Merge remote-tracking branch 'qatar/master'

* qatar/master:
  png: add missing #if HAVE_SSSE3 around function pointer assignment.
  imdct36: mark SSE functions as using all 16 XMM registers.
  png: move DSP functions to their own DSP context.
  sunrast: Add a sample request for TIFF, IFF, and Experimental Rastfile formats.
  sunrast: Cosmetics
  sunrast: Remove if (unsigned int < 0) check.
  sunrast: Replace magic number by a macro.

Conflicts:
	libavcodec/dsputil.c
	libavcodec/dsputil.h
	libavcodec/pngdec.c
	libavcodec/sunrast.c
	libavcodec/x86/Makefile
	libavcodec/x86/dsputil_mmx.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
This commit is contained in:
Michael Niedermayer 2012-01-30 05:20:58 +01:00
commit e1492151fb
10 changed files with 192 additions and 136 deletions

View File

@ -322,7 +322,7 @@ OBJS-$(CONFIG_PGMYUV_DECODER) += pnmdec.o pnm.o
OBJS-$(CONFIG_PGMYUV_ENCODER) += pnmenc.o pnm.o
OBJS-$(CONFIG_PGSSUB_DECODER) += pgssubdec.o
OBJS-$(CONFIG_PICTOR_DECODER) += pictordec.o cga_data.o
OBJS-$(CONFIG_PNG_DECODER) += png.o pngdec.o
OBJS-$(CONFIG_PNG_DECODER) += png.o pngdec.o pngdsp.o
OBJS-$(CONFIG_PNG_ENCODER) += png.o pngenc.o
OBJS-$(CONFIG_PPM_DECODER) += pnmdec.o pnm.o
OBJS-$(CONFIG_PPM_ENCODER) += pnmenc.o pnm.o

View File

@ -23,9 +23,6 @@
#define AVCODEC_PNG_H
#include <stdint.h>
#include <zlib.h>
#include "avcodec.h"
#define PNG_COLOR_MASK_PALETTE 1
#define PNG_COLOR_MASK_COLOR 2
@ -72,41 +69,4 @@ int ff_png_pass_row_size(int pass, int bits_per_pixel, int width);
void ff_add_png_paeth_prediction(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp);
typedef struct PNGDecContext {
const uint8_t *bytestream;
const uint8_t *bytestream_start;
const uint8_t *bytestream_end;
AVFrame picture1, picture2;
AVFrame *current_picture, *last_picture;
int state;
int width, height;
int bit_depth;
int color_type;
int compression_type;
int interlace_type;
int filter_type;
int channels;
int bits_per_pixel;
int bpp;
uint8_t *image_buf;
int image_linesize;
uint32_t palette[256];
uint8_t *crow_buf;
uint8_t *last_row;
uint8_t *tmp_row;
int pass;
int crow_size; /* compressed row size (include filter type) */
int row_size; /* decompressed row size */
int pass_row_size; /* decompress row size of the current pass */
int y;
z_stream zstream;
void (*add_bytes_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w);
void (*add_paeth_prediction)(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp);
} PNGDecContext;
void ff_png_init_mmx(PNGDecContext *s);
#endif /* AVCODEC_PNG_H */

View File

@ -25,6 +25,7 @@
#include "avcodec.h"
#include "bytestream.h"
#include "png.h"
#include "pngdsp.h"
/* TODO:
* - add 16 bit depth support
@ -32,6 +33,42 @@
#include <zlib.h>
//#define DEBUG
typedef struct PNGDecContext {
PNGDSPContext dsp;
const uint8_t *bytestream;
const uint8_t *bytestream_start;
const uint8_t *bytestream_end;
AVFrame picture1, picture2;
AVFrame *current_picture, *last_picture;
int state;
int width, height;
int bit_depth;
int color_type;
int compression_type;
int interlace_type;
int filter_type;
int channels;
int bits_per_pixel;
int bpp;
uint8_t *image_buf;
int image_linesize;
uint32_t palette[256];
uint8_t *crow_buf;
uint8_t *last_row;
uint8_t *tmp_row;
int pass;
int crow_size; /* compressed row size (include filter type) */
int row_size; /* decompressed row size */
int pass_row_size; /* decompress row size of the current pass */
int y;
z_stream zstream;
} PNGDecContext;
/* Mask to determine which y pixels can be written in a pass */
static const uint8_t png_pass_dsp_ymask[NB_PASSES] = {
0xff, 0xff, 0x0f, 0xff, 0x33, 0xff, 0x55,
@ -114,23 +151,7 @@ static void png_put_interlaced_row(uint8_t *dst, int width,
}
}
// 0x7f7f7f7f or 0x7f7f7f7f7f7f7f7f or whatever, depending on the cpu's native arithmetic size
#define pb_7f (~0UL/255 * 0x7f)
#define pb_80 (~0UL/255 * 0x80)
static void add_bytes_l2_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w)
{
long i;
for(i=0; i<=w-sizeof(long); i+=sizeof(long)){
long a = *(long*)(src1+i);
long b = *(long*)(src2+i);
*(long*)(dst+i) = ((a&pb_7f) + (b&pb_7f)) ^ ((a^b)&pb_80);
}
for(; i<w; i++)
dst[i] = src1[i]+src2[i];
}
static void add_paeth_prediction_c(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp)
void ff_add_png_paeth_prediction(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp)
{
int i;
for(i = 0; i < w; i++) {
@ -187,7 +208,7 @@ static void add_paeth_prediction_c(uint8_t *dst, uint8_t *src, uint8_t *top, int
}
/* NOTE: 'dst' can be equal to 'last' */
static void png_filter_row(PNGDecContext *s, uint8_t *dst, int filter_type,
static void png_filter_row(PNGDSPContext *dsp, uint8_t *dst, int filter_type,
uint8_t *src, uint8_t *last, int size, int bpp)
{
int i, p, r, g, b, a;
@ -213,7 +234,7 @@ static void png_filter_row(PNGDecContext *s, uint8_t *dst, int filter_type,
}
break;
case PNG_FILTER_VALUE_UP:
s->add_bytes_l2(dst, src, last, size);
dsp->add_bytes_l2(dst, src, last, size);
break;
case PNG_FILTER_VALUE_AVG:
for(i = 0; i < bpp; i++) {
@ -231,10 +252,10 @@ static void png_filter_row(PNGDecContext *s, uint8_t *dst, int filter_type,
if(bpp > 2 && size > 4) {
// would write off the end of the array if we let it process the last pixel with bpp=3
int w = bpp==4 ? size : size-3;
s->add_paeth_prediction(dst+i, src+i, last+i, w-i, bpp);
dsp->add_paeth_prediction(dst+i, src+i, last+i, w-i, bpp);
i = w;
}
add_paeth_prediction_c(dst+i, src+i, last+i, size-i, bpp);
ff_add_png_paeth_prediction(dst+i, src+i, last+i, size-i, bpp);
break;
}
}
@ -704,14 +725,7 @@ static av_cold int png_dec_init(AVCodecContext *avctx)
avcodec_get_frame_defaults(&s->picture1);
avcodec_get_frame_defaults(&s->picture2);
#if HAVE_MMX
ff_png_init_mmx(s);
#endif
if (!s->add_paeth_prediction)
s->add_paeth_prediction = add_paeth_prediction_c;
if (!s->add_bytes_l2)
s->add_bytes_l2 = add_bytes_l2_c;
ff_pngdsp_init(&s->dsp);
return 0;
}

48
libavcodec/pngdsp.c Normal file
View File

@ -0,0 +1,48 @@
/*
* PNG image format
* Copyright (c) 2008 Loren Merrit <lorenm@u.washington.edu>
*
* 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
*/
#include "libavutil/common.h"
#include "png.h"
#include "pngdsp.h"
// 0x7f7f7f7f or 0x7f7f7f7f7f7f7f7f or whatever, depending on the cpu's native arithmetic size
#define pb_7f (~0UL/255 * 0x7f)
#define pb_80 (~0UL/255 * 0x80)
static void add_bytes_l2_c(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w)
{
long i;
for (i = 0; i <= w - sizeof(long); i += sizeof(long)) {
long a = *(long *)(src1 + i);
long b = *(long *)(src2 + i);
*(long *)(dst + i) = ((a & pb_7f) + (b & pb_7f)) ^ ((a ^ b) & pb_80);
}
for (; i < w; i++)
dst[i] = src1[i] + src2[i];
}
void ff_pngdsp_init(PNGDSPContext *dsp)
{
dsp->add_bytes_l2 = add_bytes_l2_c;
dsp->add_paeth_prediction = ff_add_png_paeth_prediction;
if (HAVE_MMX) ff_pngdsp_init_x86(dsp);
}

40
libavcodec/pngdsp.h Normal file
View File

@ -0,0 +1,40 @@
/*
* PNG image format
* Copyright (c) 2008 Loren Merrit <lorenm@u.washington.edu>
*
* 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_PNGDSP_H
#define AVCODEC_PNGDSP_H
#include <stdint.h>
typedef struct PNGDSPContext {
void (*add_bytes_l2)(uint8_t *dst /* align 16 */,
uint8_t *src1 /* align 16 */,
uint8_t *src2 /* align 16 */, int w);
/* this might write to dst[w] */
void (*add_paeth_prediction)(uint8_t *dst, uint8_t *src,
uint8_t *top, int w, int bpp);
} PNGDSPContext;
void ff_pngdsp_init(PNGDSPContext *dsp);
void ff_pngdsp_init_x86(PNGDSPContext *dsp);
#endif /* AVCDODEC_PNGDSP_H */

View File

@ -23,6 +23,8 @@
#include "libavutil/imgutils.h"
#include "avcodec.h"
#define RAS_MAGIC 0x59a66a95
/* The Old and Standard format types indicate that the image data is
* uncompressed. There is no difference between the two formats. */
#define RT_OLD 0
@ -55,18 +57,18 @@ static av_cold int sunrast_init(AVCodecContext *avctx) {
SUNRASTContext *s = avctx->priv_data;
avcodec_get_frame_defaults(&s->picture);
avctx->coded_frame= &s->picture;
avctx->coded_frame = &s->picture;
return 0;
}
static int sunrast_decode_frame(AVCodecContext *avctx, void *data,
int *data_size, AVPacket *avpkt) {
const uint8_t *buf = avpkt->data;
const uint8_t *buf_end = avpkt->data + avpkt->size;
const uint8_t *buf = avpkt->data;
const uint8_t *buf_end = avpkt->data + avpkt->size;
SUNRASTContext * const s = avctx->priv_data;
AVFrame *picture = data;
AVFrame * const p = &s->picture;
AVFrame *picture = data;
AVFrame * const p = &s->picture;
unsigned int w, h, depth, type, maptype, maplength, stride, x, y, len, alen;
uint8_t *ptr, *ptr2 = NULL;
const uint8_t *bufstart = buf;
@ -74,22 +76,22 @@ static int sunrast_decode_frame(AVCodecContext *avctx, void *data,
if (avpkt->size < 32)
return AVERROR_INVALIDDATA;
if (AV_RB32(buf) != 0x59a66a95) {
if (AV_RB32(buf) != RAS_MAGIC) {
av_log(avctx, AV_LOG_ERROR, "this is not sunras encoded data\n");
return -1;
}
w = AV_RB32(buf+4);
h = AV_RB32(buf+8);
depth = AV_RB32(buf+12);
type = AV_RB32(buf+20);
maptype = AV_RB32(buf+24);
maplength = AV_RB32(buf+28);
w = AV_RB32(buf + 4);
h = AV_RB32(buf + 8);
depth = AV_RB32(buf + 12);
type = AV_RB32(buf + 20);
maptype = AV_RB32(buf + 24);
maplength = AV_RB32(buf + 28);
buf += 32;
if (type == RT_EXPERIMENTAL) {
av_log(avctx, AV_LOG_ERROR, "unsupported (compression) type\n");
return -1;
av_log_ask_for_sample(avctx, "unsupported (compression) type\n");
return AVERROR_PATCHWELCOME;
}
if (type > RT_FORMAT_IFF) {
av_log(avctx, AV_LOG_ERROR, "invalid (compression) type\n");
@ -161,7 +163,7 @@ static int sunrast_decode_frame(AVCodecContext *avctx, void *data,
}
ptr = p->data[1];
for (x=0; x<len; x++, ptr+=4)
for (x = 0; x < len; x++, ptr += 4)
*(uint32_t *)ptr = (0xFF<<24) + (buf[x]<<16) + (buf[len+x]<<8) + buf[len+len+x];
}
@ -179,11 +181,11 @@ static int sunrast_decode_frame(AVCodecContext *avctx, void *data,
/* scanlines are aligned on 16 bit boundaries */
len = (depth * w + 7) >> 3;
alen = len + (len&1);
alen = len + (len & 1);
if (type == RT_BYTE_ENCODED) {
int value, run;
uint8_t *end = ptr + h*stride;
uint8_t *end = ptr + h * stride;
x = 0;
while (ptr != end && buf < buf_end) {
@ -208,7 +210,7 @@ static int sunrast_decode_frame(AVCodecContext *avctx, void *data,
}
}
} else {
for (y=0; y<h; y++) {
for (y = 0; y < h; y++) {
if (buf_end - buf < len)
break;
memcpy(ptr, buf, len);
@ -241,7 +243,7 @@ static int sunrast_decode_frame(AVCodecContext *avctx, void *data,
av_freep(&ptr_free);
}
*picture = s->picture;
*picture = s->picture;
*data_size = sizeof(AVFrame);
return buf - bufstart;

View File

@ -42,13 +42,13 @@ MMX-OBJS-$(CONFIG_CAVS_DECODER) += x86/cavsdsp_mmx.o
MMX-OBJS-$(CONFIG_DNXHD_ENCODER) += x86/dnxhd_mmx.o
MMX-OBJS-$(CONFIG_MPEGAUDIODSP) += x86/mpegaudiodec_mmx.o
YASM-OBJS-$(CONFIG_MPEGAUDIODSP) += x86/imdct36_sse.o
MMX-OBJS-$(CONFIG_PNG_DECODER) += x86/png_mmx.o
MMX-OBJS-$(CONFIG_ENCODERS) += x86/dsputilenc_mmx.o
YASM-OBJS-$(CONFIG_ENCODERS) += x86/dsputilenc_yasm.o
MMX-OBJS-$(CONFIG_GPL) += x86/idct_mmx.o
MMX-OBJS-$(CONFIG_LPC) += x86/lpc_mmx.o
YASM-OBJS-$(CONFIG_PRORES_LGPL_DECODER) += x86/proresdsp.o
MMX-OBJS-$(CONFIG_PRORES_LGPL_DECODER) += x86/proresdsp-init.o
MMX-OBJS-$(CONFIG_PNG_DECODER) += x86/pngdsp-init.o
YASM-OBJS-$(CONFIG_PRORES_DECODER) += x86/proresdsp.o
MMX-OBJS-$(CONFIG_PRORES_DECODER) += x86/proresdsp-init.o
MMX-OBJS-$(CONFIG_DWT) += x86/snowdsp_mmx.o \

View File

@ -2671,7 +2671,6 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
if (HAVE_AMD3DNOW && (mm_flags & AV_CPU_FLAG_3DNOW))
c->add_hfyu_median_prediction = add_hfyu_median_prediction_cmov;
#endif
} else if (HAVE_AMD3DNOW && (mm_flags & AV_CPU_FLAG_3DNOW)) {
c->prefetch = prefetch_3dnow;

View File

@ -393,7 +393,7 @@ INIT_XMM sse
%endif
%macro DEFINE_FOUR_IMDCT 0
cglobal four_imdct36_float, 5,5,8, out, buf, in, win, tmp
cglobal four_imdct36_float, 5,5,16, out, buf, in, win, tmp
movlps m0, [inq+64]
movhps m0, [inq+64 + 72]
movlps m3, [inq+64 + 2*72]

View File

@ -1,6 +1,6 @@
/*
* MMX optimized PNG utils
* Copyright (c) 2008 Loren Merritt
* x86 PNG optimizations.
* Copyright (c) 2008 Loren Merrit <lorenm@u.washington.edu>
*
* This file is part of FFmpeg.
*
@ -17,46 +17,18 @@
* 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
*
*/
#include "libavutil/cpu.h"
#include "libavutil/x86_cpu.h"
#include "libavcodec/dsputil.h"
#include "libavcodec/png.h"
#include "libavcodec/pngdsp.h"
#include "dsputil_mmx.h"
//#undef NDEBUG
//#include <assert.h>
static void add_bytes_l2_mmx(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w)
{
x86_reg i=0;
__asm__ volatile(
"jmp 2f \n\t"
"1: \n\t"
"movq (%2, %0), %%mm0 \n\t"
"movq 8(%2, %0), %%mm1 \n\t"
"paddb (%3, %0), %%mm0 \n\t"
"paddb 8(%3, %0), %%mm1 \n\t"
"movq %%mm0, (%1, %0) \n\t"
"movq %%mm1, 8(%1, %0) \n\t"
"add $16, %0 \n\t"
"2: \n\t"
"cmp %4, %0 \n\t"
" js 1b \n\t"
: "+r" (i)
: "r"(dst), "r"(src1), "r"(src2), "r"((x86_reg)w-15)
);
for(; i<w; i++)
dst[i] = src1[i] + src2[i];
}
#define PAETH(cpu, abs3)\
static void add_paeth_prediction_##cpu(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp)\
static void add_png_paeth_prediction_##cpu(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp)\
{\
x86_reg i, end;\
if(bpp>4) add_paeth_prediction_##cpu(dst+bpp/2, src+bpp/2, top+bpp/2, w-bpp/2, -bpp);\
if(bpp>4) add_png_paeth_prediction_##cpu(dst+bpp/2, src+bpp/2, top+bpp/2, w-bpp/2, -bpp);\
if(bpp<0) bpp=-bpp;\
i= -bpp;\
end = w-3;\
@ -128,16 +100,37 @@ PAETH(mmx2, ABS3_MMX2)
PAETH(ssse3, ABS3_SSSE3)
#endif
void ff_png_init_mmx(PNGDecContext *s)
static void add_bytes_l2_mmx(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w)
{
int mm_flags = av_get_cpu_flags();
if (mm_flags & AV_CPU_FLAG_MMX2) {
s->add_bytes_l2 = add_bytes_l2_mmx;
s->add_paeth_prediction = add_paeth_prediction_mmx2;
#if HAVE_SSSE3
if (mm_flags & AV_CPU_FLAG_SSSE3)
s->add_paeth_prediction = add_paeth_prediction_ssse3;
#endif
}
x86_reg i=0;
__asm__ volatile(
"jmp 2f \n\t"
"1: \n\t"
"movq (%2, %0), %%mm0 \n\t"
"movq 8(%2, %0), %%mm1 \n\t"
"paddb (%3, %0), %%mm0 \n\t"
"paddb 8(%3, %0), %%mm1 \n\t"
"movq %%mm0, (%1, %0) \n\t"
"movq %%mm1, 8(%1, %0) \n\t"
"add $16, %0 \n\t"
"2: \n\t"
"cmp %4, %0 \n\t"
" js 1b \n\t"
: "+r" (i)
: "r"(dst), "r"(src1), "r"(src2), "r"((x86_reg) w - 15)
);
for (; i < w; i++)
dst[i] = src1[i] + src2[i];
}
void ff_pngdsp_init_x86(PNGDSPContext *dsp)
{
int flags = av_get_cpu_flags();
if (flags & AV_CPU_FLAG_MMX)
dsp->add_bytes_l2 = add_bytes_l2_mmx;
if (flags & AV_CPU_FLAG_MMX2)
dsp->add_paeth_prediction = add_png_paeth_prediction_mmx2;
if (HAVE_SSSE3 && flags & AV_CPU_FLAG_SSSE3)
dsp->add_paeth_prediction = add_png_paeth_prediction_ssse3;
}