generic-library/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge remote-tracking branch 'qatar/master'

Browse Source 
* qatar/master:
  get_bits: remove x86 inline asm in A32 bitstream reader
  doc: Remove outdated information about our issue tracker
  avidec: Factor out the sync fucntionality.
  fate-aac: Expand coverage.
  ac3dsp: add x86-optimized versions of ac3dsp.extract_exponents().
  ac3dsp: simplify extract_exponents() now that it does not need to do clipping.
  ac3enc: clip coefficients after MDCT.
  ac3enc: add int32_t array clipping function to DSPUtil, including x86 versions.
  swscale: for >8bit scaling, read in native bit-depth.
  matroskadec: matroska_read_seek after after EBML_STOP leads to failure.
  doxygen: fix usage of @file directive in libavutil/{dict,file}.h
  doxygen: Help doxygen parser to understand the DECLARE_ALIGNED and offsetof macros

Conflicts:
	doc/issue_tracker.txt
	libavformat/avidec.c
	libavutil/dict.h
	libswscale/swscale.c
	libswscale/utils.c
	tests/ref/lavfi/pixfmts_scale

Merged-by: Michael Niedermayer <michaelni@gmx.at>
This commit is contained in:
Michael Niedermayer 2011-07-02 03:07:06 +02:00
commit 3074f03a07
19 changed files with 465 additions and 190 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
Doxyfile
View File

@ -1160,6 +1160,7 @@ INCLUDE_FILE_PATTERNS =
PREDEFINED = __attribute__(x)="" "RENAME(x)=x ## _TMPL" "DEF(x)=x ## _TMPL" \ PREDEFINED = __attribute__(x)="" "RENAME(x)=x ## _TMPL" "DEF(x)=x ## _TMPL" \
HAVE_AV_CONFIG_H HAVE_MMX HAVE_MMX2 HAVE_AMD3DNOW \ HAVE_AV_CONFIG_H HAVE_MMX HAVE_MMX2 HAVE_AMD3DNOW \
"DECLARE_ALIGNED(a,t,n)=t n" "offsetof(x,y)=0x42" \
# If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then # If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then
# this tag can be used to specify a list of macro names that should be expanded. # this tag can be used to specify a list of macro names that should be expanded.

15
libavcodec/ac3dsp.c
View File

@ -164,21 +164,8 @@ static void ac3_extract_exponents_c(uint8_t *exp, int32_t *coef, int nb_coefs)
int i; int i;
for (i = 0; i < nb_coefs; i++) { for (i = 0; i < nb_coefs; i++) {
int e;
int v = abs(coef[i]); int v = abs(coef[i]);
if (v == 0) exp[i] = v ? 23 - av_log2(v) : 24;
e = 24;
else {
e = 23 - av_log2(v);
if (e >= 24) {
e = 24;
coef[i] = 0;
} else if (e < 0) {
e = 0;
coef[i] = av_clip(coef[i], -16777215, 16777215);
}
}
exp[i] = e;
} }
} }

4
libavcodec/ac3enc.h
View File

@ -50,12 +50,16 @@
#if CONFIG_AC3ENC_FLOAT #if CONFIG_AC3ENC_FLOAT
#define AC3_NAME(x) ff_ac3_float_ ## x #define AC3_NAME(x) ff_ac3_float_ ## x
#define MAC_COEF(d,a,b) ((d)+=(a)*(b)) #define MAC_COEF(d,a,b) ((d)+=(a)*(b))
#define COEF_MIN (-16777215.0/16777216.0)
#define COEF_MAX ( 16777215.0/16777216.0)
typedef float SampleType; typedef float SampleType;
typedef float CoefType; typedef float CoefType;
typedef float CoefSumType; typedef float CoefSumType;
#else #else
#define AC3_NAME(x) ff_ac3_fixed_ ## x #define AC3_NAME(x) ff_ac3_fixed_ ## x
#define MAC_COEF(d,a,b) MAC64(d,a,b) #define MAC_COEF(d,a,b) MAC64(d,a,b)
#define COEF_MIN -16777215
#define COEF_MAX 16777215
typedef int16_t SampleType; typedef int16_t SampleType;
typedef int32_t CoefType; typedef int32_t CoefType;
typedef int64_t CoefSumType; typedef int64_t CoefSumType;

9
libavcodec/ac3enc_fixed.c
View File

@ -104,6 +104,15 @@ static void scale_coefficients(AC3EncodeContext *s)
} }
/**
* Clip MDCT coefficients to allowable range.
*/
static void clip_coefficients(DSPContext *dsp, int32_t *coef, unsigned int len)
{
dsp->vector_clip_int32(coef, coef, COEF_MIN, COEF_MAX, len);
}
static av_cold int ac3_fixed_encode_init(AVCodecContext *avctx) static av_cold int ac3_fixed_encode_init(AVCodecContext *avctx)
{ {
AC3EncodeContext *s = avctx->priv_data; AC3EncodeContext *s = avctx->priv_data;

9
libavcodec/ac3enc_float.c
View File

@ -111,6 +111,15 @@ static void scale_coefficients(AC3EncodeContext *s)
} }
/**
* Clip MDCT coefficients to allowable range.
*/
static void clip_coefficients(DSPContext *dsp, float *coef, unsigned int len)
{
dsp->vector_clipf(coef, coef, COEF_MIN, COEF_MAX, len);
}
#if CONFIG_AC3_ENCODER #if CONFIG_AC3_ENCODER
AVCodec ff_ac3_float_encoder = { AVCodec ff_ac3_float_encoder = {
"ac3_float", "ac3_float",

16
libavcodec/ac3enc_template.c
View File

@ -41,6 +41,8 @@ static void apply_window(DSPContext *dsp, SampleType *output,
static int normalize_samples(AC3EncodeContext *s); static int normalize_samples(AC3EncodeContext *s);
static void clip_coefficients(DSPContext *dsp, CoefType *coef, unsigned int len);
int AC3_NAME(allocate_sample_buffers)(AC3EncodeContext *s) int AC3_NAME(allocate_sample_buffers)(AC3EncodeContext *s)
{ {
@ -171,8 +173,8 @@ static void apply_channel_coupling(AC3EncodeContext *s)
cpl_coef[i] += ch_coef[i]; cpl_coef[i] += ch_coef[i];
} }
/* coefficients must be clipped to +/- 1.0 in order to be encoded */ /* coefficients must be clipped in order to be encoded */
s->dsp.vector_clipf(cpl_coef, cpl_coef, -1.0f, 1.0f, num_cpl_coefs); clip_coefficients(&s->dsp, cpl_coef, num_cpl_coefs);
/* scale coupling coefficients from float to 24-bit fixed-point */ /* scale coupling coefficients from float to 24-bit fixed-point */
s->ac3dsp.float_to_fixed24(&block->fixed_coef[CPL_CH][cpl_start], s->ac3dsp.float_to_fixed24(&block->fixed_coef[CPL_CH][cpl_start],
@ -300,6 +302,7 @@ static void apply_channel_coupling(AC3EncodeContext *s)
if (!block->cpl_in_use || !block->new_cpl_coords) if (!block->cpl_in_use || !block->new_cpl_coords)
continue; continue;
clip_coefficients(&s->dsp, cpl_coords[blk][1], s->fbw_channels * 16);
s->ac3dsp.float_to_fixed24(fixed_cpl_coords[blk][1], s->ac3dsp.float_to_fixed24(fixed_cpl_coords[blk][1],
cpl_coords[blk][1], cpl_coords[blk][1],
s->fbw_channels * 16); s->fbw_channels * 16);
@ -433,7 +436,11 @@ int AC3_NAME(encode_frame)(AVCodecContext *avctx, unsigned char *frame,
apply_mdct(s); apply_mdct(s);
scale_coefficients(s); if (s->fixed_point)
scale_coefficients(s);
clip_coefficients(&s->dsp, s->blocks[0].mdct_coef[1],
AC3_MAX_COEFS * AC3_MAX_BLOCKS * s->channels);
s->cpl_on = s->cpl_enabled; s->cpl_on = s->cpl_enabled;
ff_ac3_compute_coupling_strategy(s); ff_ac3_compute_coupling_strategy(s);
@ -443,6 +450,9 @@ int AC3_NAME(encode_frame)(AVCodecContext *avctx, unsigned char *frame,
compute_rematrixing_strategy(s); compute_rematrixing_strategy(s);
if (!s->fixed_point)
scale_coefficients(s);
ff_ac3_apply_rematrixing(s); ff_ac3_apply_rematrixing(s);
ff_ac3_process_exponents(s); ff_ac3_process_exponents(s);

17
libavcodec/dsputil.c
View File

@ -2664,6 +2664,22 @@ static void apply_window_int16_c(int16_t *output, const int16_t *input,
} }
} }
static void vector_clip_int32_c(int32_t *dst, const int32_t *src, int32_t min,
int32_t max, unsigned int len)
{
do {
*dst++ = av_clip(*src++, min, max);
*dst++ = av_clip(*src++, min, max);
*dst++ = av_clip(*src++, min, max);
*dst++ = av_clip(*src++, min, max);
*dst++ = av_clip(*src++, min, max);
*dst++ = av_clip(*src++, min, max);
*dst++ = av_clip(*src++, min, max);
*dst++ = av_clip(*src++, min, max);
len -= 8;
} while (len > 0);
}
#define W0 2048 #define W0 2048
#define W1 2841 /* 2048*sqrt (2)*cos (1*pi/16) */ #define W1 2841 /* 2048*sqrt (2)*cos (1*pi/16) */
#define W2 2676 /* 2048*sqrt (2)*cos (2*pi/16) */ #define W2 2676 /* 2048*sqrt (2)*cos (2*pi/16) */
@ -3106,6 +3122,7 @@ av_cold void dsputil_init(DSPContext* c, AVCodecContext *avctx)
c->scalarproduct_int16 = scalarproduct_int16_c; c->scalarproduct_int16 = scalarproduct_int16_c;
c->scalarproduct_and_madd_int16 = scalarproduct_and_madd_int16_c; c->scalarproduct_and_madd_int16 = scalarproduct_and_madd_int16_c;
c->apply_window_int16 = apply_window_int16_c; c->apply_window_int16 = apply_window_int16_c;
c->vector_clip_int32 = vector_clip_int32_c;
c->scalarproduct_float = scalarproduct_float_c; c->scalarproduct_float = scalarproduct_float_c;
c->butterflies_float = butterflies_float_c; c->butterflies_float = butterflies_float_c;
c->vector_fmul_scalar = vector_fmul_scalar_c; c->vector_fmul_scalar = vector_fmul_scalar_c;

16
libavcodec/dsputil.h
View File

@ -553,6 +553,22 @@ typedef struct DSPContext {
void (*apply_window_int16)(int16_t *output, const int16_t *input, void (*apply_window_int16)(int16_t *output, const int16_t *input,
const int16_t *window, unsigned int len); const int16_t *window, unsigned int len);
/**
* Clip each element in an array of int32_t to a given minimum and maximum value.
* @param dst destination array
* constraints: 16-byte aligned
* @param src source array
* constraints: 16-byte aligned
* @param min minimum value
* constraints: must in the the range [-(1<<24), 1<<24]
* @param max maximum value
* constraints: must in the the range [-(1<<24), 1<<24]
* @param len number of elements in the array
* constraints: multiple of 32 greater than zero
*/
void (*vector_clip_int32)(int32_t *dst, const int32_t *src, int32_t min,
int32_t max, unsigned int len);
/* rv30 functions */ /* rv30 functions */
qpel_mc_func put_rv30_tpel_pixels_tab[4][16]; qpel_mc_func put_rv30_tpel_pixels_tab[4][16];
qpel_mc_func avg_rv30_tpel_pixels_tab[4][16]; qpel_mc_func avg_rv30_tpel_pixels_tab[4][16];

8
libavcodec/get_bits.h
View File

@ -201,19 +201,11 @@ static inline void skip_bits_long(GetBitContext *s, int n){
} \ } \
} while (0) } while (0)
#if ARCH_X86
# define SKIP_CACHE(name, gb, num) \
__asm__("shldl %2, %1, %0 \n\t" \
"shll %2, %1 \n\t" \
: "+r" (name##_cache0), "+r" (name##_cache1) \
: "Ic" ((uint8_t)(num)))
#else
# define SKIP_CACHE(name, gb, num) do { \ # define SKIP_CACHE(name, gb, num) do { \
name##_cache0 <<= (num); \ name##_cache0 <<= (num); \
name##_cache0 |= NEG_USR32(name##_cache1,num); \ name##_cache0 |= NEG_USR32(name##_cache1,num); \
name##_cache1 <<= (num); \ name##_cache1 <<= (num); \
} while (0) } while (0)
#endif
# define SKIP_COUNTER(name, gb, num) name##_bit_count += (num) # define SKIP_COUNTER(name, gb, num) name##_bit_count += (num)

102
libavcodec/x86/ac3dsp.asm
View File

@ -32,6 +32,11 @@ cextern ac3_bap_bits
pw_bap_mul1: dw 21846, 21846, 0, 32768, 21846, 21846, 0, 32768 pw_bap_mul1: dw 21846, 21846, 0, 32768, 21846, 21846, 0, 32768
pw_bap_mul2: dw 5, 7, 0, 7, 5, 7, 0, 7 pw_bap_mul2: dw 5, 7, 0, 7, 5, 7, 0, 7
; used in ff_ac3_extract_exponents()
pd_1: times 4 dd 1
pd_151: times 4 dd 151
pb_shuf_4dwb: db 0, 4, 8, 12
SECTION .text SECTION .text
;----------------------------------------------------------------------------- ;-----------------------------------------------------------------------------
@ -346,3 +351,100 @@ cglobal ac3_compute_mantissa_size_sse2, 1,2,4, mant_cnt, sum
movd eax, m0 movd eax, m0
add eax, sumd add eax, sumd
RET RET
;------------------------------------------------------------------------------
; void ff_ac3_extract_exponents(uint8_t *exp, int32_t *coef, int nb_coefs)
;------------------------------------------------------------------------------
%macro PABSD_MMX 2 ; src/dst, tmp
pxor %2, %2
pcmpgtd %2, %1
pxor %1, %2
psubd %1, %2
%endmacro
%macro PABSD_SSSE3 1-2 ; src/dst, unused
pabsd %1, %1
%endmacro
%ifdef HAVE_AMD3DNOW
INIT_MMX
cglobal ac3_extract_exponents_3dnow, 3,3,0, exp, coef, len
add expq, lenq
lea coefq, [coefq+4*lenq]
neg lenq
movq m3, [pd_1]
movq m4, [pd_151]
.loop:
movq m0, [coefq+4*lenq ]
movq m1, [coefq+4*lenq+8]
PABSD_MMX m0, m2
PABSD_MMX m1, m2
pslld m0, 1
por m0, m3
pi2fd m2, m0
psrld m2, 23
movq m0, m4
psubd m0, m2
pslld m1, 1
por m1, m3
pi2fd m2, m1
psrld m2, 23
movq m1, m4
psubd m1, m2
packssdw m0, m0
packuswb m0, m0
packssdw m1, m1
packuswb m1, m1
punpcklwd m0, m1
movd [expq+lenq], m0
add lenq, 4
jl .loop
REP_RET
%endif
%macro AC3_EXTRACT_EXPONENTS 1
cglobal ac3_extract_exponents_%1, 3,3,5, exp, coef, len
add expq, lenq
lea coefq, [coefq+4*lenq]
neg lenq
mova m2, [pd_1]
mova m3, [pd_151]
%ifidn %1, ssse3 ;
movd m4, [pb_shuf_4dwb]
%endif
.loop:
; move 4 32-bit coefs to xmm0
mova m0, [coefq+4*lenq]
; absolute value
PABSD m0, m1
; convert to float and extract exponents
pslld m0, 1
por m0, m2
cvtdq2ps m1, m0
psrld m1, 23
mova m0, m3
psubd m0, m1
; move the lowest byte in each of 4 dwords to the low dword
%ifidn %1, ssse3
pshufb m0, m4
%else
packssdw m0, m0
packuswb m0, m0
%endif
movd [expq+lenq], m0
add lenq, 4
jl .loop
REP_RET
%endmacro
%ifdef HAVE_SSE
INIT_XMM
%define PABSD PABSD_MMX
AC3_EXTRACT_EXPONENTS sse2
%ifdef HAVE_SSSE3
%define PABSD PABSD_SSSE3
AC3_EXTRACT_EXPONENTS ssse3
%endif
%endif

9
libavcodec/x86/ac3dsp_mmx.c
View File

@ -44,6 +44,10 @@ extern void ff_float_to_fixed24_sse2 (int32_t *dst, const float *src, unsigned i
extern int ff_ac3_compute_mantissa_size_sse2(uint16_t mant_cnt[6][16]); extern int ff_ac3_compute_mantissa_size_sse2(uint16_t mant_cnt[6][16]);
extern void ff_ac3_extract_exponents_3dnow(uint8_t *exp, int32_t *coef, int nb_coefs);
extern void ff_ac3_extract_exponents_sse2 (uint8_t *exp, int32_t *coef, int nb_coefs);
extern void ff_ac3_extract_exponents_ssse3(uint8_t *exp, int32_t *coef, int nb_coefs);
av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact) av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact)
{ {
int mm_flags = av_get_cpu_flags(); int mm_flags = av_get_cpu_flags();
@ -56,6 +60,7 @@ av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact)
c->ac3_rshift_int32 = ff_ac3_rshift_int32_mmx; c->ac3_rshift_int32 = ff_ac3_rshift_int32_mmx;
} }
if (mm_flags & AV_CPU_FLAG_3DNOW && HAVE_AMD3DNOW) { if (mm_flags & AV_CPU_FLAG_3DNOW && HAVE_AMD3DNOW) {
c->extract_exponents = ff_ac3_extract_exponents_3dnow;
if (!bit_exact) { if (!bit_exact) {
c->float_to_fixed24 = ff_float_to_fixed24_3dnow; c->float_to_fixed24 = ff_float_to_fixed24_3dnow;
} }
@ -72,6 +77,7 @@ av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact)
c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_sse2; c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_sse2;
c->float_to_fixed24 = ff_float_to_fixed24_sse2; c->float_to_fixed24 = ff_float_to_fixed24_sse2;
c->compute_mantissa_size = ff_ac3_compute_mantissa_size_sse2; c->compute_mantissa_size = ff_ac3_compute_mantissa_size_sse2;
c->extract_exponents = ff_ac3_extract_exponents_sse2;
if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) { if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) {
c->ac3_lshift_int16 = ff_ac3_lshift_int16_sse2; c->ac3_lshift_int16 = ff_ac3_lshift_int16_sse2;
c->ac3_rshift_int32 = ff_ac3_rshift_int32_sse2; c->ac3_rshift_int32 = ff_ac3_rshift_int32_sse2;
@ -79,6 +85,9 @@ av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact)
} }
if (mm_flags & AV_CPU_FLAG_SSSE3 && HAVE_SSSE3) { if (mm_flags & AV_CPU_FLAG_SSSE3 && HAVE_SSSE3) {
c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_ssse3; c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_ssse3;
if (!(mm_flags & AV_CPU_FLAG_ATOM)) {
c->extract_exponents = ff_ac3_extract_exponents_ssse3;
}
} }
#endif #endif
} }

23
libavcodec/x86/dsputil_mmx.c
View File

@ -2333,6 +2333,15 @@ int ff_add_hfyu_left_prediction_sse4(uint8_t *dst, const uint8_t *src, int w, i
float ff_scalarproduct_float_sse(const float *v1, const float *v2, int order); float ff_scalarproduct_float_sse(const float *v1, const float *v2, int order);
void ff_vector_clip_int32_mmx (int32_t *dst, const int32_t *src, int32_t min,
int32_t max, unsigned int len);
void ff_vector_clip_int32_sse2 (int32_t *dst, const int32_t *src, int32_t min,
int32_t max, unsigned int len);
void ff_vector_clip_int32_sse2_int(int32_t *dst, const int32_t *src, int32_t min,
int32_t max, unsigned int len);
void ff_vector_clip_int32_sse41 (int32_t *dst, const int32_t *src, int32_t min,
int32_t max, unsigned int len);
void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx) void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
{ {
int mm_flags = av_get_cpu_flags(); int mm_flags = av_get_cpu_flags();
@ -2473,6 +2482,8 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
c->put_rv40_chroma_pixels_tab[0]= ff_put_rv40_chroma_mc8_mmx; c->put_rv40_chroma_pixels_tab[0]= ff_put_rv40_chroma_mc8_mmx;
c->put_rv40_chroma_pixels_tab[1]= ff_put_rv40_chroma_mc4_mmx; c->put_rv40_chroma_pixels_tab[1]= ff_put_rv40_chroma_mc4_mmx;
c->vector_clip_int32 = ff_vector_clip_int32_mmx;
#endif #endif
if (mm_flags & AV_CPU_FLAG_MMX2) { if (mm_flags & AV_CPU_FLAG_MMX2) {
@ -2756,6 +2767,11 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
#if HAVE_YASM #if HAVE_YASM
c->scalarproduct_int16 = ff_scalarproduct_int16_sse2; c->scalarproduct_int16 = ff_scalarproduct_int16_sse2;
c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_sse2; c->scalarproduct_and_madd_int16 = ff_scalarproduct_and_madd_int16_sse2;
if (mm_flags & AV_CPU_FLAG_ATOM) {
c->vector_clip_int32 = ff_vector_clip_int32_sse2_int;
} else {
c->vector_clip_int32 = ff_vector_clip_int32_sse2;
}
if (avctx->flags & CODEC_FLAG_BITEXACT) { if (avctx->flags & CODEC_FLAG_BITEXACT) {
c->apply_window_int16 = ff_apply_window_int16_sse2_ba; c->apply_window_int16 = ff_apply_window_int16_sse2_ba;
} else { } else {
@ -2781,6 +2797,13 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
} }
#endif #endif
} }
if (mm_flags & AV_CPU_FLAG_SSE4 && HAVE_SSE) {
#if HAVE_YASM
c->vector_clip_int32 = ff_vector_clip_int32_sse41;
#endif
}
#if HAVE_AVX && HAVE_YASM #if HAVE_AVX && HAVE_YASM
if (mm_flags & AV_CPU_FLAG_AVX) { if (mm_flags & AV_CPU_FLAG_AVX) {
if (bit_depth == 10) { if (bit_depth == 10) {

115
libavcodec/x86/dsputil_yasm.asm
View File

@ -1048,3 +1048,118 @@ emu_edge sse
%ifdef ARCH_X86_32 %ifdef ARCH_X86_32
emu_edge mmx emu_edge mmx
%endif %endif
;-----------------------------------------------------------------------------
; void ff_vector_clip_int32(int32_t *dst, const int32_t *src, int32_t min,
; int32_t max, unsigned int len)
;-----------------------------------------------------------------------------
%macro PMINSD_MMX 3 ; dst, src, tmp
mova %3, %2
pcmpgtd %3, %1
pxor %1, %2
pand %1, %3
pxor %1, %2
%endmacro
%macro PMAXSD_MMX 3 ; dst, src, tmp
mova %3, %1
pcmpgtd %3, %2
pand %1, %3
pandn %3, %2
por %1, %3
%endmacro
%macro CLIPD_MMX 3-4 ; src/dst, min, max, tmp
PMINSD_MMX %1, %3, %4
PMAXSD_MMX %1, %2, %4
%endmacro
%macro CLIPD_SSE2 3-4 ; src/dst, min (float), max (float), unused
cvtdq2ps %1, %1
minps %1, %3
maxps %1, %2
cvtps2dq %1, %1
%endmacro
%macro CLIPD_SSE41 3-4 ; src/dst, min, max, unused
pminsd %1, %3
pmaxsd %1, %2
%endmacro
%macro SPLATD_MMX 1
punpckldq %1, %1
%endmacro
%macro SPLATD_SSE2 1
pshufd %1, %1, 0
%endmacro
%macro VECTOR_CLIP_INT32 4
cglobal vector_clip_int32_%1, 5,5,%2, dst, src, min, max, len
%ifidn %1, sse2
cvtsi2ss m4, minm
cvtsi2ss m5, maxm
%else
movd m4, minm
movd m5, maxm
%endif
SPLATD m4
SPLATD m5
.loop:
%assign %%i 1
%rep %3
mova m0, [srcq+mmsize*0*%%i]
mova m1, [srcq+mmsize*1*%%i]
mova m2, [srcq+mmsize*2*%%i]
mova m3, [srcq+mmsize*3*%%i]
%if %4
mova m7, [srcq+mmsize*4*%%i]
mova m8, [srcq+mmsize*5*%%i]
mova m9, [srcq+mmsize*6*%%i]
mova m10, [srcq+mmsize*7*%%i]
%endif
CLIPD m0, m4, m5, m6
CLIPD m1, m4, m5, m6
CLIPD m2, m4, m5, m6
CLIPD m3, m4, m5, m6
%if %4
CLIPD m7, m4, m5, m6
CLIPD m8, m4, m5, m6
CLIPD m9, m4, m5, m6
CLIPD m10, m4, m5, m6
%endif
mova [dstq+mmsize*0*%%i], m0
mova [dstq+mmsize*1*%%i], m1
mova [dstq+mmsize*2*%%i], m2
mova [dstq+mmsize*3*%%i], m3
%if %4
mova [dstq+mmsize*4*%%i], m7
mova [dstq+mmsize*5*%%i], m8
mova [dstq+mmsize*6*%%i], m9
mova [dstq+mmsize*7*%%i], m10
%endif
%assign %%i %%i+1
%endrep
add srcq, mmsize*4*(%3+%4)
add dstq, mmsize*4*(%3+%4)
sub lend, mmsize*(%3+%4)
jg .loop
REP_RET
%endmacro
INIT_MMX
%define SPLATD SPLATD_MMX
%define CLIPD CLIPD_MMX
VECTOR_CLIP_INT32 mmx, 0, 1, 0
INIT_XMM
%define SPLATD SPLATD_SSE2
VECTOR_CLIP_INT32 sse2_int, 6, 1, 0
%define CLIPD CLIPD_SSE2
VECTOR_CLIP_INT32 sse2, 6, 2, 0
%define CLIPD CLIPD_SSE41
%ifdef m8
VECTOR_CLIP_INT32 sse41, 11, 1, 1
%else
VECTOR_CLIP_INT32 sse41, 6, 1, 0
%endif

244
libavformat/avidec.c
View File

@ -861,13 +861,137 @@ static int get_stream_idx(int *d){
} }
} }
static int avi_read_packet(AVFormatContext *s, AVPacket *pkt) static int avi_sync(AVFormatContext *s)
{ {
AVIContext *avi = s->priv_data; AVIContext *avi = s->priv_data;
AVIOContext *pb = s->pb; AVIOContext *pb = s->pb;
int n, d[8]; int n, d[8];
unsigned int size; unsigned int size;
int64_t i, sync; int64_t i, sync;
start_sync:
memset(d, -1, sizeof(int)*8);
for(i=sync=avio_tell(pb); !url_feof(pb); i++) {
int j;
for(j=0; j<7; j++)
d[j]= d[j+1];
d[7]= avio_r8(pb);
size= d[4] + (d[5]<<8) + (d[6]<<16) + (d[7]<<24);
n= get_stream_idx(d+2);
//av_log(s, AV_LOG_DEBUG, "%X %X %X %X %X %X %X %X %"PRId64" %d %d\n", d[0], d[1], d[2], d[3], d[4], d[5], d[6], d[7], i, size, n);
if(i + (uint64_t)size > avi->fsize || d[0]<0)
continue;
//parse ix##
if( (d[0] == 'i' && d[1] == 'x' && n < s->nb_streams)
//parse JUNK
||(d[0] == 'J' && d[1] == 'U' && d[2] == 'N' && d[3] == 'K')
||(d[0] == 'i' && d[1] == 'd' && d[2] == 'x' && d[3] == '1')){
avio_skip(pb, size);
//av_log(s, AV_LOG_DEBUG, "SKIP\n");
goto start_sync;
}
//parse stray LIST
if(d[0] == 'L' && d[1] == 'I' && d[2] == 'S' && d[3] == 'T'){
avio_skip(pb, 4);
goto start_sync;
}
n= get_stream_idx(d);
if(!((i-avi->last_pkt_pos)&1) && get_stream_idx(d+1) < s->nb_streams)
continue;
//detect ##ix chunk and skip
if(d[2] == 'i' && d[3] == 'x' && n < s->nb_streams){
avio_skip(pb, size);
goto start_sync;
}
//parse ##dc/##wb
if(n < s->nb_streams){
AVStream *st;
AVIStream *ast;
st = s->streams[n];
ast = st->priv_data;
if(s->nb_streams>=2){
AVStream *st1 = s->streams[1];
AVIStream *ast1= st1->priv_data;
//workaround for broken small-file-bug402.avi
if( d[2] == 'w' && d[3] == 'b'
&& n==0
&& st ->codec->codec_type == AVMEDIA_TYPE_VIDEO
&& st1->codec->codec_type == AVMEDIA_TYPE_AUDIO
&& ast->prefix == 'd'*256+'c'
&& (d[2]*256+d[3] == ast1->prefix || !ast1->prefix_count)
){
n=1;
st = st1;
ast = ast1;
av_log(s, AV_LOG_WARNING, "Invalid stream + prefix combination, assuming audio.\n");
}
}
if( (st->discard >= AVDISCARD_DEFAULT && size==0)
/*|| (st->discard >= AVDISCARD_NONKEY && !(pkt->flags & AV_PKT_FLAG_KEY))*/ //FIXME needs a little reordering
|| st->discard >= AVDISCARD_ALL){
ast->frame_offset += get_duration(ast, size);
avio_skip(pb, size);
goto start_sync;
}
if (d[2] == 'p' && d[3] == 'c' && size<=4*256+4) {
int k = avio_r8(pb);
int last = (k + avio_r8(pb) - 1) & 0xFF;
avio_rl16(pb); //flags
for (; k <= last; k++)
ast->pal[k] = avio_rb32(pb)>>8;// b + (g << 8) + (r << 16);
ast->has_pal= 1;
goto start_sync;
} else if( ((ast->prefix_count<5 || sync+9 > i) && d[2]<128 && d[3]<128) ||
d[2]*256+d[3] == ast->prefix /*||
(d[2] == 'd' && d[3] == 'c') ||
(d[2] == 'w' && d[3] == 'b')*/) {
//av_log(s, AV_LOG_DEBUG, "OK\n");
if(d[2]*256+d[3] == ast->prefix)
ast->prefix_count++;
else{
ast->prefix= d[2]*256+d[3];
ast->prefix_count= 0;
}
avi->stream_index= n;
ast->packet_size= size + 8;
ast->remaining= size;
if(size || !ast->sample_size){
uint64_t pos= avio_tell(pb) - 8;
if(!st->index_entries || !st->nb_index_entries || st->index_entries[st->nb_index_entries - 1].pos < pos){
av_add_index_entry(st, pos, ast->frame_offset, size, 0, AVINDEX_KEYFRAME);
}
}
return 0;
}
}
}
return AVERROR_EOF;
}
static int avi_read_packet(AVFormatContext *s, AVPacket *pkt)
{
AVIContext *avi = s->priv_data;
AVIOContext *pb = s->pb;
int err;
void* dstr; void* dstr;
if (CONFIG_DV_DEMUXER && avi->dv_demux) { if (CONFIG_DV_DEMUXER && avi->dv_demux) {
@ -1041,121 +1165,9 @@ resync:
return size; return size;
} }
memset(d, -1, sizeof(int)*8); if ((err = avi_sync(s)) < 0)
for(i=sync=avio_tell(pb); !url_feof(pb); i++) { return err;
int j; goto resync;
for(j=0; j<7; j++)
d[j]= d[j+1];
d[7]= avio_r8(pb);
size= d[4] + (d[5]<<8) + (d[6]<<16) + (d[7]<<24);
n= get_stream_idx(d+2);
//av_log(s, AV_LOG_DEBUG, "%X %X %X %X %X %X %X %X %"PRId64" %d %d\n", d[0], d[1], d[2], d[3], d[4], d[5], d[6], d[7], i, size, n);
if(i + (uint64_t)size > avi->fsize || d[0]<0)
continue;
//parse ix##
if( (d[0] == 'i' && d[1] == 'x' && n < s->nb_streams)
//parse JUNK
||(d[0] == 'J' && d[1] == 'U' && d[2] == 'N' && d[3] == 'K')
||(d[0] == 'i' && d[1] == 'd' && d[2] == 'x' && d[3] == '1')){
avio_skip(pb, size);
//av_log(s, AV_LOG_DEBUG, "SKIP\n");
goto resync;
}
//parse stray LIST
if(d[0] == 'L' && d[1] == 'I' && d[2] == 'S' && d[3] == 'T'){
avio_skip(pb, 4);
goto resync;
}
n= get_stream_idx(d);
if(!((i-avi->last_pkt_pos)&1) && get_stream_idx(d+1) < s->nb_streams)
continue;
//detect ##ix chunk and skip
if(d[2] == 'i' && d[3] == 'x' && n < s->nb_streams){
avio_skip(pb, size);
goto resync;
}
//parse ##dc/##wb
if(n < s->nb_streams){
AVStream *st;
AVIStream *ast;
st = s->streams[n];
ast = st->priv_data;
if(s->nb_streams>=2){
AVStream *st1 = s->streams[1];
AVIStream *ast1= st1->priv_data;
//workaround for broken small-file-bug402.avi
if( d[2] == 'w' && d[3] == 'b'
&& n==0
&& st ->codec->codec_type == AVMEDIA_TYPE_VIDEO
&& st1->codec->codec_type == AVMEDIA_TYPE_AUDIO
&& ast->prefix == 'd'*256+'c'
&& (d[2]*256+d[3] == ast1->prefix || !ast1->prefix_count)
){
n=1;
st = st1;
ast = ast1;
av_log(s, AV_LOG_WARNING, "Invalid stream + prefix combination, assuming audio.\n");
}
}
if( (st->discard >= AVDISCARD_DEFAULT && size==0)
/*|| (st->discard >= AVDISCARD_NONKEY && !(pkt->flags & AV_PKT_FLAG_KEY))*/ //FIXME needs a little reordering
|| st->discard >= AVDISCARD_ALL){
ast->frame_offset += get_duration(ast, size);
avio_skip(pb, size);
goto resync;
}
if (d[2] == 'p' && d[3] == 'c' && size<=4*256+4) {
int k = avio_r8(pb);
int last = (k + avio_r8(pb) - 1) & 0xFF;
avio_rl16(pb); //flags
for (; k <= last; k++)
ast->pal[k] = avio_rb32(pb)>>8;// b + (g << 8) + (r << 16);
ast->has_pal= 1;
goto resync;
} else if( ((ast->prefix_count<5 || sync+9 > i) && d[2]<128 && d[3]<128) ||
d[2]*256+d[3] == ast->prefix /*||
(d[2] == 'd' && d[3] == 'c') ||
(d[2] == 'w' && d[3] == 'b')*/) {
//av_log(s, AV_LOG_DEBUG, "OK\n");
if(d[2]*256+d[3] == ast->prefix)
ast->prefix_count++;
else{
ast->prefix= d[2]*256+d[3];
ast->prefix_count= 0;
}
avi->stream_index= n;
ast->packet_size= size + 8;
ast->remaining= size;
if(size || !ast->sample_size){
uint64_t pos= avio_tell(pb) - 8;
if(!st->index_entries || !st->nb_index_entries || st->index_entries[st->nb_index_entries - 1].pos < pos){
av_add_index_entry(st, pos, ast->frame_offset, size, 0, AVINDEX_KEYFRAME);
}
}
goto resync;
}
}
}
return AVERROR_EOF;
} }
/* XXX: We make the implicit supposition that the positions are sorted /* XXX: We make the implicit supposition that the positions are sorted

2
libavformat/matroskadec.c
View File

@ -1960,6 +1960,7 @@ static int matroska_read_seek(AVFormatContext *s, int stream_index,
if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) { if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
avio_seek(s->pb, st->index_entries[st->nb_index_entries-1].pos, SEEK_SET); avio_seek(s->pb, st->index_entries[st->nb_index_entries-1].pos, SEEK_SET);
matroska->current_id = 0;
while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) { while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
matroska_clear_queue(matroska); matroska_clear_queue(matroska);
if (matroska_parse_cluster(matroska) < 0) if (matroska_parse_cluster(matroska) < 0)
@ -1988,6 +1989,7 @@ static int matroska_read_seek(AVFormatContext *s, int stream_index,
} }
avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET); avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET);
matroska->current_id = 0;
matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY); matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
matroska->skip_to_timecode = st->index_entries[index].timestamp; matroska->skip_to_timecode = st->index_entries[index].timestamp;
matroska->done = 0; matroska->done = 0;

3
libavutil/dict.h
View File

@ -18,7 +18,8 @@
*/ */
/** /**
* @file Public dictionary API. * @file
* Public dictionary API.
* @deprecated * @deprecated
* AVDictionary is provided for compatibility with libav. It is both in * AVDictionary is provided for compatibility with libav. It is both in
* implementation as well as API inefficient. It does not scale and is * implementation as well as API inefficient. It does not scale and is

3
libavutil/file.h
View File

@ -22,7 +22,8 @@
#include "avutil.h" #include "avutil.h"
/** /**
* @file misc file utilities * @file
* Misc file utilities.
*/ */
/** /**

47
libswscale/swscale.c
View File

@ -1783,53 +1783,6 @@ static void nv21ToUV_c(uint8_t *dstU, uint8_t *dstV,
#define input_pixel(pos) (isBE(origin) ? AV_RB16(pos) : AV_RL16(pos)) #define input_pixel(pos) (isBE(origin) ? AV_RB16(pos) : AV_RL16(pos))
// FIXME Maybe dither instead.
static av_always_inline void
yuv9_OR_10ToUV_c_template(uint8_t *dstU, uint8_t *dstV,
const uint8_t *_srcU, const uint8_t *_srcV,
int width, enum PixelFormat origin, int depth)
{
int i;
const uint16_t *srcU = (const uint16_t *) _srcU;
const uint16_t *srcV = (const uint16_t *) _srcV;
for (i = 0; i < width; i++) {
dstU[i] = input_pixel(&srcU[i]) >> (depth - 8);
dstV[i] = input_pixel(&srcV[i]) >> (depth - 8);
}
}
static av_always_inline void
yuv9_or_10ToY_c_template(uint8_t *dstY, const uint8_t *_srcY,
int width, enum PixelFormat origin, int depth)
{
int i;
const uint16_t *srcY = (const uint16_t*)_srcY;
for (i = 0; i < width; i++)
dstY[i] = input_pixel(&srcY[i]) >> (depth - 8);
}
#undef input_pixel
#define YUV_NBPS(depth, BE_LE, origin) \
static void BE_LE ## depth ## ToUV_c(uint8_t *dstU, uint8_t *dstV, \
const uint8_t *srcU, const uint8_t *srcV, \
int width, uint32_t *unused) \
{ \
yuv9_OR_10ToUV_c_template(dstU, dstV, srcU, srcV, width, origin, depth); \
} \
static void BE_LE ## depth ## ToY_c(uint8_t *dstY, const uint8_t *srcY, \
int width, uint32_t *unused) \
{ \
yuv9_or_10ToY_c_template(dstY, srcY, width, origin, depth); \
}
YUV_NBPS( 9, LE, PIX_FMT_YUV420P9LE);
YUV_NBPS( 9, BE, PIX_FMT_YUV420P9BE);
YUV_NBPS(10, LE, PIX_FMT_YUV420P10LE);
YUV_NBPS(10, BE, PIX_FMT_YUV420P10BE);
static void bgr24ToY_c(int16_t *dst, const uint8_t *src, static void bgr24ToY_c(int16_t *dst, const uint8_t *src,
int width, uint32_t *unused) int width, uint32_t *unused)
{ {

12
tests/fate/aac.mak
View File

@ -2,10 +2,22 @@ FATE_AAC += fate-aac-al04_44
fate-aac-al04_44: CMD = pcm -i $(SAMPLES)/aac/al04_44.mp4 fate-aac-al04_44: CMD = pcm -i $(SAMPLES)/aac/al04_44.mp4
fate-aac-al04_44: REF = $(SAMPLES)/aac/al04_44.s16 fate-aac-al04_44: REF = $(SAMPLES)/aac/al04_44.s16
FATE_AAC += fate-aac-al05_44
fate-aac-al05_44: CMD = pcm -i $(SAMPLES)/aac/al05_44.mp4
fate-aac-al05_44: REF = $(SAMPLES)/aac/al05_44.s16
FATE_AAC += fate-aac-al06_44
fate-aac-al06_44: CMD = pcm -i $(SAMPLES)/aac/al06_44.mp4
fate-aac-al06_44: REF = $(SAMPLES)/aac/al06_44.s16
FATE_AAC += fate-aac-al07_96 FATE_AAC += fate-aac-al07_96
fate-aac-al07_96: CMD = pcm -i $(SAMPLES)/aac/al07_96.mp4 fate-aac-al07_96: CMD = pcm -i $(SAMPLES)/aac/al07_96.mp4
fate-aac-al07_96: REF = $(SAMPLES)/aac/al07_96.s16 fate-aac-al07_96: REF = $(SAMPLES)/aac/al07_96.s16
FATE_AAC += fate-aac-al17_44
fate-aac-al17_44: CMD = pcm -i $(SAMPLES)/aac/al17_44.mp4
fate-aac-al17_44: REF = $(SAMPLES)/aac/al17_44.s16
FATE_AAC += fate-aac-am00_88 FATE_AAC += fate-aac-am00_88
fate-aac-am00_88: CMD = pcm -i $(SAMPLES)/aac/am00_88.mp4 fate-aac-am00_88: CMD = pcm -i $(SAMPLES)/aac/am00_88.mp4
fate-aac-am00_88: REF = $(SAMPLES)/aac/am00_88.s16 fate-aac-am00_88: REF = $(SAMPLES)/aac/am00_88.s16