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
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
|
/*
* Copyright 2011 Google Inc.
*
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
*/
#include "SkBitmapCache.h"
#include "SkBitmapProcState.h"
#include "SkColorPriv.h"
#include "SkFilterProc.h"
#include "SkPaint.h"
#include "SkShader.h" // for tilemodes
#include "SkUtilsArm.h"
#include "SkBitmapScaler.h"
#include "SkMipMap.h"
#include "SkPixelRef.h"
#include "SkImageEncoder.h"
#include "SkResourceCache.h"
#if !SK_ARM_NEON_IS_NONE
// These are defined in src/opts/SkBitmapProcState_arm_neon.cpp
extern const SkBitmapProcState::SampleProc16 gSkBitmapProcStateSample16_neon[];
extern const SkBitmapProcState::SampleProc32 gSkBitmapProcStateSample32_neon[];
extern void S16_D16_filter_DX_neon(const SkBitmapProcState&, const uint32_t*, int, uint16_t*);
extern void Clamp_S16_D16_filter_DX_shaderproc_neon(const SkBitmapProcState&, int, int, uint16_t*, int);
extern void Repeat_S16_D16_filter_DX_shaderproc_neon(const SkBitmapProcState&, int, int, uint16_t*, int);
extern void SI8_opaque_D32_filter_DX_neon(const SkBitmapProcState&, const uint32_t*, int, SkPMColor*);
extern void SI8_opaque_D32_filter_DX_shaderproc_neon(const SkBitmapProcState&, int, int, uint32_t*, int);
extern void Clamp_SI8_opaque_D32_filter_DX_shaderproc_neon(const SkBitmapProcState&, int, int, uint32_t*, int);
#endif
extern void Clamp_S32_opaque_D32_nofilter_DX_shaderproc(const SkBitmapProcState&, int, int, uint32_t*, int);
#define NAME_WRAP(x) x
#include "SkBitmapProcState_filter.h"
#include "SkBitmapProcState_procs.h"
///////////////////////////////////////////////////////////////////////////////
// true iff the matrix contains, at most, scale and translate elements
static bool matrix_only_scale_translate(const SkMatrix& m) {
return m.getType() <= (SkMatrix::kScale_Mask | SkMatrix::kTranslate_Mask);
}
/**
* For the purposes of drawing bitmaps, if a matrix is "almost" translate
* go ahead and treat it as if it were, so that subsequent code can go fast.
*/
static bool just_trans_clamp(const SkMatrix& matrix, const SkBitmap& bitmap) {
SkASSERT(matrix_only_scale_translate(matrix));
if (matrix.getType() & SkMatrix::kScale_Mask) {
SkRect src, dst;
bitmap.getBounds(&src);
// Can't call mapRect(), since that will fix up inverted rectangles,
// e.g. when scale is negative, and we don't want to return true for
// those.
matrix.mapPoints(SkTCast<SkPoint*>(&dst),
SkTCast<const SkPoint*>(&src),
2);
// Now round all 4 edges to device space, and then compare the device
// width/height to the original. Note: we must map all 4 and subtract
// rather than map the "width" and compare, since we care about the
// phase (in pixel space) that any translate in the matrix might impart.
SkIRect idst;
dst.round(&idst);
return idst.width() == bitmap.width() && idst.height() == bitmap.height();
}
// if we got here, we're either kTranslate_Mask or identity
return true;
}
static bool just_trans_general(const SkMatrix& matrix) {
SkASSERT(matrix_only_scale_translate(matrix));
if (matrix.getType() & SkMatrix::kScale_Mask) {
const SkScalar tol = SK_Scalar1 / 32768;
if (!SkScalarNearlyZero(matrix[SkMatrix::kMScaleX] - SK_Scalar1, tol)) {
return false;
}
if (!SkScalarNearlyZero(matrix[SkMatrix::kMScaleY] - SK_Scalar1, tol)) {
return false;
}
}
// if we got here, treat us as either kTranslate_Mask or identity
return true;
}
///////////////////////////////////////////////////////////////////////////////
static bool valid_for_filtering(unsigned dimension) {
// for filtering, width and height must fit in 14bits, since we use steal
// 2 bits from each to store our 4bit subpixel data
return (dimension & ~0x3FFF) == 0;
}
// Check to see that the size of the bitmap that would be produced by
// scaling by the given inverted matrix is less than the maximum allowed.
static inline bool cache_size_okay(const SkBitmap& bm, const SkMatrix& invMat) {
size_t maximumAllocation = SkResourceCache::GetEffectiveSingleAllocationByteLimit();
if (0 == maximumAllocation) {
return true;
}
// float matrixScaleFactor = 1.0 / (invMat.scaleX * invMat.scaleY);
// return ((origBitmapSize * matrixScaleFactor) < maximumAllocationSize);
// Skip the division step:
return bm.info().getSafeSize(bm.info().minRowBytes())
< (maximumAllocation * invMat.getScaleX() * invMat.getScaleY());
}
/*
* High quality is implemented by performing up-right scale-only filtering and then
* using bilerp for any remaining transformations.
*/
void SkBitmapProcState::processHQRequest() {
SkASSERT(kHigh_SkFilterQuality == fFilterLevel);
// Our default return state is to downgrade the request to Medium, w/ or w/o setting fBitmap
// to a valid bitmap. If we succeed, we will set this to Low instead.
fFilterLevel = kMedium_SkFilterQuality;
if (kN32_SkColorType != fOrigBitmap.colorType() || !cache_size_okay(fOrigBitmap, fInvMatrix) ||
fInvMatrix.hasPerspective())
{
return; // can't handle the reqeust
}
SkScalar invScaleX = fInvMatrix.getScaleX();
SkScalar invScaleY = fInvMatrix.getScaleY();
if (fInvMatrix.getType() & SkMatrix::kAffine_Mask) {
SkSize scale;
if (!fInvMatrix.decomposeScale(&scale)) {
return;
}
invScaleX = scale.width();
invScaleY = scale.height();
}
if (SkScalarNearlyEqual(invScaleX, 1) && SkScalarNearlyEqual(invScaleY, 1)) {
return; // no need for HQ
}
SkScalar trueDestWidth = fOrigBitmap.width() / invScaleX;
SkScalar trueDestHeight = fOrigBitmap.height() / invScaleY;
SkScalar roundedDestWidth = SkScalarRoundToScalar(trueDestWidth);
SkScalar roundedDestHeight = SkScalarRoundToScalar(trueDestHeight);
if (!SkBitmapCache::Find(fOrigBitmap, roundedDestWidth, roundedDestHeight, &fScaledBitmap)) {
if (!SkBitmapScaler::Resize(&fScaledBitmap,
fOrigBitmap,
SkBitmapScaler::RESIZE_BEST,
roundedDestWidth,
roundedDestHeight,
SkResourceCache::GetAllocator())) {
return; // we failed to create fScaledBitmap
}
SkASSERT(fScaledBitmap.getPixels());
fScaledBitmap.setImmutable();
SkBitmapCache::Add(fOrigBitmap, roundedDestWidth, roundedDestHeight, fScaledBitmap);
}
SkASSERT(fScaledBitmap.getPixels());
fBitmap = &fScaledBitmap;
fInvMatrix.postScale(roundedDestWidth / fOrigBitmap.width(),
roundedDestHeight / fOrigBitmap.height());
fFilterLevel = kLow_SkFilterQuality;
}
/*
* Modulo internal errors, this should always succeed *if* the matrix is downscaling
* (in this case, we have the inverse, so it succeeds if fInvMatrix is upscaling)
*/
void SkBitmapProcState::processMediumRequest() {
SkASSERT(kMedium_SkFilterQuality == fFilterLevel);
// Our default return state is to downgrade the request to Low, w/ or w/o setting fBitmap
// to a valid bitmap.
fFilterLevel = kLow_SkFilterQuality;
SkSize invScaleSize;
if (!fInvMatrix.decomposeScale(&invScaleSize, NULL)) {
return;
}
SkScalar invScale = SkScalarSqrt(invScaleSize.width() * invScaleSize.height());
if (invScale > SK_Scalar1) {
fCurrMip.reset(SkMipMapCache::FindAndRef(fOrigBitmap));
if (NULL == fCurrMip.get()) {
fCurrMip.reset(SkMipMapCache::AddAndRef(fOrigBitmap));
if (NULL == fCurrMip.get()) {
return;
}
}
// diagnostic for a crasher...
if (NULL == fCurrMip->data()) {
sk_throw();
}
SkScalar levelScale = SkScalarInvert(invScale);
SkMipMap::Level level;
if (fCurrMip->extractLevel(levelScale, &level)) {
SkScalar invScaleFixup = level.fScale;
fInvMatrix.postScale(invScaleFixup, invScaleFixup);
const SkImageInfo info = fOrigBitmap.info().makeWH(level.fWidth, level.fHeight);
// todo: if we could wrap the fCurrMip in a pixelref, then we could just install
// that here, and not need to explicitly track it ourselves.
fScaledBitmap.installPixels(info, level.fPixels, level.fRowBytes);
fBitmap = &fScaledBitmap;
} else {
// failed to extract, so release the mipmap
fCurrMip.reset(NULL);
}
}
}
bool SkBitmapProcState::lockBaseBitmap() {
// TODO(reed): use bitmap cache here?
fScaledBitmap = fOrigBitmap;
fScaledBitmap.lockPixels();
if (NULL == fScaledBitmap.getPixels()) {
return false;
}
fBitmap = &fScaledBitmap;
return true;
}
static bool valid_for_drawing(const SkBitmap& bm) {
if (0 == bm.width() || 0 == bm.height()) {
return false; // nothing to draw
}
if (NULL == bm.pixelRef()) {
return false; // no pixels to read
}
if (bm.getTexture()) {
// we can handle texture (ugh) since lockPixels will perform a read-back
return true;
}
if (kIndex_8_SkColorType == bm.colorType()) {
SkAutoLockPixels alp(bm); // but we need to call it before getColorTable() is safe.
if (!bm.getColorTable()) {
return false;
}
}
return true;
}
/*
* Analyze filter-quality and matrix, and decide how to implement that.
*
* In general, we cascade down the request level [ High ... None ]
* - for a given level, if we can fulfill it, fine, else
* - else we downgrade to the next lower level and try again.
* We can always fulfill requests for Low and None
* - sometimes we will "ignore" Low and give None, but this is likely a legacy perf hack
* and may be removed.
*/
bool SkBitmapProcState::chooseProcs(const SkMatrix& inv, const SkPaint& paint) {
if (!valid_for_drawing(fOrigBitmap)) {
return false;
}
fBitmap = NULL;
fInvMatrix = inv;
fFilterLevel = paint.getFilterQuality();
if (kHigh_SkFilterQuality == fFilterLevel) {
this->processHQRequest();
}
SkASSERT(fFilterLevel < kHigh_SkFilterQuality);
if (kMedium_SkFilterQuality == fFilterLevel) {
this->processMediumRequest();
}
SkASSERT(fFilterLevel < kMedium_SkFilterQuality);
if (NULL == fBitmap) {
if (!this->lockBaseBitmap()) {
return false;
}
}
SkASSERT(fBitmap);
bool trivialMatrix = (fInvMatrix.getType() & ~SkMatrix::kTranslate_Mask) == 0;
bool clampClamp = SkShader::kClamp_TileMode == fTileModeX &&
SkShader::kClamp_TileMode == fTileModeY;
// Most of the scanline procs deal with "unit" texture coordinates, as this
// makes it easy to perform tiling modes (repeat = (x & 0xFFFF)). To generate
// those, we divide the matrix by its dimensions here.
//
// We don't do this if we're either trivial (can ignore the matrix) or clamping
// in both X and Y since clamping to width,height is just as easy as to 0xFFFF.
if (!(clampClamp || trivialMatrix)) {
fInvMatrix.postIDiv(fBitmap->width(), fBitmap->height());
}
// Now that all possible changes to the matrix have taken place, check
// to see if we're really close to a no-scale matrix. If so, explicitly
// set it to be so. Subsequent code may inspect this matrix to choose
// a faster path in this case.
// This code will only execute if the matrix has some scale component;
// if it's already pure translate then we won't do this inversion.
if (matrix_only_scale_translate(fInvMatrix)) {
SkMatrix forward;
if (fInvMatrix.invert(&forward)) {
if (clampClamp ? just_trans_clamp(forward, *fBitmap)
: just_trans_general(forward)) {
SkScalar tx = -SkScalarRoundToScalar(forward.getTranslateX());
SkScalar ty = -SkScalarRoundToScalar(forward.getTranslateY());
fInvMatrix.setTranslate(tx, ty);
}
}
}
fInvProc = fInvMatrix.getMapXYProc();
fInvType = fInvMatrix.getType();
fInvSx = SkScalarToFixed(fInvMatrix.getScaleX());
fInvSxFractionalInt = SkScalarToFractionalInt(fInvMatrix.getScaleX());
fInvKy = SkScalarToFixed(fInvMatrix.getSkewY());
fInvKyFractionalInt = SkScalarToFractionalInt(fInvMatrix.getSkewY());
fAlphaScale = SkAlpha255To256(paint.getAlpha());
fShaderProc32 = NULL;
fShaderProc16 = NULL;
fSampleProc32 = NULL;
fSampleProc16 = NULL;
// recompute the triviality of the matrix here because we may have
// changed it!
trivialMatrix = (fInvMatrix.getType() & ~SkMatrix::kTranslate_Mask) == 0;
if (kLow_SkFilterQuality == fFilterLevel) {
// Only try bilerp if the matrix is "interesting" and
// the image has a suitable size.
if (fInvType <= SkMatrix::kTranslate_Mask ||
!valid_for_filtering(fBitmap->width() | fBitmap->height()))
{
fFilterLevel = kNone_SkFilterQuality;
}
}
return this->chooseScanlineProcs(trivialMatrix, clampClamp, paint);
}
bool SkBitmapProcState::chooseScanlineProcs(bool trivialMatrix, bool clampClamp,
const SkPaint& paint) {
fMatrixProc = this->chooseMatrixProc(trivialMatrix);
// TODO(dominikg): SkASSERT(fMatrixProc) instead? chooseMatrixProc never returns NULL.
if (NULL == fMatrixProc) {
return false;
}
///////////////////////////////////////////////////////////////////////
const SkAlphaType at = fBitmap->alphaType();
// No need to do this if we're doing HQ sampling; if filter quality is
// still set to HQ by the time we get here, then we must have installed
// the shader procs above and can skip all this.
if (fFilterLevel < kHigh_SkFilterQuality) {
int index = 0;
if (fAlphaScale < 256) { // note: this distinction is not used for D16
index |= 1;
}
if (fInvType <= (SkMatrix::kTranslate_Mask | SkMatrix::kScale_Mask)) {
index |= 2;
}
if (fFilterLevel > kNone_SkFilterQuality) {
index |= 4;
}
// bits 3,4,5 encoding the source bitmap format
switch (fBitmap->colorType()) {
case kN32_SkColorType:
if (kPremul_SkAlphaType != at && kOpaque_SkAlphaType != at) {
return false;
}
index |= 0;
break;
case kRGB_565_SkColorType:
index |= 8;
break;
case kIndex_8_SkColorType:
if (kPremul_SkAlphaType != at && kOpaque_SkAlphaType != at) {
return false;
}
index |= 16;
break;
case kARGB_4444_SkColorType:
if (kPremul_SkAlphaType != at && kOpaque_SkAlphaType != at) {
return false;
}
index |= 24;
break;
case kAlpha_8_SkColorType:
index |= 32;
fPaintPMColor = SkPreMultiplyColor(paint.getColor());
break;
case kGray_8_SkColorType:
index |= 40;
fPaintPMColor = SkPreMultiplyColor(paint.getColor());
break;
default:
// TODO(dominikg): Should we ever get here? SkASSERT(false) instead?
return false;
}
#if !SK_ARM_NEON_IS_ALWAYS
static const SampleProc32 gSkBitmapProcStateSample32[] = {
S32_opaque_D32_nofilter_DXDY,
S32_alpha_D32_nofilter_DXDY,
S32_opaque_D32_nofilter_DX,
S32_alpha_D32_nofilter_DX,
S32_opaque_D32_filter_DXDY,
S32_alpha_D32_filter_DXDY,
S32_opaque_D32_filter_DX,
S32_alpha_D32_filter_DX,
S16_opaque_D32_nofilter_DXDY,
S16_alpha_D32_nofilter_DXDY,
S16_opaque_D32_nofilter_DX,
S16_alpha_D32_nofilter_DX,
S16_opaque_D32_filter_DXDY,
S16_alpha_D32_filter_DXDY,
S16_opaque_D32_filter_DX,
S16_alpha_D32_filter_DX,
SI8_opaque_D32_nofilter_DXDY,
SI8_alpha_D32_nofilter_DXDY,
SI8_opaque_D32_nofilter_DX,
SI8_alpha_D32_nofilter_DX,
SI8_opaque_D32_filter_DXDY,
SI8_alpha_D32_filter_DXDY,
SI8_opaque_D32_filter_DX,
SI8_alpha_D32_filter_DX,
S4444_opaque_D32_nofilter_DXDY,
S4444_alpha_D32_nofilter_DXDY,
S4444_opaque_D32_nofilter_DX,
S4444_alpha_D32_nofilter_DX,
S4444_opaque_D32_filter_DXDY,
S4444_alpha_D32_filter_DXDY,
S4444_opaque_D32_filter_DX,
S4444_alpha_D32_filter_DX,
// A8 treats alpha/opaque the same (equally efficient)
SA8_alpha_D32_nofilter_DXDY,
SA8_alpha_D32_nofilter_DXDY,
SA8_alpha_D32_nofilter_DX,
SA8_alpha_D32_nofilter_DX,
SA8_alpha_D32_filter_DXDY,
SA8_alpha_D32_filter_DXDY,
SA8_alpha_D32_filter_DX,
SA8_alpha_D32_filter_DX,
// todo: possibly specialize on opaqueness
SG8_alpha_D32_nofilter_DXDY,
SG8_alpha_D32_nofilter_DXDY,
SG8_alpha_D32_nofilter_DX,
SG8_alpha_D32_nofilter_DX,
SG8_alpha_D32_filter_DXDY,
SG8_alpha_D32_filter_DXDY,
SG8_alpha_D32_filter_DX,
SG8_alpha_D32_filter_DX
};
static const SampleProc16 gSkBitmapProcStateSample16[] = {
S32_D16_nofilter_DXDY,
S32_D16_nofilter_DX,
S32_D16_filter_DXDY,
S32_D16_filter_DX,
S16_D16_nofilter_DXDY,
S16_D16_nofilter_DX,
S16_D16_filter_DXDY,
S16_D16_filter_DX,
SI8_D16_nofilter_DXDY,
SI8_D16_nofilter_DX,
SI8_D16_filter_DXDY,
SI8_D16_filter_DX,
// Don't support 4444 -> 565
NULL, NULL, NULL, NULL,
// Don't support A8 -> 565
NULL, NULL, NULL, NULL,
// Don't support G8 -> 565 (but we could)
NULL, NULL, NULL, NULL
};
#endif
fSampleProc32 = SK_ARM_NEON_WRAP(gSkBitmapProcStateSample32)[index];
index >>= 1; // shift away any opaque/alpha distinction
fSampleProc16 = SK_ARM_NEON_WRAP(gSkBitmapProcStateSample16)[index];
// our special-case shaderprocs
if (SK_ARM_NEON_WRAP(S16_D16_filter_DX) == fSampleProc16) {
if (clampClamp) {
fShaderProc16 = SK_ARM_NEON_WRAP(Clamp_S16_D16_filter_DX_shaderproc);
} else if (SkShader::kRepeat_TileMode == fTileModeX &&
SkShader::kRepeat_TileMode == fTileModeY) {
fShaderProc16 = SK_ARM_NEON_WRAP(Repeat_S16_D16_filter_DX_shaderproc);
}
} else if (SK_ARM_NEON_WRAP(SI8_opaque_D32_filter_DX) == fSampleProc32 && clampClamp) {
fShaderProc32 = SK_ARM_NEON_WRAP(Clamp_SI8_opaque_D32_filter_DX_shaderproc);
} else if (S32_opaque_D32_nofilter_DX == fSampleProc32 && clampClamp) {
fShaderProc32 = Clamp_S32_opaque_D32_nofilter_DX_shaderproc;
}
if (NULL == fShaderProc32) {
fShaderProc32 = this->chooseShaderProc32();
}
}
// see if our platform has any accelerated overrides
this->platformProcs();
return true;
}
static void Clamp_S32_D32_nofilter_trans_shaderproc(const SkBitmapProcState& s,
int x, int y,
SkPMColor* SK_RESTRICT colors,
int count) {
SkASSERT(((s.fInvType & ~SkMatrix::kTranslate_Mask)) == 0);
SkASSERT(s.fInvKy == 0);
SkASSERT(count > 0 && colors != NULL);
SkASSERT(kNone_SkFilterQuality == s.fFilterLevel);
const int maxX = s.fBitmap->width() - 1;
const int maxY = s.fBitmap->height() - 1;
int ix = s.fFilterOneX + x;
int iy = SkClampMax(s.fFilterOneY + y, maxY);
#ifdef SK_DEBUG
{
SkPoint pt;
s.fInvProc(s.fInvMatrix, SkIntToScalar(x) + SK_ScalarHalf,
SkIntToScalar(y) + SK_ScalarHalf, &pt);
int iy2 = SkClampMax(SkScalarFloorToInt(pt.fY), maxY);
int ix2 = SkScalarFloorToInt(pt.fX);
SkASSERT(iy == iy2);
SkASSERT(ix == ix2);
}
#endif
const SkPMColor* row = s.fBitmap->getAddr32(0, iy);
// clamp to the left
if (ix < 0) {
int n = SkMin32(-ix, count);
sk_memset32(colors, row[0], n);
count -= n;
if (0 == count) {
return;
}
colors += n;
SkASSERT(-ix == n);
ix = 0;
}
// copy the middle
if (ix <= maxX) {
int n = SkMin32(maxX - ix + 1, count);
memcpy(colors, row + ix, n * sizeof(SkPMColor));
count -= n;
if (0 == count) {
return;
}
colors += n;
}
SkASSERT(count > 0);
// clamp to the right
sk_memset32(colors, row[maxX], count);
}
static inline int sk_int_mod(int x, int n) {
SkASSERT(n > 0);
if ((unsigned)x >= (unsigned)n) {
if (x < 0) {
x = n + ~(~x % n);
} else {
x = x % n;
}
}
return x;
}
static inline int sk_int_mirror(int x, int n) {
x = sk_int_mod(x, 2 * n);
if (x >= n) {
x = n + ~(x - n);
}
return x;
}
static void Repeat_S32_D32_nofilter_trans_shaderproc(const SkBitmapProcState& s,
int x, int y,
SkPMColor* SK_RESTRICT colors,
int count) {
SkASSERT(((s.fInvType & ~SkMatrix::kTranslate_Mask)) == 0);
SkASSERT(s.fInvKy == 0);
SkASSERT(count > 0 && colors != NULL);
SkASSERT(kNone_SkFilterQuality == s.fFilterLevel);
const int stopX = s.fBitmap->width();
const int stopY = s.fBitmap->height();
int ix = s.fFilterOneX + x;
int iy = sk_int_mod(s.fFilterOneY + y, stopY);
#ifdef SK_DEBUG
{
SkPoint pt;
s.fInvProc(s.fInvMatrix, SkIntToScalar(x) + SK_ScalarHalf,
SkIntToScalar(y) + SK_ScalarHalf, &pt);
int iy2 = sk_int_mod(SkScalarFloorToInt(pt.fY), stopY);
int ix2 = SkScalarFloorToInt(pt.fX);
SkASSERT(iy == iy2);
SkASSERT(ix == ix2);
}
#endif
const SkPMColor* row = s.fBitmap->getAddr32(0, iy);
ix = sk_int_mod(ix, stopX);
for (;;) {
int n = SkMin32(stopX - ix, count);
memcpy(colors, row + ix, n * sizeof(SkPMColor));
count -= n;
if (0 == count) {
return;
}
colors += n;
ix = 0;
}
}
static void S32_D32_constX_shaderproc(const SkBitmapProcState& s,
int x, int y,
SkPMColor* SK_RESTRICT colors,
int count) {
SkASSERT((s.fInvType & ~(SkMatrix::kTranslate_Mask | SkMatrix::kScale_Mask)) == 0);
SkASSERT(s.fInvKy == 0);
SkASSERT(count > 0 && colors != NULL);
SkASSERT(1 == s.fBitmap->width());
int iY0;
int iY1 SK_INIT_TO_AVOID_WARNING;
int iSubY SK_INIT_TO_AVOID_WARNING;
if (kNone_SkFilterQuality != s.fFilterLevel) {
SkBitmapProcState::MatrixProc mproc = s.getMatrixProc();
uint32_t xy[2];
mproc(s, xy, 1, x, y);
iY0 = xy[0] >> 18;
iY1 = xy[0] & 0x3FFF;
iSubY = (xy[0] >> 14) & 0xF;
} else {
int yTemp;
if (s.fInvType > SkMatrix::kTranslate_Mask) {
SkPoint pt;
s.fInvProc(s.fInvMatrix,
SkIntToScalar(x) + SK_ScalarHalf,
SkIntToScalar(y) + SK_ScalarHalf,
&pt);
// When the matrix has a scale component the setup code in
// chooseProcs multiples the inverse matrix by the inverse of the
// bitmap's width and height. Since this method is going to do
// its own tiling and sampling we need to undo that here.
if (SkShader::kClamp_TileMode != s.fTileModeX ||
SkShader::kClamp_TileMode != s.fTileModeY) {
yTemp = SkScalarFloorToInt(pt.fY * s.fBitmap->height());
} else {
yTemp = SkScalarFloorToInt(pt.fY);
}
} else {
yTemp = s.fFilterOneY + y;
}
const int stopY = s.fBitmap->height();
switch (s.fTileModeY) {
case SkShader::kClamp_TileMode:
iY0 = SkClampMax(yTemp, stopY-1);
break;
case SkShader::kRepeat_TileMode:
iY0 = sk_int_mod(yTemp, stopY);
break;
case SkShader::kMirror_TileMode:
default:
iY0 = sk_int_mirror(yTemp, stopY);
break;
}
#ifdef SK_DEBUG
{
SkPoint pt;
s.fInvProc(s.fInvMatrix,
SkIntToScalar(x) + SK_ScalarHalf,
SkIntToScalar(y) + SK_ScalarHalf,
&pt);
if (s.fInvType > SkMatrix::kTranslate_Mask &&
(SkShader::kClamp_TileMode != s.fTileModeX ||
SkShader::kClamp_TileMode != s.fTileModeY)) {
pt.fY *= s.fBitmap->height();
}
int iY2;
switch (s.fTileModeY) {
case SkShader::kClamp_TileMode:
iY2 = SkClampMax(SkScalarFloorToInt(pt.fY), stopY-1);
break;
case SkShader::kRepeat_TileMode:
iY2 = sk_int_mod(SkScalarFloorToInt(pt.fY), stopY);
break;
case SkShader::kMirror_TileMode:
default:
iY2 = sk_int_mirror(SkScalarFloorToInt(pt.fY), stopY);
break;
}
SkASSERT(iY0 == iY2);
}
#endif
}
const SkPMColor* row0 = s.fBitmap->getAddr32(0, iY0);
SkPMColor color;
if (kNone_SkFilterQuality != s.fFilterLevel) {
const SkPMColor* row1 = s.fBitmap->getAddr32(0, iY1);
if (s.fAlphaScale < 256) {
Filter_32_alpha(iSubY, *row0, *row1, &color, s.fAlphaScale);
} else {
Filter_32_opaque(iSubY, *row0, *row1, &color);
}
} else {
if (s.fAlphaScale < 256) {
color = SkAlphaMulQ(*row0, s.fAlphaScale);
} else {
color = *row0;
}
}
sk_memset32(colors, color, count);
}
static void DoNothing_shaderproc(const SkBitmapProcState&, int x, int y,
SkPMColor* SK_RESTRICT colors, int count) {
// if we get called, the matrix is too tricky, so we just draw nothing
sk_memset32(colors, 0, count);
}
bool SkBitmapProcState::setupForTranslate() {
SkPoint pt;
fInvProc(fInvMatrix, SK_ScalarHalf, SK_ScalarHalf, &pt);
/*
* if the translate is larger than our ints, we can get random results, or
* worse, we might get 0x80000000, which wreaks havoc on us, since we can't
* negate it.
*/
const SkScalar too_big = SkIntToScalar(1 << 30);
if (SkScalarAbs(pt.fX) > too_big || SkScalarAbs(pt.fY) > too_big) {
return false;
}
// Since we know we're not filtered, we re-purpose these fields allow
// us to go from device -> src coordinates w/ just an integer add,
// rather than running through the inverse-matrix
fFilterOneX = SkScalarFloorToInt(pt.fX);
fFilterOneY = SkScalarFloorToInt(pt.fY);
return true;
}
SkBitmapProcState::ShaderProc32 SkBitmapProcState::chooseShaderProc32() {
if (kN32_SkColorType != fBitmap->colorType()) {
return NULL;
}
static const unsigned kMask = SkMatrix::kTranslate_Mask | SkMatrix::kScale_Mask;
if (1 == fBitmap->width() && 0 == (fInvType & ~kMask)) {
if (kNone_SkFilterQuality == fFilterLevel &&
fInvType <= SkMatrix::kTranslate_Mask &&
!this->setupForTranslate()) {
return DoNothing_shaderproc;
}
return S32_D32_constX_shaderproc;
}
if (fAlphaScale < 256) {
return NULL;
}
if (fInvType > SkMatrix::kTranslate_Mask) {
return NULL;
}
if (kNone_SkFilterQuality != fFilterLevel) {
return NULL;
}
SkShader::TileMode tx = (SkShader::TileMode)fTileModeX;
SkShader::TileMode ty = (SkShader::TileMode)fTileModeY;
if (SkShader::kClamp_TileMode == tx && SkShader::kClamp_TileMode == ty) {
if (this->setupForTranslate()) {
return Clamp_S32_D32_nofilter_trans_shaderproc;
}
return DoNothing_shaderproc;
}
if (SkShader::kRepeat_TileMode == tx && SkShader::kRepeat_TileMode == ty) {
if (this->setupForTranslate()) {
return Repeat_S32_D32_nofilter_trans_shaderproc;
}
return DoNothing_shaderproc;
}
return NULL;
}
///////////////////////////////////////////////////////////////////////////////
#ifdef SK_DEBUG
static void check_scale_nofilter(uint32_t bitmapXY[], int count,
unsigned mx, unsigned my) {
unsigned y = *bitmapXY++;
SkASSERT(y < my);
const uint16_t* xptr = reinterpret_cast<const uint16_t*>(bitmapXY);
for (int i = 0; i < count; ++i) {
SkASSERT(xptr[i] < mx);
}
}
static void check_scale_filter(uint32_t bitmapXY[], int count,
unsigned mx, unsigned my) {
uint32_t YY = *bitmapXY++;
unsigned y0 = YY >> 18;
unsigned y1 = YY & 0x3FFF;
SkASSERT(y0 < my);
SkASSERT(y1 < my);
for (int i = 0; i < count; ++i) {
uint32_t XX = bitmapXY[i];
unsigned x0 = XX >> 18;
unsigned x1 = XX & 0x3FFF;
SkASSERT(x0 < mx);
SkASSERT(x1 < mx);
}
}
static void check_affine_nofilter(uint32_t bitmapXY[], int count,
unsigned mx, unsigned my) {
for (int i = 0; i < count; ++i) {
uint32_t XY = bitmapXY[i];
unsigned x = XY & 0xFFFF;
unsigned y = XY >> 16;
SkASSERT(x < mx);
SkASSERT(y < my);
}
}
static void check_affine_filter(uint32_t bitmapXY[], int count,
unsigned mx, unsigned my) {
for (int i = 0; i < count; ++i) {
uint32_t YY = *bitmapXY++;
unsigned y0 = YY >> 18;
unsigned y1 = YY & 0x3FFF;
SkASSERT(y0 < my);
SkASSERT(y1 < my);
uint32_t XX = *bitmapXY++;
unsigned x0 = XX >> 18;
unsigned x1 = XX & 0x3FFF;
SkASSERT(x0 < mx);
SkASSERT(x1 < mx);
}
}
void SkBitmapProcState::DebugMatrixProc(const SkBitmapProcState& state,
uint32_t bitmapXY[], int count,
int x, int y) {
SkASSERT(bitmapXY);
SkASSERT(count > 0);
state.fMatrixProc(state, bitmapXY, count, x, y);
void (*proc)(uint32_t bitmapXY[], int count, unsigned mx, unsigned my);
// There are four formats possible:
// scale -vs- affine
// filter -vs- nofilter
if (state.fInvType <= (SkMatrix::kTranslate_Mask | SkMatrix::kScale_Mask)) {
proc = state.fFilterLevel != kNone_SkFilterQuality ? check_scale_filter : check_scale_nofilter;
} else {
proc = state.fFilterLevel != kNone_SkFilterQuality ? check_affine_filter : check_affine_nofilter;
}
proc(bitmapXY, count, state.fBitmap->width(), state.fBitmap->height());
}
SkBitmapProcState::MatrixProc SkBitmapProcState::getMatrixProc() const {
return DebugMatrixProc;
}
#endif
///////////////////////////////////////////////////////////////////////////////
/*
The storage requirements for the different matrix procs are as follows,
where each X or Y is 2 bytes, and N is the number of pixels/elements:
scale/translate nofilter Y(4bytes) + N * X
affine/perspective nofilter N * (X Y)
scale/translate filter Y Y + N * (X X)
affine/perspective filter N * (Y Y X X)
*/
int SkBitmapProcState::maxCountForBufferSize(size_t bufferSize) const {
int32_t size = static_cast<int32_t>(bufferSize);
size &= ~3; // only care about 4-byte aligned chunks
if (fInvType <= (SkMatrix::kTranslate_Mask | SkMatrix::kScale_Mask)) {
size -= 4; // the shared Y (or YY) coordinate
if (size < 0) {
size = 0;
}
size >>= 1;
} else {
size >>= 2;
}
if (fFilterLevel != kNone_SkFilterQuality) {
size >>= 1;
}
return size;
}
///////////////////////
void Clamp_S32_opaque_D32_nofilter_DX_shaderproc(const SkBitmapProcState& s, int x, int y,
SkPMColor* SK_RESTRICT dst, int count) {
SkASSERT((s.fInvType & ~(SkMatrix::kTranslate_Mask |
SkMatrix::kScale_Mask)) == 0);
const unsigned maxX = s.fBitmap->width() - 1;
SkFractionalInt fx;
int dstY;
{
SkPoint pt;
s.fInvProc(s.fInvMatrix, SkIntToScalar(x) + SK_ScalarHalf, SkIntToScalar(y) + SK_ScalarHalf,
&pt);
fx = SkScalarToFractionalInt(pt.fY);
const unsigned maxY = s.fBitmap->height() - 1;
dstY = SkClampMax(SkFractionalIntToInt(fx), maxY);
fx = SkScalarToFractionalInt(pt.fX);
}
const SkPMColor* SK_RESTRICT src = s.fBitmap->getAddr32(0, dstY);
const SkFractionalInt dx = s.fInvSxFractionalInt;
// Check if we're safely inside [0...maxX] so no need to clamp each computed index.
//
if ((uint64_t)SkFractionalIntToInt(fx) <= maxX &&
(uint64_t)SkFractionalIntToInt(fx + dx * (count - 1)) <= maxX)
{
int count4 = count >> 2;
for (int i = 0; i < count4; ++i) {
SkPMColor src0 = src[SkFractionalIntToInt(fx)]; fx += dx;
SkPMColor src1 = src[SkFractionalIntToInt(fx)]; fx += dx;
SkPMColor src2 = src[SkFractionalIntToInt(fx)]; fx += dx;
SkPMColor src3 = src[SkFractionalIntToInt(fx)]; fx += dx;
dst[0] = src0;
dst[1] = src1;
dst[2] = src2;
dst[3] = src3;
dst += 4;
}
for (int i = (count4 << 2); i < count; ++i) {
unsigned index = SkFractionalIntToInt(fx);
SkASSERT(index <= maxX);
*dst++ = src[index];
fx += dx;
}
} else {
for (int i = 0; i < count; ++i) {
dst[i] = src[SkClampMax(SkFractionalIntToInt(fx), maxX)];
fx += dx;
}
}
}
|