diff options
| -rw-r--r-- | libavcodec/cabac.c | 2 | ||||
| -rw-r--r-- | libavcodec/x86/constants.c | 20 | ||||
| -rw-r--r-- | libavutil/mem.h | 17 | ||||
| -rw-r--r-- | libswscale/x86/swscale.c | 12 |
4 files changed, 34 insertions, 17 deletions
diff --git a/libavcodec/cabac.c b/libavcodec/cabac.c index dd2b057c6d..e51139de3b 100644 --- a/libavcodec/cabac.c +++ b/libavcodec/cabac.c @@ -32,7 +32,7 @@ #include "cabac.h" #include "cabac_functions.h" -const uint8_t ff_h264_cabac_tables[512 + 4*2*64 + 4*64 + 63] = { +DECLARE_ASM_ALIGNED(1, const uint8_t, ff_h264_cabac_tables)[512 + 4*2*64 + 4*64 + 63] = { 9,8,7,7,6,6,6,6,5,5,5,5,5,5,5,5, 4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4, 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, diff --git a/libavcodec/x86/constants.c b/libavcodec/x86/constants.c index 11002ee61e..47905c74a1 100644 --- a/libavcodec/x86/constants.c +++ b/libavcodec/x86/constants.c @@ -26,23 +26,23 @@ DECLARE_ALIGNED(32, const ymm_reg, ff_pw_1) = { 0x0001000100010001ULL, 0x000 0x0001000100010001ULL, 0x0001000100010001ULL }; DECLARE_ALIGNED(32, const ymm_reg, ff_pw_2) = { 0x0002000200020002ULL, 0x0002000200020002ULL, 0x0002000200020002ULL, 0x0002000200020002ULL }; -DECLARE_ALIGNED(16, const xmm_reg, ff_pw_3) = { 0x0003000300030003ULL, 0x0003000300030003ULL }; -DECLARE_ALIGNED(32, const ymm_reg, ff_pw_4) = { 0x0004000400040004ULL, 0x0004000400040004ULL, +DECLARE_ASM_ALIGNED(16, const xmm_reg, ff_pw_3) = { 0x0003000300030003ULL, 0x0003000300030003ULL }; +DECLARE_ASM_ALIGNED(32, const ymm_reg, ff_pw_4) = { 0x0004000400040004ULL, 0x0004000400040004ULL, 0x0004000400040004ULL, 0x0004000400040004ULL }; -DECLARE_ALIGNED(16, const xmm_reg, ff_pw_5) = { 0x0005000500050005ULL, 0x0005000500050005ULL }; +DECLARE_ASM_ALIGNED(16, const xmm_reg, ff_pw_5) = { 0x0005000500050005ULL, 0x0005000500050005ULL }; DECLARE_ALIGNED(16, const xmm_reg, ff_pw_8) = { 0x0008000800080008ULL, 0x0008000800080008ULL }; -DECLARE_ALIGNED(16, const xmm_reg, ff_pw_9) = { 0x0009000900090009ULL, 0x0009000900090009ULL }; +DECLARE_ASM_ALIGNED(16, const xmm_reg, ff_pw_9) = { 0x0009000900090009ULL, 0x0009000900090009ULL }; DECLARE_ALIGNED(8, const uint64_t, ff_pw_15) = 0x000F000F000F000FULL; DECLARE_ALIGNED(16, const xmm_reg, ff_pw_16) = { 0x0010001000100010ULL, 0x0010001000100010ULL }; DECLARE_ALIGNED(16, const xmm_reg, ff_pw_17) = { 0x0011001100110011ULL, 0x0011001100110011ULL }; -DECLARE_ALIGNED(16, const xmm_reg, ff_pw_18) = { 0x0012001200120012ULL, 0x0012001200120012ULL }; +DECLARE_ASM_ALIGNED(16, const xmm_reg, ff_pw_18) = { 0x0012001200120012ULL, 0x0012001200120012ULL }; DECLARE_ALIGNED(16, const xmm_reg, ff_pw_20) = { 0x0014001400140014ULL, 0x0014001400140014ULL }; DECLARE_ALIGNED(16, const xmm_reg, ff_pw_32) = { 0x0020002000200020ULL, 0x0020002000200020ULL }; -DECLARE_ALIGNED(8, const uint64_t, ff_pw_42) = 0x002A002A002A002AULL; -DECLARE_ALIGNED(8, const uint64_t, ff_pw_53) = 0x0035003500350035ULL; -DECLARE_ALIGNED(16, const xmm_reg, ff_pw_64) = { 0x0040004000400040ULL, 0x0040004000400040ULL }; -DECLARE_ALIGNED(8, const uint64_t, ff_pw_96) = 0x0060006000600060ULL; -DECLARE_ALIGNED(8, const uint64_t, ff_pw_128) = 0x0080008000800080ULL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_pw_42) = 0x002A002A002A002AULL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_pw_53) = 0x0035003500350035ULL; +DECLARE_ASM_ALIGNED(16, const xmm_reg, ff_pw_64) = { 0x0040004000400040ULL, 0x0040004000400040ULL }; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_pw_96) = 0x0060006000600060ULL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_pw_128) = 0x0080008000800080ULL; DECLARE_ALIGNED(32, const ymm_reg, ff_pw_255) = { 0x00ff00ff00ff00ffULL, 0x00ff00ff00ff00ffULL, 0x00ff00ff00ff00ffULL, 0x00ff00ff00ff00ffULL }; DECLARE_ALIGNED(32, const ymm_reg, ff_pw_256) = { 0x0100010001000100ULL, 0x0100010001000100ULL, diff --git a/libavutil/mem.h b/libavutil/mem.h index e5ca8299dc..9e344bd2c3 100644 --- a/libavutil/mem.h +++ b/libavutil/mem.h @@ -74,6 +74,19 @@ */ /** + * @def DECLARE_ASM_ALIGNED(n,t,v) + * Declare an aligned variable appropriate for use in inline assembly code. + * + * @code{.c} + * DECLARE_ASM_ALIGNED(16, uint64_t, pw_08) = UINT64_C(0x0008000800080008); + * @endcode + * + * @param n Minimum alignment in bytes + * @param t Type of the variable (or array element) + * @param v Name of the variable + */ + +/** * @def DECLARE_ASM_CONST(n,t,v) * Declare a static constant aligned variable appropriate for use in inline * assembly code. @@ -89,18 +102,22 @@ #if defined(__INTEL_COMPILER) && __INTEL_COMPILER < 1110 || defined(__SUNPRO_C) #define DECLARE_ALIGNED(n,t,v) t __attribute__ ((aligned (n))) v + #define DECLARE_ASM_ALIGNED(n,t,v) t __attribute__ ((aligned (n))) v #define DECLARE_ASM_CONST(n,t,v) const t __attribute__ ((aligned (n))) v #elif defined(__DJGPP__) #define DECLARE_ALIGNED(n,t,v) t __attribute__ ((aligned (FFMIN(n, 16)))) v #define DECLARE_ASM_CONST(n,t,v) static const t av_used __attribute__ ((aligned (FFMIN(n, 16)))) v #elif defined(__GNUC__) || defined(__clang__) #define DECLARE_ALIGNED(n,t,v) t __attribute__ ((aligned (n))) v + #define DECLARE_ASM_ALIGNED(n,t,v) t av_used __attribute__ ((aligned (n))) v #define DECLARE_ASM_CONST(n,t,v) static const t av_used __attribute__ ((aligned (n))) v #elif defined(_MSC_VER) #define DECLARE_ALIGNED(n,t,v) __declspec(align(n)) t v + #define DECLARE_ASM_ALIGNED(n,t,v) __declspec(align(n)) t v #define DECLARE_ASM_CONST(n,t,v) __declspec(align(n)) static const t v #else #define DECLARE_ALIGNED(n,t,v) t v + #define DECLARE_ASM_ALIGNED(n,t,v) t v #define DECLARE_ASM_CONST(n,t,v) static const t v #endif diff --git a/libswscale/x86/swscale.c b/libswscale/x86/swscale.c index 869e7fb176..7dc2d70574 100644 --- a/libswscale/x86/swscale.c +++ b/libswscale/x86/swscale.c @@ -53,13 +53,13 @@ DECLARE_ASM_CONST(8, uint64_t, b15Mask)= 0x001F001F001F001FLL; DECLARE_ASM_CONST(8, uint64_t, g15Mask)= 0x03E003E003E003E0LL; DECLARE_ASM_CONST(8, uint64_t, r15Mask)= 0x7C007C007C007C00LL; -DECLARE_ALIGNED(8, const uint64_t, ff_M24A) = 0x00FF0000FF0000FFLL; -DECLARE_ALIGNED(8, const uint64_t, ff_M24B) = 0xFF0000FF0000FF00LL; -DECLARE_ALIGNED(8, const uint64_t, ff_M24C) = 0x0000FF0000FF0000LL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_M24A) = 0x00FF0000FF0000FFLL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_M24B) = 0xFF0000FF0000FF00LL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_M24C) = 0x0000FF0000FF0000LL; -DECLARE_ALIGNED(8, const uint64_t, ff_bgr2YOffset) = 0x1010101010101010ULL; -DECLARE_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL; -DECLARE_ALIGNED(8, const uint64_t, ff_w1111) = 0x0001000100010001ULL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_bgr2YOffset) = 0x1010101010101010ULL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_bgr2UVOffset) = 0x8080808080808080ULL; +DECLARE_ASM_ALIGNED(8, const uint64_t, ff_w1111) = 0x0001000100010001ULL; //MMX versions |