Index: include/sse.h =================================================================== --- include/sse.h (revision 1172) +++ include/sse.h (working copy) @@ -23,9 +23,9 @@ #ifndef LIBMPEG2_SSE_H #define LIBMPEG2_SSE_H -typedef union { +DECLARE_ALIGNED(16, typedef union { float sf[4]; /* Single-precision (32-bit) value */ -} ATTR_ALIGN(16) sse_t; /* On a 16 byte (128-bit) boundary */ +}, sse_t); /* On a 16 byte (128-bit) boundary */ #define sse_i2r(op, imm, reg) \ Index: include/mmx.h =================================================================== --- include/mmx.h (revision 1172) +++ include/mmx.h (working copy) @@ -30,7 +30,7 @@ * values by ULL, lest they be truncated by the compiler) */ -typedef union { +DECLARE_ALIGNED(8, typedef union { long long q; /* Quadword (64-bit) value */ unsigned long long uq; /* Unsigned Quadword */ int d[2]; /* 2 Doubleword (32-bit) values */ @@ -40,7 +40,7 @@ char b[8]; /* 8 Byte (8-bit) values */ unsigned char ub[8]; /* 8 Unsigned Byte */ float s[2]; /* Single-precision (32-bit) value */ -} ATTR_ALIGN(8) mmx_t; /* On an 8-byte (64-bit) boundary */ +}, mmx_t); /* On an 8-byte (64-bit) boundary */ #define mmx_i2r(op,imm,reg) \ @@ -263,7 +263,7 @@ /* SSE2 */ -typedef union { +DECLARE_ALIGNED(16, typedef union { long long q[2]; /* Quadword (64-bit) value */ unsigned long long uq[2]; /* Unsigned Quadword */ int d[4]; /* 2 Doubleword (32-bit) values */ @@ -273,7 +273,7 @@ char b[16]; /* 8 Byte (8-bit) values */ unsigned char ub[16]; /* 8 Unsigned Byte */ float s[4]; /* Single-precision (32-bit) value */ -} ATTR_ALIGN(16) sse_t; /* On an 16-byte (128-bit) boundary */ +}, sse_t); /* On an 16-byte (128-bit) boundary */ #define movdqu_m2r(var,reg) mmx_m2r (movdqu, var, reg) #define movdqu_r2m(reg,var) mmx_r2m (movdqu, reg, var) Index: include/attributes.h =================================================================== --- include/attributes.h (revision 1172) +++ include/attributes.h (working copy) @@ -24,11 +24,13 @@ #ifndef LIBMPEG2_ATTRIBUTES_H #define LIBMPEG2_ATTRIBUTES_H -/* use gcc attribs to align critical data structures */ -#ifdef ATTRIBUTE_ALIGNED_MAX -#define ATTR_ALIGN(align) __attribute__ ((__aligned__ ((ATTRIBUTE_ALIGNED_MAX < align) ? ATTRIBUTE_ALIGNED_MAX : align))) +/* use GCC and MSVC attribs to align critical data structures */ +#if (defined(ATTRIBUTE_ALIGNED_MAX) && defined(__GNUC__)) +#define DECLARE_ALIGNED(n,t,v) t v __attribute__ ((__aligned__ ((ATTRIBUTE_ALIGNED_MAX < n) ? ATTRIBUTE_ALIGNED_MAX : n))) +#elif defined(_MSC_VER) +#define DECLARE_ALIGNED(n,t,v) t __declspec(align(n)) v #else -#define ATTR_ALIGN(align) +#define DECLARE_ALIGNED(n,t,v) t v #endif #ifdef HAVE_BUILTIN_EXPECT Index: vc++/config.h =================================================================== --- vc++/config.h (revision 1172) +++ vc++/config.h (working copy) @@ -1,7 +1,9 @@ /* vc++/config.h - manually adapted from include/config.h.in */ /* Suppress warnings relating to mismatched declarations */ +#ifdef _MSC_VER #pragma warning (disable:4028) +#endif /* autodetect accelerations */ #define ACCEL_DETECT @@ -165,7 +167,7 @@ /* Define as `__restrict' if that's what the C compiler calls it, or to nothing if it is not supported. */ -#define restrict __restrict +//#define restrict __restrict /* Define to `unsigned' if does not define. */ /* #undef size_t */ Index: libmpeg2/convert/rgb.c =================================================================== --- libmpeg2/convert/rgb.c (revision 1172) +++ libmpeg2/convert/rgb.c (working copy) @@ -43,7 +43,7 @@ {117579, 136230, 16907, 35559} /* SMPTE 240M (1987) */ }; -static const uint8_t dither[] ATTR_ALIGN(32) = { +DECLARE_ALIGNED(32, static const uint8_t, dither[]) = { 0, 0, 23, 54, 5, 13, 29, 68, 1, 3, 24, 58, 7, 17, 30, 71, 0, 0, 23, 54, 5, 13, 29, 68, 1, 3, 24, 58, 7, 17, 30, 71, 0, 0, 23, 54, 5, 13, 29, 68, 1, 3, 24, 58, 7, 17, 30, 71, Index: libmpeg2/convert/rgb_vis.c =================================================================== --- libmpeg2/convert/rgb_vis.c (revision 1172) +++ libmpeg2/convert/rgb_vis.c (working copy) @@ -39,16 +39,16 @@ * process both even and odd scanlines in one pass. */ -static const uint16_t const_2048[] ATTR_ALIGN(8) = {2048, 2048, 2048, 2048}; -static const uint16_t const_1024[] ATTR_ALIGN(8) = {1024, 1024, 1024, 1024}; -static const uint16_t const_128[] ATTR_ALIGN(8) = {128, 128, 128, 128}; -static const uint8_t const_Ugreen[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const uint16_t, const_2048[]) = {2048, 2048, 2048, 2048}; +DECLARE_ALIGNED(8, static const uint16_t, const_1024[]) = {1024, 1024, 1024, 1024}; +DECLARE_ALIGNED(8, static const uint16_t, const_128[]) = {128, 128, 128, 128}; +DECLARE_ALIGNED(8, static const uint8_t, const_Ugreen[]) = {0xf3, 0x00, 0xf3, 0x00, 0xf3, 0x00, 0xf3, 0x00}; -static const uint8_t const_Vgreen[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const uint8_t, const_Vgreen[]) = {0xe6, 0x00, 0xe6, 0x00, 0xe6, 0x00, 0xe6, 0x00}; -static const uint8_t const_Ublue_Vred[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const uint8_t, const_Ublue_Vred[]) = {0x41, 0x41, 0x41, 0x41, 0x33, 0x33, 0x33, 0x33}; -static const uint8_t const_Ycoeff[] ATTR_ALIGN(4) = {0x25, 0x25, 0x25, 0x25}; +DECLARE_ALIGNED(4, static const uint8_t, const_Ycoeff[]) = {0x25, 0x25, 0x25, 0x25}; #define TMP0 0 #define TMP1 1 Index: libmpeg2/motion_comp_vis.c =================================================================== --- libmpeg2/motion_comp_vis.c (revision 1172) +++ libmpeg2/motion_comp_vis.c (working copy) @@ -53,16 +53,16 @@ #define DUP4(x) {x, x, x, x} #define DUP8(x) {x, x, x, x, x, x, x, x} -static const int16_t constants1[] ATTR_ALIGN(8) = DUP4 (1); -static const int16_t constants2[] ATTR_ALIGN(8) = DUP4 (2); -static const int16_t constants3[] ATTR_ALIGN(8) = DUP4 (3); -static const int16_t constants6[] ATTR_ALIGN(8) = DUP4 (6); -static const int8_t constants_fe[] ATTR_ALIGN(8) = DUP8 (0xfe); -static const int8_t constants_7f[] ATTR_ALIGN(8) = DUP8 (0x7f); -static const int8_t constants128[] ATTR_ALIGN(8) = DUP8 (128); -static const int16_t constants256_512[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int16_t, constants1[]) = DUP4 (1); +DECLARE_ALIGNED(8, static const int16_t, constants2[]) = DUP4 (2); +DECLARE_ALIGNED(8, static const int16_t, constants3[]) = DUP4 (3); +DECLARE_ALIGNED(8, static const int16_t, constants6[]) = DUP4 (6); +DECLARE_ALIGNED(8, static const int8_t, constants_fe[]) = DUP8 (0xfe); +DECLARE_ALIGNED(8, static const int8_t, constants_7f[]) = DUP8 (0x7f); +DECLARE_ALIGNED(8, static const int8_t, constants128[]) = DUP8 (128); +DECLARE_ALIGNED(8, static const int16_t, constants256_512[]) = {256, 512, 256, 512}; -static const int16_t constants256_1024[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int16_t, constants256_1024[]) = {256, 1024, 256, 1024}; #define REF_0 0 Index: libmpeg2/idct_mmx.c =================================================================== --- libmpeg2/idct_mmx.c (revision 1172) +++ libmpeg2/idct_mmx.c (working copy) @@ -464,10 +464,10 @@ /* Almost identical to mmxext version: */ /* just do both 4x8 columns in paraller */ - static const short t1_vector[] ATTR_ALIGN(16) = {T1,T1,T1,T1,T1,T1,T1,T1}; - static const short t2_vector[] ATTR_ALIGN(16) = {T2,T2,T2,T2,T2,T2,T2,T2}; - static const short t3_vector[] ATTR_ALIGN(16) = {T3,T3,T3,T3,T3,T3,T3,T3}; - static const short c4_vector[] ATTR_ALIGN(16) = {C4,C4,C4,C4,C4,C4,C4,C4}; + DECLARE_ALIGNED(16, static const short, t1_vector[]) = {T1,T1,T1,T1,T1,T1,T1,T1}; + DECLARE_ALIGNED(16, static const short, t2_vector[]) = {T2,T2,T2,T2,T2,T2,T2,T2}; + DECLARE_ALIGNED(16, static const short, t3_vector[]) = {T3,T3,T3,T3,T3,T3,T3,T3}; + DECLARE_ALIGNED(16, static const short, c4_vector[]) = {C4,C4,C4,C4,C4,C4,C4,C4}; #if defined(__x86_64__) @@ -710,10 +710,10 @@ /* MMX column IDCT */ static inline void idct_col (int16_t * const col, const int offset) { - static const short t1_vector[] ATTR_ALIGN(8) = {T1,T1,T1,T1}; - static const short t2_vector[] ATTR_ALIGN(8) = {T2,T2,T2,T2}; - static const short t3_vector[] ATTR_ALIGN(8) = {T3,T3,T3,T3}; - static const short c4_vector[] ATTR_ALIGN(8) = {C4,C4,C4,C4}; + DECLARE_ALIGNED(8, static const short, t1_vector[]) = {T1,T1,T1,T1}; + DECLARE_ALIGNED(8, static const short, t2_vector[]) = {T2,T2,T2,T2}; + DECLARE_ALIGNED(8, static const short, t3_vector[]) = {T3,T3,T3,T3}; + DECLARE_ALIGNED(8, static const short, c4_vector[]) = {C4,C4,C4,C4}; /* column code adapted from peter gubanov */ /* http://www.elecard.com/peter/idct.shtml */ @@ -847,33 +847,33 @@ } -static const int32_t rounder0[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int32_t, rounder0[]) = rounder ((1 << (COL_SHIFT - 1)) - 0.5); -static const int32_t rounder4[] ATTR_ALIGN(8) = rounder (0); -static const int32_t rounder1[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int32_t, rounder4[]) = rounder (0); +DECLARE_ALIGNED(8, static const int32_t, rounder1[]) = rounder (1.25683487303); /* C1*(C1/C4+C1+C7)/2 */ -static const int32_t rounder7[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int32_t, rounder7[]) = rounder (-0.25); /* C1*(C7/C4+C7-C1)/2 */ -static const int32_t rounder2[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int32_t, rounder2[]) = rounder (0.60355339059); /* C2 * (C6+C2)/2 */ -static const int32_t rounder6[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int32_t, rounder6[]) = rounder (-0.25); /* C2 * (C6-C2)/2 */ -static const int32_t rounder3[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int32_t, rounder3[]) = rounder (0.087788325588); /* C3*(-C3/C4+C3+C5)/2 */ -static const int32_t rounder5[] ATTR_ALIGN(8) = +DECLARE_ALIGNED(8, static const int32_t, rounder5[]) = rounder (-0.441341716183); /* C3*(-C5/C4+C5-C3)/2 */ #define declare_idct(idct,table,idct_row_head,idct_row,idct_row_tail,idct_row_mid) \ static inline void idct (int16_t * const block) \ { \ - static const int16_t table04[] ATTR_ALIGN(16) = \ + DECLARE_ALIGNED(16, static const int16_t, table04[]) = \ table (22725, 21407, 19266, 16384, 12873, 8867, 4520); \ - static const int16_t table17[] ATTR_ALIGN(16) = \ + DECLARE_ALIGNED(16, static const int16_t, table17[]) = \ table (31521, 29692, 26722, 22725, 17855, 12299, 6270); \ - static const int16_t table26[] ATTR_ALIGN(16) = \ + DECLARE_ALIGNED(16, static const int16_t, table26[]) = \ table (29692, 27969, 25172, 21407, 16819, 11585, 5906); \ - static const int16_t table35[] ATTR_ALIGN(16) = \ + DECLARE_ALIGNED(16, static const int16_t, table35[]) = \ table (26722, 25172, 22654, 19266, 15137, 10426, 5315); \ \ idct_row_head (block, 0*8, table04); \ @@ -900,29 +900,29 @@ static inline void sse2_idct (int16_t * const block) { - static const int16_t table04[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int16_t, table04[]) = sse2_table (22725, 21407, 19266, 16384, 12873, 8867, 4520); - static const int16_t table17[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int16_t, table17[]) = sse2_table (31521, 29692, 26722, 22725, 17855, 12299, 6270); - static const int16_t table26[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int16_t, table26[]) = sse2_table (29692, 27969, 25172, 21407, 16819, 11585, 5906); - static const int16_t table35[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int16_t, table35[]) = sse2_table (26722, 25172, 22654, 19266, 15137, 10426, 5315); - static const int32_t rounder0_128[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int32_t, rounder0_128[]) = rounder_sse2 ((1 << (COL_SHIFT - 1)) - 0.5); - static const int32_t rounder4_128[] ATTR_ALIGN(16) = rounder_sse2 (0); - static const int32_t rounder1_128[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int32_t, rounder4_128[]) = rounder_sse2 (0); + DECLARE_ALIGNED(16, static const int32_t, rounder1_128[]) = rounder_sse2 (1.25683487303); /* C1*(C1/C4+C1+C7)/2 */ - static const int32_t rounder7_128[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int32_t, rounder7_128[]) = rounder_sse2 (-0.25); /* C1*(C7/C4+C7-C1)/2 */ - static const int32_t rounder2_128[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int32_t, rounder2_128[]) = rounder_sse2 (0.60355339059); /* C2 * (C6+C2)/2 */ - static const int32_t rounder6_128[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int32_t, rounder6_128[]) = rounder_sse2 (-0.25); /* C2 * (C6-C2)/2 */ - static const int32_t rounder3_128[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int32_t, rounder3_128[]) = rounder_sse2 (0.087788325588); /* C3*(-C3/C4+C3+C5)/2 */ - static const int32_t rounder5_128[] ATTR_ALIGN(16) = + DECLARE_ALIGNED(16, static const int32_t, rounder5_128[]) = rounder_sse2 (-0.441341716183); /* C3*(-C5/C4+C5-C3)/2 */ #if defined(__x86_64__) Index: libmpeg2/mpeg2_internal.h =================================================================== --- libmpeg2/mpeg2_internal.h (revision 1172) +++ libmpeg2/mpeg2_internal.h (working copy) @@ -97,7 +97,7 @@ int16_t dc_dct_pred[3]; /* DCT coefficients */ - int16_t DCTblock[64] ATTR_ALIGN(64); + DECLARE_ALIGNED(64, int16_t, DCTblock[64]); uint8_t * picture_dest[3]; void (* convert) (void * convert_id, uint8_t * const * src, Index: libmpeg2/header.c =================================================================== --- libmpeg2/header.c (revision 1172) +++ libmpeg2/header.c (working copy) @@ -40,7 +40,7 @@ #define PIC_CODING_EXT 0x100 /* default intra quant matrix, in zig-zag order */ -static const uint8_t default_intra_quantizer_matrix[64] ATTR_ALIGN(16) = { +static const DECLARE_ALIGNED(16, uint8_t, default_intra_quantizer_matrix[64]) = { 8, 16, 16, 19, 16, 19, @@ -58,7 +58,7 @@ 83 }; -uint8_t mpeg2_scan_norm[64] ATTR_ALIGN(16) = { +DECLARE_ALIGNED(16, uint8_t, mpeg2_scan_norm[64]) = { /* Zig-Zag scan pattern */ 0, 1, 8, 16, 9, 2, 3, 10, 17, 24, 32, 25, 18, 11, 4, 5, 12, 19, 26, 33, 40, 48, 41, 34, 27, 20, 13, 6, 7, 14, 21, 28, @@ -66,7 +66,7 @@ 58, 59, 52, 45, 38, 31, 39, 46, 53, 60, 61, 54, 47, 55, 62, 63 }; -uint8_t mpeg2_scan_alt[64] ATTR_ALIGN(16) = { +DECLARE_ALIGNED(16, uint8_t, mpeg2_scan_alt[64]) = { /* Alternate scan pattern */ 0, 8, 16, 24, 1, 9, 2, 10, 17, 25, 32, 40, 48, 56, 57, 49, 41, 33, 26, 18, 3, 11, 4, 12, 19, 27, 34, 42, 50, 58, 35, 43, Index: libmpeg2/idct_altivec.c =================================================================== --- libmpeg2/idct_altivec.c (revision 1172) +++ libmpeg2/idct_altivec.c (working copy) @@ -62,15 +62,15 @@ #define VEC_S16(a,b,c,d,e,f,g,h) {a, b, c, d, e, f, g, h} #endif -static const vector_s16_t constants ATTR_ALIGN(16) = +DECLARE_ALIGNED(16, static const vector_s16_t, constants) = VEC_S16 (23170, 13573, 6518, 21895, -23170, -21895, 32, 31); -static const vector_s16_t constants_1 ATTR_ALIGN(16) = +DECLARE_ALIGNED(16, static const vector_s16_t, constants_1) = VEC_S16 (16384, 22725, 21407, 19266, 16384, 19266, 21407, 22725); -static const vector_s16_t constants_2 ATTR_ALIGN(16) = +DECLARE_ALIGNED(16, static const vector_s16_t, constants_2) = VEC_S16 (16069, 22289, 20995, 18895, 16069, 18895, 20995, 22289); -static const vector_s16_t constants_3 ATTR_ALIGN(16) = +DECLARE_ALIGNED(16, static const vector_s16_t, constants_3) = VEC_S16 (21407, 29692, 27969, 25172, 21407, 25172, 27969, 29692); -static const vector_s16_t constants_4 ATTR_ALIGN(16) = +DECLARE_ALIGNED(16, static const vector_s16_t, constants_4) = VEC_S16 (13623, 18895, 17799, 16019, 13623, 16019, 17799, 18895); #define IDCT \