1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
|
#!/usr/bin/env python
'''
/**************************************************************************
*
* Copyright 2009-2010 VMware, Inc.
* All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
* "Software"), to deal in the Software without restriction, including
* without limitation the rights to use, copy, modify, merge, publish,
* distribute, sub license, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
*
* The above copyright notice and this permission notice (including the
* next paragraph) shall be included in all copies or substantial portions
* of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
* IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
* ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*
**************************************************************************/
/**
* @file
* Pixel format packing and unpacking functions.
*
* @author Jose Fonseca <jfonseca@vmware.com>
*/
'''
import sys
import math
from u_format_parse import *
def generate_format_type(format):
'''Generate a structure that describes the format.'''
print 'union util_format_%s {' % format.short_name()
if format.is_bitmask() or format.short_name() == "r11g11b10_float":
print ' uint%u_t value;' % (format.block_size(),)
print ' struct {'
for channel in format.channels:
if (format.is_bitmask() or format.is_mixed()) and not format.is_array() or format.short_name() == "r11g11b10_float":
if channel.type == VOID:
if channel.size:
print ' unsigned %s:%u;' % (channel.name, channel.size)
elif channel.type == UNSIGNED:
print ' unsigned %s:%u;' % (channel.name, channel.size)
elif channel.type in (SIGNED, FIXED):
print ' int %s:%u;' % (channel.name, channel.size)
elif channel.type == FLOAT:
if channel.size == 32:
print ' float %s;' % (channel.name)
else:
print ' unsigned %s:%u;' % (channel.name, channel.size)
else:
assert 0
else:
assert channel.size % 8 == 0 and is_pot(channel.size)
if channel.type == VOID:
if channel.size:
print ' uint%u_t %s;' % (channel.size, channel.name)
elif channel.type == UNSIGNED:
print ' uint%u_t %s;' % (channel.size, channel.name)
elif channel.type in (SIGNED, FIXED):
print ' int%u_t %s;' % (channel.size, channel.name)
elif channel.type == FLOAT:
if channel.size == 64:
print ' double %s;' % (channel.name)
elif channel.size == 32:
print ' float %s;' % (channel.name)
elif channel.size == 16:
print ' uint16_t %s;' % (channel.name)
else:
assert 0
else:
assert 0
print ' } chan;'
print '};'
print
def bswap_format(format):
'''Generate a structure that describes the format.'''
if format.is_bitmask() and not format.is_array() and format.block_size() > 8:
print '#ifdef PIPE_ARCH_BIG_ENDIAN'
print ' pixel.value = util_bswap%u(pixel.value);' % format.block_size()
print '#endif'
def is_format_supported(format):
'''Determines whether we actually have the plumbing necessary to generate the
to read/write to/from this format.'''
# FIXME: Ideally we would support any format combination here.
if format.layout != PLAIN:
return False
for i in range(4):
channel = format.channels[i]
if channel.type not in (VOID, UNSIGNED, SIGNED, FLOAT, FIXED):
return False
if channel.type == FLOAT and channel.size not in (16, 32, 64):
return False
# We can only read a color from a depth/stencil format if the depth channel is present
if format.colorspace == 'zs' and format.swizzles[0] == SWIZZLE_NONE:
return False
return True
def native_type(format):
'''Get the native appropriate for a format.'''
if format.layout == PLAIN:
if not format.is_array():
# For arithmetic pixel formats return the integer type that matches the whole pixel
return 'uint%u_t' % format.block_size()
else:
# For array pixel formats return the integer type that matches the color channel
channel = format.channels[0]
if channel.type in (UNSIGNED, VOID):
return 'uint%u_t' % channel.size
elif channel.type in (SIGNED, FIXED):
return 'int%u_t' % channel.size
elif channel.type == FLOAT:
if channel.size == 16:
return 'uint16_t'
elif channel.size == 32:
return 'float'
elif channel.size == 64:
return 'double'
else:
assert False
else:
assert False
else:
assert False
def intermediate_native_type(bits, sign):
'''Find a native type adequate to hold intermediate results of the request bit size.'''
bytes = 4 # don't use anything smaller than 32bits
while bytes * 8 < bits:
bytes *= 2
bits = bytes*8
if sign:
return 'int%u_t' % bits
else:
return 'uint%u_t' % bits
def get_one_shift(type):
'''Get the number of the bit that matches unity for this type.'''
if type.type == 'FLOAT':
assert False
if not type.norm:
return 0
if type.type == UNSIGNED:
return type.size
if type.type == SIGNED:
return type.size - 1
if type.type == FIXED:
return type.size / 2
assert False
def value_to_native(type, value):
'''Get the value of unity for this type.'''
if type.type == FLOAT:
return value
if type.type == FIXED:
return int(value * (1 << (type.size/2)))
if not type.norm:
return int(value)
if type.type == UNSIGNED:
return int(value * ((1 << type.size) - 1))
if type.type == SIGNED:
return int(value * ((1 << (type.size - 1)) - 1))
assert False
def native_to_constant(type, value):
'''Get the value of unity for this type.'''
if type.type == FLOAT:
if type.size <= 32:
return "%ff" % value
else:
return "%ff" % value
else:
return str(int(value))
def get_one(type):
'''Get the value of unity for this type.'''
return value_to_native(type, 1)
def clamp_expr(src_channel, dst_channel, dst_native_type, value):
'''Generate the expression to clamp the value in the source type to the
destination type range.'''
if src_channel == dst_channel:
return value
src_min = src_channel.min()
src_max = src_channel.max()
dst_min = dst_channel.min()
dst_max = dst_channel.max()
# Translate the destination range to the src native value
dst_min_native = value_to_native(src_channel, dst_min)
dst_max_native = value_to_native(src_channel, dst_max)
if src_min < dst_min and src_max > dst_max:
return 'CLAMP(%s, %s, %s)' % (value, dst_min_native, dst_max_native)
if src_max > dst_max:
return 'MIN2(%s, %s)' % (value, dst_max_native)
if src_min < dst_min:
return 'MAX2(%s, %s)' % (value, dst_min_native)
return value
def conversion_expr(src_channel,
dst_channel, dst_native_type,
value,
clamp=True,
src_colorspace = 'rgb',
dst_colorspace = 'rgb'):
'''Generate the expression to convert a value between two types.'''
if src_colorspace != dst_colorspace:
if src_colorspace == 'srgb':
assert src_channel.type == UNSIGNED
assert src_channel.norm
assert src_channel.size == 8
assert dst_colorspace == 'rgb'
if dst_channel.type == FLOAT:
return 'util_format_srgb_8unorm_to_linear_float(%s)' % value
else:
assert dst_channel.type == UNSIGNED
assert dst_channel.norm
assert dst_channel.size == 8
return 'util_format_srgb_to_linear_8unorm(%s)' % value
elif dst_colorspace == 'srgb':
assert dst_channel.type == UNSIGNED
assert dst_channel.norm
assert dst_channel.size == 8
assert src_colorspace == 'rgb'
if src_channel.type == FLOAT:
return 'util_format_linear_float_to_srgb_8unorm(%s)' % value
else:
assert src_channel.type == UNSIGNED
assert src_channel.norm
assert src_channel.size == 8
return 'util_format_linear_to_srgb_8unorm(%s)' % value
elif src_colorspace == 'zs':
pass
elif dst_colorspace == 'zs':
pass
else:
assert 0
if src_channel == dst_channel:
return value
src_type = src_channel.type
src_size = src_channel.size
src_norm = src_channel.norm
# Promote half to float
if src_type == FLOAT and src_size == 16:
value = 'util_half_to_float(%s)' % value
src_size = 32
if clamp:
if dst_channel.type != FLOAT or src_type != FLOAT:
value = clamp_expr(src_channel, dst_channel, dst_native_type, value)
if src_type in (SIGNED, UNSIGNED) and dst_channel.type in (SIGNED, UNSIGNED):
if not src_norm and not dst_channel.norm:
# neither is normalized -- just cast
return '(%s)%s' % (dst_native_type, value)
src_one = get_one(src_channel)
dst_one = get_one(dst_channel)
if src_one > dst_one and src_norm and dst_channel.norm:
# We can just bitshift
src_shift = get_one_shift(src_channel)
dst_shift = get_one_shift(dst_channel)
value = '(%s >> %s)' % (value, src_shift - dst_shift)
else:
# We need to rescale using an intermediate type big enough to hold the multiplication of both
tmp_native_type = intermediate_native_type(src_size + dst_channel.size, src_channel.sign and dst_channel.sign)
value = '((%s)%s)' % (tmp_native_type, value)
value = '(%s * 0x%x / 0x%x)' % (value, dst_one, src_one)
value = '(%s)%s' % (dst_native_type, value)
return value
# Promote to either float or double
if src_type != FLOAT:
if src_norm or src_type == FIXED:
one = get_one(src_channel)
if src_size <= 23:
value = '(%s * (1.0f/0x%x))' % (value, one)
if dst_channel.size <= 32:
value = '(float)%s' % value
src_size = 32
else:
# bigger than single precision mantissa, use double
value = '(%s * (1.0/0x%x))' % (value, one)
src_size = 64
src_norm = False
else:
if src_size <= 23 or dst_channel.size <= 32:
value = '(float)%s' % value
src_size = 32
else:
# bigger than single precision mantissa, use double
value = '(double)%s' % value
src_size = 64
src_type = FLOAT
# Convert double or float to non-float
if dst_channel.type != FLOAT:
if dst_channel.norm or dst_channel.type == FIXED:
dst_one = get_one(dst_channel)
if dst_channel.size <= 23:
value = '(%s * 0x%x)' % (value, dst_one)
else:
# bigger than single precision mantissa, use double
value = '(%s * (double)0x%x)' % (value, dst_one)
value = '(%s)%s' % (dst_native_type, value)
else:
# Cast double to float when converting to either half or float
if dst_channel.size <= 32 and src_size > 32:
value = '(float)%s' % value
src_size = 32
if dst_channel.size == 16:
value = 'util_float_to_half(%s)' % value
elif dst_channel.size == 64 and src_size < 64:
value = '(double)%s' % value
return value
def generate_unpack_kernel(format, dst_channel, dst_native_type):
if not is_format_supported(format):
return
assert format.layout == PLAIN
src_native_type = native_type(format)
if format.is_bitmask():
depth = format.block_size()
print ' uint%u_t value = *(const uint%u_t *)src;' % (depth, depth)
# Declare the intermediate variables
for i in range(format.nr_channels()):
src_channel = format.channels[i]
if src_channel.type == UNSIGNED:
print ' uint%u_t %s;' % (depth, src_channel.name)
elif src_channel.type == SIGNED:
print ' int%u_t %s;' % (depth, src_channel.name)
if depth > 8:
print '#ifdef PIPE_ARCH_BIG_ENDIAN'
print ' value = util_bswap%u(value);' % depth
print '#endif'
# Compute the intermediate unshifted values
shift = 0
for i in range(format.nr_channels()):
src_channel = format.channels[i]
value = 'value'
if src_channel.type == UNSIGNED:
if shift:
value = '%s >> %u' % (value, shift)
if shift + src_channel.size < depth:
value = '(%s) & 0x%x' % (value, (1 << src_channel.size) - 1)
elif src_channel.type == SIGNED:
if shift + src_channel.size < depth:
# Align the sign bit
lshift = depth - (shift + src_channel.size)
value = '%s << %u' % (value, lshift)
# Cast to signed
value = '(int%u_t)(%s) ' % (depth, value)
if src_channel.size < depth:
# Align the LSB bit
rshift = depth - src_channel.size
value = '(%s) >> %u' % (value, rshift)
else:
value = None
if value is not None:
print ' %s = %s;' % (src_channel.name, value)
shift += src_channel.size
# Convert, swizzle, and store final values
for i in range(4):
swizzle = format.swizzles[i]
if swizzle < 4:
src_channel = format.channels[swizzle]
src_colorspace = format.colorspace
if src_colorspace == 'srgb' and i == 3:
# Alpha channel is linear
src_colorspace = 'rgb'
value = src_channel.name
value = conversion_expr(src_channel,
dst_channel, dst_native_type,
value,
src_colorspace = src_colorspace)
elif swizzle == SWIZZLE_0:
value = '0'
elif swizzle == SWIZZLE_1:
value = get_one(dst_channel)
elif swizzle == SWIZZLE_NONE:
value = '0'
else:
assert False
if format.colorspace == ZS:
if i == 3:
value = get_one(dst_channel)
elif i >= 1:
value = 'dst[0]'
print ' dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
else:
print ' union util_format_%s pixel;' % format.short_name()
print ' memcpy(&pixel, src, sizeof pixel);'
bswap_format(format)
for i in range(4):
swizzle = format.swizzles[i]
if swizzle < 4:
src_channel = format.channels[swizzle]
src_colorspace = format.colorspace
if src_colorspace == 'srgb' and i == 3:
# Alpha channel is linear
src_colorspace = 'rgb'
value = 'pixel.chan.%s' % src_channel.name
value = conversion_expr(src_channel,
dst_channel, dst_native_type,
value,
src_colorspace = src_colorspace)
elif swizzle == SWIZZLE_0:
value = '0'
elif swizzle == SWIZZLE_1:
value = get_one(dst_channel)
elif swizzle == SWIZZLE_NONE:
value = '0'
else:
assert False
if format.colorspace == ZS:
if i == 3:
value = get_one(dst_channel)
elif i >= 1:
value = 'dst[0]'
print ' dst[%u] = %s; /* %s */' % (i, value, 'rgba'[i])
def generate_pack_kernel(format, src_channel, src_native_type):
if not is_format_supported(format):
return
dst_native_type = native_type(format)
assert format.layout == PLAIN
inv_swizzle = format.inv_swizzles()
if format.is_bitmask():
depth = format.block_size()
print ' uint%u_t value = 0;' % depth
shift = 0
for i in range(4):
dst_channel = format.channels[i]
if inv_swizzle[i] is not None:
value ='src[%u]' % inv_swizzle[i]
dst_colorspace = format.colorspace
if dst_colorspace == 'srgb' and inv_swizzle[i] == 3:
# Alpha channel is linear
dst_colorspace = 'rgb'
value = conversion_expr(src_channel,
dst_channel, dst_native_type,
value,
dst_colorspace = dst_colorspace)
if format.colorspace == ZS:
if i == 3:
value = get_one(dst_channel)
elif i >= 1:
value = '0'
if dst_channel.type in (UNSIGNED, SIGNED):
if shift + dst_channel.size < depth:
value = '(%s) & 0x%x' % (value, (1 << dst_channel.size) - 1)
if shift:
value = '(%s) << %u' % (value, shift)
if dst_channel.type == SIGNED:
# Cast to unsigned
value = '(uint%u_t)(%s) ' % (depth, value)
else:
value = None
if value is not None:
print ' value |= %s;' % (value)
shift += dst_channel.size
if depth > 8:
print '#ifdef PIPE_ARCH_BIG_ENDIAN'
print ' value = util_bswap%u(value);' % depth
print '#endif'
print ' *(uint%u_t *)dst = value;' % depth
else:
print ' union util_format_%s pixel;' % format.short_name()
for i in range(4):
dst_channel = format.channels[i]
width = dst_channel.size
if inv_swizzle[i] is None:
continue
dst_colorspace = format.colorspace
if dst_colorspace == 'srgb' and inv_swizzle[i] == 3:
# Alpha channel is linear
dst_colorspace = 'rgb'
value ='src[%u]' % inv_swizzle[i]
value = conversion_expr(src_channel,
dst_channel, dst_native_type,
value,
dst_colorspace = dst_colorspace)
if format.colorspace == ZS:
if i == 3:
value = get_one(dst_channel)
elif i >= 1:
value = '0'
print ' pixel.chan.%s = %s;' % (dst_channel.name, value)
bswap_format(format)
print ' memcpy(dst, &pixel, sizeof pixel);'
def generate_format_unpack(format, dst_channel, dst_native_type, dst_suffix):
'''Generate the function to unpack pixels from a particular format'''
name = format.short_name()
print 'static INLINE void'
print 'util_format_%s_unpack_%s(%s *dst_row, unsigned dst_stride, const uint8_t *src_row, unsigned src_stride, unsigned width, unsigned height)' % (name, dst_suffix, dst_native_type)
print '{'
if is_format_supported(format):
print ' unsigned x, y;'
print ' for(y = 0; y < height; y += %u) {' % (format.block_height,)
print ' %s *dst = dst_row;' % (dst_native_type)
print ' const uint8_t *src = src_row;'
print ' for(x = 0; x < width; x += %u) {' % (format.block_width,)
generate_unpack_kernel(format, dst_channel, dst_native_type)
print ' src += %u;' % (format.block_size() / 8,)
print ' dst += 4;'
print ' }'
print ' src_row += src_stride;'
print ' dst_row += dst_stride/sizeof(*dst_row);'
print ' }'
print '}'
print
def generate_format_pack(format, src_channel, src_native_type, src_suffix):
'''Generate the function to pack pixels to a particular format'''
name = format.short_name()
print 'static INLINE void'
print 'util_format_%s_pack_%s(uint8_t *dst_row, unsigned dst_stride, const %s *src_row, unsigned src_stride, unsigned width, unsigned height)' % (name, src_suffix, src_native_type)
print '{'
if is_format_supported(format):
print ' unsigned x, y;'
print ' for(y = 0; y < height; y += %u) {' % (format.block_height,)
print ' const %s *src = src_row;' % (src_native_type)
print ' uint8_t *dst = dst_row;'
print ' for(x = 0; x < width; x += %u) {' % (format.block_width,)
generate_pack_kernel(format, src_channel, src_native_type)
print ' src += 4;'
print ' dst += %u;' % (format.block_size() / 8,)
print ' }'
print ' dst_row += dst_stride;'
print ' src_row += src_stride/sizeof(*src_row);'
print ' }'
print '}'
print
def generate_format_fetch(format, dst_channel, dst_native_type, dst_suffix):
'''Generate the function to unpack pixels from a particular format'''
name = format.short_name()
print 'static INLINE void'
print 'util_format_%s_fetch_%s(%s *dst, const uint8_t *src, unsigned i, unsigned j)' % (name, dst_suffix, dst_native_type)
print '{'
if is_format_supported(format):
generate_unpack_kernel(format, dst_channel, dst_native_type)
print '}'
print
def is_format_hand_written(format):
return format.layout in ('s3tc', 'subsampled')
def generate(formats):
print
print '#include "pipe/p_compiler.h"'
print '#include "u_math.h"'
print '#include "u_format.h"'
print '#include "u_format_srgb.h"'
print '#include "u_format_yuv.h"'
print '#include "u_half.h"'
print
for format in formats:
if is_format_supported(format):
generate_format_type(format)
if not is_format_hand_written(format):
channel = Channel(FLOAT, False, 32)
native_type = 'float'
suffix = 'float'
generate_format_unpack(format, channel, native_type, suffix)
generate_format_pack(format, channel, native_type, suffix)
generate_format_fetch(format, channel, native_type, suffix)
channel = Channel(UNSIGNED, True, 8)
native_type = 'uint8_t'
suffix = '8unorm'
generate_format_unpack(format, channel, native_type, suffix)
generate_format_pack(format, channel, native_type, suffix)
|