From 4454dc1b6f07e3a81ec179ca08cbedece5a5af9a Mon Sep 17 00:00:00 2001 From: John Dalgliesh Date: Sat, 12 Aug 2006 16:37:31 +0000 Subject: [PATCH] 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 Originally committed as revision 5990 to svn://svn.ffmpeg.org/ffmpeg/trunk --- configure | 14 ++++++++++++++ libavcodec/i386/dsputil_mmx.c | 12 ++++++------ libavcodec/i386/dsputil_mmx_avg.h | 2 +- libavcodec/i386/dsputil_mmx_rnd.h | 16 ++++++++-------- libavcodec/i386/motion_est_mmx.c | 12 ++++++------ libavcodec/i386/mpegvideo_mmx.c | 12 ++++++------ libavcodec/i386/mpegvideo_mmx_template.c | 4 ++-- libavcodec/i386/simple_idct_mmx.c | 14 +++++++------- 8 files changed, 50 insertions(+), 36 deletions(-) diff --git a/configure b/configure index 0891a0a4e4..17a60664b4 100755 --- a/configure +++ b/configure @@ -468,6 +468,7 @@ pthreads="no" swscaler="no" gpl="no" memalignhack="no" +asmalign_pot="unknown" # OS specific targetos=`uname -s` @@ -1469,6 +1470,12 @@ if test "$gprof" = "yes" ; then LDFLAGS="$LDFLAGS -p" fi +# find if .align arg is power-of-two or not +if test $asmalign_pot = "unknown"; then + asmalign_pot="no" + echo 'asm (".align 3");' | check_cc && asmalign_pot="yes" +fi + echo "install prefix $PREFIX" echo "source path $source_path" echo "C compiler $cc" @@ -1535,6 +1542,7 @@ echo "network support $network" if test "$network" = "yes" ; then echo "IPv6 support $ipv6" fi +echo ".align is power-of-two" $asmalign_pot if test "$gpl" = "no" ; then echo "License: LGPL" else @@ -2096,6 +2104,12 @@ if test "$amr_if2" = "yes" ; then echo "AMR_CFLAGS=-DIF2=1" >> config.mak fi +if test "$asmalign_pot" = "yes" ; then + echo '#define ASMALIGN(ZEROBITS) ".align " #ZEROBITS "\n\t"' >> $TMPH +else + echo '#define ASMALIGN(ZEROBITS) ".align 1<<" #ZEROBITS "\n\t"' >> $TMPH +fi + for codec in $DECODER_LIST $ENCODER_LIST $PARSER_LIST $DEMUXER_LIST $MUXER_LIST; do echo "#define CONFIG_`echo $codec | tr a-z A-Z` 1" >> $TMPH diff --git a/libavcodec/i386/dsputil_mmx.c b/libavcodec/i386/dsputil_mmx.c index 8ffa90b078..cf67ca4dda 100644 --- a/libavcodec/i386/dsputil_mmx.c +++ b/libavcodec/i386/dsputil_mmx.c @@ -56,7 +56,7 @@ static const uint64_t ff_pw_15 attribute_used __attribute__ ((aligned(8))) = 0x0 static const uint64_t ff_pb_3F attribute_used __attribute__ ((aligned(8))) = 0x3F3F3F3F3F3F3F3FULL; static const uint64_t ff_pb_FC attribute_used __attribute__ ((aligned(8))) = 0xFCFCFCFCFCFCFCFCULL; -#define JUMPALIGN() __asm __volatile (".balign 8"::) +#define JUMPALIGN() __asm __volatile (ASMALIGN(3)::) #define MOVQ_ZERO(regd) __asm __volatile ("pxor %%" #regd ", %%" #regd ::) #define MOVQ_WONE(regd) \ @@ -204,7 +204,7 @@ static void get_pixels_mmx(DCTELEM *block, const uint8_t *pixels, int line_size) asm volatile( "mov $-128, %%"REG_a" \n\t" "pxor %%mm7, %%mm7 \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%0), %%mm0 \n\t" "movq (%0, %2), %%mm2 \n\t" @@ -232,7 +232,7 @@ static inline void diff_pixels_mmx(DCTELEM *block, const uint8_t *s1, const uint asm volatile( "pxor %%mm7, %%mm7 \n\t" "mov $-128, %%"REG_a" \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%0), %%mm0 \n\t" "movq (%1), %%mm2 \n\t" @@ -375,7 +375,7 @@ static void put_pixels4_mmx(uint8_t *block, const uint8_t *pixels, int line_size { __asm __volatile( "lea (%3, %3), %%"REG_a" \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movd (%1), %%mm0 \n\t" "movd (%1, %3), %%mm1 \n\t" @@ -401,7 +401,7 @@ static void put_pixels8_mmx(uint8_t *block, const uint8_t *pixels, int line_size { __asm __volatile( "lea (%3, %3), %%"REG_a" \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1), %%mm0 \n\t" "movq (%1, %3), %%mm1 \n\t" @@ -427,7 +427,7 @@ static void put_pixels16_mmx(uint8_t *block, const uint8_t *pixels, int line_siz { __asm __volatile( "lea (%3, %3), %%"REG_a" \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1), %%mm0 \n\t" "movq 8(%1), %%mm4 \n\t" diff --git a/libavcodec/i386/dsputil_mmx_avg.h b/libavcodec/i386/dsputil_mmx_avg.h index 440c5bb9c0..8692fb0487 100644 --- a/libavcodec/i386/dsputil_mmx_avg.h +++ b/libavcodec/i386/dsputil_mmx_avg.h @@ -754,7 +754,7 @@ static void DEF(avg_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line "lea (%3, %3), %%"REG_a" \n\t" "movq (%1), %%mm0 \n\t" PAVGB" 1(%1), %%mm0 \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1, %%"REG_a"), %%mm2 \n\t" "movq (%1, %3), %%mm1 \n\t" diff --git a/libavcodec/i386/dsputil_mmx_rnd.h b/libavcodec/i386/dsputil_mmx_rnd.h index 3ecd776b82..1ab28f171a 100644 --- a/libavcodec/i386/dsputil_mmx_rnd.h +++ b/libavcodec/i386/dsputil_mmx_rnd.h @@ -28,7 +28,7 @@ static void DEF(put, pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line MOVQ_BFE(mm6); __asm __volatile( "lea (%3, %3), %%"REG_a" \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1), %%mm0 \n\t" "movq 1(%1), %%mm1 \n\t" @@ -69,7 +69,7 @@ static void attribute_unused DEF(put, pixels8_l2)(uint8_t *dst, uint8_t *src1, u "movq %%mm4, (%3) \n\t" "add %5, %3 \n\t" "decl %0 \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1), %%mm0 \n\t" "movq (%2), %%mm1 \n\t" @@ -110,7 +110,7 @@ static void DEF(put, pixels16_x2)(uint8_t *block, const uint8_t *pixels, int lin MOVQ_BFE(mm6); __asm __volatile( "lea (%3, %3), %%"REG_a" \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1), %%mm0 \n\t" "movq 1(%1), %%mm1 \n\t" @@ -168,7 +168,7 @@ static void attribute_unused DEF(put, pixels16_l2)(uint8_t *dst, uint8_t *src1, "movq %%mm5, 8(%3) \n\t" "add %5, %3 \n\t" "decl %0 \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1), %%mm0 \n\t" "movq (%2), %%mm1 \n\t" @@ -206,7 +206,7 @@ static void DEF(put, pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line __asm __volatile( "lea (%3, %3), %%"REG_a" \n\t" "movq (%1), %%mm0 \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1, %3), %%mm1 \n\t" "movq (%1, %%"REG_a"),%%mm2 \n\t" @@ -246,7 +246,7 @@ static void DEF(put, pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int lin "paddusw %%mm1, %%mm5 \n\t" "xor %%"REG_a", %%"REG_a" \n\t" "add %3, %1 \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1, %%"REG_a"), %%mm0 \n\t" "movq 1(%1, %%"REG_a"), %%mm2 \n\t" @@ -458,7 +458,7 @@ static void DEF(avg, pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line __asm __volatile( "lea (%3, %3), %%"REG_a" \n\t" "movq (%1), %%mm0 \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1, %3), %%mm1 \n\t" "movq (%1, %%"REG_a"), %%mm2 \n\t" @@ -509,7 +509,7 @@ static void DEF(avg, pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int lin "paddusw %%mm1, %%mm5 \n\t" "xor %%"REG_a", %%"REG_a" \n\t" "add %3, %1 \n\t" - ".balign 8 \n\t" + ASMALIGN(3) "1: \n\t" "movq (%1, %%"REG_a"), %%mm0 \n\t" "movq 1(%1, %%"REG_a"), %%mm2 \n\t" diff --git a/libavcodec/i386/motion_est_mmx.c b/libavcodec/i386/motion_est_mmx.c index edcabcf387..0f7817338c 100644 --- a/libavcodec/i386/motion_est_mmx.c +++ b/libavcodec/i386/motion_est_mmx.c @@ -34,7 +34,7 @@ static inline void sad8_1_mmx(uint8_t *blk1, uint8_t *blk2, int stride, int h) { long len= -(stride*h); asm volatile( - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%1, %%"REG_a"), %%mm0 \n\t" "movq (%2, %%"REG_a"), %%mm2 \n\t" @@ -70,7 +70,7 @@ static inline void sad8_1_mmx2(uint8_t *blk1, uint8_t *blk2, int stride, int h) { long len= -(stride*h); asm volatile( - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%1, %%"REG_a"), %%mm0 \n\t" "movq (%2, %%"REG_a"), %%mm2 \n\t" @@ -92,7 +92,7 @@ static inline void sad8_2_mmx2(uint8_t *blk1a, uint8_t *blk1b, uint8_t *blk2, in { long len= -(stride*h); asm volatile( - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%1, %%"REG_a"), %%mm0 \n\t" "movq (%2, %%"REG_a"), %%mm2 \n\t" @@ -118,7 +118,7 @@ static inline void sad8_4_mmx2(uint8_t *blk1, uint8_t *blk2, int stride, int h) { //FIXME reuse src long len= -(stride*h); asm volatile( - ".balign 16 \n\t" + ASMALIGN(4) "movq "MANGLE(bone)", %%mm5 \n\t" "1: \n\t" "movq (%1, %%"REG_a"), %%mm0 \n\t" @@ -155,7 +155,7 @@ static inline void sad8_2_mmx(uint8_t *blk1a, uint8_t *blk1b, uint8_t *blk2, int { long len= -(stride*h); asm volatile( - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%1, %%"REG_a"), %%mm0 \n\t" "movq (%2, %%"REG_a"), %%mm1 \n\t" @@ -193,7 +193,7 @@ static inline void sad8_4_mmx(uint8_t *blk1, uint8_t *blk2, int stride, int h) { long len= -(stride*h); asm volatile( - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%1, %%"REG_a"), %%mm0 \n\t" "movq (%2, %%"REG_a"), %%mm1 \n\t" diff --git a/libavcodec/i386/mpegvideo_mmx.c b/libavcodec/i386/mpegvideo_mmx.c index c00a602bdd..1c3a957da6 100644 --- a/libavcodec/i386/mpegvideo_mmx.c +++ b/libavcodec/i386/mpegvideo_mmx.c @@ -66,7 +66,7 @@ asm volatile( "packssdw %%mm5, %%mm5 \n\t" "psubw %%mm5, %%mm7 \n\t" "pxor %%mm4, %%mm4 \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%0, %3), %%mm0 \n\t" "movq 8(%0, %3), %%mm1 \n\t" @@ -129,7 +129,7 @@ asm volatile( "packssdw %%mm5, %%mm5 \n\t" "psubw %%mm5, %%mm7 \n\t" "pxor %%mm4, %%mm4 \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%0, %3), %%mm0 \n\t" "movq 8(%0, %3), %%mm1 \n\t" @@ -222,7 +222,7 @@ asm volatile( "packssdw %%mm6, %%mm6 \n\t" "packssdw %%mm6, %%mm6 \n\t" "mov %3, %%"REG_a" \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%0, %%"REG_a"), %%mm0 \n\t" "movq 8(%0, %%"REG_a"), %%mm1 \n\t" @@ -285,7 +285,7 @@ asm volatile( "packssdw %%mm6, %%mm6 \n\t" "packssdw %%mm6, %%mm6 \n\t" "mov %3, %%"REG_a" \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%0, %%"REG_a"), %%mm0 \n\t" "movq 8(%0, %%"REG_a"), %%mm1 \n\t" @@ -357,7 +357,7 @@ asm volatile( "packssdw %%mm6, %%mm6 \n\t" "packssdw %%mm6, %%mm6 \n\t" "mov %3, %%"REG_a" \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%0, %%"REG_a"), %%mm0 \n\t" "movq 8(%0, %%"REG_a"), %%mm1 \n\t" @@ -418,7 +418,7 @@ asm volatile( "packssdw %%mm6, %%mm6 \n\t" "packssdw %%mm6, %%mm6 \n\t" "mov %3, %%"REG_a" \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "movq (%0, %%"REG_a"), %%mm0 \n\t" "movq 8(%0, %%"REG_a"), %%mm1 \n\t" diff --git a/libavcodec/i386/mpegvideo_mmx_template.c b/libavcodec/i386/mpegvideo_mmx_template.c index de2ef08e54..f96582fdf6 100644 --- a/libavcodec/i386/mpegvideo_mmx_template.c +++ b/libavcodec/i386/mpegvideo_mmx_template.c @@ -112,7 +112,7 @@ static int RENAME(dct_quantize)(MpegEncContext *s, "pxor %%mm6, %%mm6 \n\t" "psubw (%3), %%mm6 \n\t" // -bias[0] "mov $-128, %%"REG_a" \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "pxor %%mm1, %%mm1 \n\t" // 0 "movq (%1, %%"REG_a"), %%mm0 \n\t" // block[i] @@ -156,7 +156,7 @@ static int RENAME(dct_quantize)(MpegEncContext *s, "pxor %%mm7, %%mm7 \n\t" // 0 "pxor %%mm4, %%mm4 \n\t" // 0 "mov $-128, %%"REG_a" \n\t" - ".balign 16 \n\t" + ASMALIGN(4) "1: \n\t" "pxor %%mm1, %%mm1 \n\t" // 0 "movq (%1, %%"REG_a"), %%mm0 \n\t" // block[i] diff --git a/libavcodec/i386/simple_idct_mmx.c b/libavcodec/i386/simple_idct_mmx.c index fcff3bf140..2f8d156fd3 100644 --- a/libavcodec/i386/simple_idct_mmx.c +++ b/libavcodec/i386/simple_idct_mmx.c @@ -785,7 +785,7 @@ IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20) IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20) "jmp 9f \n\t" - "#.balign 16 \n\t"\ + "#" ASMALIGN(4) \ "4: \n\t" Z_COND_IDCT( 64(%0), 72(%0), 80(%0), 88(%0), 64(%1),paddd (%2), 11, 6f) Z_COND_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1),paddd (%2), 11, 5f) @@ -860,7 +860,7 @@ IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20) IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20) "jmp 9f \n\t" - "#.balign 16 \n\t"\ + "#" ASMALIGN(4) \ "6: \n\t" Z_COND_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1),paddd (%2), 11, 7f) @@ -926,7 +926,7 @@ IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20) IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20) "jmp 9f \n\t" - "#.balign 16 \n\t"\ + "#" ASMALIGN(4) \ "2: \n\t" Z_COND_IDCT( 96(%0),104(%0),112(%0),120(%0), 96(%1),paddd (%2), 11, 3f) @@ -1003,7 +1003,7 @@ IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20) IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20) "jmp 9f \n\t" - "#.balign 16 \n\t"\ + "#" ASMALIGN(4) \ "3: \n\t" #undef IDCT #define IDCT(src0, src4, src1, src5, dst, shift) \ @@ -1067,7 +1067,7 @@ IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20) IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20) "jmp 9f \n\t" - "#.balign 16 \n\t"\ + "#" ASMALIGN(4) \ "5: \n\t" #undef IDCT #define IDCT(src0, src4, src1, src5, dst, shift) \ @@ -1132,7 +1132,7 @@ IDCT( 16(%1), 80(%1), 48(%1), 112(%1), 8(%0), 20) "jmp 9f \n\t" - "#.balign 16 \n\t"\ + "#" ASMALIGN(4) \ "1: \n\t" #undef IDCT #define IDCT(src0, src4, src1, src5, dst, shift) \ @@ -1206,7 +1206,7 @@ IDCT( 24(%1), 88(%1), 56(%1), 120(%1), 12(%0), 20) "jmp 9f \n\t" - "#.balign 16 \n\t" + "#" ASMALIGN(4) "7: \n\t" #undef IDCT #define IDCT(src0, src4, src1, src5, dst, shift) \