 case PIPE_FORMAT_R8G8B8_USCALED:
478 return &emit_R8G8B8_USCALED;
479 case PIPE_FORMAT_R8G8B8A8_USCALED:
480 return &emit_R8G8B8A8_USCALED;
481
482 case PIPE_FORMAT_R8_SNORM:
483 return &emit_R8_SNORM;
484 case PIPE_FORMAT_R8G8_SNORM:
485 return &emit_R8G8_SNORM;
486 case PIPE_FORMAT_R8G8B8_SNORM:
487 return &emit_R8G8B8_SNORM;
488 case PIPE_FORMAT_R8G8B8A8_SNORM:
489 return &emit_R8G8B8A8_SNORM;
490
491 case PIPE_FORMAT_R8_SSCALED:
492 return &emit_R8_SSCALED;
493 case PIPE_FORMAT_R8G8_SSCALED:
494 return &emit_R8G8_SSCALED;
495 case PIPE_FORMAT_R8G8B8_SSCALED:
496 return &emit_R8G8B8_SSCALED;
497 case PIPE_FORMAT_R8G8B8A8_SSCALED:
498 return &emit_R8G8B8A8_SSCALED;
499
500 case PIPE_FORMAT_B8G8R8A8_UNORM:
501 return &emit_B8G8R8A8_UNORM;
502
503 case PIPE_FORMAT_A8R8G8B8_UNORM:
504 return &emit_A8R8G8B8_UNORM;
505
506 case PIPE_FORMAT_R32_UINT:
507 return &emit_R32_UINT;
508 case PIPE_FORMAT_R32G32_UINT:
509 return &emit_R32G32_UINT;
510 case PIPE_FORMAT_R32G32B32_UINT:
511 return &emit_R32G32B32_UINT;
512 case PIPE_FORMAT_R32G32B32A32_UINT:
513 return &emit_R32G32B32A32_UINT;
514
515 case PIPE_FORMAT_R16_UINT:
516 return &emit_R16_UINT;
517 case PIPE_FORMAT_R16G16_UINT:
518 return &emit_R16G16_UINT;
519 case PIPE_FORMAT_R16G16B16_UINT:
520 return &emit_R16G16B16_UINT;
521 case PIPE_FORMAT_R16G16B16A16_UINT:
522 return &emit_R16G16B16A16_UINT;
523
524 case PIPE_FORMAT_R8_UINT:
525 return &emit_R8_UINT;
526 case PIPE_FORMAT_R8G8_UINT:
527 return &emit_R8G8_UINT;
528 case PIPE_FORMAT_R8G8B8_UINT:
529 return &emit_R8G8B8_UINT;
530 case PIPE_FORMAT_R8G8B8A8_UINT:
531 return &emit_R8G8B8A8_UINT;
532
533 case PIPE_FORMAT_R32_SINT:
534 return &emit_R32_SINT;
535 case PIPE_FORMAT_R32G32_SINT:
536 return &emit_R32G32_SINT;
537 case PIPE_FORMAT_R32G32B32_SINT:
538 return &emit_R32G32B32_SINT;
539 case PIPE_FORMAT_R32G32B32A32_SINT:
540 return &emit_R32G32B32A32_SINT;
541
542 case PIPE_FORMAT_R16_SINT:
543 return &emit_R16_SINT;
544 case PIPE_FORMAT_R16G16_SINT:
545 return &emit_R16G16_SINT;
546 case PIPE_FORMAT_R16G16B16_SINT:
547 return &emit_R16G16B16_SINT;
548 case PIPE_FORMAT_R16G16B16A16_SINT:
549 return &emit_R16G16B16A16_SINT;
550
551 case PIPE_FORMAT_R8_SINT:
552 return &emit_R8_SINT;
553 case PIPE_FORMAT_R8G8_SINT:
554 return &emit_R8G8_SINT;
555 case PIPE_FORMAT_R8G8B8_SINT:
556 return &emit_R8G8B8_SINT;
557 case PIPE_FORMAT_R8G8B8A8_SINT:
558 return &emit_R8G8B8A8_SINT;
559
560 case PIPE_FORMAT_B10G10R10A2_UNORM:
561 return &emit_B10G10R10A2_UNORM;
562 case PIPE_FORMAT_B10G10R10A2_USCALED:
563 return &emit_B10G10R10A2_USCALED;
564 case PIPE_FORMAT_B10G10R10A2_SNORM:
565 return &emit_B10G10R10A2_SNORM;
566 case PIPE_FORMAT_B10G10R10A2_SSCALED:
567 return &emit_B10G10R10A2_SSCALED;
568
569 case PIPE_FORMAT_R10G10B10A2_UNORM:
570 return &emit_R10G10B10A2_UNORM;
571 case PIPE_FORMAT_R10G10B10A2_USCALED:
572 return &emit_R10G10B10A2_USCALED;
573 case PIPE_FORMAT_R10G10B10A2_SNORM:
574 return &emit_R10G10B10A2_SNORM;
575 case PIPE_FORMAT_R10G10B10A2_SSCALED:
576 return &emit_R10G10B10A2_SSCALED;
577
578 default:
579 assert(0);
580 return &emit_NULL;
581 }
582 }
583
generic_run_one(struct translate_generic * tg,unsigned elt,unsigned start_instance,unsigned instance_id,void * vert)584 static ALWAYS_INLINE void PIPE_CDECL generic_run_one( struct translate_generic *tg,
585 unsigned elt,
586 unsigned start_instance,
587 unsigned instance_id,
588 void *vert )
589 {
590 unsigned nr_attrs = tg->nr_attrib;
591 unsigned attr;
592
593 for (attr = 0; attr < nr_attrs; attr++) {
594 float data[4];
595 uint8_t *dst = (uint8_t *)vert + tg->attrib[attr].output_offset;
596
597 if (tg->attrib[attr].type == TRANSLATE_ELEMENT_NORMAL) {
598 const uint8_t *src;
599 unsigned index;
600 int copy_size;
601
602 if (tg->attrib[attr].instance_divisor) {
603 index = start_instance;
604 index += (instance_id / tg->attrib[attr].instance_divisor);
605 /* XXX we need to clamp the index here too, but to a
606 * per-array max value, not the draw->pt.max_index value
607 * that's being given to us via translate->set_buffer().
608 */
609 }
610 else {
611 index = elt;
612 /* clamp to avoid going out of bounds */
613 index = MIN2(index, tg->attrib[attr].max_index);
614 }
615
616 src = tg->attrib[attr].input_ptr +
617 (ptrdiff_t)tg->attrib[attr].input_stride * index;
618
619 copy_size = tg->attrib[attr].copy_size;
620 if(likely(copy_size >= 0))
621 memcpy(dst, src, copy_size);
622 else
623 {
624 tg->attrib[attr].fetch( data, src, 0, 0 );
625
626 if (0)
627 debug_printf("Fetch linear attr %d from %p stride %d index %d: "
628 " %f, %f, %f, %f \n",
629 attr,
630 tg->attrib[attr].input_ptr,
631 tg->attrib[attr].input_stride,
632 index,
633 data[0], data[1],data[2], data[3]);
634
635 tg->attrib[attr].emit( data, dst );
636 }
637 } else {
638 if(likely(tg->attrib[attr].copy_size >= 0))
639 memcpy(data, &instance_id, 4);
640 else
641 {
642 data[0] = (float)instance_id;
643 tg->attrib[attr].emit( data, dst );
644 }
645 }
646 }
647 }
648
649 /**
650 * Fetch vertex attributes for 'count' vertices.
651 */
generic_run_elts(struct translate * translate,const unsigned * elts,unsigned count,unsigned start_instance,unsigned instance_id,void * output_buffer)652 static void PIPE_CDECL generic_run_elts( struct translate *translate,
653 const unsigned *elts,
654 unsigned count,
655 unsigned start_instance,
656 unsigned instance_id,
657 void *output_buffer )
658 {
659 struct translate_generic *tg = translate_generic(translate);
660 char *vert = output_buffer;
661 unsigned i;
662
663 for (i = 0; i < count; i++) {
664 generic_run_one(tg, *elts++, start_instance, instance_id, vert);
665 vert += tg->translate.key.output_stride;
666 }
667 }
668
generic_run_elts16(struct translate * translate,const uint16_t * elts,unsigned count,unsigned start_instance,unsigned instance_id,void * output_buffer)669 static void PIPE_CDECL generic_run_elts16( struct translate *translate,
670 const uint16_t *elts,
671 unsigned count,
672 unsigned start_instance,
673 unsigned instance_id,
674 void *output_buffer )
675 {
676 struct translate_generic *tg = translate_generic(translate);
677 char *vert = output_buffer;
678 unsigned i;
679
680 for (i = 0; i < count; i++) {
681 generic_run_one(tg, *elts++, start_instance, instance_id, vert);
682 vert += tg->translate.key.output_stride;
683 }
684 }
685
generic_run_elts8(struct translate * translate,const uint8_t * elts,unsigned count,unsigned start_instance,unsigned instance_id,void * output_buffer)686 static void PIPE_CDECL generic_run_elts8( struct translate *translate,
687 const uint8_t *elts,
688 unsigned count,
689 unsigned start_instance,
690 unsigned instance_id,
691 void *output_buffer )
692 {
693 struct translate_generic *tg = translate_generic(translate);
694 char *vert = output_buffer;
695 unsigned i;
696
697 for (i = 0; i < count; i++) {
698 generic_run_one(tg, *elts++, start_instance, instance_id, vert);
699 vert += tg->translate.key.output_stride;
700 }
701 }
702
generic_run(struct translate * translate,unsigned start,unsigned count,unsigned start_instance,unsigned instance_id,void * output_buffer)703 static void PIPE_CDECL generic_run( struct translate *translate,
704 unsigned start,
705 unsigned count,
706 unsigned start_instance,
707 unsigned instance_id,
708 void *output_buffer )
709 {
710 struct translate_generic *tg = translate_generic(translate);
711 char *vert = output_buffer;
712 unsigned i;
713
714 for (i = 0; i < count; i++) {
715 generic_run_one(tg, start + i, start_instance, instance_id, vert);
716 vert += tg->translate.key.output_stride;
717 }
718 }
719
720
721
generic_set_buffer(struct translate * translate,unsigned buf,const void * ptr,unsigned stride,unsigned max_index)722 static void generic_set_buffer( struct translate *translate,
723 unsigned buf,
724 const void *ptr,
725 unsigned stride,
726 unsigned max_index )
727 {
728 struct translate_generic *tg = translate_generic(translate);
729 unsigned i;
730
731 for (i = 0; i < tg->nr_attrib; i++) {
732 if (tg->attrib[i].buffer == buf) {
733 tg->attrib[i].input_ptr = ((const uint8_t *)ptr +
734 tg->attrib[i].input_offset);
735 tg->attrib[i].input_stride = stride;
736 tg->attrib[i].max_index = max_index;
737 }
738 }
739 }
740
741
generic_release(struct translate * translate)742 static void generic_release( struct translate *translate )
743 {
744 /* Refcount?
745 */
746 FREE(translate);
747 }
748
749 static boolean
is_legal_int_format_combo(const struct util_format_description * src,const struct util_format_description * dst)750 is_legal_int_format_combo( const struct util_format_description *src,
751 const struct util_format_description *dst )
752 {
753 unsigned i;
754 unsigned nr = MIN2(src->nr_channels, dst->nr_channels);
755
756 for (i = 0; i < nr; i++) {
757 /* The signs must match. */
758 if (src->channel[i].type != dst->channel[i].type) {
759 return FALSE;
760 }
761
762 /* Integers must not lose precision at any point in the pipeline. */
763 if (src->channel[i].size > dst->channel[i].size) {
764 return FALSE;
765 }
766 }
767 return TRUE;
768 }
769
translate_generic_create(const struct translate_key * key)770 struct translate *translate_generic_create( const struct translate_key *key )
771 {
772 struct translate_generic *tg = CALLOC_STRUCT(translate_generic);
773 unsigned i;
774
775 if (!tg)
776 return NULL;
777
778 assert(key->nr_elements <= TRANSLATE_MAX_ATTRIBS);
779
780 tg->translate.key = *key;
781 tg->translate.release = generic_release;
782 tg->translate.set_buffer = generic_set_buffer;
783 tg->translate.run_elts = generic_run_elts;
784 tg->translate.run_elts16 = generic_run_elts16;
785 tg->translate.run_elts8 = generic_run_elts8;
786 tg->translate.run = generic_run;
787
788 for (i = 0; i < key->nr_elements; i++) {
789 const struct util_format_description *format_desc =
790 util_format_description(key->element[i].input_format);
791
792 assert(format_desc);
793
794 tg->attrib[i].type = key->element[i].type;
795
796 if (format_desc->channel[0].pure_integer) {
797 const struct util_format_description *out_format_desc =
798 util_format_description(key->element[i].output_format);
799
800 if (!is_legal_int_format_combo(format_desc, out_format_desc)) {
801 FREE(tg);
802 return NULL;
803 }
804
805 if (format_desc->channel[0].type == UTIL_FORMAT_TYPE_SIGNED) {
806 assert(format_desc->fetch_rgba_sint);
807 tg->attrib[i].fetch = (fetch_func)format_desc->fetch_rgba_sint;
808 } else {
809 assert(format_desc->fetch_rgba_uint);
810 tg->attrib[i].fetch = (fetch_func)format_desc->fetch_rgba_uint;
811 }
812 } else {
813 assert(format_desc->fetch_rgba_float);
814 tg->attrib[i].fetch = (fetch_func)format_desc->fetch_rgba_float;
815 }
816
817 tg->attrib[i].buffer = key->element[i].input_buffer;
818 tg->attrib[i].input_offset = key->element[i].input_offset;
819 tg->attrib[i].instance_divisor = key->element[i].instance_divisor;
820
821 tg->attrib[i].output_offset = key->element[i].output_offset;
822
823 tg->attrib[i].copy_size = -1;
824 if (tg->attrib[i].type == TRANSLATE_ELEMENT_INSTANCE_ID)
825 {
826 if(key->element[i].output_format == PIPE_FORMAT_R32_USCALED
827 || key->element[i].output_format == PIPE_FORMAT_R32_SSCALED)
828 tg->attrib[i].copy_size = 4;
829 }
830 else
831 {
832 if(key->element[i].input_format == key->element[i].output_format
833 && format_desc->block.width == 1
834 && format_desc->block.height == 1
835 && !(format_desc->block.bits & 7))
836 tg->attrib[i].copy_size = format_desc->block.bits >> 3;
837 }
838
839 if(tg->attrib[i].copy_size < 0)
840 tg->attrib[i].emit = get_emit_func(key->element[i].output_format);
841 else
842 tg->attrib[i].emit = NULL;
843 }
844
845 tg->nr_attrib = key->nr_elements;
846
847
848 return &tg->translate;
849 }
850
translate_generic_is_output_format_supported(enum pipe_format format)851 boolean translate_generic_is_output_format_supported(enum pipe_format format)
852 {
853 switch(format)
854 {
855 case PIPE_FORMAT_R64G64B64A64_FLOAT: return TRUE;
856 case PIPE_FORMAT_R64G64B64_FLOAT: return TRUE;
857 case PIPE_FORMAT_R64G64_FLOAT: return TRUE;
858 case PIPE_FORMAT_R64_FLOAT: return TRUE;
859
860 case PIPE_FORMAT_R32G32B32A32_FLOAT: return TRUE;
861 case PIPE_FORMAT_R32G32B32_FLOAT: return TRUE;
862 case PIPE_FORMAT_R32G32_FLOAT: return TRUE;
863 case PIPE_FORMAT_R32_FLOAT: return TRUE;
864
865 case PIPE_FORMAT_R16G16B16A16_FLOAT: return TRUE;
866 case PIPE_FORMAT_R16G16B16_FLOAT: return TRUE;
867 case PIPE_FORMAT_R16G16_FLOAT: return TRUE;
868 case PIPE_FORMAT_R16_FLOAT: return TRUE;
869
870 case PIPE_FORMAT_R32G32B32A32_USCALED: return TRUE;
871 case PIPE_FORMAT_R32G32B32_USCALED: return TRUE;
872 case PIPE_FORMAT_R32G32_USCALED: return TRUE;
873 case PIPE_FORMAT_R32_USCALED: return TRUE;
874
875 case PIPE_FORMAT_R32G32B32A32_SSCALED: return TRUE;
876 case PIPE_FORMAT_R32G32B32_SSCALED: return TRUE;
877 case PIPE_FORMAT_R32G32_SSCALED: return TRUE;
878 case PIPE_FORMAT_R32_SSCALED: return TRUE;
879
880 case PIPE_FORMAT_R32G32B32A32_UNORM: return TRUE;
881 case PIPE_FORMAT_R32G32B32_UNORM: return TRUE;
882 case PIPE_FORMAT_R32G32_UNORM: return TRUE;
883 case PIPE_FORMAT_R32_UNORM: return TRUE;
884
885 case PIPE_FORMAT_R32G32B32A32_SNORM: return TRUE;
886 case PIPE_FORMAT_R32G32B32_SNORM: return TRUE;
887 case PIPE_FORMAT_R32G32_SNORM: return TRUE;
888 case PIPE_FORMAT_R32_SNORM: return TRUE;
889
890 case PIPE_FORMAT_R16G16B16A16_USCALED: return TRUE;
891 case PIPE_FORMAT_R16G16B16_USCALED: return TRUE;
892 case PIPE_FORMAT_R16G16_USCALED: return TRUE;
893 case PIPE_FORMAT_R16_USCALED: return TRUE;
894
895 case PIPE_FORMAT_R16G16B16A16_SSCALED: return TRUE;
896 case PIPE_FORMAT_R16G16B16_SSCALED: return TRUE;
897 case PIPE_FORMAT_R16G16_SSCALED: return TRUE;
898 case PIPE_FORMAT_R16_SSCALED: return TRUE;
899
900 case PIPE_FORMAT_R16G16B16A16_UNORM: return TRUE;
901 case PIPE_FORMAT_R16G16B16_UNORM: return TRUE;
902 case PIPE_FORMAT_R16G16_UNORM: return TRUE;
903 case PIPE_FORMAT_R16_UNORM: return TRUE;
904
905 case PIPE_FORMAT_R16G16B16A16_SNORM: return TRUE;
906 case PIPE_FORMAT_R16G16B16_SNORM: return TRUE;
907 case PIPE_FORMAT_R16G16_SNORM: return TRUE;
908 case PIPE_FORMAT_R16_SNORM: return TRUE;
909
910 case PIPE_FORMAT_R8G8B8A8_USCALED: return TRUE;
911 case PIPE_FORMAT_R8G8B8_USCALED: return TRUE;
912 case PIPE_FORMAT_R8G8_USCALED: return TRUE;
913 case PIPE_FORMAT_R8_USCALED: return TRUE;
914
915 case PIPE_FORMAT_R8G8B8A8_SSCALED: return TRUE;
916 case PIPE_FORMAT_R8G8B8_SSCALED: return TRUE;
917 case PIPE_FORMAT_R8G8_SSCALED: return TRUE;
918 case PIPE_FORMAT_R8_SSCALED: return TRUE;
919
920 case PIPE_FORMAT_R8G8B8A8_UNORM: return TRUE;
921 case PIPE_FORMAT_R8G8B8_UNORM: return TRUE;
922 case PIPE_FORMAT_R8G8_UNORM: return TRUE;
923 case PIPE_FORMAT_R8_UNORM: return TRUE;
924
925 case PIPE_FORMAT_R8G8B8A8_SNORM: return TRUE;
926 case PIPE_FORMAT_R8G8B8_SNORM: return TRUE;
927 case PIPE_FORMAT_R8G8_SNORM: return TRUE;
928 case PIPE_FORMAT_R8_SNORM: return TRUE;
929
930 case PIPE_FORMAT_A8R8G8B8_UNORM: return TRUE;
931 case PIPE_FORMAT_B8G8R8A8_UNORM: return TRUE;
932
933 case PIPE_FORMAT_R32G32B32A32_UINT: return TRUE;
934 case PIPE_FORMAT_R32G32B32_UINT: return TRUE;
935 case PIPE_FORMAT_R32G32_UINT: return TRUE;
936 case PIPE_FORMAT_R32_UINT: return TRUE;
937
938 case PIPE_FORMAT_R16G16B16A16_UINT: return TRUE;
939 case PIPE_FORMAT_R16G16B16_UINT: return TRUE;
940 case PIPE_FORMAT_R16G16_UINT: return TRUE;
941 case PIPE_FORMAT_R16_UINT: return TRUE;
942
943 case PIPE_FORMAT_R8G8B8A8_UINT: return TRUE;
944 case PIPE_FORMAT_R8G8B8_UINT: return TRUE;
945 case PIPE_FORMAT_R8G8_UINT: return TRUE;
946 case PIPE_FORMAT_R8_UINT: return TRUE;
947
948 case PIPE_FORMAT_R32G32B32A32_SINT: return TRUE;
949 case PIPE_FORMAT_R32G32B32_SINT: return TRUE;
950 case PIPE_FORMAT_R32G32_SINT: return TRUE;
951 case PIPE_FORMAT_R32_SINT: return TRUE;
952
953 case PIPE_FORMAT_R16G16B16A16_SINT: return TRUE;
954 case PIPE_FORMAT_R16G16B16_SINT: return TRUE;
955 case PIPE_FORMAT_R16G16_SINT: return TRUE;
956 case PIPE_FORMAT_R16_SINT: return TRUE;
957
958 case PIPE_FORMAT_R8G8B8A8_SINT: return TRUE;
959 case PIPE_FORMAT_R8G8B8_SINT: return TRUE;
960 case PIPE_FORMAT_R8G8_SINT: return TRUE;
961 case PIPE_FORMAT_R8_SINT: return TRUE;
962
963 case PIPE_FORMAT_B10G10R10A2_UNORM: return TRUE;
964 case PIPE_FORMAT_B10G10R10A2_USCALED: return TRUE;
965 case PIPE_FORMAT_B10G10R10A2_SNORM: return TRUE;
966 case PIPE_FORMAT_B10G10R10A2_SSCALED: return TRUE;
967
968 case PIPE_FORMAT_R10G10B10A2_UNORM: return TRUE;
969 case PIPE_FORMAT_R10G10B10A2_USCALED: return TRUE;
970 case PIPE_FORMAT_R10G10B10A2_SNORM: return TRUE;
971 case PIPE_FORMAT_R10G10B10A2_SSCALED: return TRUE;
972
973 default: return FALSE;
974 }
975 }
976