falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
Browse Source

Merge remote-tracking branch 'qatar/master' 

* qatar/master:
  dsputil: fix invalid array indexing
  configure: add libavresample to rpath
  build: icc: silence some warnings
  fft-test: add option to set cpuflag mask
  cpu: recognise only cpu flag names pertinent to the architecture
  avutil: add av_parse_cpu_flags() function
  vp8: armv6: fix non-armv6t2 build
  vp8: armv6 optimisations
  vp8: arm: separate ARMv6 functions from NEON
  ARM: add some compatibility macros
  mov: support eac3 audio
  avf: fix faulty check in has_duration

Conflicts:
	configure
	doc/APIchanges
	ffmpeg.c
	libavcodec/arm/Makefile
	libavcodec/arm/asm.S
	libavcodec/arm/vp8dsp_armv6.S
	libavcodec/arm/vp8dsp_init_arm.c
	libavutil/avutil.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
tags/n0.11
Michael Niedermayer 13 years ago
parent
f1f5b39866 0a07f2b346
commit
e890b68962
17 changed files with 1529 additions and 2100 deletions
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +3
    -2
      configure
  2. +3
    -0
      doc/APIchanges
  3. +3
    -1
      libavcodec/arm/Makefile
  4. +8
    -1
      libavcodec/arm/asm.S
  5. +78
    -0
      libavcodec/arm/vp8dsp.h
  6. +1090
    -1790
      libavcodec/arm/vp8dsp_armv6.S
  7. +5
    -285
      libavcodec/arm/vp8dsp_init_arm.c
  8. +118
    -0
      libavcodec/arm/vp8dsp_init_armv6.c
  9. +114
    -0
      libavcodec/arm/vp8dsp_init_neon.c
  10. +10
    -6
      libavcodec/dsputil.c
  11. +9
    -1
      libavcodec/fft-test.c
  12. +1
    -0
      libavformat/isom.c
  13. +1
    -1
      libavformat/utils.c
  14. +1
    -1
      libavutil/avutil.h
  15. +64
    -0
      libavutil/cpu.c
  16. +7
    -0
      libavutil/cpu.h
  17. +14
    -12
      tests/fate/fft.mak

+ 3
- 2
configure View File

@@ -3358,7 +3358,7 @@ enabled extra_warnings && check_cflags -Winline

# add some linker flags
check_ldflags -Wl,--warn-common
check_ldflags -Wl,-rpath-link=libpostproc:libswresample:libswscale:libavfilter:libavdevice:libavformat:libavcodec:libavutil
check_ldflags -Wl,-rpath-link=libpostproc:libswresample:libswscale:libavfilter:libavdevice:libavformat:libavcodec:libavutil:libavresample
test_ldflags -Wl,-Bsymbolic && append SHFLAGS -Wl,-Bsymbolic

enabled xmm_clobber_test && \
@@ -3407,10 +3407,11 @@ if enabled icc; then
# -wd: Disable following warnings
# 144, 167, 556: -Wno-pointer-sign
# 1292: attribute "foo" ignored
# 1419: external declaration in primary source file
# 10006: ignoring unknown option -fno-signed-zeros
# 10148: ignoring unknown option -Wno-parentheses
# 10156: ignoring option '-W'; no argument required
check_cflags -wd144,167,556,1292,10006,10148,10156
check_cflags -wd144,167,556,1292,1419,10006,10148,10156
# 11030: Warning unknown option --as-needed
# 10156: ignoring option '-export'; no argument required
check_ldflags -wd10156,11030


+ 3
- 0
doc/APIchanges View File

@@ -23,6 +23,9 @@ API changes, most recent first:
2012-03-26 - a67d9cf - lavfi 2.66.100
Add avfilter_fill_frame_from_{audio_,}buffer_ref() functions.

2012-04-25 - xxxxxxx - lavu 51.29.0 - cpu.h
Add av_parse_cpu_flags()

2012-xx-xx - xxxxxxx - lavr 0.0.0
Add libavresample audio conversion library



+ 3
- 1
libavcodec/arm/Makefile View File

@@ -14,6 +14,7 @@ OBJS-$(CONFIG_VP5_DECODER) += arm/vp56dsp_init_arm.o
OBJS-$(CONFIG_VP6_DECODER) += arm/vp56dsp_init_arm.o
OBJS-$(CONFIG_VP8_DECODER) += arm/vp8dsp_init_arm.o
ARMV6-OBJS-$(CONFIG_VP8_DECODER) += arm/vp8_armv6.o \
arm/vp8dsp_init_armv6.o \
arm/vp8dsp_armv6.o

OBJS-$(CONFIG_H264DSP) += arm/h264dsp_init_arm.o
@@ -80,7 +81,8 @@ NEON-OBJS-$(CONFIG_VP5_DECODER) += arm/vp56dsp_neon.o \
NEON-OBJS-$(CONFIG_VP6_DECODER) += arm/vp56dsp_neon.o \
arm/vp3dsp_neon.o \

NEON-OBJS-$(CONFIG_VP8_DECODER) += arm/vp8dsp_neon.o
NEON-OBJS-$(CONFIG_VP8_DECODER) += arm/vp8dsp_init_neon.o \
arm/vp8dsp_neon.o

NEON-OBJS += arm/dsputil_init_neon.o \
arm/dsputil_neon.o \


+ 8
- 1
libavcodec/arm/asm.S View File

@@ -88,6 +88,13 @@ ELF .size \name, . - \name
\name:
.endm

#if !HAVE_ARMV6T2
.macro movw rd, val
mov \rd, \val & 255
orr \rd, \val & ~255
.endm
#endif

.macro mov32 rd, val
#if HAVE_ARMV6T2
movw \rd, #(\val) & 0xffff
@@ -120,7 +127,7 @@ T sub \rn, \rn, \rm
T ldr \rt, [\rn]
.endm

.macro ldr_dpren rt, rn, rm:vararg
.macro ldr_nreg rt, rn, rm:vararg
A ldr \rt, [\rn, -\rm]
T sub \rt, \rn, \rm
T ldr \rt, [\rt]


+ 78
- 0
libavcodec/arm/vp8dsp.h View File

@@ -0,0 +1,78 @@
/*
* This file is part of Libav.
*
* Libav 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.
*
* Libav 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 Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/

#ifndef AVCODEC_ARM_VP8DSP_H
#define AVCODEC_ARM_VP8DSP_H

#include "libavcodec/vp8dsp.h"

void ff_vp8dsp_init_armv6(VP8DSPContext *dsp);
void ff_vp8dsp_init_neon(VP8DSPContext *dsp);

#define VP8_LF_Y(hv, inner, opt) \
void ff_vp8_##hv##_loop_filter16##inner##_##opt(uint8_t *dst, \
ptrdiff_t stride, \
int flim_E, int flim_I, \
int hev_thresh)

#define VP8_LF_UV(hv, inner, opt) \
void ff_vp8_##hv##_loop_filter8uv##inner##_##opt(uint8_t *dstU, \
uint8_t *dstV, \
ptrdiff_t stride, \
int flim_E, int flim_I, \
int hev_thresh)

#define VP8_LF_SIMPLE(hv, opt) \
void ff_vp8_##hv##_loop_filter16_simple_##opt(uint8_t *dst, \
ptrdiff_t stride, \
int flim)

#define VP8_LF_HV(inner, opt) \
VP8_LF_Y(h, inner, opt); \
VP8_LF_Y(v, inner, opt); \
VP8_LF_UV(h, inner, opt); \
VP8_LF_UV(v, inner, opt)

#define VP8_LF(opt) \
VP8_LF_HV(, opt); \
VP8_LF_HV(_inner, opt); \
VP8_LF_SIMPLE(h, opt); \
VP8_LF_SIMPLE(v, opt)

#define VP8_MC(n, opt) \
void ff_put_vp8_##n##_##opt(uint8_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int h, int x, int y)

#define VP8_EPEL(w, opt) \
VP8_MC(pixels ## w, opt); \
VP8_MC(epel ## w ## _h4, opt); \
VP8_MC(epel ## w ## _h6, opt); \
VP8_MC(epel ## w ## _v4, opt); \
VP8_MC(epel ## w ## _h4v4, opt); \
VP8_MC(epel ## w ## _h6v4, opt); \
VP8_MC(epel ## w ## _v6, opt); \
VP8_MC(epel ## w ## _h4v6, opt); \
VP8_MC(epel ## w ## _h6v6, opt)

#define VP8_BILIN(w, opt) \
VP8_MC(bilin ## w ## _h, opt); \
VP8_MC(bilin ## w ## _v, opt); \
VP8_MC(bilin ## w ## _hv, opt)

#endif /* AVCODEC_ARM_VP8DSP_H */

+ 1090
- 1790
libavcodec/arm/vp8dsp_armv6.S
File diff suppressed because it is too large
View File


+ 5
- 285
libavcodec/arm/vp8dsp_init_arm.c View File

@@ -20,294 +20,14 @@

#include "libavutil/arm/cpu.h"
#include "libavcodec/vp8dsp.h"

void ff_vp8_luma_dc_wht_dc_armv6(DCTELEM block[4][4][16], DCTELEM dc[16]);

#define idct_funcs(opt) \
void ff_vp8_luma_dc_wht_ ## opt(DCTELEM block[4][4][16], DCTELEM dc[16]); \
void ff_vp8_idct_add_ ## opt(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride); \
void ff_vp8_idct_dc_add_ ## opt(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride); \
void ff_vp8_idct_dc_add4y_ ## opt(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride); \
void ff_vp8_idct_dc_add4uv_ ## opt(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride)

idct_funcs(neon);
idct_funcs(armv6);

void ff_vp8_v_loop_filter16_neon(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh);
void ff_vp8_h_loop_filter16_neon(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh);
void ff_vp8_v_loop_filter8uv_neon(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh);
void ff_vp8_h_loop_filter8uv_neon(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh);

void ff_vp8_v_loop_filter16_inner_neon(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh);
void ff_vp8_h_loop_filter16_inner_neon(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh);
void ff_vp8_v_loop_filter8uv_inner_neon(uint8_t *dstU, uint8_t *dstV,
ptrdiff_t stride, int flim_E, int flim_I,
int hev_thresh);
void ff_vp8_h_loop_filter8uv_inner_neon(uint8_t *dstU, uint8_t *dstV,
ptrdiff_t stride, int flim_E, int flim_I,
int hev_thresh);

void ff_vp8_v_loop_filter_inner_armv6(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I,
int hev_thresh, int count);
void ff_vp8_h_loop_filter_inner_armv6(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I,
int hev_thresh, int count);
void ff_vp8_v_loop_filter_armv6(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I,
int hev_thresh, int count);
void ff_vp8_h_loop_filter_armv6(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I,
int hev_thresh, int count);

static void ff_vp8_v_loop_filter16_armv6(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh)
{
ff_vp8_v_loop_filter_armv6(dst, stride, flim_E, flim_I, hev_thresh, 4);
}

static void ff_vp8_h_loop_filter16_armv6(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh)
{
ff_vp8_h_loop_filter_armv6(dst, stride, flim_E, flim_I, hev_thresh, 4);
}

static void ff_vp8_v_loop_filter8uv_armv6(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh)
{
ff_vp8_v_loop_filter_armv6(dstU, stride, flim_E, flim_I, hev_thresh, 2);
ff_vp8_v_loop_filter_armv6(dstV, stride, flim_E, flim_I, hev_thresh, 2);
}

static void ff_vp8_h_loop_filter8uv_armv6(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh)
{
ff_vp8_h_loop_filter_armv6(dstU, stride, flim_E, flim_I, hev_thresh, 2);
ff_vp8_h_loop_filter_armv6(dstV, stride, flim_E, flim_I, hev_thresh, 2);
}

static void ff_vp8_v_loop_filter16_inner_armv6(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh)
{
ff_vp8_v_loop_filter_inner_armv6(dst, stride, flim_E, flim_I, hev_thresh, 4);
}

static void ff_vp8_h_loop_filter16_inner_armv6(uint8_t *dst, ptrdiff_t stride,
int flim_E, int flim_I, int hev_thresh)
{
ff_vp8_h_loop_filter_inner_armv6(dst, stride, flim_E, flim_I, hev_thresh, 4);
}

static void ff_vp8_v_loop_filter8uv_inner_armv6(uint8_t *dstU, uint8_t *dstV,
ptrdiff_t stride, int flim_E, int flim_I,
int hev_thresh)
{
ff_vp8_v_loop_filter_inner_armv6(dstU, stride, flim_E, flim_I, hev_thresh, 2);
ff_vp8_v_loop_filter_inner_armv6(dstV, stride, flim_E, flim_I, hev_thresh, 2);
}

static void ff_vp8_h_loop_filter8uv_inner_armv6(uint8_t *dstU, uint8_t *dstV,
ptrdiff_t stride, int flim_E, int flim_I,
int hev_thresh)
{
ff_vp8_h_loop_filter_inner_armv6(dstU, stride, flim_E, flim_I, hev_thresh, 2);
ff_vp8_h_loop_filter_inner_armv6(dstV, stride, flim_E, flim_I, hev_thresh, 2);
}

#define simple_lf_funcs(opt) \
void ff_vp8_v_loop_filter16_simple_ ## opt(uint8_t *dst, ptrdiff_t stride, int flim); \
void ff_vp8_h_loop_filter16_simple_ ## opt(uint8_t *dst, ptrdiff_t stride, int flim)

simple_lf_funcs(neon);
simple_lf_funcs(armv6);

#define VP8_MC_OPT(n, opt) \
void ff_put_vp8_##n##_##opt(uint8_t *dst, ptrdiff_t dststride, \
uint8_t *src, ptrdiff_t srcstride, \
int h, int x, int y)

#define VP8_MC(n) \
VP8_MC_OPT(n, neon)

#define VP8_EPEL(w) \
VP8_MC(epel ## w ## _h4); \
VP8_MC(epel ## w ## _h6); \
VP8_MC(epel ## w ## _h4v4); \
VP8_MC(epel ## w ## _h6v4); \
VP8_MC(epel ## w ## _v4); \
VP8_MC(epel ## w ## _v6); \
VP8_MC(epel ## w ## _h4v6); \
VP8_MC(epel ## w ## _h6v6)

VP8_EPEL(16);
VP8_MC(pixels16);
VP8_MC_OPT(pixels16, armv6);
VP8_EPEL(8);
VP8_MC(pixels8);
VP8_MC_OPT(pixels8, armv6);
VP8_EPEL(4);
VP8_MC_OPT(pixels4, armv6);

VP8_MC(bilin16_h);
VP8_MC(bilin16_v);
VP8_MC(bilin16_hv);
VP8_MC(bilin8_h);
VP8_MC(bilin8_v);
VP8_MC(bilin8_hv);
VP8_MC(bilin4_h);
VP8_MC(bilin4_v);
VP8_MC(bilin4_hv);

#define VP8_V6_MC(n) \
void ff_put_vp8_##n##_armv6(uint8_t *dst, int dststride, uint8_t *src, \
int srcstride, int w, int h, int mxy)

VP8_V6_MC(epel_v6);
VP8_V6_MC(epel_h6);
VP8_V6_MC(epel_v4);
VP8_V6_MC(epel_h4);
VP8_V6_MC(bilin_v);
VP8_V6_MC(bilin_h);

#define VP8_EPEL_HV(SIZE, TAPNUMX, TAPNUMY, NAME, HNAME, VNAME, MAXHEIGHT) \
static void ff_put_vp8_##NAME##SIZE##_##HNAME##VNAME##_armv6( \
uint8_t *dst, int dststride, uint8_t *src, \
int srcstride, int h, int mx, int my) \
{ \
DECLARE_ALIGNED(4, uint8_t, tmp)[SIZE * (MAXHEIGHT + TAPNUMY - 1)]; \
uint8_t *tmpptr = tmp + SIZE * (TAPNUMY / 2 - 1); \
src -= srcstride * (TAPNUMY / 2 - 1); \
ff_put_vp8_ ## NAME ## _ ## HNAME ## _armv6(tmp, SIZE, src, srcstride, \
SIZE, h + TAPNUMY - 1, mx); \
ff_put_vp8_ ## NAME ## _ ## VNAME ## _armv6(dst, dststride, tmpptr, SIZE, \
SIZE, h, my); \
}

VP8_EPEL_HV(16, 6, 6, epel, h6, v6, 16);
VP8_EPEL_HV(16, 2, 2, bilin, h, v, 16);
VP8_EPEL_HV(8, 6, 6, epel, h6, v6, 16);
VP8_EPEL_HV(8, 4, 6, epel, h4, v6, 16);
VP8_EPEL_HV(8, 6, 4, epel, h6, v4, 16);
VP8_EPEL_HV(8, 4, 4, epel, h4, v4, 16);
VP8_EPEL_HV(8, 2, 2, bilin, h, v, 16);
VP8_EPEL_HV(4, 6, 6, epel, h6, v6, 8);
VP8_EPEL_HV(4, 4, 6, epel, h4, v6, 8);
VP8_EPEL_HV(4, 6, 4, epel, h6, v4, 8);
VP8_EPEL_HV(4, 4, 4, epel, h4, v4, 8);
VP8_EPEL_HV(4, 2, 2, bilin, h, v, 8);

extern void put_vp8_epel4_v6_c(uint8_t *dst, int d, uint8_t *src, int s, int h, int mx, int my);
#undef printf
#define VP8_EPEL_H_OR_V(SIZE, NAME, HV) \
static void ff_put_vp8_##NAME##SIZE##_##HV##_armv6( \
uint8_t *dst, int dststride, uint8_t *src, \
int srcstride, int h, int mx, int my) \
{ \
ff_put_vp8_## NAME ## _ ## HV ## _armv6(dst, dststride, src, srcstride, \
SIZE, h, mx | my); \
}

VP8_EPEL_H_OR_V(4, epel, h6);
VP8_EPEL_H_OR_V(4, epel, h4);
VP8_EPEL_H_OR_V(4, epel, v6);
VP8_EPEL_H_OR_V(4, epel, v4);
VP8_EPEL_H_OR_V(4, bilin, v);
VP8_EPEL_H_OR_V(4, bilin, h);
VP8_EPEL_H_OR_V(8, epel, h6);
VP8_EPEL_H_OR_V(8, epel, h4);
VP8_EPEL_H_OR_V(8, epel, v6);
VP8_EPEL_H_OR_V(8, epel, v4);
VP8_EPEL_H_OR_V(8, bilin, v);
VP8_EPEL_H_OR_V(8, bilin, h);
VP8_EPEL_H_OR_V(16, epel, h6);
VP8_EPEL_H_OR_V(16, epel, v6);
VP8_EPEL_H_OR_V(16, bilin, v);
VP8_EPEL_H_OR_V(16, bilin, h);
#include "vp8dsp.h"

av_cold void ff_vp8dsp_init_arm(VP8DSPContext *dsp)
{
#define set_func_ptrs(opt) \
dsp->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_##opt; \
dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_armv6; \
\
dsp->vp8_idct_add = ff_vp8_idct_add_##opt; \
dsp->vp8_idct_dc_add = ff_vp8_idct_dc_add_##opt; \
dsp->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_##opt; \
dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_##opt; \
\
dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_##opt; \
dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_##opt; \
dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_##opt; \
dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_##opt; \
\
dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_##opt; \
dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_##opt; \
dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_##opt; \
dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_##opt; \
\
dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_##opt; \
dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_##opt; \
\
dsp->put_vp8_epel_pixels_tab[0][0][0] = ff_put_vp8_pixels16_##opt; \
dsp->put_vp8_epel_pixels_tab[0][0][2] = ff_put_vp8_epel16_h6_##opt; \
dsp->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_##opt; \
dsp->put_vp8_epel_pixels_tab[0][2][2] = ff_put_vp8_epel16_h6v6_##opt; \
\
dsp->put_vp8_epel_pixels_tab[1][0][0] = ff_put_vp8_pixels8_##opt; \
dsp->put_vp8_epel_pixels_tab[1][0][1] = ff_put_vp8_epel8_h4_##opt; \
dsp->put_vp8_epel_pixels_tab[1][0][2] = ff_put_vp8_epel8_h6_##opt; \
dsp->put_vp8_epel_pixels_tab[1][1][0] = ff_put_vp8_epel8_v4_##opt; \
dsp->put_vp8_epel_pixels_tab[1][1][1] = ff_put_vp8_epel8_h4v4_##opt; \
dsp->put_vp8_epel_pixels_tab[1][1][2] = ff_put_vp8_epel8_h6v4_##opt; \
dsp->put_vp8_epel_pixels_tab[1][2][0] = ff_put_vp8_epel8_v6_##opt; \
dsp->put_vp8_epel_pixels_tab[1][2][1] = ff_put_vp8_epel8_h4v6_##opt; \
dsp->put_vp8_epel_pixels_tab[1][2][2] = ff_put_vp8_epel8_h6v6_##opt; \
\
dsp->put_vp8_epel_pixels_tab[2][0][0] = ff_put_vp8_pixels4_armv6; \
dsp->put_vp8_epel_pixels_tab[2][0][1] = ff_put_vp8_epel4_h4_##opt; \
dsp->put_vp8_epel_pixels_tab[2][0][2] = ff_put_vp8_epel4_h6_##opt; \
dsp->put_vp8_epel_pixels_tab[2][1][0] = ff_put_vp8_epel4_v4_##opt; \
dsp->put_vp8_epel_pixels_tab[2][1][1] = ff_put_vp8_epel4_h4v4_##opt; \
dsp->put_vp8_epel_pixels_tab[2][1][2] = ff_put_vp8_epel4_h6v4_##opt; \
dsp->put_vp8_epel_pixels_tab[2][2][0] = ff_put_vp8_epel4_v6_##opt; \
dsp->put_vp8_epel_pixels_tab[2][2][1] = ff_put_vp8_epel4_h4v6_##opt; \
dsp->put_vp8_epel_pixels_tab[2][2][2] = ff_put_vp8_epel4_h6v6_##opt; \
\
dsp->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_##opt; \
dsp->put_vp8_bilinear_pixels_tab[0][0][2] = ff_put_vp8_bilin16_h_##opt; \
dsp->put_vp8_bilinear_pixels_tab[0][2][0] = ff_put_vp8_bilin16_v_##opt; \
dsp->put_vp8_bilinear_pixels_tab[0][2][2] = ff_put_vp8_bilin16_hv_##opt; \
\
dsp->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_##opt; \
dsp->put_vp8_bilinear_pixels_tab[1][0][1] = ff_put_vp8_bilin8_h_##opt; \
dsp->put_vp8_bilinear_pixels_tab[1][0][2] = ff_put_vp8_bilin8_h_##opt; \
dsp->put_vp8_bilinear_pixels_tab[1][1][0] = ff_put_vp8_bilin8_v_##opt; \
dsp->put_vp8_bilinear_pixels_tab[1][1][1] = ff_put_vp8_bilin8_hv_##opt; \
dsp->put_vp8_bilinear_pixels_tab[1][1][2] = ff_put_vp8_bilin8_hv_##opt; \
dsp->put_vp8_bilinear_pixels_tab[1][2][0] = ff_put_vp8_bilin8_v_##opt; \
dsp->put_vp8_bilinear_pixels_tab[1][2][1] = ff_put_vp8_bilin8_hv_##opt; \
dsp->put_vp8_bilinear_pixels_tab[1][2][2] = ff_put_vp8_bilin8_hv_##opt; \
\
dsp->put_vp8_bilinear_pixels_tab[2][0][0] = ff_put_vp8_pixels4_armv6; \
dsp->put_vp8_bilinear_pixels_tab[2][0][1] = ff_put_vp8_bilin4_h_##opt; \
dsp->put_vp8_bilinear_pixels_tab[2][0][2] = ff_put_vp8_bilin4_h_##opt; \
dsp->put_vp8_bilinear_pixels_tab[2][1][0] = ff_put_vp8_bilin4_v_##opt; \
dsp->put_vp8_bilinear_pixels_tab[2][1][1] = ff_put_vp8_bilin4_hv_##opt; \
dsp->put_vp8_bilinear_pixels_tab[2][1][2] = ff_put_vp8_bilin4_hv_##opt; \
dsp->put_vp8_bilinear_pixels_tab[2][2][0] = ff_put_vp8_bilin4_v_##opt; \
dsp->put_vp8_bilinear_pixels_tab[2][2][1] = ff_put_vp8_bilin4_hv_##opt; \
dsp->put_vp8_bilinear_pixels_tab[2][2][2] = ff_put_vp8_bilin4_hv_##opt
int cpu_flags = av_get_cpu_flags();

if (have_neon(cpu_flags)) {
set_func_ptrs(neon);
} else if (HAVE_ARMV6) {
set_func_ptrs(armv6);
}
if (have_armv6(cpu_flags))
ff_vp8dsp_init_armv6(dsp);
if (have_neon(cpu_flags))
ff_vp8dsp_init_neon(dsp);
}

+ 118
- 0
libavcodec/arm/vp8dsp_init_armv6.c View File

@@ -0,0 +1,118 @@
/*
* This file is part of Libav.
*
* Libav 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.
*
* Libav 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 Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/

#include <stdint.h>
#include "libavcodec/vp8dsp.h"
#include "vp8dsp.h"

void ff_vp8_luma_dc_wht_armv6(DCTELEM block[4][4][16], DCTELEM dc[16]);
void ff_vp8_luma_dc_wht_dc_armv6(DCTELEM block[4][4][16], DCTELEM dc[16]);

void ff_vp8_idct_add_armv6(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
void ff_vp8_idct_dc_add_armv6(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
void ff_vp8_idct_dc_add4y_armv6(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
void ff_vp8_idct_dc_add4uv_armv6(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);

VP8_LF(armv6);

VP8_EPEL(16, armv6);
VP8_EPEL(8, armv6);
VP8_EPEL(4, armv6);

VP8_BILIN(16, armv6);
VP8_BILIN(8, armv6);
VP8_BILIN(4, armv6);

av_cold void ff_vp8dsp_init_armv6(VP8DSPContext *dsp)
{
dsp->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_armv6;
dsp->vp8_luma_dc_wht_dc = ff_vp8_luma_dc_wht_dc_armv6;

dsp->vp8_idct_add = ff_vp8_idct_add_armv6;
dsp->vp8_idct_dc_add = ff_vp8_idct_dc_add_armv6;
dsp->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_armv6;
dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_armv6;

dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_armv6;
dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_armv6;
dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_armv6;
dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_armv6;

dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_armv6;
dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_armv6;
dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_armv6;
dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_armv6;

dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_armv6;
dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_armv6;

dsp->put_vp8_epel_pixels_tab[0][0][0] = ff_put_vp8_pixels16_armv6;
dsp->put_vp8_epel_pixels_tab[0][0][2] = ff_put_vp8_epel16_h6_armv6;
dsp->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_armv6;
dsp->put_vp8_epel_pixels_tab[0][2][2] = ff_put_vp8_epel16_h6v6_armv6;

dsp->put_vp8_epel_pixels_tab[1][0][0] = ff_put_vp8_pixels8_armv6;
dsp->put_vp8_epel_pixels_tab[1][0][1] = ff_put_vp8_epel8_h4_armv6;
dsp->put_vp8_epel_pixels_tab[1][0][2] = ff_put_vp8_epel8_h6_armv6;
dsp->put_vp8_epel_pixels_tab[1][1][0] = ff_put_vp8_epel8_v4_armv6;
dsp->put_vp8_epel_pixels_tab[1][1][1] = ff_put_vp8_epel8_h4v4_armv6;
dsp->put_vp8_epel_pixels_tab[1][1][2] = ff_put_vp8_epel8_h6v4_armv6;
dsp->put_vp8_epel_pixels_tab[1][2][0] = ff_put_vp8_epel8_v6_armv6;
dsp->put_vp8_epel_pixels_tab[1][2][1] = ff_put_vp8_epel8_h4v6_armv6;
dsp->put_vp8_epel_pixels_tab[1][2][2] = ff_put_vp8_epel8_h6v6_armv6;

dsp->put_vp8_epel_pixels_tab[2][0][0] = ff_put_vp8_pixels4_armv6;
dsp->put_vp8_epel_pixels_tab[2][0][1] = ff_put_vp8_epel4_h4_armv6;
dsp->put_vp8_epel_pixels_tab[2][0][2] = ff_put_vp8_epel4_h6_armv6;
dsp->put_vp8_epel_pixels_tab[2][1][0] = ff_put_vp8_epel4_v4_armv6;
dsp->put_vp8_epel_pixels_tab[2][1][1] = ff_put_vp8_epel4_h4v4_armv6;
dsp->put_vp8_epel_pixels_tab[2][1][2] = ff_put_vp8_epel4_h6v4_armv6;
dsp->put_vp8_epel_pixels_tab[2][2][0] = ff_put_vp8_epel4_v6_armv6;
dsp->put_vp8_epel_pixels_tab[2][2][1] = ff_put_vp8_epel4_h4v6_armv6;
dsp->put_vp8_epel_pixels_tab[2][2][2] = ff_put_vp8_epel4_h6v6_armv6;

dsp->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_armv6;
dsp->put_vp8_bilinear_pixels_tab[0][0][1] = ff_put_vp8_bilin16_h_armv6;
dsp->put_vp8_bilinear_pixels_tab[0][0][2] = ff_put_vp8_bilin16_h_armv6;
dsp->put_vp8_bilinear_pixels_tab[0][1][0] = ff_put_vp8_bilin16_v_armv6;
dsp->put_vp8_bilinear_pixels_tab[0][1][1] = ff_put_vp8_bilin16_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[0][1][2] = ff_put_vp8_bilin16_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[0][2][0] = ff_put_vp8_bilin16_v_armv6;
dsp->put_vp8_bilinear_pixels_tab[0][2][1] = ff_put_vp8_bilin16_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[0][2][2] = ff_put_vp8_bilin16_hv_armv6;

dsp->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_armv6;
dsp->put_vp8_bilinear_pixels_tab[1][0][1] = ff_put_vp8_bilin8_h_armv6;
dsp->put_vp8_bilinear_pixels_tab[1][0][2] = ff_put_vp8_bilin8_h_armv6;
dsp->put_vp8_bilinear_pixels_tab[1][1][0] = ff_put_vp8_bilin8_v_armv6;
dsp->put_vp8_bilinear_pixels_tab[1][1][1] = ff_put_vp8_bilin8_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[1][1][2] = ff_put_vp8_bilin8_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[1][2][0] = ff_put_vp8_bilin8_v_armv6;
dsp->put_vp8_bilinear_pixels_tab[1][2][1] = ff_put_vp8_bilin8_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[1][2][2] = ff_put_vp8_bilin8_hv_armv6;

dsp->put_vp8_bilinear_pixels_tab[2][0][0] = ff_put_vp8_pixels4_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][0][1] = ff_put_vp8_bilin4_h_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][0][2] = ff_put_vp8_bilin4_h_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][1][0] = ff_put_vp8_bilin4_v_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][1][1] = ff_put_vp8_bilin4_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][1][2] = ff_put_vp8_bilin4_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][2][0] = ff_put_vp8_bilin4_v_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][2][1] = ff_put_vp8_bilin4_hv_armv6;
dsp->put_vp8_bilinear_pixels_tab[2][2][2] = ff_put_vp8_bilin4_hv_armv6;
}

+ 114
- 0
libavcodec/arm/vp8dsp_init_neon.c View File

@@ -0,0 +1,114 @@
/*
* This file is part of Libav.
*
* Libav 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.
*
* Libav 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 Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/

#include <stdint.h>
#include "libavcodec/vp8dsp.h"
#include "vp8dsp.h"

void ff_vp8_luma_dc_wht_neon(DCTELEM block[4][4][16], DCTELEM dc[16]);

void ff_vp8_idct_add_neon(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
void ff_vp8_idct_dc_add_neon(uint8_t *dst, DCTELEM block[16], ptrdiff_t stride);
void ff_vp8_idct_dc_add4y_neon(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);
void ff_vp8_idct_dc_add4uv_neon(uint8_t *dst, DCTELEM block[4][16], ptrdiff_t stride);

VP8_LF(neon);

VP8_EPEL(16, neon);
VP8_EPEL(8, neon);
VP8_EPEL(4, neon);

VP8_BILIN(16, neon);
VP8_BILIN(8, neon);
VP8_BILIN(4, neon);

av_cold void ff_vp8dsp_init_neon(VP8DSPContext *dsp)
{
dsp->vp8_luma_dc_wht = ff_vp8_luma_dc_wht_neon;

dsp->vp8_idct_add = ff_vp8_idct_add_neon;
dsp->vp8_idct_dc_add = ff_vp8_idct_dc_add_neon;
dsp->vp8_idct_dc_add4y = ff_vp8_idct_dc_add4y_neon;
dsp->vp8_idct_dc_add4uv = ff_vp8_idct_dc_add4uv_neon;

dsp->vp8_v_loop_filter16y = ff_vp8_v_loop_filter16_neon;
dsp->vp8_h_loop_filter16y = ff_vp8_h_loop_filter16_neon;
dsp->vp8_v_loop_filter8uv = ff_vp8_v_loop_filter8uv_neon;
dsp->vp8_h_loop_filter8uv = ff_vp8_h_loop_filter8uv_neon;

dsp->vp8_v_loop_filter16y_inner = ff_vp8_v_loop_filter16_inner_neon;
dsp->vp8_h_loop_filter16y_inner = ff_vp8_h_loop_filter16_inner_neon;
dsp->vp8_v_loop_filter8uv_inner = ff_vp8_v_loop_filter8uv_inner_neon;
dsp->vp8_h_loop_filter8uv_inner = ff_vp8_h_loop_filter8uv_inner_neon;

dsp->vp8_v_loop_filter_simple = ff_vp8_v_loop_filter16_simple_neon;
dsp->vp8_h_loop_filter_simple = ff_vp8_h_loop_filter16_simple_neon;

dsp->put_vp8_epel_pixels_tab[0][0][0] = ff_put_vp8_pixels16_neon;
dsp->put_vp8_epel_pixels_tab[0][0][2] = ff_put_vp8_epel16_h6_neon;
dsp->put_vp8_epel_pixels_tab[0][2][0] = ff_put_vp8_epel16_v6_neon;
dsp->put_vp8_epel_pixels_tab[0][2][2] = ff_put_vp8_epel16_h6v6_neon;

dsp->put_vp8_epel_pixels_tab[1][0][0] = ff_put_vp8_pixels8_neon;
dsp->put_vp8_epel_pixels_tab[1][0][1] = ff_put_vp8_epel8_h4_neon;
dsp->put_vp8_epel_pixels_tab[1][0][2] = ff_put_vp8_epel8_h6_neon;
dsp->put_vp8_epel_pixels_tab[1][1][0] = ff_put_vp8_epel8_v4_neon;
dsp->put_vp8_epel_pixels_tab[1][1][1] = ff_put_vp8_epel8_h4v4_neon;
dsp->put_vp8_epel_pixels_tab[1][1][2] = ff_put_vp8_epel8_h6v4_neon;
dsp->put_vp8_epel_pixels_tab[1][2][0] = ff_put_vp8_epel8_v6_neon;
dsp->put_vp8_epel_pixels_tab[1][2][1] = ff_put_vp8_epel8_h4v6_neon;
dsp->put_vp8_epel_pixels_tab[1][2][2] = ff_put_vp8_epel8_h6v6_neon;

dsp->put_vp8_epel_pixels_tab[2][0][1] = ff_put_vp8_epel4_h4_neon;
dsp->put_vp8_epel_pixels_tab[2][0][2] = ff_put_vp8_epel4_h6_neon;
dsp->put_vp8_epel_pixels_tab[2][1][0] = ff_put_vp8_epel4_v4_neon;
dsp->put_vp8_epel_pixels_tab[2][1][1] = ff_put_vp8_epel4_h4v4_neon;
dsp->put_vp8_epel_pixels_tab[2][1][2] = ff_put_vp8_epel4_h6v4_neon;
dsp->put_vp8_epel_pixels_tab[2][2][0] = ff_put_vp8_epel4_v6_neon;
dsp->put_vp8_epel_pixels_tab[2][2][1] = ff_put_vp8_epel4_h4v6_neon;
dsp->put_vp8_epel_pixels_tab[2][2][2] = ff_put_vp8_epel4_h6v6_neon;

dsp->put_vp8_bilinear_pixels_tab[0][0][0] = ff_put_vp8_pixels16_neon;
dsp->put_vp8_bilinear_pixels_tab[0][0][1] = ff_put_vp8_bilin16_h_neon;
dsp->put_vp8_bilinear_pixels_tab[0][0][2] = ff_put_vp8_bilin16_h_neon;
dsp->put_vp8_bilinear_pixels_tab[0][1][0] = ff_put_vp8_bilin16_v_neon;
dsp->put_vp8_bilinear_pixels_tab[0][1][1] = ff_put_vp8_bilin16_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[0][1][2] = ff_put_vp8_bilin16_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[0][2][0] = ff_put_vp8_bilin16_v_neon;
dsp->put_vp8_bilinear_pixels_tab[0][2][1] = ff_put_vp8_bilin16_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[0][2][2] = ff_put_vp8_bilin16_hv_neon;

dsp->put_vp8_bilinear_pixels_tab[1][0][0] = ff_put_vp8_pixels8_neon;
dsp->put_vp8_bilinear_pixels_tab[1][0][1] = ff_put_vp8_bilin8_h_neon;
dsp->put_vp8_bilinear_pixels_tab[1][0][2] = ff_put_vp8_bilin8_h_neon;
dsp->put_vp8_bilinear_pixels_tab[1][1][0] = ff_put_vp8_bilin8_v_neon;
dsp->put_vp8_bilinear_pixels_tab[1][1][1] = ff_put_vp8_bilin8_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[1][1][2] = ff_put_vp8_bilin8_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[1][2][0] = ff_put_vp8_bilin8_v_neon;
dsp->put_vp8_bilinear_pixels_tab[1][2][1] = ff_put_vp8_bilin8_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[1][2][2] = ff_put_vp8_bilin8_hv_neon;

dsp->put_vp8_bilinear_pixels_tab[2][0][1] = ff_put_vp8_bilin4_h_neon;
dsp->put_vp8_bilinear_pixels_tab[2][0][2] = ff_put_vp8_bilin4_h_neon;
dsp->put_vp8_bilinear_pixels_tab[2][1][0] = ff_put_vp8_bilin4_v_neon;
dsp->put_vp8_bilinear_pixels_tab[2][1][1] = ff_put_vp8_bilin4_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[2][1][2] = ff_put_vp8_bilin4_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[2][2][0] = ff_put_vp8_bilin4_v_neon;
dsp->put_vp8_bilinear_pixels_tab[2][2][1] = ff_put_vp8_bilin4_hv_neon;
dsp->put_vp8_bilinear_pixels_tab[2][2][2] = ff_put_vp8_bilin4_hv_neon;
}

+ 10
- 6
libavcodec/dsputil.c View File

@@ -2819,7 +2819,7 @@ int ff_check_alignment(void){

av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
{
int i;
int i, j;

ff_check_alignment();

@@ -3178,11 +3178,15 @@ av_cold void ff_dsputil_init(DSPContext* c, AVCodecContext *avctx)
if (ARCH_SH4) ff_dsputil_init_sh4 (c, avctx);
if (ARCH_BFIN) ff_dsputil_init_bfin (c, avctx);

for(i=0; i<64; i++){
if(!c->put_2tap_qpel_pixels_tab[0][i])
c->put_2tap_qpel_pixels_tab[0][i]= c->put_h264_qpel_pixels_tab[0][i];
if(!c->avg_2tap_qpel_pixels_tab[0][i])
c->avg_2tap_qpel_pixels_tab[0][i]= c->avg_h264_qpel_pixels_tab[0][i];
for (i = 0; i < 4; i++) {
for (j = 0; j < 16; j++) {
if(!c->put_2tap_qpel_pixels_tab[i][j])
c->put_2tap_qpel_pixels_tab[i][j] =
c->put_h264_qpel_pixels_tab[i][j];
if(!c->avg_2tap_qpel_pixels_tab[i][j])
c->avg_2tap_qpel_pixels_tab[i][j] =
c->avg_h264_qpel_pixels_tab[i][j];
}
}

ff_init_scantable_permutation(c->idct_permutation,


+ 9
- 1
libavcodec/fft-test.c View File

@@ -23,6 +23,7 @@
* FFT and MDCT tests.
*/

#include "libavutil/cpu.h"
#include "libavutil/mathematics.h"
#include "libavutil/lfg.h"
#include "libavutil/log.h"
@@ -240,6 +241,7 @@ int main(int argc, char **argv)
FFTComplex *tab, *tab1, *tab_ref;
FFTSample *tab2;
int it, i, c;
int cpuflags;
int do_speed = 0;
int err = 1;
enum tf_transform transform = TRANSFORM_FFT;
@@ -258,7 +260,7 @@ int main(int argc, char **argv)

fft_nbits = 9;
for(;;) {
c = getopt(argc, argv, "hsimrdn:f:");
c = getopt(argc, argv, "hsimrdn:f:c:");
if (c == -1)
break;
switch(c) {
@@ -286,6 +288,12 @@ int main(int argc, char **argv)
case 'f':
scale = atof(optarg);
break;
case 'c':
cpuflags = av_parse_cpu_flags(optarg);
if (cpuflags < 0)
return 1;
av_set_cpu_flags_mask(cpuflags);
break;
}
}



+ 1
- 0
libavformat/isom.c View File

@@ -249,6 +249,7 @@ const AVCodecTag ff_codec_movaudio_tags[] = {
{ CODEC_ID_DTS, MKTAG('d', 't', 's', 'h') }, /* DTS-HD audio formats */
{ CODEC_ID_DTS, MKTAG('d', 't', 's', 'l') }, /* DTS-HD Lossless formats */
{ CODEC_ID_DTS, MKTAG('D', 'T', 'S', ' ') }, /* non-standard */
{ CODEC_ID_EAC3, MKTAG('e', 'c', '-', '3') }, /* ETSI TS 102 366 Annex F (only valid in ISOBMFF) */
{ CODEC_ID_DVAUDIO, MKTAG('v', 'd', 'v', 'a') },
{ CODEC_ID_DVAUDIO, MKTAG('d', 'v', 'c', 'a') },
{ CODEC_ID_EAC3, MKTAG('e', 'c', '-', '3') }, /* ETSI TS 102 366 Annex F */


+ 1
- 1
libavformat/utils.c View File

@@ -1985,7 +1985,7 @@ static int has_duration(AVFormatContext *ic)
if (st->duration != AV_NOPTS_VALUE)
return 1;
}
if (ic->duration)
if (ic->duration != AV_NOPTS_VALUE)
return 1;
return 0;
}


+ 1
- 1
libavutil/avutil.h View File

@@ -153,7 +153,7 @@
*/

#define LIBAVUTIL_VERSION_MAJOR 51
#define LIBAVUTIL_VERSION_MINOR 47
#define LIBAVUTIL_VERSION_MINOR 48
#define LIBAVUTIL_VERSION_MICRO 100

#define LIBAVUTIL_VERSION_INT AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \


+ 64
- 0
libavutil/cpu.c View File

@@ -18,6 +18,7 @@

#include "cpu.h"
#include "config.h"
#include "opt.h"

static int flags, checked;

@@ -46,6 +47,69 @@ void av_set_cpu_flags_mask(int mask)
checked = 1;
}

int av_parse_cpu_flags(const char *s)
{
#define CPUFLAG_MMX2 (AV_CPU_FLAG_MMX | AV_CPU_FLAG_MMX2)
#define CPUFLAG_3DNOW (AV_CPU_FLAG_3DNOW | AV_CPU_FLAG_MMX)
#define CPUFLAG_3DNOWEXT (AV_CPU_FLAG_3DNOWEXT | CPUFLAG_3DNOW)
#define CPUFLAG_SSE (AV_CPU_FLAG_SSE | CPUFLAG_MMX2)
#define CPUFLAG_SSE2 (AV_CPU_FLAG_SSE2 | CPUFLAG_SSE)
#define CPUFLAG_SSE2SLOW (AV_CPU_FLAG_SSE2SLOW | CPUFLAG_SSE2)
#define CPUFLAG_SSE3 (AV_CPU_FLAG_SSE3 | CPUFLAG_SSE2)
#define CPUFLAG_SSE3SLOW (AV_CPU_FLAG_SSE3SLOW | CPUFLAG_SSE3)
#define CPUFLAG_SSSE3 (AV_CPU_FLAG_SSSE3 | CPUFLAG_SSE3)
#define CPUFLAG_SSE4 (AV_CPU_FLAG_SSE4 | CPUFLAG_SSSE3)
#define CPUFLAG_SSE42 (AV_CPU_FLAG_SSE42 | CPUFLAG_SSE4)
#define CPUFLAG_AVX (AV_CPU_FLAG_AVX | CPUFLAG_SSE42)
#define CPUFLAG_XOP (AV_CPU_FLAG_XOP | CPUFLAG_AVX)
#define CPUFLAG_FMA4 (AV_CPU_FLAG_FMA4 | CPUFLAG_AVX)
static const AVOption cpuflags_opts[] = {
{ "flags" , NULL, 0, AV_OPT_TYPE_FLAGS, { 0 }, INT64_MIN, INT64_MAX, .unit = "flags" },
#if ARCH_PPC
{ "altivec" , NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_ALTIVEC }, .unit = "flags" },
#elif ARCH_X86
{ "mmx" , NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_MMX }, .unit = "flags" },
{ "mmx2" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_MMX2 }, .unit = "flags" },
{ "sse" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE }, .unit = "flags" },
{ "sse2" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE2 }, .unit = "flags" },
{ "sse2slow", NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE2SLOW }, .unit = "flags" },
{ "sse3" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE3 }, .unit = "flags" },
{ "sse3slow", NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE3SLOW }, .unit = "flags" },
{ "ssse3" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSSE3 }, .unit = "flags" },
{ "atom" , NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_ATOM }, .unit = "flags" },
{ "sse4.1" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE4 }, .unit = "flags" },
{ "sse4.2" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_SSE42 }, .unit = "flags" },
{ "avx" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_AVX }, .unit = "flags" },
{ "xop" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_XOP }, .unit = "flags" },
{ "fma4" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_FMA4 }, .unit = "flags" },
{ "3dnow" , NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_3DNOW }, .unit = "flags" },
{ "3dnowext", NULL, 0, AV_OPT_TYPE_CONST, { CPUFLAG_3DNOWEXT }, .unit = "flags" },
#elif ARCH_ARM
{ "armv5te", NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_ARMV5TE }, .unit = "flags" },
{ "armv6", NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_ARMV6 }, .unit = "flags" },
{ "armv6t2", NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_ARMV6T2 }, .unit = "flags" },
{ "vfp", NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_VFP }, .unit = "flags" },
{ "vfpv3", NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_VFPV3 }, .unit = "flags" },
{ "neon", NULL, 0, AV_OPT_TYPE_CONST, { AV_CPU_FLAG_NEON }, .unit = "flags" },
#endif
{ NULL },
};
static const AVClass class = {
.class_name = "cpuflags",
.item_name = av_default_item_name,
.option = cpuflags_opts,
.version = LIBAVUTIL_VERSION_INT,
};

int flags = 0, ret;
const AVClass *pclass = &class;

if ((ret = av_opt_eval_flags(&pclass, &cpuflags_opts[0], s, &flags)) < 0)
return ret;

return flags & INT_MAX;
}

#ifdef TEST

#undef printf


+ 7
- 0
libavutil/cpu.h View File

@@ -72,6 +72,13 @@ void av_force_cpu_flags(int flags);
*/
attribute_deprecated void av_set_cpu_flags_mask(int mask);

/**
* Parse CPU flags from a string.
*
* @return a combination of AV_CPU_* flags, negative on error.
*/
int av_parse_cpu_flags(const char *s);

/* The following CPU-specific functions shall not be called directly. */
int ff_get_cpu_flags_arm(void);
int ff_get_cpu_flags_ppc(void);


+ 14
- 12
tests/fate/fft.mak View File

@@ -4,36 +4,38 @@ FATE_FFT += fate-fft-$(1) fate-ifft-$(1) \
fate-rdft-$(1) fate-irdft-$(1) \
fate-dct1d-$(1) fate-idct1d-$(1)

fate-fft-$(N): CMD = run libavcodec/fft-test -n$(1)
fate-ifft-$(N): CMD = run libavcodec/fft-test -n$(1) -i
fate-mdct-$(N): CMD = run libavcodec/fft-test -n$(1) -m
fate-imdct-$(N): CMD = run libavcodec/fft-test -n$(1) -m -i
fate-rdft-$(N): CMD = run libavcodec/fft-test -n$(1) -r
fate-irdft-$(N): CMD = run libavcodec/fft-test -n$(1) -r -i
fate-dct1d-$(N): CMD = run libavcodec/fft-test -n$(1) -d
fate-idct1d-$(N): CMD = run libavcodec/fft-test -n$(1) -d -i
fate-fft-$(N): ARGS = -n$(1)
fate-ifft-$(N): ARGS = -n$(1) -i
fate-mdct-$(N): ARGS = -n$(1) -m
fate-imdct-$(N): ARGS = -n$(1) -m -i
fate-rdft-$(N): ARGS = -n$(1) -r
fate-irdft-$(N): ARGS = -n$(1) -r -i
fate-dct1d-$(N): ARGS = -n$(1) -d
fate-idct1d-$(N): ARGS = -n$(1) -d -i
endef

$(foreach N, 4 5 6 7 8 9 10 11 12, $(eval $(call DEF_FFT,$(N))))

fate-fft-test: $(FATE_FFT)
$(FATE_FFT): libavcodec/fft-test$(EXESUF)
$(FATE_FFT): CMD = run libavcodec/fft-test $(CPUFLAGS:%=-c%) $(ARGS)
$(FATE_FFT): REF = /dev/null

define DEF_FFT_FIXED
FATE_FFT_FIXED += fate-fft-fixed-$(1) fate-ifft-fixed-$(1) \
fate-mdct-fixed-$(1) fate-imdct-fixed-$(1)

fate-fft-fixed-$(1): CMD = run libavcodec/fft-fixed-test -n$(1)
fate-ifft-fixed-$(1): CMD = run libavcodec/fft-fixed-test -n$(1) -i
fate-mdct-fixed-$(1): CMD = run libavcodec/fft-fixed-test -n$(1) -m
fate-imdct-fixed-$(1): CMD = run libavcodec/fft-fixed-test -n$(1) -m -i
fate-fft-fixed-$(1): ARGS = -n$(1)
fate-ifft-fixed-$(1): ARGS = -n$(1) -i
fate-mdct-fixed-$(1): ARGS = -n$(1) -m
fate-imdct-fixed-$(1): ARGS = -n$(1) -m -i
endef

$(foreach N, 4 5 6 7 8 9 10 11 12, $(eval $(call DEF_FFT_FIXED,$(N))))

fate-fft-fixed-test: $(FATE_FFT_FIXED)
$(FATE_FFT_FIXED): libavcodec/fft-fixed-test$(EXESUF)
$(FATE_FFT_FIXED): CMD = run libavcodec/fft-fixed-test $(CPUFLAGS:%=-c%) $(ARGS)
$(FATE_FFT_FIXED): REF = /dev/null

FATE_TESTS += $(FATE_FFT) $(FATE_FFT_FIXED)


Write Preview
Loading…
Cancel
Save