diff --git a/libavcodec/ppc/dsputil_altivec.c b/libavcodec/ppc/dsputil_altivec.c index e183432754..8a1cd44376 100644 --- a/libavcodec/ppc/dsputil_altivec.c +++ b/libavcodec/ppc/dsputil_altivec.c @@ -20,10 +20,11 @@ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ +#include "config.h" +#if HAVE_ALTIVEC_H +#include +#endif #include "libavcodec/dsputil.h" - -#include "gcc_fixes.h" - #include "dsputil_ppc.h" #include "util_altivec.h" #include "types_altivec.h" @@ -1124,70 +1125,70 @@ xlc goes to around 660 on the regular C code... static int hadamard8_diff16x8_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h) { int sum; register vector signed short - temp0 REG_v(v0), - temp1 REG_v(v1), - temp2 REG_v(v2), - temp3 REG_v(v3), - temp4 REG_v(v4), - temp5 REG_v(v5), - temp6 REG_v(v6), - temp7 REG_v(v7); + temp0 __asm__ ("v0"), + temp1 __asm__ ("v1"), + temp2 __asm__ ("v2"), + temp3 __asm__ ("v3"), + temp4 __asm__ ("v4"), + temp5 __asm__ ("v5"), + temp6 __asm__ ("v6"), + temp7 __asm__ ("v7"); register vector signed short - temp0S REG_v(v8), - temp1S REG_v(v9), - temp2S REG_v(v10), - temp3S REG_v(v11), - temp4S REG_v(v12), - temp5S REG_v(v13), - temp6S REG_v(v14), - temp7S REG_v(v15); - register const vector unsigned char vzero REG_v(v31)= + temp0S __asm__ ("v8"), + temp1S __asm__ ("v9"), + temp2S __asm__ ("v10"), + temp3S __asm__ ("v11"), + temp4S __asm__ ("v12"), + temp5S __asm__ ("v13"), + temp6S __asm__ ("v14"), + temp7S __asm__ ("v15"); + register const vector unsigned char vzero __asm__ ("v31") = (const vector unsigned char)vec_splat_u8(0); { - register const vector signed short vprod1 REG_v(v16)= + register const vector signed short vprod1 __asm__ ("v16") = (const vector signed short){ 1,-1, 1,-1, 1,-1, 1,-1 }; - register const vector signed short vprod2 REG_v(v17)= + register const vector signed short vprod2 __asm__ ("v17") = (const vector signed short){ 1, 1,-1,-1, 1, 1,-1,-1 }; - register const vector signed short vprod3 REG_v(v18)= + register const vector signed short vprod3 __asm__ ("v18") = (const vector signed short){ 1, 1, 1, 1,-1,-1,-1,-1 }; - register const vector unsigned char perm1 REG_v(v19)= + register const vector unsigned char perm1 __asm__ ("v19") = (const vector unsigned char) {0x02, 0x03, 0x00, 0x01, 0x06, 0x07, 0x04, 0x05, 0x0A, 0x0B, 0x08, 0x09, 0x0E, 0x0F, 0x0C, 0x0D}; - register const vector unsigned char perm2 REG_v(v20)= + register const vector unsigned char perm2 __asm__ ("v20") = (const vector unsigned char) {0x04, 0x05, 0x06, 0x07, 0x00, 0x01, 0x02, 0x03, 0x0C, 0x0D, 0x0E, 0x0F, 0x08, 0x09, 0x0A, 0x0B}; - register const vector unsigned char perm3 REG_v(v21)= + register const vector unsigned char perm3 __asm__ ("v21") = (const vector unsigned char) {0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07}; -#define ONEITERBUTTERFLY(i, res1, res2) \ - { \ - register vector unsigned char src1 REG_v(v22), \ - src2 REG_v(v23), \ - dst1 REG_v(v24), \ - dst2 REG_v(v25), \ - srcO REG_v(v22), \ - dstO REG_v(v23); \ +#define ONEITERBUTTERFLY(i, res1, res2) \ + { \ + register vector unsigned char src1 __asm__ ("v22"), \ + src2 __asm__ ("v23"), \ + dst1 __asm__ ("v24"), \ + dst2 __asm__ ("v25"), \ + srcO __asm__ ("v22"), \ + dstO __asm__ ("v23"); \ \ - register vector signed short srcV REG_v(v24), \ - dstV REG_v(v25), \ - srcW REG_v(v26), \ - dstW REG_v(v27), \ - but0 REG_v(v28), \ - but0S REG_v(v29), \ - op1 REG_v(v30), \ - but1 REG_v(v22), \ - op1S REG_v(v23), \ - but1S REG_v(v24), \ - op2 REG_v(v25), \ - but2 REG_v(v26), \ - op2S REG_v(v27), \ - but2S REG_v(v28), \ - op3 REG_v(v29), \ - op3S REG_v(v30); \ + register vector signed short srcV __asm__ ("v24"), \ + dstV __asm__ ("v25"), \ + srcW __asm__ ("v26"), \ + dstW __asm__ ("v27"), \ + but0 __asm__ ("v28"), \ + but0S __asm__ ("v29"), \ + op1 __asm__ ("v30"), \ + but1 __asm__ ("v22"), \ + op1S __asm__ ("v23"), \ + but1S __asm__ ("v24"), \ + op2 __asm__ ("v25"), \ + but2 __asm__ ("v26"), \ + op2S __asm__ ("v27"), \ + but2S __asm__ ("v28"), \ + op3 __asm__ ("v29"), \ + op3S __asm__ ("v30"); \ \ src1 = vec_ld(stride * i, src); \ src2 = vec_ld((stride * i) + 16, src); \ diff --git a/libavcodec/ppc/fdct_altivec.c b/libavcodec/ppc/fdct_altivec.c index 41bebd184a..3dfd59ca29 100644 --- a/libavcodec/ppc/fdct_altivec.c +++ b/libavcodec/ppc/fdct_altivec.c @@ -18,11 +18,13 @@ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ - +#include "config.h" +#if HAVE_ALTIVEC_H +#include +#endif #include "libavutil/common.h" #include "libavcodec/dsputil.h" #include "dsputil_ppc.h" -#include "gcc_fixes.h" #define vs16(v) ((vector signed short)(v)) diff --git a/libavcodec/ppc/float_altivec.c b/libavcodec/ppc/float_altivec.c index fd4aa5375c..096f75f8b9 100644 --- a/libavcodec/ppc/float_altivec.c +++ b/libavcodec/ppc/float_altivec.c @@ -20,8 +20,6 @@ #include "libavcodec/dsputil.h" -#include "gcc_fixes.h" - #include "dsputil_altivec.h" #include "util_altivec.h" diff --git a/libavcodec/ppc/gcc_fixes.h b/libavcodec/ppc/gcc_fixes.h deleted file mode 100644 index a97921f1f5..0000000000 --- a/libavcodec/ppc/gcc_fixes.h +++ /dev/null @@ -1,102 +0,0 @@ -/* - * gcc fixes for altivec. - * Used to workaround broken gcc (FSF gcc-3 pre gcc-3.3) - * and to stay somewhat compatible with Darwin. - * - * This file is part of FFmpeg. - * - * FFmpeg is free software; you can redistribute it and/or - * modify it under the terms of the GNU Lesser General Public - * License as published by the Free Software Foundation; either - * version 2.1 of the License, or (at your option) any later version. - * - * FFmpeg is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * Lesser General Public License for more details. - * - * You should have received a copy of the GNU Lesser General Public - * License along with FFmpeg; if not, write to the Free Software - * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - */ - -#ifndef AVCODEC_PPC_GCC_FIXES_H -#define AVCODEC_PPC_GCC_FIXES_H - -#include "config.h" - -#if HAVE_ALTIVEC_H -#include -#endif - -#if (__GNUC__ < 4) -# define REG_v(a) -#else -# define REG_v(a) __asm__ ( #a ) -#endif - -#if (__GNUC__ == 3 && __GNUC_MINOR__ < 3) - -/* This code was provided to me by Bartosch Pixa - * as a separate header file (broken_mergel.h). - * thanks to lu_zero for the workaround. - * - * See this mail for more information: - * http://gcc.gnu.org/ml/gcc/2003-04/msg00967.html - */ - -static inline vector signed char ff_vmrglb (vector signed char const A, - vector signed char const B) -{ - static const vector unsigned char lowbyte = { - 0x08, 0x18, 0x09, 0x19, 0x0a, 0x1a, 0x0b, 0x1b, - 0x0c, 0x1c, 0x0d, 0x1d, 0x0e, 0x1e, 0x0f, 0x1f - }; - return vec_perm (A, B, lowbyte); -} - -static inline vector signed short ff_vmrglh (vector signed short const A, - vector signed short const B) -{ - static const vector unsigned char lowhalf = { - 0x08, 0x09, 0x18, 0x19, 0x0a, 0x0b, 0x1a, 0x1b, - 0x0c, 0x0d, 0x1c, 0x1d, 0x0e, 0x0f, 0x1e, 0x1f - }; - return vec_perm (A, B, lowhalf); -} - -static inline vector signed int ff_vmrglw (vector signed int const A, - vector signed int const B) -{ - static const vector unsigned char lowword = { - 0x08, 0x09, 0x0a, 0x0b, 0x18, 0x19, 0x1a, 0x1b, - 0x0c, 0x0d, 0x0e, 0x0f, 0x1c, 0x1d, 0x1e, 0x1f - }; - return vec_perm (A, B, lowword); -} -/*#define ff_vmrglb ff_vmrglb -#define ff_vmrglh ff_vmrglh -#define ff_vmrglw ff_vmrglw -*/ -#undef vec_mergel - -#define vec_mergel(a1, a2) \ -__ch (__bin_args_eq (vector signed char, (a1), vector signed char, (a2)), \ - ((vector signed char) ff_vmrglb ((vector signed char) (a1), (vector signed char) (a2))), \ -__ch (__bin_args_eq (vector unsigned char, (a1), vector unsigned char, (a2)), \ - ((vector unsigned char) ff_vmrglb ((vector signed char) (a1), (vector signed char) (a2))), \ -__ch (__bin_args_eq (vector signed short, (a1), vector signed short, (a2)), \ - ((vector signed short) ff_vmrglh ((vector signed short) (a1), (vector signed short) (a2))), \ -__ch (__bin_args_eq (vector unsigned short, (a1), vector unsigned short, (a2)), \ - ((vector unsigned short) ff_vmrglh ((vector signed short) (a1), (vector signed short) (a2))), \ -__ch (__bin_args_eq (vector float, (a1), vector float, (a2)), \ - ((vector float) ff_vmrglw ((vector signed int) (a1), (vector signed int) (a2))), \ -__ch (__bin_args_eq (vector signed int, (a1), vector signed int, (a2)), \ - ((vector signed int) ff_vmrglw ((vector signed int) (a1), (vector signed int) (a2))), \ -__ch (__bin_args_eq (vector unsigned int, (a1), vector unsigned int, (a2)), \ - ((vector unsigned int) ff_vmrglw ((vector signed int) (a1), (vector signed int) (a2))), \ - __altivec_link_error_invalid_argument ()))))))) - -#endif /* (__GNUC__ == 3 && __GNUC_MINOR__ < 3) */ - -#endif /* AVCODEC_PPC_GCC_FIXES_H */ diff --git a/libavcodec/ppc/h264_altivec.c b/libavcodec/ppc/h264_altivec.c index 5248a2f9c3..4efd13cf12 100644 --- a/libavcodec/ppc/h264_altivec.c +++ b/libavcodec/ppc/h264_altivec.c @@ -21,8 +21,6 @@ #include "libavcodec/dsputil.h" #include "libavcodec/h264data.h" -#include "gcc_fixes.h" - #include "dsputil_ppc.h" #include "dsputil_altivec.h" #include "util_altivec.h" diff --git a/libavcodec/ppc/idct_altivec.c b/libavcodec/ppc/idct_altivec.c index 19cc93a980..6043a172bf 100644 --- a/libavcodec/ppc/idct_altivec.c +++ b/libavcodec/ppc/idct_altivec.c @@ -37,9 +37,11 @@ #include /* malloc(), free() */ #include +#include "config.h" +#if HAVE_ALTIVEC_H +#include +#endif #include "libavcodec/dsputil.h" - -#include "gcc_fixes.h" #include "types_altivec.h" #include "dsputil_ppc.h" diff --git a/libavcodec/ppc/mpegvideo_altivec.c b/libavcodec/ppc/mpegvideo_altivec.c index 28ada7a316..74775f05fd 100644 --- a/libavcodec/ppc/mpegvideo_altivec.c +++ b/libavcodec/ppc/mpegvideo_altivec.c @@ -26,8 +26,6 @@ #include "libavcodec/dsputil.h" #include "libavcodec/mpegvideo.h" -#include "gcc_fixes.h" - #include "dsputil_ppc.h" #include "util_altivec.h" // Swaps two variables (used for altivec registers)