-
Notifications
You must be signed in to change notification settings - Fork 54
/
pixman-mmx.c
4084 lines (3385 loc) · 96.5 KB
/
pixman-mmx.c
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
/*
* Copyright © 2004, 2005 Red Hat, Inc.
* Copyright © 2004 Nicholas Miell
* Copyright © 2005 Trolltech AS
*
* Permission to use, copy, modify, distribute, and sell this software and its
* documentation for any purpose is hereby granted without fee, provided that
* the above copyright notice appear in all copies and that both that
* copyright notice and this permission notice appear in supporting
* documentation, and that the name of Red Hat not be used in advertising or
* publicity pertaining to distribution of the software without specific,
* written prior permission. Red Hat makes no representations about the
* suitability of this software for any purpose. It is provided "as is"
* without express or implied warranty.
*
* THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
* SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
* FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
* SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
* AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
* OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
* SOFTWARE.
*
* Author: Søren Sandmann (sandmann@redhat.com)
* Minor Improvements: Nicholas Miell (nmiell@gmail.com)
* MMX code paths for fbcompose.c by Lars Knoll (lars@trolltech.com)
*
* Based on work by Owen Taylor
*/
#ifdef HAVE_CONFIG_H
#include <config.h>
#endif
#if defined USE_X86_MMX || defined USE_ARM_IWMMXT || defined USE_LOONGSON_MMI
#ifdef USE_LOONGSON_MMI
#include <loongson-mmintrin.h>
#else
#include <mmintrin.h>
#endif
#include "pixman-private.h"
#include "pixman-combine32.h"
#include "pixman-inlines.h"
#define no_vERBOSE
#ifdef VERBOSE
#define CHECKPOINT() error_f ("at %s %d\n", __FUNCTION__, __LINE__)
#else
#define CHECKPOINT()
#endif
#if defined USE_ARM_IWMMXT && __GNUC__ == 4 && __GNUC_MINOR__ < 8
/* Empty the multimedia state. For some reason, ARM's mmintrin.h doesn't provide this. */
extern __inline void __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_empty (void)
{
}
#endif
#ifdef USE_X86_MMX
# if (defined(__SUNPRO_C) || defined(_MSC_VER) || defined(_WIN64))
# include <xmmintrin.h>
# else
/* We have to compile with -msse to use xmmintrin.h, but that causes SSE
* instructions to be generated that we don't want. Just duplicate the
* functions we want to use. */
extern __inline int __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_movemask_pi8 (__m64 __A)
{
int ret;
asm ("pmovmskb %1, %0\n\t"
: "=r" (ret)
: "y" (__A)
);
return ret;
}
extern __inline __m64 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_mulhi_pu16 (__m64 __A, __m64 __B)
{
asm ("pmulhuw %1, %0\n\t"
: "+y" (__A)
: "y" (__B)
);
return __A;
}
# ifdef __OPTIMIZE__
extern __inline __m64 __attribute__((__gnu_inline__, __always_inline__, __artificial__))
_mm_shuffle_pi16 (__m64 __A, int8_t const __N)
{
__m64 ret;
asm ("pshufw %2, %1, %0\n\t"
: "=y" (ret)
: "y" (__A), "K" (__N)
);
return ret;
}
# else
# define _mm_shuffle_pi16(A, N) \
({ \
__m64 ret; \
\
asm ("pshufw %2, %1, %0\n\t" \
: "=y" (ret) \
: "y" (A), "K" ((const int8_t)N) \
); \
\
ret; \
})
# endif
# endif
#endif
#ifndef _MSC_VER
#define _MM_SHUFFLE(fp3,fp2,fp1,fp0) \
(((fp3) << 6) | ((fp2) << 4) | ((fp1) << 2) | (fp0))
#endif
/* Notes about writing mmx code
*
* give memory operands as the second operand. If you give it as the
* first, gcc will first load it into a register, then use that
* register
*
* ie. use
*
* _mm_mullo_pi16 (x, mmx_constant);
*
* not
*
* _mm_mullo_pi16 (mmx_constant, x);
*
* Also try to minimize dependencies. i.e. when you need a value, try
* to calculate it from a value that was calculated as early as
* possible.
*/
/* --------------- MMX primitives ------------------------------------- */
/* If __m64 is defined as a struct or union, then define M64_MEMBER to be
* the name of the member used to access the data.
* If __m64 requires using mm_cvt* intrinsics functions to convert between
* uint64_t and __m64 values, then define USE_CVT_INTRINSICS.
* If __m64 and uint64_t values can just be cast to each other directly,
* then define USE_M64_CASTS.
* If __m64 is a double datatype, then define USE_M64_DOUBLE.
*/
#ifdef _MSC_VER
# define M64_MEMBER m64_u64
#elif defined(__ICC)
# define USE_CVT_INTRINSICS
#elif defined(USE_LOONGSON_MMI)
# define USE_M64_DOUBLE
#elif defined(__GNUC__)
# define USE_M64_CASTS
#elif defined(__SUNPRO_C)
# if (__SUNPRO_C >= 0x5120) && !defined(__NOVECTORSIZE__)
/* Solaris Studio 12.3 (Sun C 5.12) introduces __attribute__(__vector_size__)
* support, and defaults to using it to define __m64, unless __NOVECTORSIZE__
* is defined. If it is used, then the mm_cvt* intrinsics must be used.
*/
# define USE_CVT_INTRINSICS
# else
/* For Studio 12.2 or older, or when __attribute__(__vector_size__) is
* disabled, __m64 is defined as a struct containing "unsigned long long l_".
*/
# define M64_MEMBER l_
# endif
#endif
#if defined(USE_M64_CASTS) || defined(USE_CVT_INTRINSICS) || defined(USE_M64_DOUBLE)
typedef uint64_t mmxdatafield;
#else
typedef __m64 mmxdatafield;
#endif
typedef struct
{
mmxdatafield mmx_4x00ff;
mmxdatafield mmx_4x0080;
mmxdatafield mmx_565_rgb;
mmxdatafield mmx_565_unpack_multiplier;
mmxdatafield mmx_565_pack_multiplier;
mmxdatafield mmx_565_r;
mmxdatafield mmx_565_g;
mmxdatafield mmx_565_b;
mmxdatafield mmx_packed_565_rb;
mmxdatafield mmx_packed_565_g;
mmxdatafield mmx_expand_565_g;
mmxdatafield mmx_expand_565_b;
mmxdatafield mmx_expand_565_r;
#ifndef USE_LOONGSON_MMI
mmxdatafield mmx_mask_0;
mmxdatafield mmx_mask_1;
mmxdatafield mmx_mask_2;
mmxdatafield mmx_mask_3;
#endif
mmxdatafield mmx_full_alpha;
mmxdatafield mmx_4x0101;
mmxdatafield mmx_ff000000;
} mmx_data_t;
#if defined(_MSC_VER)
# define MMXDATA_INIT(field, val) { val ## UI64 }
#elif defined(M64_MEMBER) /* __m64 is a struct, not an integral type */
# define MMXDATA_INIT(field, val) field = { val ## ULL }
#else /* mmxdatafield is an integral type */
# define MMXDATA_INIT(field, val) field = val ## ULL
#endif
static const mmx_data_t c =
{
MMXDATA_INIT (.mmx_4x00ff, 0x00ff00ff00ff00ff),
MMXDATA_INIT (.mmx_4x0080, 0x0080008000800080),
MMXDATA_INIT (.mmx_565_rgb, 0x000001f0003f001f),
MMXDATA_INIT (.mmx_565_unpack_multiplier, 0x0000008404100840),
MMXDATA_INIT (.mmx_565_pack_multiplier, 0x2000000420000004),
MMXDATA_INIT (.mmx_565_r, 0x000000f800000000),
MMXDATA_INIT (.mmx_565_g, 0x0000000000fc0000),
MMXDATA_INIT (.mmx_565_b, 0x00000000000000f8),
MMXDATA_INIT (.mmx_packed_565_rb, 0x00f800f800f800f8),
MMXDATA_INIT (.mmx_packed_565_g, 0x0000fc000000fc00),
MMXDATA_INIT (.mmx_expand_565_g, 0x07e007e007e007e0),
MMXDATA_INIT (.mmx_expand_565_b, 0x001f001f001f001f),
MMXDATA_INIT (.mmx_expand_565_r, 0xf800f800f800f800),
#ifndef USE_LOONGSON_MMI
MMXDATA_INIT (.mmx_mask_0, 0xffffffffffff0000),
MMXDATA_INIT (.mmx_mask_1, 0xffffffff0000ffff),
MMXDATA_INIT (.mmx_mask_2, 0xffff0000ffffffff),
MMXDATA_INIT (.mmx_mask_3, 0x0000ffffffffffff),
#endif
MMXDATA_INIT (.mmx_full_alpha, 0x00ff000000000000),
MMXDATA_INIT (.mmx_4x0101, 0x0101010101010101),
MMXDATA_INIT (.mmx_ff000000, 0xff000000ff000000),
};
#ifdef USE_CVT_INTRINSICS
# define MC(x) to_m64 (c.mmx_ ## x)
#elif defined(USE_M64_CASTS)
# define MC(x) ((__m64)c.mmx_ ## x)
#elif defined(USE_M64_DOUBLE)
# define MC(x) (*(__m64 *)&c.mmx_ ## x)
#else
# define MC(x) c.mmx_ ## x
#endif
static force_inline __m64
to_m64 (uint64_t x)
{
#ifdef USE_CVT_INTRINSICS
return _mm_cvtsi64_m64 (x);
#elif defined M64_MEMBER /* __m64 is a struct, not an integral type */
__m64 res;
res.M64_MEMBER = x;
return res;
#elif defined USE_M64_DOUBLE
return *(__m64 *)&x;
#else /* USE_M64_CASTS */
return (__m64)x;
#endif
}
static force_inline uint64_t
to_uint64 (__m64 x)
{
#ifdef USE_CVT_INTRINSICS
return _mm_cvtm64_si64 (x);
#elif defined M64_MEMBER /* __m64 is a struct, not an integral type */
uint64_t res = x.M64_MEMBER;
return res;
#elif defined USE_M64_DOUBLE
return *(uint64_t *)&x;
#else /* USE_M64_CASTS */
return (uint64_t)x;
#endif
}
static force_inline __m64
shift (__m64 v,
int s)
{
if (s > 0)
return _mm_slli_si64 (v, s);
else if (s < 0)
return _mm_srli_si64 (v, -s);
else
return v;
}
static force_inline __m64
negate (__m64 mask)
{
return _mm_xor_si64 (mask, MC (4x00ff));
}
static force_inline __m64
pix_multiply (__m64 a, __m64 b)
{
__m64 res;
res = _mm_mullo_pi16 (a, b);
res = _mm_adds_pu16 (res, MC (4x0080));
res = _mm_mulhi_pu16 (res, MC (4x0101));
return res;
}
static force_inline __m64
pix_add (__m64 a, __m64 b)
{
return _mm_adds_pu8 (a, b);
}
static force_inline __m64
expand_alpha (__m64 pixel)
{
return _mm_shuffle_pi16 (pixel, _MM_SHUFFLE (3, 3, 3, 3));
}
static force_inline __m64
expand_alpha_rev (__m64 pixel)
{
return _mm_shuffle_pi16 (pixel, _MM_SHUFFLE (0, 0, 0, 0));
}
static force_inline __m64
invert_colors (__m64 pixel)
{
return _mm_shuffle_pi16 (pixel, _MM_SHUFFLE (3, 0, 1, 2));
}
static force_inline __m64
over (__m64 src,
__m64 srca,
__m64 dest)
{
return _mm_adds_pu8 (src, pix_multiply (dest, negate (srca)));
}
static force_inline __m64
over_rev_non_pre (__m64 src, __m64 dest)
{
__m64 srca = expand_alpha (src);
__m64 srcfaaa = _mm_or_si64 (srca, MC (full_alpha));
return over (pix_multiply (invert_colors (src), srcfaaa), srca, dest);
}
static force_inline __m64
in (__m64 src, __m64 mask)
{
return pix_multiply (src, mask);
}
#ifndef _MSC_VER
static force_inline __m64
in_over (__m64 src, __m64 srca, __m64 mask, __m64 dest)
{
return over (in (src, mask), pix_multiply (srca, mask), dest);
}
#else
#define in_over(src, srca, mask, dest) \
over (in (src, mask), pix_multiply (srca, mask), dest)
#endif
/* Elemental unaligned loads */
static force_inline __m64 ldq_u(__m64 *p)
{
#ifdef USE_X86_MMX
/* x86's alignment restrictions are very relaxed. */
return *(__m64 *)p;
#elif defined USE_ARM_IWMMXT
int align = (uintptr_t)p & 7;
__m64 *aligned_p;
if (align == 0)
return *p;
aligned_p = (__m64 *)((uintptr_t)p & ~7);
return (__m64) _mm_align_si64 (aligned_p[0], aligned_p[1], align);
#else
struct __una_u64 { __m64 x __attribute__((packed)); };
const struct __una_u64 *ptr = (const struct __una_u64 *) p;
return (__m64) ptr->x;
#endif
}
static force_inline uint32_t ldl_u(const uint32_t *p)
{
#ifdef USE_X86_MMX
/* x86's alignment restrictions are very relaxed. */
return *p;
#else
struct __una_u32 { uint32_t x __attribute__((packed)); };
const struct __una_u32 *ptr = (const struct __una_u32 *) p;
return ptr->x;
#endif
}
static force_inline __m64
load (const uint32_t *v)
{
#ifdef USE_LOONGSON_MMI
__m64 ret;
asm ("lwc1 %0, %1\n\t"
: "=f" (ret)
: "m" (*v)
);
return ret;
#else
return _mm_cvtsi32_si64 (*v);
#endif
}
static force_inline __m64
load8888 (const uint32_t *v)
{
#ifdef USE_LOONGSON_MMI
return _mm_unpacklo_pi8_f (*(__m32 *)v, _mm_setzero_si64 ());
#else
return _mm_unpacklo_pi8 (load (v), _mm_setzero_si64 ());
#endif
}
static force_inline __m64
load8888u (const uint32_t *v)
{
uint32_t l = ldl_u (v);
return load8888 (&l);
}
static force_inline __m64
pack8888 (__m64 lo, __m64 hi)
{
return _mm_packs_pu16 (lo, hi);
}
static force_inline void
store (uint32_t *dest, __m64 v)
{
#ifdef USE_LOONGSON_MMI
asm ("swc1 %1, %0\n\t"
: "=m" (*dest)
: "f" (v)
: "memory"
);
#else
*dest = _mm_cvtsi64_si32 (v);
#endif
}
static force_inline void
store8888 (uint32_t *dest, __m64 v)
{
v = pack8888 (v, _mm_setzero_si64 ());
store (dest, v);
}
static force_inline pixman_bool_t
is_equal (__m64 a, __m64 b)
{
#ifdef USE_LOONGSON_MMI
/* __m64 is double, we can compare directly. */
return a == b;
#else
return _mm_movemask_pi8 (_mm_cmpeq_pi8 (a, b)) == 0xff;
#endif
}
static force_inline pixman_bool_t
is_opaque (__m64 v)
{
#ifdef USE_LOONGSON_MMI
return is_equal (_mm_and_si64 (v, MC (full_alpha)), MC (full_alpha));
#else
__m64 ffs = _mm_cmpeq_pi8 (v, v);
return (_mm_movemask_pi8 (_mm_cmpeq_pi8 (v, ffs)) & 0x40);
#endif
}
static force_inline pixman_bool_t
is_zero (__m64 v)
{
return is_equal (v, _mm_setzero_si64 ());
}
/* Expand 16 bits positioned at @pos (0-3) of a mmx register into
*
* 00RR00GG00BB
*
* --- Expanding 565 in the low word ---
*
* m = (m << (32 - 3)) | (m << (16 - 5)) | m;
* m = m & (01f0003f001f);
* m = m * (008404100840);
* m = m >> 8;
*
* Note the trick here - the top word is shifted by another nibble to
* avoid it bumping into the middle word
*/
static force_inline __m64
expand565 (__m64 pixel, int pos)
{
__m64 p = pixel;
__m64 t1, t2;
/* move pixel to low 16 bit and zero the rest */
#ifdef USE_LOONGSON_MMI
p = loongson_extract_pi16 (p, pos);
#else
p = shift (shift (p, (3 - pos) * 16), -48);
#endif
t1 = shift (p, 36 - 11);
t2 = shift (p, 16 - 5);
p = _mm_or_si64 (t1, p);
p = _mm_or_si64 (t2, p);
p = _mm_and_si64 (p, MC (565_rgb));
pixel = _mm_mullo_pi16 (p, MC (565_unpack_multiplier));
return _mm_srli_pi16 (pixel, 8);
}
/* Expand 4 16 bit pixels in an mmx register into two mmx registers of
*
* AARRGGBBRRGGBB
*/
static force_inline void
expand_4xpacked565 (__m64 vin, __m64 *vout0, __m64 *vout1, int full_alpha)
{
__m64 t0, t1, alpha = _mm_setzero_si64 ();
__m64 r = _mm_and_si64 (vin, MC (expand_565_r));
__m64 g = _mm_and_si64 (vin, MC (expand_565_g));
__m64 b = _mm_and_si64 (vin, MC (expand_565_b));
if (full_alpha)
alpha = _mm_cmpeq_pi32 (alpha, alpha);
/* Replicate high bits into empty low bits. */
r = _mm_or_si64 (_mm_srli_pi16 (r, 8), _mm_srli_pi16 (r, 13));
g = _mm_or_si64 (_mm_srli_pi16 (g, 3), _mm_srli_pi16 (g, 9));
b = _mm_or_si64 (_mm_slli_pi16 (b, 3), _mm_srli_pi16 (b, 2));
r = _mm_packs_pu16 (r, _mm_setzero_si64 ()); /* 00 00 00 00 R3 R2 R1 R0 */
g = _mm_packs_pu16 (g, _mm_setzero_si64 ()); /* 00 00 00 00 G3 G2 G1 G0 */
b = _mm_packs_pu16 (b, _mm_setzero_si64 ()); /* 00 00 00 00 B3 B2 B1 B0 */
t1 = _mm_unpacklo_pi8 (r, alpha); /* A3 R3 A2 R2 A1 R1 A0 R0 */
t0 = _mm_unpacklo_pi8 (b, g); /* G3 B3 G2 B2 G1 B1 G0 B0 */
*vout0 = _mm_unpacklo_pi16 (t0, t1); /* A1 R1 G1 B1 A0 R0 G0 B0 */
*vout1 = _mm_unpackhi_pi16 (t0, t1); /* A3 R3 G3 B3 A2 R2 G2 B2 */
}
static force_inline __m64
expand8888 (__m64 in, int pos)
{
if (pos == 0)
return _mm_unpacklo_pi8 (in, _mm_setzero_si64 ());
else
return _mm_unpackhi_pi8 (in, _mm_setzero_si64 ());
}
static force_inline __m64
expandx888 (__m64 in, int pos)
{
return _mm_or_si64 (expand8888 (in, pos), MC (full_alpha));
}
static force_inline void
expand_4x565 (__m64 vin, __m64 *vout0, __m64 *vout1, __m64 *vout2, __m64 *vout3, int full_alpha)
{
__m64 v0, v1;
expand_4xpacked565 (vin, &v0, &v1, full_alpha);
*vout0 = expand8888 (v0, 0);
*vout1 = expand8888 (v0, 1);
*vout2 = expand8888 (v1, 0);
*vout3 = expand8888 (v1, 1);
}
static force_inline __m64
pack_565 (__m64 pixel, __m64 target, int pos)
{
__m64 p = pixel;
__m64 t = target;
__m64 r, g, b;
r = _mm_and_si64 (p, MC (565_r));
g = _mm_and_si64 (p, MC (565_g));
b = _mm_and_si64 (p, MC (565_b));
#ifdef USE_LOONGSON_MMI
r = shift (r, -(32 - 8));
g = shift (g, -(16 - 3));
b = shift (b, -(0 + 3));
p = _mm_or_si64 (r, g);
p = _mm_or_si64 (p, b);
return loongson_insert_pi16 (t, p, pos);
#else
r = shift (r, -(32 - 8) + pos * 16);
g = shift (g, -(16 - 3) + pos * 16);
b = shift (b, -(0 + 3) + pos * 16);
if (pos == 0)
t = _mm_and_si64 (t, MC (mask_0));
else if (pos == 1)
t = _mm_and_si64 (t, MC (mask_1));
else if (pos == 2)
t = _mm_and_si64 (t, MC (mask_2));
else if (pos == 3)
t = _mm_and_si64 (t, MC (mask_3));
p = _mm_or_si64 (r, t);
p = _mm_or_si64 (g, p);
return _mm_or_si64 (b, p);
#endif
}
static force_inline __m64
pack_4xpacked565 (__m64 a, __m64 b)
{
__m64 rb0 = _mm_and_si64 (a, MC (packed_565_rb));
__m64 rb1 = _mm_and_si64 (b, MC (packed_565_rb));
__m64 t0 = _mm_madd_pi16 (rb0, MC (565_pack_multiplier));
__m64 t1 = _mm_madd_pi16 (rb1, MC (565_pack_multiplier));
__m64 g0 = _mm_and_si64 (a, MC (packed_565_g));
__m64 g1 = _mm_and_si64 (b, MC (packed_565_g));
t0 = _mm_or_si64 (t0, g0);
t1 = _mm_or_si64 (t1, g1);
t0 = shift(t0, -5);
#ifdef USE_ARM_IWMMXT
t1 = shift(t1, -5);
return _mm_packs_pu32 (t0, t1);
#else
t1 = shift(t1, -5 + 16);
return _mm_shuffle_pi16 (_mm_or_si64 (t0, t1), _MM_SHUFFLE (3, 1, 2, 0));
#endif
}
#ifndef _MSC_VER
static force_inline __m64
pack_4x565 (__m64 v0, __m64 v1, __m64 v2, __m64 v3)
{
return pack_4xpacked565 (pack8888 (v0, v1), pack8888 (v2, v3));
}
static force_inline __m64
pix_add_mul (__m64 x, __m64 a, __m64 y, __m64 b)
{
x = pix_multiply (x, a);
y = pix_multiply (y, b);
return pix_add (x, y);
}
#else
/* MSVC only handles a "pass by register" of up to three SSE intrinsics */
#define pack_4x565(v0, v1, v2, v3) \
pack_4xpacked565 (pack8888 (v0, v1), pack8888 (v2, v3))
#define pix_add_mul(x, a, y, b) \
( x = pix_multiply (x, a), \
y = pix_multiply (y, b), \
pix_add (x, y) )
#endif
/* --------------- MMX code patch for fbcompose.c --------------------- */
static force_inline __m64
combine (const uint32_t *src, const uint32_t *mask)
{
__m64 vsrc = load8888 (src);
if (mask)
{
__m64 m = load8888 (mask);
m = expand_alpha (m);
vsrc = pix_multiply (vsrc, m);
}
return vsrc;
}
static force_inline __m64
core_combine_over_u_pixel_mmx (__m64 vsrc, __m64 vdst)
{
vsrc = _mm_unpacklo_pi8 (vsrc, _mm_setzero_si64 ());
if (is_opaque (vsrc))
{
return vsrc;
}
else if (!is_zero (vsrc))
{
return over (vsrc, expand_alpha (vsrc),
_mm_unpacklo_pi8 (vdst, _mm_setzero_si64 ()));
}
return _mm_unpacklo_pi8 (vdst, _mm_setzero_si64 ());
}
static void
mmx_combine_over_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end = dest + width;
while (dest < end)
{
__m64 vsrc = combine (src, mask);
if (is_opaque (vsrc))
{
store8888 (dest, vsrc);
}
else if (!is_zero (vsrc))
{
__m64 sa = expand_alpha (vsrc);
store8888 (dest, over (vsrc, sa, load8888 (dest)));
}
++dest;
++src;
if (mask)
++mask;
}
_mm_empty ();
}
static void
mmx_combine_over_reverse_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end = dest + width;
while (dest < end)
{
__m64 d, da;
__m64 s = combine (src, mask);
d = load8888 (dest);
da = expand_alpha (d);
store8888 (dest, over (d, da, s));
++dest;
++src;
if (mask)
mask++;
}
_mm_empty ();
}
static void
mmx_combine_in_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end = dest + width;
while (dest < end)
{
__m64 a;
__m64 x = combine (src, mask);
a = load8888 (dest);
a = expand_alpha (a);
x = pix_multiply (x, a);
store8888 (dest, x);
++dest;
++src;
if (mask)
mask++;
}
_mm_empty ();
}
static void
mmx_combine_in_reverse_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end = dest + width;
while (dest < end)
{
__m64 a = combine (src, mask);
__m64 x;
x = load8888 (dest);
a = expand_alpha (a);
x = pix_multiply (x, a);
store8888 (dest, x);
++dest;
++src;
if (mask)
mask++;
}
_mm_empty ();
}
static void
mmx_combine_out_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end = dest + width;
while (dest < end)
{
__m64 a;
__m64 x = combine (src, mask);
a = load8888 (dest);
a = expand_alpha (a);
a = negate (a);
x = pix_multiply (x, a);
store8888 (dest, x);
++dest;
++src;
if (mask)
mask++;
}
_mm_empty ();
}
static void
mmx_combine_out_reverse_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end = dest + width;
while (dest < end)
{
__m64 a = combine (src, mask);
__m64 x;
x = load8888 (dest);
a = expand_alpha (a);
a = negate (a);
x = pix_multiply (x, a);
store8888 (dest, x);
++dest;
++src;
if (mask)
mask++;
}
_mm_empty ();
}
static void
mmx_combine_atop_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end = dest + width;
while (dest < end)
{
__m64 da, d, sia;
__m64 s = combine (src, mask);
d = load8888 (dest);
sia = expand_alpha (s);
sia = negate (sia);
da = expand_alpha (d);
s = pix_add_mul (s, da, d, sia);
store8888 (dest, s);
++dest;
++src;
if (mask)
mask++;
}
_mm_empty ();
}
static void
mmx_combine_atop_reverse_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end;
end = dest + width;
while (dest < end)
{
__m64 dia, d, sa;
__m64 s = combine (src, mask);
d = load8888 (dest);
sa = expand_alpha (s);
dia = expand_alpha (d);
dia = negate (dia);
s = pix_add_mul (s, dia, d, sa);
store8888 (dest, s);
++dest;
++src;
if (mask)
mask++;
}
_mm_empty ();
}
static void
mmx_combine_xor_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
const uint32_t *end = dest + width;
while (dest < end)
{
__m64 dia, d, sia;
__m64 s = combine (src, mask);
d = load8888 (dest);
sia = expand_alpha (s);
dia = expand_alpha (d);
sia = negate (sia);
dia = negate (dia);
s = pix_add_mul (s, dia, d, sia);
store8888 (dest, s);
++dest;
++src;
if (mask)
mask++;
}
_mm_empty ();
}
static void
mmx_combine_add_u (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{