2 @ ARMv4L optimized DSP utils
3 @ Copyright (c) 2004 AGAWA Koji <i (AT) atty (DOT) jp>
5 @ This file is part of FFmpeg.
7 @ FFmpeg is free software; you can redistribute it and/or
8 @ modify it under the terms of the GNU Lesser General Public
9 @ License as published by the Free Software Foundation; either
10 @ version 2.1 of the License, or (at your option) any later version.
12 @ FFmpeg is distributed in the hope that it will be useful,
13 @ but WITHOUT ANY WARRANTY; without even the implied warranty of
14 @ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 @ Lesser General Public License for more details.
17 @ You should have received a copy of the GNU Lesser General Public
18 @ License along with FFmpeg; if not, write to the Free Software
19 @ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
22 .macro ADJ_ALIGN_QUADWORD_D shift, Rd0, Rd1, Rd2, Rd3, Rn0, Rn1, Rn2, Rn3, Rn4
23 mov \Rd0, \Rn0, lsr #(\shift * 8)
24 mov \Rd1, \Rn1, lsr #(\shift * 8)
25 mov \Rd2, \Rn2, lsr #(\shift * 8)
26 mov \Rd3, \Rn3, lsr #(\shift * 8)
27 orr \Rd0, \Rd0, \Rn1, lsl #(32 - \shift * 8)
28 orr \Rd1, \Rd1, \Rn2, lsl #(32 - \shift * 8)
29 orr \Rd2, \Rd2, \Rn3, lsl #(32 - \shift * 8)
30 orr \Rd3, \Rd3, \Rn4, lsl #(32 - \shift * 8)
32 .macro ADJ_ALIGN_DOUBLEWORD shift, R0, R1, R2
33 mov \R0, \R0, lsr #(\shift * 8)
34 orr \R0, \R0, \R1, lsl #(32 - \shift * 8)
35 mov \R1, \R1, lsr #(\shift * 8)
36 orr \R1, \R1, \R2, lsl #(32 - \shift * 8)
38 .macro ADJ_ALIGN_DOUBLEWORD_D shift, Rdst0, Rdst1, Rsrc0, Rsrc1, Rsrc2
39 mov \Rdst0, \Rsrc0, lsr #(\shift * 8)
40 mov \Rdst1, \Rsrc1, lsr #(\shift * 8)
41 orr \Rdst0, \Rdst0, \Rsrc1, lsl #(32 - (\shift * 8))
42 orr \Rdst1, \Rdst1, \Rsrc2, lsl #(32 - (\shift * 8))
45 .macro RND_AVG32 Rd0, Rd1, Rn0, Rn1, Rm0, Rm1, Rmask
46 @ Rd = (Rn | Rm) - (((Rn ^ Rm) & ~0x01010101) >> 1)
53 and \Rd0, \Rd0, \Rmask
54 and \Rd1, \Rd1, \Rmask
55 sub \Rd0, \Rn0, \Rd0, lsr #1
56 sub \Rd1, \Rn1, \Rd1, lsr #1
59 .macro NO_RND_AVG32 Rd0, Rd1, Rn0, Rn1, Rm0, Rm1, Rmask
60 @ Rd = (Rn & Rm) - (((Rn ^ Rm) & ~0x01010101) >> 1)
67 and \Rd0, \Rd0, \Rmask
68 and \Rd1, \Rd1, \Rmask
69 add \Rd0, \Rn0, \Rd0, lsr #1
70 add \Rd1, \Rn1, \Rd1, lsr #1
73 @ ----------------------------------------------------------------
75 .global put_pixels16_arm
77 @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
78 @ block = word aligned, pixles = unaligned
80 stmfd sp!, {r4-r11, lr} @ R14 is also called LR
84 add r5, r5, r4, lsl #2
94 ldmfd sp!, {r4-r11, pc}
99 ADJ_ALIGN_QUADWORD_D 1, r9, r10, r11, r12, r4, r5, r6, r7, r8
105 ldmfd sp!, {r4-r11, pc}
110 ADJ_ALIGN_QUADWORD_D 2, r9, r10, r11, r12, r4, r5, r6, r7, r8
116 ldmfd sp!, {r4-r11, pc}
121 ADJ_ALIGN_QUADWORD_D 3, r9, r10, r11, r12, r4, r5, r6, r7, r8
127 ldmfd sp!, {r4-r11,pc}
135 @ ----------------------------------------------------------------
137 .global put_pixels8_arm
139 @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
140 @ block = word aligned, pixles = unaligned
142 stmfd sp!, {r4-r5,lr} @ R14 is also called LR
146 add r5, r5, r4, lsl #2
156 ldmfd sp!, {r4-r5,pc}
159 ldmia r1, {r4-r5, r12}
161 ADJ_ALIGN_DOUBLEWORD 1, r4, r5, r12
167 ldmfd sp!, {r4-r5,pc}
170 ldmia r1, {r4-r5, r12}
172 ADJ_ALIGN_DOUBLEWORD 2, r4, r5, r12
178 ldmfd sp!, {r4-r5,pc}
181 ldmia r1, {r4-r5, r12}
183 ADJ_ALIGN_DOUBLEWORD 3, r4, r5, r12
189 ldmfd sp!, {r4-r5,pc}
197 @ ----------------------------------------------------------------
199 .global put_pixels8_x2_arm
201 @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
202 @ block = word aligned, pixles = unaligned
204 stmfd sp!, {r4-r10,lr} @ R14 is also called LR
208 add r5, r5, r4, lsl #2
212 ldmia r1, {r4-r5, r10}
214 ADJ_ALIGN_DOUBLEWORD_D 1, r6, r7, r4, r5, r10
216 RND_AVG32 r8, r9, r4, r5, r6, r7, r12
221 ldmfd sp!, {r4-r10,pc}
224 ldmia r1, {r4-r5, r10}
226 ADJ_ALIGN_DOUBLEWORD_D 1, r6, r7, r4, r5, r10
227 ADJ_ALIGN_DOUBLEWORD_D 2, r8, r9, r4, r5, r10
229 RND_AVG32 r4, r5, r6, r7, r8, r9, r12
234 ldmfd sp!, {r4-r10,pc}
237 ldmia r1, {r4-r5, r10}
239 ADJ_ALIGN_DOUBLEWORD_D 2, r6, r7, r4, r5, r10
240 ADJ_ALIGN_DOUBLEWORD_D 3, r8, r9, r4, r5, r10
242 RND_AVG32 r4, r5, r6, r7, r8, r9, r12
247 ldmfd sp!, {r4-r10,pc}
250 ldmia r1, {r4-r5, r10}
252 ADJ_ALIGN_DOUBLEWORD_D 3, r6, r7, r4, r5, r10
254 RND_AVG32 r8, r9, r6, r7, r5, r10, r12
259 ldmfd sp!, {r4-r10,pc} @@ update PC with LR content.
268 .global put_no_rnd_pixels8_x2_arm
269 put_no_rnd_pixels8_x2_arm:
270 @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
271 @ block = word aligned, pixles = unaligned
273 stmfd sp!, {r4-r10,lr} @ R14 is also called LR
277 add r5, r5, r4, lsl #2
281 ldmia r1, {r4-r5, r10}
283 ADJ_ALIGN_DOUBLEWORD_D 1, r6, r7, r4, r5, r10
285 NO_RND_AVG32 r8, r9, r4, r5, r6, r7, r12
290 ldmfd sp!, {r4-r10,pc}
293 ldmia r1, {r4-r5, r10}
295 ADJ_ALIGN_DOUBLEWORD_D 1, r6, r7, r4, r5, r10
296 ADJ_ALIGN_DOUBLEWORD_D 2, r8, r9, r4, r5, r10
298 NO_RND_AVG32 r4, r5, r6, r7, r8, r9, r12
303 ldmfd sp!, {r4-r10,pc}
306 ldmia r1, {r4-r5, r10}
308 ADJ_ALIGN_DOUBLEWORD_D 2, r6, r7, r4, r5, r10
309 ADJ_ALIGN_DOUBLEWORD_D 3, r8, r9, r4, r5, r10
311 NO_RND_AVG32 r4, r5, r6, r7, r8, r9, r12
316 ldmfd sp!, {r4-r10,pc}
319 ldmia r1, {r4-r5, r10}
321 ADJ_ALIGN_DOUBLEWORD_D 3, r6, r7, r4, r5, r10
323 NO_RND_AVG32 r8, r9, r6, r7, r5, r10, r12
328 ldmfd sp!, {r4-r10,pc} @@ update PC with LR content.
337 @ ----------------------------------------------------------------
339 .global put_pixels8_y2_arm
341 @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
342 @ block = word aligned, pixles = unaligned
344 stmfd sp!, {r4-r11,lr} @ R14 is also called LR
349 add r5, r5, r4, lsl #2
358 RND_AVG32 r8, r9, r4, r5, r6, r7, r12
364 RND_AVG32 r8, r9, r6, r7, r4, r5, r12
369 ldmfd sp!, {r4-r11,pc}
375 ADJ_ALIGN_DOUBLEWORD 1, r4, r5, r6
379 ADJ_ALIGN_DOUBLEWORD 1, r7, r8, r9
380 RND_AVG32 r10, r11, r4, r5, r7, r8, r12
386 ADJ_ALIGN_DOUBLEWORD 1, r4, r5, r6
388 RND_AVG32 r10, r11, r7, r8, r4, r5, r12
392 ldmfd sp!, {r4-r11,pc}
398 ADJ_ALIGN_DOUBLEWORD 2, r4, r5, r6
402 ADJ_ALIGN_DOUBLEWORD 2, r7, r8, r9
403 RND_AVG32 r10, r11, r4, r5, r7, r8, r12
409 ADJ_ALIGN_DOUBLEWORD 2, r4, r5, r6
411 RND_AVG32 r10, r11, r7, r8, r4, r5, r12
415 ldmfd sp!, {r4-r11,pc}
421 ADJ_ALIGN_DOUBLEWORD 3, r4, r5, r6
425 ADJ_ALIGN_DOUBLEWORD 3, r7, r8, r9
426 RND_AVG32 r10, r11, r4, r5, r7, r8, r12
432 ADJ_ALIGN_DOUBLEWORD 3, r4, r5, r6
434 RND_AVG32 r10, r11, r7, r8, r4, r5, r12
438 ldmfd sp!, {r4-r11,pc}
448 .global put_no_rnd_pixels8_y2_arm
449 put_no_rnd_pixels8_y2_arm:
450 @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
451 @ block = word aligned, pixles = unaligned
453 stmfd sp!, {r4-r11,lr} @ R14 is also called LR
458 add r5, r5, r4, lsl #2
467 NO_RND_AVG32 r8, r9, r4, r5, r6, r7, r12
473 NO_RND_AVG32 r8, r9, r6, r7, r4, r5, r12
478 ldmfd sp!, {r4-r11,pc}
484 ADJ_ALIGN_DOUBLEWORD 1, r4, r5, r6
488 ADJ_ALIGN_DOUBLEWORD 1, r7, r8, r9
489 NO_RND_AVG32 r10, r11, r4, r5, r7, r8, r12
495 ADJ_ALIGN_DOUBLEWORD 1, r4, r5, r6
497 NO_RND_AVG32 r10, r11, r7, r8, r4, r5, r12
501 ldmfd sp!, {r4-r11,pc}
507 ADJ_ALIGN_DOUBLEWORD 2, r4, r5, r6
511 ADJ_ALIGN_DOUBLEWORD 2, r7, r8, r9
512 NO_RND_AVG32 r10, r11, r4, r5, r7, r8, r12
518 ADJ_ALIGN_DOUBLEWORD 2, r4, r5, r6
520 NO_RND_AVG32 r10, r11, r7, r8, r4, r5, r12
524 ldmfd sp!, {r4-r11,pc}
530 ADJ_ALIGN_DOUBLEWORD 3, r4, r5, r6
534 ADJ_ALIGN_DOUBLEWORD 3, r7, r8, r9
535 NO_RND_AVG32 r10, r11, r4, r5, r7, r8, r12
541 ADJ_ALIGN_DOUBLEWORD 3, r4, r5, r6
543 NO_RND_AVG32 r10, r11, r7, r8, r4, r5, r12
547 ldmfd sp!, {r4-r11,pc}
555 @ ----------------------------------------------------------------
556 .macro RND_XY2_IT align, rnd
557 @ l1= (a & 0x03030303) + (b & 0x03030303) ?(+ 0x02020202)
558 @ h1= ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2)
569 ADJ_ALIGN_DOUBLEWORD_D 1, r4, r5, r6, r7, r8
571 ADJ_ALIGN_DOUBLEWORD_D 1, r4, r5, r8, r9, r10
572 ADJ_ALIGN_DOUBLEWORD_D 2, r6, r7, r8, r9, r10
574 ADJ_ALIGN_DOUBLEWORD_D 2, r4, r5, r8, r9, r10
575 ADJ_ALIGN_DOUBLEWORD_D 3, r6, r7, r8, r9, r10
577 ADJ_ALIGN_DOUBLEWORD_D 3, r4, r5, r5, r6, r7
579 ldr r14, [r12, #0] @ 0x03030303
586 ldreq r14, [r12, #16] @ 0x02020202
588 ldreq r14, [r12, #28] @ 0x01010101
594 ldr r14, [r12, #20] @ 0xFCFCFCFC >> 2
595 and r4, r14, r4, lsr #2
596 and r5, r14, r5, lsr #2
597 and r6, r14, r6, lsr #2
598 and r7, r14, r7, lsr #2
603 .macro RND_XY2_EXPAND align, rnd
604 RND_XY2_IT \align, \rnd
605 6: stmfd sp!, {r8-r11}
606 RND_XY2_IT \align, \rnd
612 ldr r14, [r12, #24] @ 0x0F0F0F0F
613 and r4, r14, r4, lsr #2
614 and r5, r14, r5, lsr #2
621 ldmfd sp!, {r4-r11,pc}
625 .global put_pixels8_xy2_arm
627 @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
628 @ block = word aligned, pixles = unaligned
630 stmfd sp!, {r4-r11,lr} @ R14 is also called LR
633 add r5, r12, r4, lsl #2
657 .word 0xFCFCFCFC >> 2
662 .global put_no_rnd_pixels8_xy2_arm
663 put_no_rnd_pixels8_xy2_arm:
664 @ void func(uint8_t *block, const uint8_t *pixels, int line_size, int h)
665 @ block = word aligned, pixles = unaligned
667 stmfd sp!, {r4-r11,lr} @ R14 is also called LR
670 add r5, r12, r4, lsl #2
694 .word 0xFCFCFCFC >> 2