Mercurial > libavcodec.hg
comparison i386/dsputil_mmx.c @ 3576:f7125bf10892 libavcodec
Support for MacIntel, last part: balign directives
Determines whether .align's arg is power-of-two or not, then defines ASMALIGN appropriately in config.h. Changes all .baligns to ASMALIGNs.
Patch by John Dalgliesh % johnd AH defyne P org %
Original thread:
Date: Aug 11, 2006 8:00 AM
Subject: Re: [Ffmpeg-devel] Mac OS X Intel last part: balign directives
| author | gpoirier |
|---|---|
| date | Sat, 12 Aug 2006 16:37:31 +0000 |
| parents | f549d1e685f7 |
| children | 47821be55b6c |
comparison
equal
deleted
inserted
replaced
| 3575:4baed6f6577b | 3576:f7125bf10892 |
|---|---|
| 54 static const uint64_t ff_pw_15 attribute_used __attribute__ ((aligned(8))) = 0x000F000F000F000FULL; | 54 static const uint64_t ff_pw_15 attribute_used __attribute__ ((aligned(8))) = 0x000F000F000F000FULL; |
| 55 | 55 |
| 56 static const uint64_t ff_pb_3F attribute_used __attribute__ ((aligned(8))) = 0x3F3F3F3F3F3F3F3FULL; | 56 static const uint64_t ff_pb_3F attribute_used __attribute__ ((aligned(8))) = 0x3F3F3F3F3F3F3F3FULL; |
| 57 static const uint64_t ff_pb_FC attribute_used __attribute__ ((aligned(8))) = 0xFCFCFCFCFCFCFCFCULL; | 57 static const uint64_t ff_pb_FC attribute_used __attribute__ ((aligned(8))) = 0xFCFCFCFCFCFCFCFCULL; |
| 58 | 58 |
| 59 #define JUMPALIGN() __asm __volatile (".balign 8"::) | 59 #define JUMPALIGN() __asm __volatile (ASMALIGN(3)::) |
| 60 #define MOVQ_ZERO(regd) __asm __volatile ("pxor %%" #regd ", %%" #regd ::) | 60 #define MOVQ_ZERO(regd) __asm __volatile ("pxor %%" #regd ", %%" #regd ::) |
| 61 | 61 |
| 62 #define MOVQ_WONE(regd) \ | 62 #define MOVQ_WONE(regd) \ |
| 63 __asm __volatile ( \ | 63 __asm __volatile ( \ |
| 64 "pcmpeqd %%" #regd ", %%" #regd " \n\t" \ | 64 "pcmpeqd %%" #regd ", %%" #regd " \n\t" \ |
| 202 static void get_pixels_mmx(DCTELEM *block, const uint8_t *pixels, int line_size) | 202 static void get_pixels_mmx(DCTELEM *block, const uint8_t *pixels, int line_size) |
| 203 { | 203 { |
| 204 asm volatile( | 204 asm volatile( |
| 205 "mov $-128, %%"REG_a" \n\t" | 205 "mov $-128, %%"REG_a" \n\t" |
| 206 "pxor %%mm7, %%mm7 \n\t" | 206 "pxor %%mm7, %%mm7 \n\t" |
| 207 ".balign 16 \n\t" | 207 ASMALIGN(4) |
| 208 "1: \n\t" | 208 "1: \n\t" |
| 209 "movq (%0), %%mm0 \n\t" | 209 "movq (%0), %%mm0 \n\t" |
| 210 "movq (%0, %2), %%mm2 \n\t" | 210 "movq (%0, %2), %%mm2 \n\t" |
| 211 "movq %%mm0, %%mm1 \n\t" | 211 "movq %%mm0, %%mm1 \n\t" |
| 212 "movq %%mm2, %%mm3 \n\t" | 212 "movq %%mm2, %%mm3 \n\t" |
| 230 static inline void diff_pixels_mmx(DCTELEM *block, const uint8_t *s1, const uint8_t *s2, int stride) | 230 static inline void diff_pixels_mmx(DCTELEM *block, const uint8_t *s1, const uint8_t *s2, int stride) |
| 231 { | 231 { |
| 232 asm volatile( | 232 asm volatile( |
| 233 "pxor %%mm7, %%mm7 \n\t" | 233 "pxor %%mm7, %%mm7 \n\t" |
| 234 "mov $-128, %%"REG_a" \n\t" | 234 "mov $-128, %%"REG_a" \n\t" |
| 235 ".balign 16 \n\t" | 235 ASMALIGN(4) |
| 236 "1: \n\t" | 236 "1: \n\t" |
| 237 "movq (%0), %%mm0 \n\t" | 237 "movq (%0), %%mm0 \n\t" |
| 238 "movq (%1), %%mm2 \n\t" | 238 "movq (%1), %%mm2 \n\t" |
| 239 "movq %%mm0, %%mm1 \n\t" | 239 "movq %%mm0, %%mm1 \n\t" |
| 240 "movq %%mm2, %%mm3 \n\t" | 240 "movq %%mm2, %%mm3 \n\t" |
| 373 | 373 |
| 374 static void put_pixels4_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h) | 374 static void put_pixels4_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h) |
| 375 { | 375 { |
| 376 __asm __volatile( | 376 __asm __volatile( |
| 377 "lea (%3, %3), %%"REG_a" \n\t" | 377 "lea (%3, %3), %%"REG_a" \n\t" |
| 378 ".balign 8 \n\t" | 378 ASMALIGN(3) |
| 379 "1: \n\t" | 379 "1: \n\t" |
| 380 "movd (%1), %%mm0 \n\t" | 380 "movd (%1), %%mm0 \n\t" |
| 381 "movd (%1, %3), %%mm1 \n\t" | 381 "movd (%1, %3), %%mm1 \n\t" |
| 382 "movd %%mm0, (%2) \n\t" | 382 "movd %%mm0, (%2) \n\t" |
| 383 "movd %%mm1, (%2, %3) \n\t" | 383 "movd %%mm1, (%2, %3) \n\t" |
| 399 | 399 |
| 400 static void put_pixels8_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h) | 400 static void put_pixels8_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h) |
| 401 { | 401 { |
| 402 __asm __volatile( | 402 __asm __volatile( |
| 403 "lea (%3, %3), %%"REG_a" \n\t" | 403 "lea (%3, %3), %%"REG_a" \n\t" |
| 404 ".balign 8 \n\t" | 404 ASMALIGN(3) |
| 405 "1: \n\t" | 405 "1: \n\t" |
| 406 "movq (%1), %%mm0 \n\t" | 406 "movq (%1), %%mm0 \n\t" |
| 407 "movq (%1, %3), %%mm1 \n\t" | 407 "movq (%1, %3), %%mm1 \n\t" |
| 408 "movq %%mm0, (%2) \n\t" | 408 "movq %%mm0, (%2) \n\t" |
| 409 "movq %%mm1, (%2, %3) \n\t" | 409 "movq %%mm1, (%2, %3) \n\t" |
| 425 | 425 |
| 426 static void put_pixels16_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h) | 426 static void put_pixels16_mmx(uint8_t *block, const uint8_t *pixels, int line_size, int h) |
| 427 { | 427 { |
| 428 __asm __volatile( | 428 __asm __volatile( |
| 429 "lea (%3, %3), %%"REG_a" \n\t" | 429 "lea (%3, %3), %%"REG_a" \n\t" |
| 430 ".balign 8 \n\t" | 430 ASMALIGN(3) |
| 431 "1: \n\t" | 431 "1: \n\t" |
| 432 "movq (%1), %%mm0 \n\t" | 432 "movq (%1), %%mm0 \n\t" |
| 433 "movq 8(%1), %%mm4 \n\t" | 433 "movq 8(%1), %%mm4 \n\t" |
| 434 "movq (%1, %3), %%mm1 \n\t" | 434 "movq (%1, %3), %%mm1 \n\t" |
| 435 "movq 8(%1, %3), %%mm5 \n\t" | 435 "movq 8(%1, %3), %%mm5 \n\t" |