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

Merge remote branch 'internal/upstream' into HEAD

Browse Source 
Conflicts:
	vp8/common/alloccommon.c
	vp8/common/onyxc_int.h
	vp8/vp8_cx_iface.c
	vpxenc.c
This commit is contained in:
John Koleszar
2010-11-04 21:50:37 -04:00
parent 6804199073 507eb4b577
commit 362f763cfe
226 changed files with 16995 additions and 9616 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
.gitignore vendored
View File

@@ -1,19 +1,16 @@
*~
*.a
*.asm.s
*.d
*.o
*-old
*-new
*.mk
*.asm
TAGS
.bins
.libs
.deps
/.cflags-new
/.cflags-old
*~
/*-*.mk
/*.asm
/*.doxy
/.bins
/.deps
/.docs
/.install-*
/.libs
/Makefile
/config.err
/config.mk
@@ -23,14 +20,11 @@ TAGS
/decode_with_drops
/decode_with_drops.c
/decode_with_drops.dox
/docs-generic-gnu.mk
/docs/
/doxyfile
/error_resilient
/error_resilient.c
/error_resilient.dox
/examples-generic-gnu.mk
/examples.doxy
/force_keyframe
/force_keyframe.c
/force_keyframe.dox
@@ -38,8 +32,7 @@ TAGS
/ivfdec.dox
/ivfenc
/ivfenc.dox
/libs-generic-gnu.mk
/libs.doxy
/obj_int_extract
/postproc
/postproc.c
/postproc.dox
@@ -66,3 +59,4 @@ TAGS
/vpx_config.c
/vpx_config.h
/vpx_version.h
TAGS

2
AUTHORS
View File

@@ -1,6 +1,7 @@
# This file is automatically generated from the git commit history
# by tools/gen_authors.sh.
Aaron Watry <awatry@gmail.com>
Adrian Grange <agrange@google.com>
Alex Converse <alex.converse@gmail.com>
Andres Mejia <mcitadel@gmail.com>
@@ -20,6 +21,7 @@ Justin Clift <justin@salasaga.org>
Justin Lebar <justin.lebar@gmail.com>
Luca Barbato <lu_zero@gentoo.org>
Makoto Kato <makoto.kt@gmail.com>
Martin Ettl <ettl.martin78@googlemail.com>
Michael Kohler <michaelkohler@live.com>
Paul Wilkins <paulwilkins@google.com>
Pavol Rusnak <stick@gk2.sk>

108
CHANGELOG
View File

@@ -1,3 +1,111 @@
2010-10-28 v0.9.5 "Aylesbury"
Our first named release, focused on a faster decoder, and a better encoder.
- Upgrading:
This release incorporates backwards-incompatible changes to the
ivfenc and ivfdec tools. These tools are now called vpxenc and vpxdec.
vpxdec
* the -q (quiet) option has been removed, and replaced with
-v (verbose). the output is quiet by default. Use -v to see
the version number of the binary.
* The default behavior is now to write output to a single file
instead of individual frames. The -y option has been removed.
Y4M output is the default.
* For raw I420/YV12 output instead of Y4M, the --i420 or --yv12
options must be specified.
$ ivfdec -o OUTPUT INPUT
$ vpxdec --i420 -o OUTPUT INPUT
* If an output file is not specified, the default is to write
Y4M to stdout. This makes piping more natural.
$ ivfdec -y -o - INPUT | ...
$ vpxdec INPUT | ...
* The output file has additional flexibility for formatting the
filename. It supports escape characters for constructing a
filename from the width, height, and sequence number. This
replaces the -p option. To get the equivalent:
$ ivfdec -p frame INPUT
$ vpxdec --i420 -o frame-%wx%h-%4.i420 INPUT
vpxenc
* The output file must be specified with -o, rather than as the
last argument.
$ ivfenc <options> INPUT OUTPUT
$ vpxenc <options> -o OUTPUT INPUT
* The output defaults to webm. To get IVF output, use the --ivf
option.
$ ivfenc <options> INPUT OUTPUT.ivf
$ vpxenc <options> -o OUTPUT.ivf --ivf INPUT
- Enhancements:
ivfenc and ivfdec have been renamed to vpxenc, vpxdec.
vpxdec supports .webm input
vpxdec writes .y4m by default
vpxenc writes .webm output by default
vpxenc --psnr now shows the average/overall PSNR at the end
ARM platforms now support runtime cpu detection
vpxdec visualizations added for motion vectors, block modes, references
vpxdec now silent by default
vpxdec --progress shows frame-by-frame timing information
vpxenc supports the distinction between --fps and --timebase
NASM is now a supported assembler
configure: enable PIC for shared libs by default
configure: add --enable-small
configure: support for ppc32-linux-gcc
configure: support for sparc-solaris-gcc
- Bugs:
Improve handling of invalid frames
Fix valgrind errors in the NEON loop filters.
Fix loopfilter delta zero transitions
Fix valgrind errors in vp8_sixtap_predict8x4_armv6().
Build fixes for darwin-icc
- Speed:
20-40% (average 28%) improvement in libvpx decoder speed,
including:
Rewrite vp8_short_walsh4x4_sse2()
Optimizations on the loopfilters.
Miscellaneous improvements for Atom
Add 4-tap version of 2nd-pass ARMv6 MC filter.
Improved multithread utilization
Better instruction choices on x86
reorder data to use wider instructions
Update NEON wide idcts
Make block access to frame buffer sequential
Improved subset block search
Bilinear subpixel optimizations for ssse3.
Decrease memory footprint
Encoder speed improvements (percentage gain not measured):
Skip unnecessary search of identical frames
Add SSE2 subtract functions
Improve bounds checking in vp8_diamond_search_sadx4()
Added vp8_fast_quantize_b_sse2
- Quality:
Over 7% overall PSNR improvement (6.3% SSIM) in "best" quality
encoding mode, and up to 60% improvement on very noisy, still
or slow moving source video
Motion compensated temporal filter for Alt-Ref Noise Reduction
Improved use of trellis quantization on 2nd order Y blocks
Tune effect of motion on KF/GF boost in two pass
Allow coefficient optimization for good quality speed 0.
Improved control of active min quantizer for two pass.
Enable ARFs for non-lagged compress
2010-09-02 v0.9.2
- Enhancements:
Disable frame dropping by default

2
README
View File

@@ -89,7 +89,7 @@ COMPILING THE APPLICATIONS/LIBRARIES:
toolchain, the following command could be used (note, POSIX SH syntax, adapt
to your shell as necessary):
$ CROSS=mipsel-linux-uclibc- ../libvpx/src/configure
$ CROSS=mipsel-linux-uclibc- ../libvpx/configure
In addition, the executables to be invoked can be overridden by specifying the
environment variables: CC, AR, LD, AS, STRIP, NM. Additional flags can be

8
args.c
View File

@@ -120,9 +120,13 @@ void arg_show_usage(FILE *fp, const struct arg_def *const *defs)
char *long_val = def->has_val ? "=<arg>" : "";
if (def->short_name && def->long_name)
snprintf(option_text, 37, "-%s%s, --%s%s",
def->short_name, short_val,
{
char *comma = def->has_val ? "," : ", ";
snprintf(option_text, 37, "-%s%s%s --%s%6s",
def->short_name, short_val, comma,
def->long_name, long_val);
}
else if (def->short_name)
snprintf(option_text, 37, "-%s%s",
def->short_name, short_val);

2
build/make/Makefile
View File

@@ -65,7 +65,7 @@ endif
BUILD_ROOT?=.
VPATH=$(SRC_PATH_BARE)
CFLAGS+=-I$(BUILD_PFX)$(BUILD_ROOT) -I$(SRC_PATH)
ASFLAGS+=-I$(BUILD_PFX)$(BUILD_ROOT) -I$(SRC_PATH)
ASFLAGS+=-I$(BUILD_PFX)$(BUILD_ROOT)/ -I$(SRC_PATH)/
DIST_DIR?=dist
HOSTCC?=gcc
TGT_ISA:=$(word 1, $(subst -, ,$(TOOLCHAIN)))

57
build/make/configure.sh
View File

@@ -255,9 +255,10 @@ TMP_H="${TMPDIRx}/vpx-conf-$$-${RANDOM}.h"
TMP_C="${TMPDIRx}/vpx-conf-$$-${RANDOM}.c"
TMP_O="${TMPDIRx}/vpx-conf-$$-${RANDOM}.o"
TMP_X="${TMPDIRx}/vpx-conf-$$-${RANDOM}.x"
TMP_ASM="${TMPDIRx}/vpx-conf-$$-${RANDOM}.asm"
clean_temp_files() {
rm -f ${TMP_C} ${TMP_H} ${TMP_O} ${TMP_X}
rm -f ${TMP_C} ${TMP_H} ${TMP_O} ${TMP_X} ${TMP_ASM}
}
#
@@ -322,6 +323,21 @@ check_add_ldflags() {
add_ldflags "$@"
}
check_asm_align() {
log check_asm_align "$@"
cat >${TMP_ASM} <<EOF
section .rodata
align 16
EOF
log_file ${TMP_ASM}
check_cmd ${AS} ${ASFLAGS} -o ${TMP_O} ${TMP_ASM}
readelf -WS ${TMP_O} >${TMP_X}
log_file ${TMP_X}
if ! grep -q '\.rodata .* 16$' ${TMP_X}; then
die "${AS} ${ASFLAGS} does not support section alignment (nasm <=2.08?)"
fi
}
write_common_config_banner() {
echo '# This file automatically generated by configure. Do not edit!' > config.mk
echo "TOOLCHAIN := ${toolchain}" >> config.mk
@@ -440,13 +456,18 @@ process_common_cmdline() {
disable builtin_libc
alt_libc="${optval}"
;;
--as=*)
[ "${optval}" = yasm -o "${optval}" = nasm -o "${optval}" = auto ] \
|| die "Must be yasm, nasm or auto: ${optval}"
alt_as="${optval}"
;;
--prefix=*)
prefix="${optval}"
;;
--libdir=*)
libdir="${optval}"
;;
--libc|--prefix|--libdir)
--libc|--as|--prefix|--libdir)
die "Option ${opt} requires argument"
;;
--help|-h) show_help
@@ -511,6 +532,9 @@ process_common_toolchain() {
*powerpc*)
tgt_isa=ppc32
;;
*sparc*)
tgt_isa=sparc
;;
esac
# detect tgt_os
@@ -530,6 +554,9 @@ process_common_toolchain() {
*linux*|*bsd*)
tgt_os=linux
;;
*solaris2.10)
tgt_os=solaris
;;
esac
if [ -n "$tgt_isa" ] && [ -n "$tgt_os" ]; then
@@ -581,6 +608,13 @@ process_common_toolchain() {
;;
esac
# Handle Solaris variants. Solaris 10 needs -lposix4
case ${toolchain} in
*-solaris-*)
add_extralibs -lposix4
;;
esac
# Process ARM architecture variants
case ${toolchain} in
arm*|iwmmxt*)
@@ -790,6 +824,7 @@ process_common_toolchain() {
soft_enable sse2
soft_enable sse3
soft_enable ssse3
soft_enable sse4_1
case ${tgt_os} in
win*)
@@ -802,6 +837,7 @@ process_common_toolchain() {
;;
esac
AS="${alt_as:-${AS:-auto}}"
case ${tgt_cc} in
icc*)
CC=${CC:-icc}
@@ -830,7 +866,16 @@ process_common_toolchain() {
;;
esac
AS=yasm
case "${AS}" in
auto|"")
which nasm >/dev/null 2>&1 && AS=nasm
which yasm >/dev/null 2>&1 && AS=yasm
[ "${AS}" = auto -o -z "${AS}" ] \
&& die "Neither yasm nor nasm have been found"
;;
esac
log_echo " using $AS"
[ "${AS##*/}" = nasm ] && add_asflags -Ox
AS_SFX=.asm
case ${tgt_os} in
win*)
@@ -839,7 +884,9 @@ process_common_toolchain() {
;;
linux*|solaris*)
add_asflags -f elf${bits}
enabled debug && add_asflags -g dwarf2
enabled debug && [ "${AS}" = yasm ] && add_asflags -g dwarf2
enabled debug && [ "${AS}" = nasm ] && add_asflags -g
[ "${AS##*/}" = nasm ] && check_asm_align
;;
darwin*)
add_asflags -f macho${bits}
@@ -852,7 +899,7 @@ process_common_toolchain() {
# enabled icc && ! enabled pic && add_cflags -fno-pic -mdynamic-no-pic
enabled icc && ! enabled pic && add_cflags -fno-pic
;;
*) log "Warning: Unknown os $tgt_os while setting up yasm flags"
*) log "Warning: Unknown os $tgt_os while setting up $AS flags"
;;
esac
;;

4
configure vendored
View File

@@ -23,6 +23,7 @@ Advanced options:
${toggle_libs} don't build libraries
${toggle_examples} don't build examples
--libc=PATH path to alternate libc
--as={yasm|nasm|auto} use specified assembler [auto, yasm preferred]
${toggle_fast_unaligned} don't use unaligned accesses, even when
supported by hardware [auto]
${toggle_codec_srcs} in/exclude codec library source code
@@ -100,6 +101,7 @@ all_platforms="${all_platforms} ppc32-linux-gcc"
all_platforms="${all_platforms} ppc64-darwin8-gcc"
all_platforms="${all_platforms} ppc64-darwin9-gcc"
all_platforms="${all_platforms} ppc64-linux-gcc"
all_platforms="${all_platforms} sparc-solaris-gcc"
all_platforms="${all_platforms} x86-darwin8-gcc"
all_platforms="${all_platforms} x86-darwin8-icc"
all_platforms="${all_platforms} x86-darwin9-gcc"
@@ -197,6 +199,7 @@ ARCH_EXT_LIST="
sse2
sse3
ssse3
sse4_1
altivec
"
@@ -271,6 +274,7 @@ CMDLINE_SELECT="
libs
examples
libc
as
fast_unaligned
codec_srcs
debug_libs

47
examples.mk
View File

@@ -12,19 +12,40 @@
# List of examples to build. UTILS are files that are taken from the source
# tree directly, and GEN_EXAMPLES are files that are created from the
# examples folder.
UTILS-$(CONFIG_DECODERS) += ivfdec.c
ivfdec.SRCS += md5_utils.c md5_utils.h
ivfdec.SRCS += vpx_ports/vpx_timer.h
ivfdec.SRCS += vpx/vpx_integer.h
ivfdec.SRCS += args.c args.h vpx_ports/config.h
ivfdec.GUID = BA5FE66F-38DD-E034-F542-B1578C5FB950
ivfdec.DESCRIPTION = Full featured decoder
UTILS-$(CONFIG_ENCODERS) += ivfenc.c
ivfenc.SRCS += args.c args.h y4minput.c y4minput.h
ivfenc.SRCS += vpx_ports/config.h vpx_ports/mem_ops.h
ivfenc.SRCS += vpx_ports/mem_ops_aligned.h
ivfenc.GUID = 548DEC74-7A15-4B2B-AFC3-AA102E7C25C1
ivfenc.DESCRIPTION = Full featured encoder
UTILS-$(CONFIG_DECODERS) += vpxdec.c
vpxdec.SRCS += md5_utils.c md5_utils.h
vpxdec.SRCS += vpx_ports/vpx_timer.h
vpxdec.SRCS += vpx/vpx_integer.h
vpxdec.SRCS += args.c args.h vpx_ports/config.h
vpxdec.SRCS += tools_common.c tools_common.h
vpxdec.SRCS += nestegg/halloc/halloc.h
vpxdec.SRCS += nestegg/halloc/src/align.h
vpxdec.SRCS += nestegg/halloc/src/halloc.c
vpxdec.SRCS += nestegg/halloc/src/hlist.h
vpxdec.SRCS += nestegg/halloc/src/macros.h
vpxdec.SRCS += nestegg/include/nestegg/nestegg.h
vpxdec.SRCS += nestegg/src/nestegg.c
vpxdec.GUID = BA5FE66F-38DD-E034-F542-B1578C5FB950
vpxdec.DESCRIPTION = Full featured decoder
UTILS-$(CONFIG_ENCODERS) += vpxenc.c
vpxenc.SRCS += args.c args.h y4minput.c y4minput.h
vpxenc.SRCS += tools_common.c tools_common.h
vpxenc.SRCS += vpx_ports/config.h vpx_ports/mem_ops.h
vpxenc.SRCS += vpx_ports/mem_ops_aligned.h
vpxenc.SRCS += libmkv/EbmlIDs.h
vpxenc.SRCS += libmkv/EbmlWriter.c
vpxenc.SRCS += libmkv/EbmlWriter.h
vpxenc.GUID = 548DEC74-7A15-4B2B-AFC3-AA102E7C25C1
vpxenc.DESCRIPTION = Full featured encoder
# Clean up old ivfenc, ivfdec binaries.
ifeq ($(CONFIG_MSVS),yes)
CLEAN-OBJS += $(foreach p,$(VS_PLATFORMS),$(p)/Release/ivfenc.exe)
CLEAN-OBJS += $(foreach p,$(VS_PLATFORMS),$(p)/Release/ivfdec.exe)
else
CLEAN-OBJS += ivfenc{.c.o,.c.d,.dox,.exe,}
CLEAN-OBJS += ivfdec{.c.o,.c.d,.dox,.exe,}
endif
# XMA example disabled for now, not used in VP8
#UTILS-$(CONFIG_DECODERS) += example_xma.c

4
examples/decoder_tmpl.c
View File

@@ -61,8 +61,8 @@ int main(int argc, char **argv) {
die("Failed to open %s for writing", argv[2]);
/* Read file header */
fread(file_hdr, 1, IVF_FILE_HDR_SZ, infile);
if(!(file_hdr[0]=='D' && file_hdr[1]=='K' && file_hdr[2]=='I'
if(!(fread(file_hdr, 1, IVF_FILE_HDR_SZ, infile) == IVF_FILE_HDR_SZ
&& file_hdr[0]=='D' && file_hdr[1]=='K' && file_hdr[2]=='I'
&& file_hdr[3]=='F'))
die("%s is not an IVF file.", argv[1]);

2
examples/decoder_tmpl.txt
View File

@@ -48,7 +48,7 @@ for(plane=0; plane < 3; plane++) {
unsigned char *buf =img->planes[plane];
for(y=0; y<img->d_h >> (plane?1:0); y++) {
fwrite(buf, 1, img->d_w >> (plane?1:0), outfile);
if(fwrite(buf, 1, img->d_w >> (plane?1:0), outfile));
buf += img->stride[plane];
}
}

4
examples/encoder_tmpl.c
View File

@@ -85,7 +85,7 @@ static void write_ivf_file_header(FILE *outfile,
mem_put_le32(header+24, frame_cnt); /* length */
mem_put_le32(header+28, 0); /* unused */
fwrite(header, 1, 32, outfile);
if(fwrite(header, 1, 32, outfile));
}
@@ -103,7 +103,7 @@ static void write_ivf_frame_header(FILE *outfile,
mem_put_le32(header+4, pts&0xFFFFFFFF);
mem_put_le32(header+8, pts >> 32);
fwrite(header, 1, 12, outfile);
if(fwrite(header, 1, 12, outfile));
}
int main(int argc, char **argv) {

4
examples/encoder_tmpl.txt
View File

@@ -61,8 +61,8 @@ if(vpx_codec_encode(&codec, frame_avail? &raw : NULL, frame_cnt,
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_FRAME
case VPX_CODEC_CX_FRAME_PKT:
write_ivf_frame_header(outfile, pkt);
fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
outfile);
if(fwrite(pkt->data.frame.buf, 1, pkt->data.frame.sz,
outfile));
break;
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ PROCESS_FRAME

640
ivfdec.c
View File

@@ -1,640 +0,0 @@
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
/* This is a simple program that reads ivf files and decodes them
* using the new interface. Decoded frames are output as YV12 raw.
*/
#include <stdio.h>
#include <stdlib.h>
#include <stdarg.h>
#include <string.h>
#define VPX_CODEC_DISABLE_COMPAT 1
#include "vpx_config.h"
#include "vpx/vpx_decoder.h"
#include "vpx_ports/vpx_timer.h"
#if CONFIG_VP8_DECODER
#include "vpx/vp8dx.h"
#endif
#if CONFIG_MD5
#include "md5_utils.h"
#endif
static const char *exec_name;
static const struct
{
char const *name;
const vpx_codec_iface_t *iface;
unsigned int fourcc;
unsigned int fourcc_mask;
} ifaces[] =
{
#if CONFIG_VP8_DECODER
{"vp8", &vpx_codec_vp8_dx_algo, 0x00385056, 0x00FFFFFF},
#endif
};
#include "args.h"
static const arg_def_t codecarg = ARG_DEF(NULL, "codec", 1,
"Codec to use");
static const arg_def_t prefixarg = ARG_DEF("p", "prefix", 1,
"Prefix to use when saving frames");
static const arg_def_t use_yv12 = ARG_DEF(NULL, "yv12", 0,
"Output file is YV12 ");
static const arg_def_t use_i420 = ARG_DEF(NULL, "i420", 0,
"Output file is I420 (default)");
static const arg_def_t flipuvarg = ARG_DEF(NULL, "flipuv", 0,
"Synonym for --yv12");
static const arg_def_t noblitarg = ARG_DEF(NULL, "noblit", 0,
"Don't process the decoded frames");
static const arg_def_t progressarg = ARG_DEF(NULL, "progress", 0,
"Show progress after each frame decodes");
static const arg_def_t limitarg = ARG_DEF(NULL, "limit", 1,
"Stop decoding after n frames");
static const arg_def_t postprocarg = ARG_DEF(NULL, "postproc", 0,
"Postprocess decoded frames");
static const arg_def_t summaryarg = ARG_DEF(NULL, "summary", 0,
"Show timing summary");
static const arg_def_t outputfile = ARG_DEF("o", "output", 1,
"Output raw yv12 file instead of images");
static const arg_def_t usey4marg = ARG_DEF("y", "y4m", 0,
"Output file is YUV4MPEG2");
static const arg_def_t threadsarg = ARG_DEF("t", "threads", 1,
"Max threads to use");
static const arg_def_t quietarg = ARG_DEF("q", "quiet", 0,
"Suppress version string");
#if CONFIG_MD5
static const arg_def_t md5arg = ARG_DEF(NULL, "md5", 0,
"Compute the MD5 sum of the decoded frame");
#endif
static const arg_def_t *all_args[] =
{
&codecarg, &prefixarg, &use_yv12, &use_i420, &flipuvarg, &noblitarg,
&progressarg, &limitarg, &postprocarg, &summaryarg, &outputfile,
&usey4marg, &threadsarg, &quietarg,
#if CONFIG_MD5
&md5arg,
#endif
NULL
};
#if CONFIG_VP8_DECODER
static const arg_def_t addnoise_level = ARG_DEF(NULL, "noise-level", 1,
"Enable VP8 postproc add noise");
static const arg_def_t deblock = ARG_DEF(NULL, "deblock", 0,
"Enable VP8 deblocking");
static const arg_def_t demacroblock_level = ARG_DEF(NULL, "demacroblock-level", 1,
"Enable VP8 demacroblocking, w/ level");
static const arg_def_t pp_debug_info = ARG_DEF(NULL, "pp-debug-info", 1,
"Enable VP8 visible debug info");
static const arg_def_t *vp8_pp_args[] =
{
&addnoise_level, &deblock, &demacroblock_level, &pp_debug_info,
NULL
};
#endif
static void usage_exit()
{
int i;
fprintf(stderr, "Usage: %s <options> filename\n\n"
"Options:\n", exec_name);
arg_show_usage(stderr, all_args);
#if CONFIG_VP8_DECODER
fprintf(stderr, "\nvp8 Postprocessing Options:\n");
arg_show_usage(stderr, vp8_pp_args);
#endif
fprintf(stderr, "\nIncluded decoders:\n\n");
for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++)
fprintf(stderr, " %-6s - %s\n",
ifaces[i].name,
vpx_codec_iface_name(ifaces[i].iface));
exit(EXIT_FAILURE);
}
void die(const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
vfprintf(stderr, fmt, ap);
fprintf(stderr, "\n");
usage_exit();
}
static unsigned int mem_get_le16(const void *vmem)
{
unsigned int val;
const unsigned char *mem = (const unsigned char *)vmem;
val = mem[1] << 8;
val |= mem[0];
return val;
}
static unsigned int mem_get_le32(const void *vmem)
{
unsigned int val;
const unsigned char *mem = (const unsigned char *)vmem;
val = mem[3] << 24;
val |= mem[2] << 16;
val |= mem[1] << 8;
val |= mem[0];
return val;
}
#define IVF_FRAME_HDR_SZ (sizeof(uint32_t) + sizeof(uint64_t))
#define RAW_FRAME_HDR_SZ (sizeof(uint32_t))
static int read_frame(FILE *infile,
uint8_t **buf,
uint32_t *buf_sz,
uint32_t *buf_alloc_sz,
int is_ivf)
{
char raw_hdr[IVF_FRAME_HDR_SZ];
uint32_t new_buf_sz;
/* For both the raw and ivf formats, the frame size is the first 4 bytes
* of the frame header. We just need to special case on the header
* size.
*/
if (fread(raw_hdr, is_ivf ? IVF_FRAME_HDR_SZ : RAW_FRAME_HDR_SZ, 1,
infile) != 1)
{
if (!feof(infile))
fprintf(stderr, "Failed to read frame size\n");
new_buf_sz = 0;
}
else
{
new_buf_sz = mem_get_le32(raw_hdr);
if (new_buf_sz > 256 * 1024 * 1024)
{
fprintf(stderr, "Error: Read invalid frame size (%u)\n",
new_buf_sz);
new_buf_sz = 0;
}
if (!is_ivf && new_buf_sz > 256 * 1024)
fprintf(stderr, "Warning: Read invalid frame size (%u)"
" - not a raw file?\n", new_buf_sz);
if (new_buf_sz > *buf_alloc_sz)
{
uint8_t *new_buf = realloc(*buf, 2 * new_buf_sz);
if (new_buf)
{
*buf = new_buf;
*buf_alloc_sz = 2 * new_buf_sz;
}
else
{
fprintf(stderr, "Failed to allocate compressed data buffer\n");
new_buf_sz = 0;
}
}
}
*buf_sz = new_buf_sz;
if (*buf_sz)
{
if (fread(*buf, 1, *buf_sz, infile) != *buf_sz)
{
fprintf(stderr, "Failed to read full frame\n");
return 1;
}
return 0;
}
return 1;
}
void *out_open(const char *out_fn, int do_md5)
{
void *out = NULL;
if (do_md5)
{
#if CONFIG_MD5
MD5Context *md5_ctx = out = malloc(sizeof(MD5Context));
(void)out_fn;
MD5Init(md5_ctx);
#endif
}
else
{
FILE *outfile = out = strcmp("-", out_fn) ? fopen(out_fn, "wb") : stdout;
if (!outfile)
{
fprintf(stderr, "Failed to output file");
exit(EXIT_FAILURE);
}
}
return out;
}
void out_put(void *out, const uint8_t *buf, unsigned int len, int do_md5)
{
if (do_md5)
{
#if CONFIG_MD5
MD5Update(out, buf, len);
#endif
}
else
{
fwrite(buf, 1, len, out);
}
}
void out_close(void *out, const char *out_fn, int do_md5)
{
if (do_md5)
{
#if CONFIG_MD5
uint8_t md5[16];
int i;
MD5Final(md5, out);
free(out);
for (i = 0; i < 16; i++)
printf("%02x", md5[i]);
printf(" %s\n", out_fn);
#endif
}
else
{
fclose(out);
}
}
unsigned int file_is_ivf(FILE *infile,
unsigned int *fourcc,
unsigned int *width,
unsigned int *height,
unsigned int *timebase_num,
unsigned int *timebase_den)
{
char raw_hdr[32];
int is_ivf = 0;
if (fread(raw_hdr, 1, 32, infile) == 32)
{
if (raw_hdr[0] == 'D' && raw_hdr[1] == 'K'
&& raw_hdr[2] == 'I' && raw_hdr[3] == 'F')
{
is_ivf = 1;
if (mem_get_le16(raw_hdr + 4) != 0)
fprintf(stderr, "Error: Unrecognized IVF version! This file may not"
" decode properly.");
*fourcc = mem_get_le32(raw_hdr + 8);
*width = mem_get_le16(raw_hdr + 12);
*height = mem_get_le16(raw_hdr + 14);
*timebase_den = mem_get_le32(raw_hdr + 16);
*timebase_num = mem_get_le32(raw_hdr + 20);
}
}
if (!is_ivf)
rewind(infile);
return is_ivf;
}
int main(int argc, const char **argv_)
{
vpx_codec_ctx_t decoder;
char *prefix = NULL, *fn = NULL;
int i;
uint8_t *buf = NULL;
uint32_t buf_sz = 0, buf_alloc_sz = 0;
FILE *infile;
int frame_in = 0, frame_out = 0, flipuv = 0, noblit = 0, do_md5 = 0, progress = 0;
int stop_after = 0, postproc = 0, summary = 0, quiet = 0;
vpx_codec_iface_t *iface = NULL;
unsigned int is_ivf, fourcc;
unsigned long dx_time = 0;
struct arg arg;
char **argv, **argi, **argj;
const char *fn2 = 0;
int use_y4m = 0;
unsigned int width;
unsigned int height;
unsigned int timebase_num;
unsigned int timebase_den;
void *out = NULL;
vpx_codec_dec_cfg_t cfg = {0};
#if CONFIG_VP8_DECODER
vp8_postproc_cfg_t vp8_pp_cfg = {0};
#endif
/* Parse command line */
exec_name = argv_[0];
argv = argv_dup(argc - 1, argv_ + 1);
for (argi = argj = argv; (*argj = *argi); argi += arg.argv_step)
{
memset(&arg, 0, sizeof(arg));
arg.argv_step = 1;
if (arg_match(&arg, &codecarg, argi))
{
int j, k = -1;
for (j = 0; j < sizeof(ifaces) / sizeof(ifaces[0]); j++)
if (!strcmp(ifaces[j].name, arg.val))
k = j;
if (k >= 0)
iface = ifaces[k].iface;
else
die("Error: Unrecognized argument (%s) to --codec\n",
arg.val);
}
else if (arg_match(&arg, &outputfile, argi))
fn2 = arg.val;
else if (arg_match(&arg, &usey4marg, argi))
use_y4m = 1;
else if (arg_match(&arg, &prefixarg, argi))
prefix = strdup(arg.val);
else if (arg_match(&arg, &use_yv12, argi))
flipuv = 1;
else if (arg_match(&arg, &use_i420, argi))
flipuv = 0;
else if (arg_match(&arg, &flipuvarg, argi))
flipuv = 1;
else if (arg_match(&arg, &noblitarg, argi))
noblit = 1;
else if (arg_match(&arg, &progressarg, argi))
progress = 1;
else if (arg_match(&arg, &limitarg, argi))
stop_after = arg_parse_uint(&arg);
else if (arg_match(&arg, &postprocarg, argi))
postproc = 1;
else if (arg_match(&arg, &md5arg, argi))
do_md5 = 1;
else if (arg_match(&arg, &summaryarg, argi))
summary = 1;
else if (arg_match(&arg, &threadsarg, argi))
cfg.threads = arg_parse_uint(&arg);
else if (arg_match(&arg, &quietarg, argi))
quiet = 1;
#if CONFIG_VP8_DECODER
else if (arg_match(&arg, &addnoise_level, argi))
{
postproc = 1;
vp8_pp_cfg.post_proc_flag |= VP8_ADDNOISE;
vp8_pp_cfg.noise_level = arg_parse_uint(&arg);
}
else if (arg_match(&arg, &demacroblock_level, argi))
{
postproc = 1;
vp8_pp_cfg.post_proc_flag |= VP8_DEMACROBLOCK;
vp8_pp_cfg.deblocking_level = arg_parse_uint(&arg);
}
else if (arg_match(&arg, &deblock, argi))
{
postproc = 1;
vp8_pp_cfg.post_proc_flag |= VP8_DEBLOCK;
}
else if (arg_match(&arg, &pp_debug_info, argi))
{
unsigned int level = arg_parse_uint(&arg);
postproc = 1;
vp8_pp_cfg.post_proc_flag &= ~0x7;
if (level)
vp8_pp_cfg.post_proc_flag |= 8 << (level - 1);
}
#endif
else
argj++;
}
/* Check for unrecognized options */
for (argi = argv; *argi; argi++)
if (argi[0][0] == '-' && strlen(argi[0]) > 1)
die("Error: Unrecognized option %s\n", *argi);
/* Handle non-option arguments */
fn = argv[0];
if (!fn)
usage_exit();
if (!prefix)
prefix = strdup("img");
/* Open file */
infile = strcmp(fn, "-") ? fopen(fn, "rb") : stdin;
if (!infile)
{
fprintf(stderr, "Failed to open file");
return EXIT_FAILURE;
}
if (fn2)
out = out_open(fn2, do_md5);
is_ivf = file_is_ivf(infile, &fourcc, &width, &height,
&timebase_num, &timebase_den);
if (is_ivf)
{
if (use_y4m)
{
char buffer[128];
if (!fn2)
{
fprintf(stderr, "YUV4MPEG2 output only supported with -o.\n");
return EXIT_FAILURE;
}
/*Correct for the factor of 2 applied to the timebase in the
encoder.*/
if(timebase_den&1)timebase_num<<=1;
else timebase_den>>=1;
/*Note: We can't output an aspect ratio here because IVF doesn't
store one, and neither does VP8.
That will have to wait until these tools support WebM natively.*/
sprintf(buffer, "YUV4MPEG2 C%s W%u H%u F%u:%u I%c\n",
"420jpeg", width, height, timebase_den, timebase_num, 'p');
out_put(out, (unsigned char *)buffer, strlen(buffer), do_md5);
}
/* Try to determine the codec from the fourcc. */
for (i = 0; i < sizeof(ifaces) / sizeof(ifaces[0]); i++)
if ((fourcc & ifaces[i].fourcc_mask) == ifaces[i].fourcc)
{
vpx_codec_iface_t *ivf_iface = ifaces[i].iface;
if (iface && iface != ivf_iface)
fprintf(stderr, "Notice -- IVF header indicates codec: %s\n",
ifaces[i].name);
else
iface = ivf_iface;
break;
}
}
else if(use_y4m)
{
fprintf(stderr, "YUV4MPEG2 output only supported from IVF input.\n");
return EXIT_FAILURE;
}
if (vpx_codec_dec_init(&decoder, iface ? iface : ifaces[0].iface, &cfg,
postproc ? VPX_CODEC_USE_POSTPROC : 0))
{
fprintf(stderr, "Failed to initialize decoder: %s\n", vpx_codec_error(&decoder));
return EXIT_FAILURE;
}
if (!quiet)
fprintf(stderr, "%s\n", decoder.name);
#if CONFIG_VP8_DECODER
if (vp8_pp_cfg.post_proc_flag
&& vpx_codec_control(&decoder, VP8_SET_POSTPROC, &vp8_pp_cfg))
{
fprintf(stderr, "Failed to configure postproc: %s\n", vpx_codec_error(&decoder));
return EXIT_FAILURE;
}
#endif
/* Decode file */
while (!read_frame(infile, &buf, &buf_sz, &buf_alloc_sz, is_ivf))
{
vpx_codec_iter_t iter = NULL;
vpx_image_t *img;
struct vpx_usec_timer timer;
vpx_usec_timer_start(&timer);
if (vpx_codec_decode(&decoder, buf, buf_sz, NULL, 0))
{
const char *detail = vpx_codec_error_detail(&decoder);
fprintf(stderr, "Failed to decode frame: %s\n", vpx_codec_error(&decoder));
if (detail)
fprintf(stderr, " Additional information: %s\n", detail);
goto fail;
}
vpx_usec_timer_mark(&timer);
dx_time += vpx_usec_timer_elapsed(&timer);
++frame_in;
if (progress)
fprintf(stderr, "decoded frame %d.\n", frame_in);
if ((img = vpx_codec_get_frame(&decoder, &iter)))
++frame_out;
if (!noblit)
{
if (img)
{
unsigned int y;
char out_fn[128+24];
uint8_t *buf;
const char *sfx = flipuv ? "yv12" : "i420";
if (!fn2)
{
sprintf(out_fn, "%s-%dx%d-%04d.%s",
prefix, img->d_w, img->d_h, frame_in, sfx);
out = out_open(out_fn, do_md5);
}
else if(use_y4m)
out_put(out, (unsigned char *)"FRAME\n", 6, do_md5);
buf = img->planes[VPX_PLANE_Y];
for (y = 0; y < img->d_h; y++)
{
out_put(out, buf, img->d_w, do_md5);
buf += img->stride[VPX_PLANE_Y];
}
buf = img->planes[flipuv?VPX_PLANE_V:VPX_PLANE_U];
for (y = 0; y < (1 + img->d_h) / 2; y++)
{
out_put(out, buf, (1 + img->d_w) / 2, do_md5);
buf += img->stride[VPX_PLANE_U];
}
buf = img->planes[flipuv?VPX_PLANE_U:VPX_PLANE_V];
for (y = 0; y < (1 + img->d_h) / 2; y++)
{
out_put(out, buf, (1 + img->d_w) / 2, do_md5);
buf += img->stride[VPX_PLANE_V];
}
if (!fn2)
out_close(out, out_fn, do_md5);
}
}
if (stop_after && frame_in >= stop_after)
break;
}
if (summary)
{
fprintf(stderr, "%d decoded frames/%d showed frames in %lu us (%.2f fps)\n",
frame_in, frame_out, dx_time, (float)frame_out * 1000000.0 / (float)dx_time);
}
fail:
if (vpx_codec_destroy(&decoder))
{
fprintf(stderr, "Failed to destroy decoder: %s\n", vpx_codec_error(&decoder));
return EXIT_FAILURE;
}
if (fn2)
out_close(out, fn2, do_md5);
free(buf);
fclose(infile);
free(prefix);
free(argv);
return EXIT_SUCCESS;
}

60
libmkv/EbmlBufferWriter.c Normal file
View File

@@ -0,0 +1,60 @@
//#include <strmif.h>
#include "EbmlBufferWriter.h"
#include "EbmlWriter.h"
//#include <cassert>
//#include <limits>
//#include <malloc.h> //_alloca
#include <stdlib.h>
#include <wchar.h>
#include <string.h>
void Ebml_Write(EbmlGlobal *glob, const void *buffer_in, unsigned long len)
{
unsigned char *src = glob->buf;
src += glob->offset;
memcpy(src, buffer_in, len);
glob->offset += len;
}
static void _Serialize(EbmlGlobal *glob, const unsigned char *p, const unsigned char *q)
{
while (q != p)
{
--q;
unsigned long cbWritten;
memcpy(&(glob->buf[glob->offset]), q, 1);
glob->offset ++;
}
}
void Ebml_Serialize(EbmlGlobal *glob, const void *buffer_in, unsigned long len)
{
//assert(buf);
const unsigned char *const p = (const unsigned char *)(buffer_in);
const unsigned char *const q = p + len;
_Serialize(glob, p, q);
}
void Ebml_StartSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc, unsigned long class_id)
{
Ebml_WriteID(glob, class_id);
ebmlLoc->offset = glob->offset;
//todo this is always taking 8 bytes, this may need later optimization
unsigned long long unknownLen = 0x01FFFFFFFFFFFFFFLLU;
Ebml_Serialize(glob, (void *)&unknownLen, 8); //this is a key that says lenght unknown
}
void Ebml_EndSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc)
{
unsigned long long size = glob->offset - ebmlLoc->offset - 8;
unsigned long long curOffset = glob->offset;
glob->offset = ebmlLoc->offset;
size |= 0x0100000000000000LLU;
Ebml_Serialize(glob, &size, 8);
glob->offset = curOffset;
}

21
libmkv/EbmlBufferWriter.h Normal file
View File

@@ -0,0 +1,21 @@
#ifndef EBMLBUFFERWRITER_HPP
#define EBMLBUFFERWRITER_HPP
typedef struct
{
unsigned long long offset;
} EbmlLoc;
typedef struct
{
unsigned char *buf;
unsigned int length;
unsigned int offset;
} EbmlGlobal;
void Ebml_StartSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc, unsigned long class_id);
void Ebml_EndSubElement(EbmlGlobal *glob, EbmlLoc *ebmlLoc);
#endif

231
libmkv/EbmlIDs.h Normal file
View File

@@ -0,0 +1,231 @@
// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the LICENSE file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
#ifndef MKV_DEFS_HPP
#define MKV_DEFS_HPP 1
//Commenting out values not available in webm, but available in matroska
enum mkv
{
EBML = 0x1A45DFA3,
EBMLVersion = 0x4286,
EBMLReadVersion = 0x42F7,
EBMLMaxIDLength = 0x42F2,
EBMLMaxSizeLength = 0x42F3,
DocType = 0x4282,
DocTypeVersion = 0x4287,
DocTypeReadVersion = 0x4285,
// CRC_32 = 0xBF,
Void = 0xEC,
SignatureSlot = 0x1B538667,
SignatureAlgo = 0x7E8A,
SignatureHash = 0x7E9A,
SignaturePublicKey = 0x7EA5,
Signature = 0x7EB5,
SignatureElements = 0x7E5B,
SignatureElementList = 0x7E7B,
SignedElement = 0x6532,
//segment
Segment = 0x18538067,
//Meta Seek Information
SeekHead = 0x114D9B74,
Seek = 0x4DBB,
SeekID = 0x53AB,
SeekPosition = 0x53AC,
//Segment Information
Info = 0x1549A966,
// SegmentUID = 0x73A4,
// SegmentFilename = 0x7384,
// PrevUID = 0x3CB923,
// PrevFilename = 0x3C83AB,
// NextUID = 0x3EB923,
// NextFilename = 0x3E83BB,
// SegmentFamily = 0x4444,
// ChapterTranslate = 0x6924,
// ChapterTranslateEditionUID = 0x69FC,
// ChapterTranslateCodec = 0x69BF,
// ChapterTranslateID = 0x69A5,
TimecodeScale = 0x2AD7B1,
Segment_Duration = 0x4489,
DateUTC = 0x4461,
// Title = 0x7BA9,
MuxingApp = 0x4D80,
WritingApp = 0x5741,
//Cluster
Cluster = 0x1F43B675,
Timecode = 0xE7,
// SilentTracks = 0x5854,
// SilentTrackNumber = 0x58D7,
// Position = 0xA7,
PrevSize = 0xAB,
BlockGroup = 0xA0,
Block = 0xA1,
// BlockVirtual = 0xA2,
// BlockAdditions = 0x75A1,
// BlockMore = 0xA6,
// BlockAddID = 0xEE,
// BlockAdditional = 0xA5,
BlockDuration = 0x9B,
// ReferencePriority = 0xFA,
ReferenceBlock = 0xFB,
// ReferenceVirtual = 0xFD,
// CodecState = 0xA4,
// Slices = 0x8E,
// TimeSlice = 0xE8,
LaceNumber = 0xCC,
// FrameNumber = 0xCD,
// BlockAdditionID = 0xCB,
// MkvDelay = 0xCE,
// Cluster_Duration = 0xCF,
SimpleBlock = 0xA3,
// EncryptedBlock = 0xAF,
//Track
Tracks = 0x1654AE6B,
TrackEntry = 0xAE,
TrackNumber = 0xD7,
TrackUID = 0x73C5,
TrackType = 0x83,
FlagEnabled = 0xB9,
FlagDefault = 0x88,
FlagForced = 0x55AA,
FlagLacing = 0x9C,
// MinCache = 0x6DE7,
// MaxCache = 0x6DF8,
DefaultDuration = 0x23E383,
// TrackTimecodeScale = 0x23314F,
// TrackOffset = 0x537F,
// MaxBlockAdditionID = 0x55EE,
Name = 0x536E,
Language = 0x22B59C,
CodecID = 0x86,
CodecPrivate = 0x63A2,
CodecName = 0x258688,
// AttachmentLink = 0x7446,
// CodecSettings = 0x3A9697,
// CodecInfoURL = 0x3B4040,
// CodecDownloadURL = 0x26B240,
// CodecDecodeAll = 0xAA,
// TrackOverlay = 0x6FAB,
// TrackTranslate = 0x6624,
// TrackTranslateEditionUID = 0x66FC,
// TrackTranslateCodec = 0x66BF,
// TrackTranslateTrackID = 0x66A5,
//video
Video = 0xE0,
FlagInterlaced = 0x9A,
// StereoMode = 0x53B8,
PixelWidth = 0xB0,
PixelHeight = 0xBA,
PixelCropBottom = 0x54AA,
PixelCropTop = 0x54BB,
PixelCropLeft = 0x54CC,
PixelCropRight = 0x54DD,
DisplayWidth = 0x54B0,
DisplayHeight = 0x54BA,
DisplayUnit = 0x54B2,
AspectRatioType = 0x54B3,
// ColourSpace = 0x2EB524,
// GammaValue = 0x2FB523,
FrameRate = 0x2383E3,
//end video
//audio
Audio = 0xE1,
SamplingFrequency = 0xB5,
OutputSamplingFrequency = 0x78B5,
Channels = 0x9F,
// ChannelPositions = 0x7D7B,
BitDepth = 0x6264,
//end audio
//content encoding
// ContentEncodings = 0x6d80,
// ContentEncoding = 0x6240,
// ContentEncodingOrder = 0x5031,
// ContentEncodingScope = 0x5032,
// ContentEncodingType = 0x5033,
// ContentCompression = 0x5034,
// ContentCompAlgo = 0x4254,
// ContentCompSettings = 0x4255,
// ContentEncryption = 0x5035,
// ContentEncAlgo = 0x47e1,
// ContentEncKeyID = 0x47e2,
// ContentSignature = 0x47e3,
// ContentSigKeyID = 0x47e4,
// ContentSigAlgo = 0x47e5,
// ContentSigHashAlgo = 0x47e6,
//end content encoding
//Cueing Data
Cues = 0x1C53BB6B,
CuePoint = 0xBB,
CueTime = 0xB3,
CueTrackPositions = 0xB7,
CueTrack = 0xF7,
CueClusterPosition = 0xF1,
CueBlockNumber = 0x5378,
// CueCodecState = 0xEA,
// CueReference = 0xDB,
// CueRefTime = 0x96,
// CueRefCluster = 0x97,
// CueRefNumber = 0x535F,
// CueRefCodecState = 0xEB,
//Attachment
// Attachments = 0x1941A469,
// AttachedFile = 0x61A7,
// FileDescription = 0x467E,
// FileName = 0x466E,
// FileMimeType = 0x4660,
// FileData = 0x465C,
// FileUID = 0x46AE,
// FileReferral = 0x4675,
//Chapters
// Chapters = 0x1043A770,
// EditionEntry = 0x45B9,
// EditionUID = 0x45BC,
// EditionFlagHidden = 0x45BD,
// EditionFlagDefault = 0x45DB,
// EditionFlagOrdered = 0x45DD,
// ChapterAtom = 0xB6,
// ChapterUID = 0x73C4,
// ChapterTimeStart = 0x91,
// ChapterTimeEnd = 0x92,
// ChapterFlagHidden = 0x98,
// ChapterFlagEnabled = 0x4598,
// ChapterSegmentUID = 0x6E67,
// ChapterSegmentEditionUID = 0x6EBC,
// ChapterPhysicalEquiv = 0x63C3,
// ChapterTrack = 0x8F,
// ChapterTrackNumber = 0x89,
// ChapterDisplay = 0x80,
// ChapString = 0x85,
// ChapLanguage = 0x437C,
// ChapCountry = 0x437E,
// ChapProcess = 0x6944,
// ChapProcessCodecID = 0x6955,
// ChapProcessPrivate = 0x450D,
// ChapProcessCommand = 0x6911,
// ChapProcessTime = 0x6922,
// ChapProcessData = 0x6933,
//Tagging
// Tags = 0x1254C367,
// Tag = 0x7373,
// Targets = 0x63C0,
// TargetTypeValue = 0x68CA,
// TargetType = 0x63CA,
// Tagging_TrackUID = 0x63C5,
// Tagging_EditionUID = 0x63C9,
// Tagging_ChapterUID = 0x63C4,
// AttachmentUID = 0x63C6,
// SimpleTag = 0x67C8,
// TagName = 0x45A3,
// TagLanguage = 0x447A,
// TagDefault = 0x4484,
// TagString = 0x4487,
// TagBinary = 0x4485,
};
#endif

166
libmkv/EbmlWriter.c Normal file
View File

@@ -0,0 +1,166 @@
// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the LICENSE file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
#include "EbmlWriter.h"
#include <stdlib.h>
#include <wchar.h>
#include <string.h>
#if defined(_MSC_VER)
#define LITERALU64(n) n
#else
#define LITERALU64(n) n##LLU
#endif
void Ebml_WriteLen(EbmlGlobal *glob, long long val)
{
//TODO check and make sure we are not > than 0x0100000000000000LLU
unsigned char size = 8; //size in bytes to output
unsigned long long minVal = LITERALU64(0x00000000000000ff); //mask to compare for byte size
for (size = 1; size < 8; size ++)
{
if (val < minVal)
break;
minVal = (minVal << 7);
}
val |= (LITERALU64(0x000000000000080) << ((size - 1) * 7));
Ebml_Serialize(glob, (void *) &val, size);
}
void Ebml_WriteString(EbmlGlobal *glob, const char *str)
{
const size_t size_ = strlen(str);
const unsigned long long size = size_;
Ebml_WriteLen(glob, size);
//TODO: it's not clear from the spec whether the nul terminator
//should be serialized too. For now we omit the null terminator.
Ebml_Write(glob, str, size);
}
void Ebml_WriteUTF8(EbmlGlobal *glob, const wchar_t *wstr)
{
const size_t strlen = wcslen(wstr);
//TODO: it's not clear from the spec whether the nul terminator
//should be serialized too. For now we include it.
const unsigned long long size = strlen;
Ebml_WriteLen(glob, size);
Ebml_Write(glob, wstr, size);
}
void Ebml_WriteID(EbmlGlobal *glob, unsigned long class_id)
{
if (class_id >= 0x01000000)
Ebml_Serialize(glob, (void *)&class_id, 4);
else if (class_id >= 0x00010000)
Ebml_Serialize(glob, (void *)&class_id, 3);
else if (class_id >= 0x00000100)
Ebml_Serialize(glob, (void *)&class_id, 2);
else
Ebml_Serialize(glob, (void *)&class_id, 1);
}
void Ebml_SerializeUnsigned64(EbmlGlobal *glob, unsigned long class_id, uint64_t ui)
{
unsigned char sizeSerialized = 8 | 0x80;
Ebml_WriteID(glob, class_id);
Ebml_Serialize(glob, &sizeSerialized, 1);
Ebml_Serialize(glob, &ui, 8);
}
void Ebml_SerializeUnsigned(EbmlGlobal *glob, unsigned long class_id, unsigned long ui)
{
unsigned char size = 8; //size in bytes to output
unsigned char sizeSerialized = 0;
unsigned long minVal;
Ebml_WriteID(glob, class_id);
minVal = 0x7fLU; //mask to compare for byte size
for (size = 1; size < 4; size ++)
{
if (ui < minVal)
{
break;
}
minVal <<= 7;
}
sizeSerialized = 0x80 | size;
Ebml_Serialize(glob, &sizeSerialized, 1);
Ebml_Serialize(glob, &ui, size);
}
//TODO: perhaps this is a poor name for this id serializer helper function
void Ebml_SerializeBinary(EbmlGlobal *glob, unsigned long class_id, unsigned long bin)
{
int size;
for (size=4; size > 1; size--)
{
if (bin & 0x000000ff << ((size-1) * 8))
break;
}
Ebml_WriteID(glob, class_id);
Ebml_WriteLen(glob, size);
Ebml_WriteID(glob, bin);
}
void Ebml_SerializeFloat(EbmlGlobal *glob, unsigned long class_id, double d)
{
unsigned char len = 0x88;
Ebml_WriteID(glob, class_id);
Ebml_Serialize(glob, &len, 1);
Ebml_Serialize(glob, &d, 8);
}
void Ebml_WriteSigned16(EbmlGlobal *glob, short val)
{
signed long out = ((val & 0x003FFFFF) | 0x00200000) << 8;
Ebml_Serialize(glob, &out, 3);
}
void Ebml_SerializeString(EbmlGlobal *glob, unsigned long class_id, const char *s)
{
Ebml_WriteID(glob, class_id);
Ebml_WriteString(glob, s);
}
void Ebml_SerializeUTF8(EbmlGlobal *glob, unsigned long class_id, wchar_t *s)
{
Ebml_WriteID(glob, class_id);
Ebml_WriteUTF8(glob, s);
}
void Ebml_SerializeData(EbmlGlobal *glob, unsigned long class_id, unsigned char *data, unsigned long data_length)
{
unsigned char size = 4;
Ebml_WriteID(glob, class_id);
Ebml_WriteLen(glob, data_length);
Ebml_Write(glob, data, data_length);
}
void Ebml_WriteVoid(EbmlGlobal *glob, unsigned long vSize)
{
unsigned char tmp = 0;
unsigned long i = 0;
Ebml_WriteID(glob, 0xEC);
Ebml_WriteLen(glob, vSize);
for (i = 0; i < vSize; i++)
{
Ebml_Write(glob, &tmp, 1);
}
}
//TODO Serialize Date

38
libmkv/EbmlWriter.h Normal file
View File

@@ -0,0 +1,38 @@
#ifndef EBMLWRITER_HPP
#define EBMLWRITER_HPP
// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the LICENSE file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
//note: you must define write and serialize functions as well as your own EBML_GLOBAL
//These functions MUST be implemented
#include <stddef.h>
#include "vpx/vpx_integer.h"
typedef struct EbmlGlobal EbmlGlobal;
void Ebml_Serialize(EbmlGlobal *glob, const void *, unsigned long);
void Ebml_Write(EbmlGlobal *glob, const void *, unsigned long);
/////
void Ebml_WriteLen(EbmlGlobal *glob, long long val);
void Ebml_WriteString(EbmlGlobal *glob, const char *str);
void Ebml_WriteUTF8(EbmlGlobal *glob, const wchar_t *wstr);
void Ebml_WriteID(EbmlGlobal *glob, unsigned long class_id);
void Ebml_SerializeUnsigned64(EbmlGlobal *glob, unsigned long class_id, uint64_t ui);
void Ebml_SerializeUnsigned(EbmlGlobal *glob, unsigned long class_id, unsigned long ui);
void Ebml_SerializeBinary(EbmlGlobal *glob, unsigned long class_id, unsigned long ui);
void Ebml_SerializeFloat(EbmlGlobal *glob, unsigned long class_id, double d);
//TODO make this more generic to signed
void Ebml_WriteSigned16(EbmlGlobal *glob, short val);
void Ebml_SerializeString(EbmlGlobal *glob, unsigned long class_id, const char *s);
void Ebml_SerializeUTF8(EbmlGlobal *glob, unsigned long class_id, wchar_t *s);
void Ebml_SerializeData(EbmlGlobal *glob, unsigned long class_id, unsigned char *data, unsigned long data_length);
void Ebml_WriteVoid(EbmlGlobal *glob, unsigned long vSize);
//TODO need date function
#endif

25
libmkv/Makefile Normal file
View File

@@ -0,0 +1,25 @@
#Variables
CC=gcc
LINKER=gcc
FLAGS=
#Build Targets
EbmlWriter.o: EbmlWriter.c EbmlWriter.h
$(CC) $(FLAGS) -c EbmlWriter.c
EbmlBufferWriter.o: EbmlBufferWriter.c EbmlBufferWriter.h
$(CC) $(FLAGS) -c EbmlBufferWriter.c
MkvElement.o: MkvElement.c WebMElement.h
$(CC) $(FLAGS) -c MkvElement.c
testlibmkv.o: testlibmkv.c
$(CC) $(FLAGS) -c testlibmkv.c
testlibmkv: testlibmkv.o MkvElement.o EbmlBufferWriter.o EbmlWriter.o
$(LINKER) $(FLAGS) testlibmkv.o MkvElement.o EbmlBufferWriter.o EbmlWriter.o -o testlibmkv
clean:
rm -rf *.o testlibmkv

220
libmkv/WebMElement.c Normal file
View File

@@ -0,0 +1,220 @@
// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the LICENSE file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
#include "EbmlBufferWriter.h"
#include "EbmlIDs.h"
#include "WebMElement.h"
#include <stdio.h>
#define kVorbisPrivateMaxSize 4000
void writeHeader(EbmlGlobal *glob)
{
EbmlLoc start;
Ebml_StartSubElement(glob, &start, EBML);
Ebml_SerializeUnsigned(glob, EBMLVersion, 1);
Ebml_SerializeUnsigned(glob, EBMLReadVersion, 1); //EBML Read Version
Ebml_SerializeUnsigned(glob, EBMLMaxIDLength, 4); //EBML Max ID Length
Ebml_SerializeUnsigned(glob, EBMLMaxSizeLength, 8); //EBML Max Size Length
Ebml_SerializeString(glob, DocType, "webm"); //Doc Type
Ebml_SerializeUnsigned(glob, DocTypeVersion, 2); //Doc Type Version
Ebml_SerializeUnsigned(glob, DocTypeReadVersion, 2); //Doc Type Read Version
Ebml_EndSubElement(glob, &start);
}
void writeSimpleBlock(EbmlGlobal *glob, unsigned char trackNumber, short timeCode,
int isKeyframe, unsigned char lacingFlag, int discardable,
unsigned char *data, unsigned long dataLength)
{
Ebml_WriteID(glob, SimpleBlock);
unsigned long blockLength = 4 + dataLength;
blockLength |= 0x10000000; //TODO check length < 0x0FFFFFFFF
Ebml_Serialize(glob, &blockLength, 4);
trackNumber |= 0x80; //TODO check track nubmer < 128
Ebml_Write(glob, &trackNumber, 1);
//Ebml_WriteSigned16(glob, timeCode,2); //this is 3 bytes
Ebml_Serialize(glob, &timeCode, 2);
unsigned char flags = 0x00 | (isKeyframe ? 0x80 : 0x00) | (lacingFlag << 1) | discardable;
Ebml_Write(glob, &flags, 1);
Ebml_Write(glob, data, dataLength);
}
static UInt64 generateTrackID(unsigned int trackNumber)
{
UInt64 t = time(NULL) * trackNumber;
UInt64 r = rand();
r = r << 32;
r += rand();
UInt64 rval = t ^ r;
return rval;
}
void writeVideoTrack(EbmlGlobal *glob, unsigned int trackNumber, int flagLacing,
char *codecId, unsigned int pixelWidth, unsigned int pixelHeight,
double frameRate)
{
EbmlLoc start;
Ebml_StartSubElement(glob, &start, TrackEntry);
Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber);
UInt64 trackID = generateTrackID(trackNumber);
Ebml_SerializeUnsigned(glob, TrackUID, trackID);
Ebml_SerializeString(glob, CodecName, "VP8"); //TODO shouldn't be fixed
Ebml_SerializeUnsigned(glob, TrackType, 1); //video is always 1
Ebml_SerializeString(glob, CodecID, codecId);
{
EbmlLoc videoStart;
Ebml_StartSubElement(glob, &videoStart, Video);
Ebml_SerializeUnsigned(glob, PixelWidth, pixelWidth);
Ebml_SerializeUnsigned(glob, PixelHeight, pixelHeight);
Ebml_SerializeFloat(glob, FrameRate, frameRate);
Ebml_EndSubElement(glob, &videoStart); //Video
}
Ebml_EndSubElement(glob, &start); //Track Entry
}
void writeAudioTrack(EbmlGlobal *glob, unsigned int trackNumber, int flagLacing,
char *codecId, double samplingFrequency, unsigned int channels,
unsigned char *private, unsigned long privateSize)
{
EbmlLoc start;
Ebml_StartSubElement(glob, &start, TrackEntry);
Ebml_SerializeUnsigned(glob, TrackNumber, trackNumber);
UInt64 trackID = generateTrackID(trackNumber);
Ebml_SerializeUnsigned(glob, TrackUID, trackID);
Ebml_SerializeUnsigned(glob, TrackType, 2); //audio is always 2
//I am using defaults for thesed required fields
/* Ebml_SerializeUnsigned(glob, FlagEnabled, 1);
Ebml_SerializeUnsigned(glob, FlagDefault, 1);
Ebml_SerializeUnsigned(glob, FlagForced, 1);
Ebml_SerializeUnsigned(glob, FlagLacing, flagLacing);*/
Ebml_SerializeString(glob, CodecID, codecId);
Ebml_SerializeData(glob, CodecPrivate, private, privateSize);
Ebml_SerializeString(glob, CodecName, "VORBIS"); //fixed for now
{
EbmlLoc AudioStart;
Ebml_StartSubElement(glob, &AudioStart, Audio);
Ebml_SerializeFloat(glob, SamplingFrequency, samplingFrequency);
Ebml_SerializeUnsigned(glob, Channels, channels);
Ebml_EndSubElement(glob, &AudioStart);
}
Ebml_EndSubElement(glob, &start);
}
void writeSegmentInformation(EbmlGlobal *ebml, EbmlLoc* startInfo, unsigned long timeCodeScale, double duration)
{
Ebml_StartSubElement(ebml, startInfo, Info);
Ebml_SerializeUnsigned(ebml, TimecodeScale, timeCodeScale);
Ebml_SerializeFloat(ebml, Segment_Duration, duration * 1000.0); //Currently fixed to using milliseconds
Ebml_SerializeString(ebml, 0x4D80, "QTmuxingAppLibWebM-0.0.1");
Ebml_SerializeString(ebml, 0x5741, "QTwritingAppLibWebM-0.0.1");
Ebml_EndSubElement(ebml, startInfo);
}
/*
void Mkv_InitializeSegment(Ebml& ebml_out, EbmlLoc& ebmlLoc)
{
Ebml_StartSubElement(ebml_out, ebmlLoc, 0x18538067);
}
void Mkv_InitializeSeek(Ebml& ebml_out, EbmlLoc& ebmlLoc)
{
Ebml_StartSubElement(ebml_out, ebmlLoc, 0x114d9b74);
}
void Mkv_WriteSeekInformation(Ebml& ebml_out, SeekStruct& seekInformation)
{
EbmlLoc ebmlLoc;
Ebml_StartSubElement(ebml_out, ebmlLoc, 0x4dbb);
Ebml_SerializeString(ebml_out, 0x53ab, seekInformation.SeekID);
Ebml_SerializeUnsigned(ebml_out, 0x53ac, seekInformation.SeekPosition);
Ebml_EndSubElement(ebml_out, ebmlLoc);
}
void Mkv_WriteSegmentInformation(Ebml& ebml_out, SegmentInformationStruct& segmentInformation)
{
Ebml_SerializeUnsigned(ebml_out, 0x73a4, segmentInformation.segmentUID);
if (segmentInformation.filename != 0)
Ebml_SerializeString(ebml_out, 0x7384, segmentInformation.filename);
Ebml_SerializeUnsigned(ebml_out, 0x2AD7B1, segmentInformation.TimecodeScale);
Ebml_SerializeUnsigned(ebml_out, 0x4489, segmentInformation.Duration);
//TODO date
Ebml_SerializeWString(ebml_out, 0x4D80, L"MKVMUX");
Ebml_SerializeWString(ebml_out, 0x5741, segmentInformation.WritingApp);
}
void Mkv_InitializeTrack(Ebml& ebml_out, EbmlLoc& ebmlLoc)
{
Ebml_StartSubElement(ebml_out, ebmlLoc, 0x1654AE6B);
}
static void Mkv_WriteGenericTrackData(Ebml& ebml_out, TrackStruct& track)
{
Ebml_SerializeUnsigned(ebml_out, 0xD7, track.TrackNumber);
Ebml_SerializeUnsigned(ebml_out, 0x73C5, track.TrackUID);
Ebml_SerializeUnsigned(ebml_out, 0x83, track.TrackType);
Ebml_SerializeUnsigned(ebml_out, 0xB9, track.FlagEnabled ? 1 :0);
Ebml_SerializeUnsigned(ebml_out, 0x88, track.FlagDefault ? 1 :0);
Ebml_SerializeUnsigned(ebml_out, 0x55AA, track.FlagForced ? 1 :0);
if (track.Language != 0)
Ebml_SerializeString(ebml_out, 0x22B59C, track.Language);
if (track.CodecID != 0)
Ebml_SerializeString(ebml_out, 0x86, track.CodecID);
if (track.CodecPrivate != 0)
Ebml_SerializeData(ebml_out, 0x63A2, track.CodecPrivate, track.CodecPrivateLength);
if (track.CodecName != 0)
Ebml_SerializeWString(ebml_out, 0x258688, track.CodecName);
}
void Mkv_WriteVideoTrack(Ebml& ebml_out, TrackStruct & track, VideoTrackStruct& video)
{
EbmlLoc trackHeadLoc, videoHeadLoc;
Ebml_StartSubElement(ebml_out, trackHeadLoc, 0xAE); //start Track
Mkv_WriteGenericTrackData(ebml_out, track);
Ebml_StartSubElement(ebml_out, videoHeadLoc, 0xE0); //start Video
Ebml_SerializeUnsigned(ebml_out, 0x9A, video.FlagInterlaced ? 1 :0);
Ebml_SerializeUnsigned(ebml_out, 0xB0, video.PixelWidth);
Ebml_SerializeUnsigned(ebml_out, 0xBA, video.PixelHeight);
Ebml_SerializeUnsigned(ebml_out, 0x54B0, video.PixelDisplayWidth);
Ebml_SerializeUnsigned(ebml_out, 0x54BA, video.PixelDisplayHeight);
Ebml_SerializeUnsigned(ebml_out, 0x54B2, video.displayUnit);
Ebml_SerializeFloat(ebml_out, 0x2383E3, video.FrameRate);
Ebml_EndSubElement(ebml_out, videoHeadLoc);
Ebml_EndSubElement(ebml_out, trackHeadLoc);
}
void Mkv_WriteAudioTrack(Ebml& ebml_out, TrackStruct & track, AudioTrackStruct& video)
{
EbmlLoc trackHeadLoc, audioHeadLoc;
Ebml_StartSubElement(ebml_out, trackHeadLoc, 0xAE);
Mkv_WriteGenericTrackData(ebml_out, track);
Ebml_StartSubElement(ebml_out, audioHeadLoc, 0xE0); //start Audio
Ebml_SerializeFloat(ebml_out, 0xB5, video.SamplingFrequency);
Ebml_SerializeUnsigned(ebml_out, 0x9F, video.Channels);
Ebml_SerializeUnsigned(ebml_out, 0x6264, video.BitDepth);
Ebml_EndSubElement(ebml_out, audioHeadLoc); // end audio
Ebml_EndSubElement(ebml_out, trackHeadLoc);
}
void Mkv_WriteEbmlClusterHead(Ebml& ebml_out, EbmlLoc& ebmlLoc, ClusterHeadStruct & clusterHead)
{
Ebml_StartSubElement(ebml_out, ebmlLoc, 0x1F43B675);
Ebml_SerializeUnsigned(ebml_out, 0x6264, clusterHead.TimeCode);
}
void Mkv_WriteSimpleBlockHead(Ebml& ebml_out, EbmlLoc& ebmlLoc, SimpleBlockStruct& block)
{
Ebml_StartSubElement(ebml_out, ebmlLoc, 0xA3);
Ebml_Write1UInt(ebml_out, block.TrackNumber);
Ebml_WriteSigned16(ebml_out,block.TimeCode);
unsigned char flags = 0x00 | (block.iskey ? 0x80:0x00) | (block.lacing << 1) | block.discardable;
Ebml_Write1UInt(ebml_out, flags); //TODO this may be the wrong function
Ebml_Serialize(ebml_out, block.data, block.dataLength);
Ebml_EndSubElement(ebml_out,ebmlLoc);
}
*/

35
libmkv/WebMElement.h Normal file
View File

@@ -0,0 +1,35 @@
// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the LICENSE file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
#ifndef MKV_CONTEXT_HPP
#define MKV_CONTEXT_HPP 1
void writeSimpleBock(EbmlGlobal *ebml, unsigned char trackNumber, unsigned short timeCode,
int isKeyframe, unsigned char lacingFlag, int discardable,
unsigned char *data, unsigned long dataLength);
// these are helper functions
void writeHeader(EbmlGlobal *ebml);
void writeSegmentInformation(EbmlGlobal *ebml, EbmlLoc* startInfo , unsigned long timeCodeScale, double duration);
//this function is a helper only, it assumes a lot of defaults
void writeVideoTrack(EbmlGlobal *ebml, unsigned int trackNumber, int flagLacing,
char *codecId, unsigned int pixelWidth, unsigned int pixelHeight,
double frameRate);
void writeAudioTrack(EbmlGlobal *glob, unsigned int trackNumber, int flagLacing,
char *codecId, double samplingFrequency, unsigned int channels,
unsigned char *private, unsigned long privateSize);
void writeSimpleBlock(EbmlGlobal *ebml, unsigned char trackNumber, short timeCode,
int isKeyframe, unsigned char lacingFlag, int discardable,
unsigned char *data, unsigned long dataLength);
#endif

63
libmkv/testlibmkv.c Normal file
View File

@@ -0,0 +1,63 @@
// Copyright (c) 2010 The WebM project authors. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the LICENSE file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
#include "EbmlIDs.h"
#include "EbmlBufferWriter.h"
#include "WebMElement.h"
#include <stdio.h>
int main(int argc, char *argv[])
{
//init the datatype we're using for ebml output
unsigned char data[8192];
EbmlGlobal ebml;
ebml.buf = data;
ebml.offset = 0;
ebml.length = 8192;
writeHeader(&ebml);
{
EbmlLoc startSegment;
Ebml_StartSubElement(&ebml, &startSegment, Segment); //segment
{
//segment info
EbmlLoc startInfo;
Ebml_StartSubElement(&ebml, &startInfo, Info);
Ebml_SerializeString(&ebml, 0x4D80, "muxingAppLibMkv");
Ebml_SerializeString(&ebml, 0x5741, "writingAppLibMkv");
Ebml_EndSubElement(&ebml, &startInfo);
}
{
EbmlLoc trackStart;
Ebml_StartSubElement(&ebml, &trackStart, Tracks);
writeVideoTrack(&ebml, 1, 1, "V_MS/VFW/FOURCC", 320, 240, 29.97);
//writeAudioTrack(&ebml,2,1, "A_VORBIS", 32000, 1, NULL, 0);
Ebml_EndSubElement(&ebml, &trackStart);
}
{
EbmlLoc clusterStart;
Ebml_StartSubElement(&ebml, &clusterStart, Cluster); //cluster
Ebml_SerializeUnsigned(&ebml, Timecode, 0);
unsigned char someData[4] = {1, 2, 3, 4};
writeSimpleBlock(&ebml, 1, 0, 1, 0, 0, someData, 4);
Ebml_EndSubElement(&ebml, &clusterStart);
} //end cluster
Ebml_EndSubElement(&ebml, &startSegment);
}
//dump ebml stuff to the file
FILE *file_out = fopen("test.mkv", "wb");
size_t bytesWritten = fwrite(data, 1, ebml.offset, file_out);
fclose(file_out);
return 0;
}

2
libs.mk
View File

@@ -91,7 +91,9 @@ ifeq ($(ARCH_X86)$(ARCH_X86_64),yes)
CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/emms.asm
CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86.h
CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86_abi_support.asm
CODEC_SRCS-$(BUILD_LIBVPX) += vpx_ports/x86_cpuid.c
endif
CODEC_SRCS-$(ARCH_ARM) += vpx_ports/arm_cpudetect.c
CODEC_SRCS-$(ARCH_ARM) += $(BUILD_PFX)vpx_config.asm
CODEC_EXPORTS-$(BUILD_LIBVPX) += vpx/exports_com
CODEC_EXPORTS-$(CONFIG_ENCODERS) += vpx/exports_enc

40
nestegg/.gitignore vendored Normal file
View File

@@ -0,0 +1,40 @@
*.lo
*.o
*.swp
*~
.deps
.dirstamp
.libs
Makefile
Makefile.in
_stdint.h
aclocal.m4
autom4te.cache
compile
config.guess
config.h
config.h.in
config.log
config.status
config.sub
configure
depcomp
docs/Doxyfile
docs/doxygen-build.stamp
docs/html
install-sh
libtool
ltmain.sh
m4/libtool.m4
m4/ltoptions.m4
m4/ltsugar.m4
m4/ltversion.m4
m4/lt~obsolete.m4
missing
nestegg-uninstalled.pc
nestegg.pc
src/.dirstamp
src/libnestegg.la
stamp-h1
test/test
include/nestegg/nestegg-stdint.h

1
nestegg/AUTHORS Normal file
View File

@@ -0,0 +1 @@
Matthew Gregan <kinetik@flim.org>

8
nestegg/INSTALL Normal file
View File

@@ -0,0 +1,8 @@
Build instructions for libnestegg
=================================
0. Change directory into the source directory.
1. Run |autoreconf --install| to generate configure.
2. Run |./configure| to configure the build.
3. Run |make| to build.
4. Run |make check| to run the test suite.

13
nestegg/LICENSE Normal file
View File

@@ -0,0 +1,13 @@
Copyright © 2010 Mozilla Foundation
Permission to use, copy, modify, and distribute this software for any
purpose with or without fee is hereby granted, provided that the above
copyright notice and this permission notice appear in all copies.
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.

51
nestegg/Makefile.am Normal file
View File

@@ -0,0 +1,51 @@
AUTOMAKE_OPTIONS = foreign 1.11 no-dist-gzip dist-bzip2 subdir-objects
ACLOCAL_AMFLAGS = -I m4
INCLUDES = -I$(top_srcdir)/include -I. -I$(top_srcdir)/halloc
AM_CFLAGS = -ansi -pedantic -Wall -Wextra -Wno-long-long -O0 -g
SUBDIRS = docs
EXTRA_DIST = \
AUTHORS README LICENSE \
nestegg-uninstalled.pc.in \
m4/as-ac-expand.m4 \
m4/pkg.m4 \
m4/ax_create_stdint_h.m4 \
halloc/src/halloc.c \
halloc/halloc.h \
halloc/src/align.h \
halloc/src/hlist.h \
halloc/src/macros.h
pkgconfigdir = $(libdir)/pkgconfig
pkgconfig_DATA = nestegg.pc
nesteggincludedir = $(includedir)/nestegg
nestegginclude_HEADERS = include/nestegg/nestegg.h include/nestegg/nestegg-stdint.h
lib_LTLIBRARIES = src/libnestegg.la
src_libnestegg_la_SOURCES = \
src/nestegg.c \
halloc/src/halloc.c \
halloc/halloc.h \
halloc/src/align.h \
halloc/src/hlist.h \
halloc/src/macros.h
check_PROGRAMS = test/test
test_test_SOURCES = test/test.c
test_test_LDADD = src/libnestegg.la
DISTCLEANFILES = include/nestegg/nestegg-stdint.h
dist-hook:
find $(distdir) -type d -name '.git' | xargs rm -rf
debug:
$(MAKE) all CFLAGS="@DEBUG@"
profile:
$(MAKE) all CFLAGS="@PROFILE@"

6
nestegg/README Normal file
View File

@@ -0,0 +1,6 @@
See INSTALL for build instructions.
Licensed under an ISC-style license. See LICENSE for details.
The source under the halloc/ directory is licensed under a BSD license. See
halloc/halloc.h for details.

21
nestegg/TODO Normal file
View File

@@ -0,0 +1,21 @@
- Document when read, seek, tell callbacks are used.
- Add an automated testsuite.
- Test (and fix, if necessary) support for unknown sizes.
- Test (and fix, if necessary) support for large files.
- Read past unknown elements rather than seeking.
- Try to handle unknown elements with unknown sizes.
- Formalize handling of default element values.
- Try to resynchronize stream when read_block fails so that failure to parse
a single block can be treated as non-fatal.
- Make logging more useful to API users.
- Avoid reparsing Cues and ignore any SeekHead at end of file.
- Optionally build a Cue index as Clusters are parsed.
- Support seeking without Cues.
- Avoid building a list of Clusters as they are parsed and retain only the
last one parsed.
- Add an asynchronous error code to struct nestegg and ensure that API calls
continue to fail safely one a fatal error has been returned.
- Modify parser/data structures to provide a clean separation. Perhaps the
parser should return a generic tree of nodes that a second pass uses to
initialize the main data structures.
- Use pool allocator for all allocations.

124
nestegg/configure.ac Normal file
View File

@@ -0,0 +1,124 @@
dnl ------------------------------------------------
dnl Initialization and Versioning
dnl ------------------------------------------------
AC_INIT(libnestegg,[0.1git])
AC_CANONICAL_HOST
AC_CANONICAL_TARGET
AC_CONFIG_MACRO_DIR([m4])
AM_CONFIG_HEADER([config.h])
AC_CONFIG_SRCDIR([src/nestegg.c])
AM_INIT_AUTOMAKE
m4_ifdef([AM_SILENT_RULES], [AM_SILENT_RULES([yes])])
dnl Library versioning
dnl CURRENT, REVISION, AGE
dnl - library source changed -> increment REVISION
dnl - interfaces added/removed/changed -> increment CURRENT, REVISION = 0
dnl - interfaces added -> increment AGE
dnl - interfaces removed -> AGE = 0
NESTEGG_CURRENT=0
NESTEGG_REVISION=0
NESTEGG_AGE=1
AC_SUBST(NESTEGG_CURRENT)
AC_SUBST(NESTEGG_REVISION)
AC_SUBST(NESTEGG_AGE)
dnl --------------------------------------------------
dnl Check for programs
dnl --------------------------------------------------
dnl save $CFLAGS since AC_PROG_CC likes to insert "-g -O2"
dnl if $CFLAGS is blank
cflags_save="$CFLAGS"
AC_PROG_CC
AC_PROG_CPP
CFLAGS="$cflags_save"
AM_PROG_CC_C_O
AC_LIBTOOL_WIN32_DLL
AM_PROG_LIBTOOL
dnl Check for doxygen
AC_ARG_ENABLE([doc],
AS_HELP_STRING([--enable-doc], [Build API documentation]),
[ac_enable_doc=$enableval], [ac_enable_doc=auto])
if test "x$ac_enable_doc" != "xno"; then
AC_CHECK_PROG(HAVE_DOXYGEN, doxygen, true, false)
if test "x$HAVE_DOXYGEN" = "xfalse" -a "x$ac_enable_doc" = "xyes"; then
AC_MSG_ERROR([*** API documentation explicitly requested but Doxygen not found])
fi
else
HAVE_DOXYGEN=false
fi
AM_CONDITIONAL(HAVE_DOXYGEN,$HAVE_DOXYGEN)
if test $HAVE_DOXYGEN = "false"; then
AC_MSG_WARN([*** doxygen not found, API documentation will not be built])
fi
# Generate portable stdint.h replacement
AX_CREATE_STDINT_H(include/nestegg/nestegg-stdint.h)
# Test whenever ld supports -version-script
AC_PROG_LD
AC_PROG_LD_GNU
AC_MSG_CHECKING([how to control symbol export])
dnl --------------------------------------------------
dnl Do substitutions
dnl --------------------------------------------------
AC_SUBST(DEBUG)
AC_SUBST(PROFILE)
AC_OUTPUT([
Makefile
docs/Makefile
docs/Doxyfile
nestegg.pc
nestegg-uninstalled.pc
])
AS_AC_EXPAND(LIBDIR, ${libdir})
AS_AC_EXPAND(INCLUDEDIR, ${includedir})
AS_AC_EXPAND(BINDIR, ${bindir})
AS_AC_EXPAND(DOCDIR, ${docdir})
if test $HAVE_DOXYGEN = "false"; then
doc_build="no"
else
doc_build="yes"
fi
AC_MSG_RESULT([
------------------------------------------------------------------------
$PACKAGE $VERSION: Automatic configuration OK.
General configuration:
API Documentation: .......... ${doc_build}
Installation paths:
libnestegg: .................. ${LIBDIR}
C header files: .............. ${INCLUDEDIR}/nestegg
Documentation: ............... ${DOCDIR}
Building:
Type 'make' to compile $PACKAGE.
Type 'make install' to install $PACKAGE.
Example programs will be built but not installed.
------------------------------------------------------------------------
])

1551
nestegg/docs/Doxyfile.in Normal file
View File

File diff suppressed because it is too large Load Diff

38
nestegg/docs/Makefile.am Normal file
View File

@@ -0,0 +1,38 @@
doc_DATA = doxygen-build.stamp
EXTRA_DIST = Doxyfile.in
if HAVE_DOXYGEN
doxygen-build.stamp: Doxyfile
doxygen
touch doxygen-build.stamp
else
doxygen-build.stamp:
echo "*** Warning: Doxygen not found; documentation will not be built."
touch doxygen-build.stamp
endif
dist_docdir = $(distdir)/libnestegg
dist-hook:
if test -d html; then \
mkdir $(dist_docdir); \
echo -n "copying built documenation..."; \
cp -rp html $(dist_docdir)/html; \
echo "OK"; \
fi
install-data-local: doxygen-build.stamp
$(mkinstalldirs) $(DESTDIR)$(docdir)
if test -d html; then \
cp -rp html $(DESTDIR)$(docdir)/html; \
fi
uninstall-local:
rm -rf $(DESTDIR)$(docdir)
clean-local:
if test -d html; then rm -rf html; fi
if test -f doxygen-build.stamp; then rm -f doxygen-build.stamp; fi

45
nestegg/halloc/README Normal file
View File

@@ -0,0 +1,45 @@
halloc 1.2.1
============
Hierarchical memory heap interface - an extension to standard
malloc/free interface that simplifies tasks of memory disposal
when allocated structures exhibit hierarchical properties.
http://swapped.cc/halloc
=
To build libhalloc.a with GNU tools run
make
To install in /usr/include and /usr/lib
make install
To cleanup the build files
make clean
=
halloc-1.2.1
* fixed a double-free bug in _set_allocator() as per
Matthew Gregan comments
* switched to using NULL instead of 0 where applicable
halloc-1.2.0
* added missing <string.h> include to halloc.c
* improved standard compliance thanks to the feedback
received from Stan Tobias. Two things were fixed -
- hblock_t structure no longer uses zero-sized 'data'
array, which happened to be common, but non-standard
extension;
- secondly, added the code to test the behaviour of
realloc(ptr, 0). Standard allows it NOT to act as
free(), in which case halloc will use its own version
of allocator calling free() when neccessary.
halloc-1.1.0
* initial public release (rewrite of hhmalloc library)
=============================================================================
Copyright (c) 2004-2010, Alex Pankratov (ap@swapped.cc). All rights reserved.

43
nestegg/halloc/halloc.h Normal file
View File

@@ -0,0 +1,43 @@
/*
* Copyright (c) 2004-2010 Alex Pankratov. All rights reserved.
*
* Hierarchical memory allocator, 1.2.1
* http://swapped.cc/halloc
*/
/*
* The program is distributed under terms of BSD license.
* You can obtain the copy of the license by visiting:
*
* http://www.opensource.org/licenses/bsd-license.php
*/
#ifndef _LIBP_HALLOC_H_
#define _LIBP_HALLOC_H_
#include <stddef.h> /* size_t */
/*
* Core API
*/
void * halloc (void * block, size_t len);
void hattach(void * block, void * parent);
/*
* standard malloc/free api
*/
void * h_malloc (size_t len);
void * h_calloc (size_t n, size_t len);
void * h_realloc(void * p, size_t len);
void h_free (void * p);
char * h_strdup (const char * str);
/*
* the underlying allocator
*/
typedef void * (* realloc_t)(void * ptr, size_t len);
extern realloc_t halloc_allocator;
#endif

36
nestegg/halloc/src/align.h Normal file
View File

@@ -0,0 +1,36 @@
/*
* Copyright (c) 2004-2010 Alex Pankratov. All rights reserved.
*
* Hierarchical memory allocator, 1.2.1
* http://swapped.cc/halloc
*/
/*
* The program is distributed under terms of BSD license.
* You can obtain the copy of the license by visiting:
*
* http://www.opensource.org/licenses/bsd-license.php
*/
#ifndef _LIBP_ALIGN_H_
#define _LIBP_ALIGN_H_
/*
* a type with the most strict alignment requirements
*/
union max_align
{
char c;
short s;
long l;
int i;
float f;
double d;
void * v;
void (*q)(void);
};
typedef union max_align max_align_t;
#endif

254
nestegg/halloc/src/halloc.c Normal file
View File

@@ -0,0 +1,254 @@
/*
* Copyright (c) 2004i-2010 Alex Pankratov. All rights reserved.
*
* Hierarchical memory allocator, 1.2.1
* http://swapped.cc/halloc
*/
/*
* The program is distributed under terms of BSD license.
* You can obtain the copy of the license by visiting:
*
* http://www.opensource.org/licenses/bsd-license.php
*/
#include <stdlib.h> /* realloc */
#include <string.h> /* memset & co */
#include "../halloc.h"
#include "align.h"
#include "hlist.h"
/*
* block control header
*/
typedef struct hblock
{
#ifndef NDEBUG
#define HH_MAGIC 0x20040518L
long magic;
#endif
hlist_item_t siblings; /* 2 pointers */
hlist_head_t children; /* 1 pointer */
max_align_t data[1]; /* not allocated, see below */
} hblock_t;
#define sizeof_hblock offsetof(hblock_t, data)
/*
*
*/
realloc_t halloc_allocator = NULL;
#define allocator halloc_allocator
/*
* static methods
*/
static void _set_allocator(void);
static void * _realloc(void * ptr, size_t n);
static int _relate(hblock_t * b, hblock_t * p);
static void _free_children(hblock_t * p);
/*
* Core API
*/
void * halloc(void * ptr, size_t len)
{
hblock_t * p;
/* set up default allocator */
if (! allocator)
{
_set_allocator();
assert(allocator);
}
/* calloc */
if (! ptr)
{
if (! len)
return NULL;
p = allocator(0, len + sizeof_hblock);
if (! p)
return NULL;
#ifndef NDEBUG
p->magic = HH_MAGIC;
#endif
hlist_init(&p->children);
hlist_init_item(&p->siblings);
return p->data;
}
p = structof(ptr, hblock_t, data);
assert(p->magic == HH_MAGIC);
/* realloc */
if (len)
{
p = allocator(p, len + sizeof_hblock);
if (! p)
return NULL;
hlist_relink(&p->siblings);
hlist_relink_head(&p->children);
return p->data;
}
/* free */
_free_children(p);
hlist_del(&p->siblings);
allocator(p, 0);
return NULL;
}
void hattach(void * block, void * parent)
{
hblock_t * b, * p;
if (! block)
{
assert(! parent);
return;
}
/* detach */
b = structof(block, hblock_t, data);
assert(b->magic == HH_MAGIC);
hlist_del(&b->siblings);
if (! parent)
return;
/* attach */
p = structof(parent, hblock_t, data);
assert(p->magic == HH_MAGIC);
/* sanity checks */
assert(b != p); /* trivial */
assert(! _relate(p, b)); /* heavy ! */
hlist_add(&p->children, &b->siblings);
}
/*
* malloc/free api
*/
void * h_malloc(size_t len)
{
return halloc(0, len);
}
void * h_calloc(size_t n, size_t len)
{
void * ptr = halloc(0, len*=n);
return ptr ? memset(ptr, 0, len) : NULL;
}
void * h_realloc(void * ptr, size_t len)
{
return halloc(ptr, len);
}
void h_free(void * ptr)
{
halloc(ptr, 0);
}
char * h_strdup(const char * str)
{
size_t len = strlen(str);
char * ptr = halloc(0, len + 1);
return ptr ? (ptr[len] = 0, memcpy(ptr, str, len)) : NULL;
}
/*
* static stuff
*/
static void _set_allocator(void)
{
void * p;
assert(! allocator);
/*
* the purpose of the test below is to check the behaviour
* of realloc(ptr, 0), which is defined in the standard
* as an implementation-specific. if it returns zero,
* then it's equivalent to free(). it can however return
* non-zero, in which case it cannot be used for freeing
* memory blocks and we'll need to supply our own version
*
* Thanks to Stan Tobias for pointing this tricky part out.
*/
allocator = realloc;
if (! (p = malloc(1)))
/* hmm */
return;
if ((p = realloc(p, 0)))
{
/* realloc cannot be used as free() */
allocator = _realloc;
free(p);
}
}
static void * _realloc(void * ptr, size_t n)
{
/*
* free'ing realloc()
*/
if (n)
return realloc(ptr, n);
free(ptr);
return NULL;
}
static int _relate(hblock_t * b, hblock_t * p)
{
hlist_item_t * i;
if (!b || !p)
return 0;
/*
* since there is no 'parent' pointer, which would've allowed
* O(log(n)) upward traversal, the check must use O(n) downward
* iteration of the entire hierarchy; and this can be VERY SLOW
*/
hlist_for_each(i, &p->children)
{
hblock_t * q = structof(i, hblock_t, siblings);
if (q == b || _relate(b, q))
return 1;
}
return 0;
}
static void _free_children(hblock_t * p)
{
hlist_item_t * i, * tmp;
#ifndef NDEBUG
/*
* this catches loops in hierarchy with almost zero
* overhead (compared to _relate() running time)
*/
assert(p && p->magic == HH_MAGIC);
p->magic = 0;
#endif
hlist_for_each_safe(i, tmp, &p->children)
{
hblock_t * q = structof(i, hblock_t, siblings);
_free_children(q);
allocator(q, 0);
}
}

136
nestegg/halloc/src/hlist.h Normal file
View File

@@ -0,0 +1,136 @@
/*
* Copyright (c) 2004-2010 Alex Pankratov. All rights reserved.
*
* Hierarchical memory allocator, 1.2.1
* http://swapped.cc/halloc
*/
/*
* The program is distributed under terms of BSD license.
* You can obtain the copy of the license by visiting:
*
* http://www.opensource.org/licenses/bsd-license.php
*/
#ifndef _LIBP_HLIST_H_
#define _LIBP_HLIST_H_
#include <assert.h>
#include "macros.h" /* static_inline */
/*
* weak double-linked list w/ tail sentinel
*/
typedef struct hlist_head hlist_head_t;
typedef struct hlist_item hlist_item_t;
/*
*
*/
struct hlist_head
{
hlist_item_t * next;
};
struct hlist_item
{
hlist_item_t * next;
hlist_item_t ** prev;
};
/*
* shared tail sentinel
*/
struct hlist_item hlist_null;
/*
*
*/
#define __hlist_init(h) { &hlist_null }
#define __hlist_init_item(i) { &hlist_null, &(i).next }
static_inline void hlist_init(hlist_head_t * h);
static_inline void hlist_init_item(hlist_item_t * i);
/* static_inline void hlist_purge(hlist_head_t * h); */
/* static_inline bool_t hlist_empty(const hlist_head_t * h); */
/* static_inline hlist_item_t * hlist_head(const hlist_head_t * h); */
/* static_inline hlist_item_t * hlist_next(const hlist_item_t * i); */
/* static_inline hlist_item_t * hlist_prev(const hlist_item_t * i,
const hlist_head_t * h); */
static_inline void hlist_add(hlist_head_t * h, hlist_item_t * i);
/* static_inline void hlist_add_prev(hlist_item_t * l, hlist_item_t * i); */
/* static_inline void hlist_add_next(hlist_item_t * l, hlist_item_t * i); */
static_inline void hlist_del(hlist_item_t * i);
static_inline void hlist_relink(hlist_item_t * i);
static_inline void hlist_relink_head(hlist_head_t * h);
#define hlist_for_each(i, h) \
for (i = (h)->next; i != &hlist_null; i = i->next)
#define hlist_for_each_safe(i, tmp, h) \
for (i = (h)->next, tmp = i->next; \
i!= &hlist_null; \
i = tmp, tmp = i->next)
/*
* static
*/
static_inline void hlist_init(hlist_head_t * h)
{
assert(h);
h->next = &hlist_null;
}
static_inline void hlist_init_item(hlist_item_t * i)
{
assert(i);
i->prev = &i->next;
i->next = &hlist_null;
}
static_inline void hlist_add(hlist_head_t * h, hlist_item_t * i)
{
hlist_item_t * next;
assert(h && i);
next = i->next = h->next;
next->prev = &i->next;
h->next = i;
i->prev = &h->next;
}
static_inline void hlist_del(hlist_item_t * i)
{
hlist_item_t * next;
assert(i);
next = i->next;
next->prev = i->prev;
*i->prev = next;
hlist_init_item(i);
}
static_inline void hlist_relink(hlist_item_t * i)
{
assert(i);
*i->prev = i;
i->next->prev = &i->next;
}
static_inline void hlist_relink_head(hlist_head_t * h)
{
assert(h);
h->next->prev = &h->next;
}
#endif

36
nestegg/halloc/src/macros.h Normal file
View File

@@ -0,0 +1,36 @@
/*
* Copyright (c) 2004-2010 Alex Pankratov. All rights reserved.
*
* Hierarchical memory allocator, 1.2.1
* http://swapped.cc/halloc
*/
/*
* The program is distributed under terms of BSD license.
* You can obtain the copy of the license by visiting:
*
* http://www.opensource.org/licenses/bsd-license.php
*/
#ifndef _LIBP_MACROS_H_
#define _LIBP_MACROS_H_
#include <stddef.h> /* offsetof */
/*
restore pointer to the structure by a pointer to its field
*/
#define structof(p,t,f) ((t*)(- offsetof(t,f) + (char*)(p)))
/*
* redefine for the target compiler
*/
#ifdef _WIN32
#define static_inline static __inline
#else
#define static_inline static __inline__
#endif
#endif

292
nestegg/include/nestegg/nestegg.h Normal file
View File

@@ -0,0 +1,292 @@
/*
* Copyright © 2010 Mozilla Foundation
*
* This program is made available under an ISC-style license. See the
* accompanying file LICENSE for details.
*/
#ifndef NESTEGG_671cac2a_365d_ed69_d7a3_4491d3538d79
#define NESTEGG_671cac2a_365d_ed69_d7a3_4491d3538d79
#include "vpx/vpx_integer.h"
#ifdef __cplusplus
extern "C" {
#endif
/** @mainpage
@section intro Introduction
This is the documentation fot the <tt>libnestegg</tt> C API.
<tt>libnestegg</tt> is a demultiplexing library for <a
href="http://www.matroska.org/">Matroska</a> and <a
href="http://www.webmproject.org/">WebMedia</a> media files.
@section example Example code
@code
nestegg * demux_ctx;
nestegg_init(&demux_ctx, io, NULL);
nestegg_packet * pkt;
while ((r = nestegg_read_packet(demux_ctx, &pkt)) > 0) {
unsigned int track;
nestegg_packet_track(pkt, &track);
// This example decodes the first track only.
if (track == 0) {
unsigned int chunk, chunks;
nestegg_packet_count(pkt, &chunks);
// Decode each chunk of data.
for (chunk = 0; chunk < chunks; ++chunk) {
unsigned char * data;
size_t data_size;
nestegg_packet_data(pkt, chunk, &data, &data_size);
example_codec_decode(codec_ctx, data, data_size);
}
}
nestegg_free_packet(pkt);
}
nestegg_destroy(demux_ctx);
@endcode
*/
/** @file
The <tt>libnestegg</tt> C API. */
#define NESTEGG_TRACK_VIDEO 0 /**< Track is of type video. */
#define NESTEGG_TRACK_AUDIO 1 /**< Track is of type audio. */
#define NESTEGG_CODEC_VP8 0 /**< Track uses Google On2 VP8 codec. */
#define NESTEGG_CODEC_VORBIS 1 /**< Track uses Xiph Vorbis codec. */
#define NESTEGG_SEEK_SET 0 /**< Seek offset relative to beginning of stream. */
#define NESTEGG_SEEK_CUR 1 /**< Seek offset relative to current position in stream. */
#define NESTEGG_SEEK_END 2 /**< Seek offset relative to end of stream. */
#define NESTEGG_LOG_DEBUG 1 /**< Debug level log message. */
#define NESTEGG_LOG_INFO 10 /**< Informational level log message. */
#define NESTEGG_LOG_WARNING 100 /**< Warning level log message. */
#define NESTEGG_LOG_ERROR 1000 /**< Error level log message. */
#define NESTEGG_LOG_CRITICAL 10000 /**< Critical level log message. */
typedef struct nestegg nestegg; /**< Opaque handle referencing the stream state. */
typedef struct nestegg_packet nestegg_packet; /**< Opaque handle referencing a packet of data. */
/** User supplied IO context. */
typedef struct {
/** User supplied read callback.
@param buffer Buffer to read data into.
@param length Length of supplied buffer in bytes.
@param userdata The #userdata supplied by the user.
@retval 1 Read succeeded.
@retval 0 End of stream.
@retval -1 Error. */
int (* read)(void * buffer, size_t length, void * userdata);
/** User supplied seek callback.
@param offset Offset within the stream to seek to.
@param whence Seek direction. One of #NESTEGG_SEEK_SET,
#NESTEGG_SEEK_CUR, or #NESTEGG_SEEK_END.
@param userdata The #userdata supplied by the user.
@retval 0 Seek succeeded.
@retval -1 Error. */
int (* seek)(int64_t offset, int whence, void * userdata);
/** User supplied tell callback.
@param userdata The #userdata supplied by the user.
@returns Current position within the stream.
@retval -1 Error. */
int64_t (* tell)(void * userdata);
/** User supplied pointer to be passed to the IO callbacks. */
void * userdata;
} nestegg_io;
/** Parameters specific to a video track. */
typedef struct {
unsigned int width; /**< Width of the video frame in pixels. */
unsigned int height; /**< Height of the video frame in pixels. */
unsigned int display_width; /**< Display width of the video frame in pixels. */
unsigned int display_height; /**< Display height of the video frame in pixels. */
unsigned int crop_bottom; /**< Pixels to crop from the bottom of the frame. */
unsigned int crop_top; /**< Pixels to crop from the top of the frame. */
unsigned int crop_left; /**< Pixels to crop from the left of the frame. */
unsigned int crop_right; /**< Pixels to crop from the right of the frame. */
} nestegg_video_params;
/** Parameters specific to an audio track. */
typedef struct {
double rate; /**< Sampling rate in Hz. */
unsigned int channels; /**< Number of audio channels. */
unsigned int depth; /**< Bits per sample. */
} nestegg_audio_params;
/** Logging callback function pointer. */
typedef void (* nestegg_log)(nestegg * context, unsigned int severity, char const * format, ...);
/** Initialize a nestegg context. During initialization the parser will
read forward in the stream processing all elements until the first
block of media is reached. All track metadata has been processed at this point.
@param context Storage for the new nestegg context. @see nestegg_destroy
@param io User supplied IO context.
@param callback Optional logging callback function pointer. May be NULL.
@retval 0 Success.
@retval -1 Error. */
int nestegg_init(nestegg ** context, nestegg_io io, nestegg_log callback);
/** Destroy a nestegg context and free associated memory.
@param context #nestegg context to be freed. @see nestegg_init */
void nestegg_destroy(nestegg * context);
/** Query the duration of the media stream in nanoseconds.
@param context Stream context initialized by #nestegg_init.
@param duration Storage for the queried duration.
@retval 0 Success.
@retval -1 Error. */
int nestegg_duration(nestegg * context, uint64_t * duration);
/** Query the tstamp scale of the media stream in nanoseconds.
Timecodes presented by nestegg have been scaled by this value
before presentation to the caller.
@param context Stream context initialized by #nestegg_init.
@param scale Storage for the queried scale factor.
@retval 0 Success.
@retval -1 Error. */
int nestegg_tstamp_scale(nestegg * context, uint64_t * scale);
/** Query the number of tracks in the media stream.
@param context Stream context initialized by #nestegg_init.
@param tracks Storage for the queried track count.
@retval 0 Success.
@retval -1 Error. */
int nestegg_track_count(nestegg * context, unsigned int * tracks);
/** Seek @a track to @a tstamp. Stream seek will terminate at the earliest
key point in the stream at or before @a tstamp. Other tracks in the
stream will output packets with unspecified but nearby timestamps.
@param context Stream context initialized by #nestegg_init.
@param track Zero based track number.
@param tstamp Absolute timestamp in nanoseconds.
@retval 0 Success.
@retval -1 Error. */
int nestegg_track_seek(nestegg * context, unsigned int track, uint64_t tstamp);
/** Query the type specified by @a track.
@param context Stream context initialized by #nestegg_init.
@param track Zero based track number.
@retval #NESTEGG_TRACK_VIDEO Track type is video.
@retval #NESTEGG_TRACK_AUDIO Track type is audio.
@retval -1 Error. */
int nestegg_track_type(nestegg * context, unsigned int track);
/** Query the codec ID specified by @a track.
@param context Stream context initialized by #nestegg_init.
@param track Zero based track number.
@retval #NESTEGG_CODEC_VP8 Track codec is VP8.
@retval #NESTEGG_CODEC_VORBIS Track codec is Vorbis.
@retval -1 Error. */
int nestegg_track_codec_id(nestegg * context, unsigned int track);
/** Query the number of codec initialization chunks for @a track. Each
chunk of data should be passed to the codec initialization functions in
the order returned.
@param context Stream context initialized by #nestegg_init.
@param track Zero based track number.
@param count Storage for the queried chunk count.
@retval 0 Success.
@retval -1 Error. */
int nestegg_track_codec_data_count(nestegg * context, unsigned int track,
unsigned int * count);
/** Get a pointer to chunk number @a item of codec initialization data for
@a track.
@param context Stream context initialized by #nestegg_init.
@param track Zero based track number.
@param item Zero based chunk item number.
@param data Storage for the queried data pointer.
The data is owned by the #nestegg context.
@param length Storage for the queried data size.
@retval 0 Success.
@retval -1 Error. */
int nestegg_track_codec_data(nestegg * context, unsigned int track, unsigned int item,
unsigned char ** data, size_t * length);
/** Query the video parameters specified by @a track.
@param context Stream context initialized by #nestegg_init.
@param track Zero based track number.
@param params Storage for the queried video parameters.
@retval 0 Success.
@retval -1 Error. */
int nestegg_track_video_params(nestegg * context, unsigned int track,
nestegg_video_params * params);
/** Query the audio parameters specified by @a track.
@param context Stream context initialized by #nestegg_init.
@param track Zero based track number.
@param params Storage for the queried audio parameters.
@retval 0 Success.
@retval -1 Error. */
int nestegg_track_audio_params(nestegg * context, unsigned int track,
nestegg_audio_params * params);
/** Read a packet of media data. A packet consists of one or more chunks of
data associated with a single track. nestegg_read_packet should be
called in a loop while the return value is 1 to drive the stream parser
forward. @see nestegg_free_packet
@param context Context returned by #nestegg_init.
@param packet Storage for the returned nestegg_packet.
@retval 1 Additional packets may be read in subsequent calls.
@retval 0 End of stream.
@retval -1 Error. */
int nestegg_read_packet(nestegg * context, nestegg_packet ** packet);
/** Destroy a nestegg_packet and free associated memory.
@param packet #nestegg_packet to be freed. @see nestegg_read_packet */
void nestegg_free_packet(nestegg_packet * packet);
/** Query the track number of @a packet.
@param packet Packet initialized by #nestegg_read_packet.
@param track Storage for the queried zero based track index.
@retval 0 Success.
@retval -1 Error. */
int nestegg_packet_track(nestegg_packet * packet, unsigned int * track);
/** Query the time stamp in nanoseconds of @a packet.
@param packet Packet initialized by #nestegg_read_packet.
@param tstamp Storage for the queried timestamp in nanoseconds.
@retval 0 Success.
@retval -1 Error. */
int nestegg_packet_tstamp(nestegg_packet * packet, uint64_t * tstamp);
/** Query the number of data chunks contained in @a packet.
@param packet Packet initialized by #nestegg_read_packet.
@param count Storage for the queried timestamp in nanoseconds.
@retval 0 Success.
@retval -1 Error. */
int nestegg_packet_count(nestegg_packet * packet, unsigned int * count);
/** Get a pointer to chunk number @a item of packet data.
@param packet Packet initialized by #nestegg_read_packet.
@param item Zero based chunk item number.
@param data Storage for the queried data pointer.
The data is owned by the #nestegg_packet packet.
@param length Storage for the queried data size.
@retval 0 Success.
@retval -1 Error. */
int nestegg_packet_data(nestegg_packet * packet, unsigned int item,
unsigned char ** data, size_t * length);
#ifdef __cplusplus
}
#endif
#endif /* NESTEGG_671cac2a_365d_ed69_d7a3_4491d3538d79 */

43
nestegg/m4/as-ac-expand.m4 Normal file
View File

@@ -0,0 +1,43 @@
dnl as-ac-expand.m4 0.2.0
dnl autostars m4 macro for expanding directories using configure's prefix
dnl thomas@apestaart.org
dnl AS_AC_EXPAND(VAR, CONFIGURE_VAR)
dnl example
dnl AS_AC_EXPAND(SYSCONFDIR, $sysconfdir)
dnl will set SYSCONFDIR to /usr/local/etc if prefix=/usr/local
AC_DEFUN([AS_AC_EXPAND],
[
EXP_VAR=[$1]
FROM_VAR=[$2]
dnl first expand prefix and exec_prefix if necessary
prefix_save=$prefix
exec_prefix_save=$exec_prefix
dnl if no prefix given, then use /usr/local, the default prefix
if test "x$prefix" = "xNONE"; then
prefix="$ac_default_prefix"
fi
dnl if no exec_prefix given, then use prefix
if test "x$exec_prefix" = "xNONE"; then
exec_prefix=$prefix
fi
full_var="$FROM_VAR"
dnl loop until it doesn't change anymore
while true; do
new_full_var="`eval echo $full_var`"
if test "x$new_full_var" = "x$full_var"; then break; fi
full_var=$new_full_var
done
dnl clean up
full_var=$new_full_var
AC_SUBST([$1], "$full_var")
dnl restore prefix and exec_prefix
prefix=$prefix_save
exec_prefix=$exec_prefix_save
])

695
nestegg/m4/ax_create_stdint_h.m4 Normal file
View File

@@ -0,0 +1,695 @@
dnl @synopsis AX_CREATE_STDINT_H [( HEADER-TO-GENERATE [, HEDERS-TO-CHECK])]
dnl
dnl the "ISO C9X: 7.18 Integer types <stdint.h>" section requires the
dnl existence of an include file <stdint.h> that defines a set of
dnl typedefs, especially uint8_t,int32_t,uintptr_t. Many older
dnl installations will not provide this file, but some will have the
dnl very same definitions in <inttypes.h>. In other enviroments we can
dnl use the inet-types in <sys/types.h> which would define the typedefs
dnl int8_t and u_int8_t respectivly.
dnl
dnl This macros will create a local "_stdint.h" or the headerfile given
dnl as an argument. In many cases that file will just "#include
dnl <stdint.h>" or "#include <inttypes.h>", while in other environments
dnl it will provide the set of basic 'stdint's definitions/typedefs:
dnl
dnl int8_t,uint8_t,int16_t,uint16_t,int32_t,uint32_t,intptr_t,uintptr_t
dnl int_least32_t.. int_fast32_t.. intmax_t
dnl
dnl which may or may not rely on the definitions of other files, or
dnl using the AC_CHECK_SIZEOF macro to determine the actual sizeof each
dnl type.
dnl
dnl if your header files require the stdint-types you will want to
dnl create an installable file mylib-int.h that all your other
dnl installable header may include. So if you have a library package
dnl named "mylib", just use
dnl
dnl AX_CREATE_STDINT_H(mylib-int.h)
dnl
dnl in configure.ac and go to install that very header file in
dnl Makefile.am along with the other headers (mylib.h) - and the
dnl mylib-specific headers can simply use "#include <mylib-int.h>" to
dnl obtain the stdint-types.
dnl
dnl Remember, if the system already had a valid <stdint.h>, the
dnl generated file will include it directly. No need for fuzzy
dnl HAVE_STDINT_H things... (oops, GCC 4.2.x has deliberatly disabled
dnl its stdint.h for non-c99 compilation and the c99-mode is not the
dnl default. Therefore this macro will not use the compiler's stdint.h
dnl - please complain to the GCC developers).
dnl
dnl @category C
dnl @author Guido U. Draheim <guidod@gmx.de>
dnl @version 2006-10-13
dnl @license GPLWithACException
AC_DEFUN([AX_CHECK_DATA_MODEL],[
AC_CHECK_SIZEOF(char)
AC_CHECK_SIZEOF(short)
AC_CHECK_SIZEOF(int)
AC_CHECK_SIZEOF(long)
AC_CHECK_SIZEOF(void*)
ac_cv_char_data_model=""
ac_cv_char_data_model="$ac_cv_char_data_model$ac_cv_sizeof_char"
ac_cv_char_data_model="$ac_cv_char_data_model$ac_cv_sizeof_short"
ac_cv_char_data_model="$ac_cv_char_data_model$ac_cv_sizeof_int"
ac_cv_long_data_model=""
ac_cv_long_data_model="$ac_cv_long_data_model$ac_cv_sizeof_int"
ac_cv_long_data_model="$ac_cv_long_data_model$ac_cv_sizeof_long"
ac_cv_long_data_model="$ac_cv_long_data_model$ac_cv_sizeof_voidp"
AC_MSG_CHECKING([data model])
case "$ac_cv_char_data_model/$ac_cv_long_data_model" in
122/242) ac_cv_data_model="IP16" ; n="standard 16bit machine" ;;
122/244) ac_cv_data_model="LP32" ; n="standard 32bit machine" ;;
122/*) ac_cv_data_model="i16" ; n="unusual int16 model" ;;
124/444) ac_cv_data_model="ILP32" ; n="standard 32bit unixish" ;;
124/488) ac_cv_data_model="LP64" ; n="standard 64bit unixish" ;;
124/448) ac_cv_data_model="LLP64" ; n="unusual 64bit unixish" ;;
124/*) ac_cv_data_model="i32" ; n="unusual int32 model" ;;
128/888) ac_cv_data_model="ILP64" ; n="unusual 64bit numeric" ;;
128/*) ac_cv_data_model="i64" ; n="unusual int64 model" ;;
222/*2) ac_cv_data_model="DSP16" ; n="strict 16bit dsptype" ;;
333/*3) ac_cv_data_model="DSP24" ; n="strict 24bit dsptype" ;;
444/*4) ac_cv_data_model="DSP32" ; n="strict 32bit dsptype" ;;
666/*6) ac_cv_data_model="DSP48" ; n="strict 48bit dsptype" ;;
888/*8) ac_cv_data_model="DSP64" ; n="strict 64bit dsptype" ;;
222/*|333/*|444/*|666/*|888/*) :
ac_cv_data_model="iDSP" ; n="unusual dsptype" ;;
*) ac_cv_data_model="none" ; n="very unusual model" ;;
esac
AC_MSG_RESULT([$ac_cv_data_model ($ac_cv_long_data_model, $n)])
])
dnl AX_CHECK_HEADER_STDINT_X([HEADERLIST][,ACTION-IF])
AC_DEFUN([AX_CHECK_HEADER_STDINT_X],[
AC_CACHE_CHECK([for stdint uintptr_t], [ac_cv_header_stdint_x],[
ac_cv_header_stdint_x="" # the 1997 typedefs (inttypes.h)
AC_MSG_RESULT([(..)])
for i in m4_ifval([$1],[$1],[stdint.h inttypes.h sys/inttypes.h sys/types.h])
do
unset ac_cv_type_uintptr_t
unset ac_cv_type_uint64_t
AC_CHECK_TYPE(uintptr_t,[ac_cv_header_stdint_x=$i],continue,[#include <$i>])
AC_CHECK_TYPE(uint64_t,[and64="/uint64_t"],[and64=""],[#include<$i>])
m4_ifvaln([$1],[$1]) break
done
AC_MSG_CHECKING([for stdint uintptr_t])
])
])
AC_DEFUN([AX_CHECK_HEADER_STDINT_O],[
AC_CACHE_CHECK([for stdint uint32_t], [ac_cv_header_stdint_o],[
ac_cv_header_stdint_o="" # the 1995 typedefs (sys/inttypes.h)
AC_MSG_RESULT([(..)])
for i in m4_ifval([$1],[$1],[inttypes.h sys/inttypes.h sys/types.h stdint.h])
do
unset ac_cv_type_uint32_t
unset ac_cv_type_uint64_t
AC_CHECK_TYPE(uint32_t,[ac_cv_header_stdint_o=$i],continue,[#include <$i>])
AC_CHECK_TYPE(uint64_t,[and64="/uint64_t"],[and64=""],[#include<$i>])
m4_ifvaln([$1],[$1]) break
break;
done
AC_MSG_CHECKING([for stdint uint32_t])
])
])
AC_DEFUN([AX_CHECK_HEADER_STDINT_U],[
AC_CACHE_CHECK([for stdint u_int32_t], [ac_cv_header_stdint_u],[
ac_cv_header_stdint_u="" # the BSD typedefs (sys/types.h)
AC_MSG_RESULT([(..)])
for i in m4_ifval([$1],[$1],[sys/types.h inttypes.h sys/inttypes.h]) ; do
unset ac_cv_type_u_int32_t
unset ac_cv_type_u_int64_t
AC_CHECK_TYPE(u_int32_t,[ac_cv_header_stdint_u=$i],continue,[#include <$i>])
AC_CHECK_TYPE(u_int64_t,[and64="/u_int64_t"],[and64=""],[#include<$i>])
m4_ifvaln([$1],[$1]) break
break;
done
AC_MSG_CHECKING([for stdint u_int32_t])
])
])
AC_DEFUN([AX_CREATE_STDINT_H],
[# ------ AX CREATE STDINT H -------------------------------------
AC_MSG_CHECKING([for stdint types])
ac_stdint_h=`echo ifelse($1, , _stdint.h, $1)`
# try to shortcircuit - if the default include path of the compiler
# can find a "stdint.h" header then we assume that all compilers can.
AC_CACHE_VAL([ac_cv_header_stdint_t],[
old_CXXFLAGS="$CXXFLAGS" ; CXXFLAGS=""
old_CPPFLAGS="$CPPFLAGS" ; CPPFLAGS=""
old_CFLAGS="$CFLAGS" ; CFLAGS=""
AC_TRY_COMPILE([#include <stdint.h>],[int_least32_t v = 0;],
[ac_cv_stdint_result="(assuming C99 compatible system)"
ac_cv_header_stdint_t="stdint.h"; ],
[ac_cv_header_stdint_t=""])
if test "$GCC" = "yes" && test ".$ac_cv_header_stdint_t" = "."; then
CFLAGS="-std=c99"
AC_TRY_COMPILE([#include <stdint.h>],[int_least32_t v = 0;],
[AC_MSG_WARN(your GCC compiler has a defunct stdint.h for its default-mode)])
fi
CXXFLAGS="$old_CXXFLAGS"
CPPFLAGS="$old_CPPFLAGS"
CFLAGS="$old_CFLAGS" ])
v="... $ac_cv_header_stdint_h"
if test "$ac_stdint_h" = "stdint.h" ; then
AC_MSG_RESULT([(are you sure you want them in ./stdint.h?)])
elif test "$ac_stdint_h" = "inttypes.h" ; then
AC_MSG_RESULT([(are you sure you want them in ./inttypes.h?)])
elif test "_$ac_cv_header_stdint_t" = "_" ; then
AC_MSG_RESULT([(putting them into $ac_stdint_h)$v])
else
ac_cv_header_stdint="$ac_cv_header_stdint_t"
AC_MSG_RESULT([$ac_cv_header_stdint (shortcircuit)])
fi
if test "_$ac_cv_header_stdint_t" = "_" ; then # can not shortcircuit..
dnl .....intro message done, now do a few system checks.....
dnl btw, all old CHECK_TYPE macros do automatically "DEFINE" a type,
dnl therefore we use the autoconf implementation detail CHECK_TYPE_NEW
dnl instead that is triggered with 3 or more arguments (see types.m4)
inttype_headers=`echo $2 | sed -e 's/,/ /g'`
ac_cv_stdint_result="(no helpful system typedefs seen)"
AX_CHECK_HEADER_STDINT_X(dnl
stdint.h inttypes.h sys/inttypes.h $inttype_headers,
ac_cv_stdint_result="(seen uintptr_t$and64 in $i)")
if test "_$ac_cv_header_stdint_x" = "_" ; then
AX_CHECK_HEADER_STDINT_O(dnl,
inttypes.h sys/inttypes.h stdint.h $inttype_headers,
ac_cv_stdint_result="(seen uint32_t$and64 in $i)")
fi
if test "_$ac_cv_header_stdint_x" = "_" ; then
if test "_$ac_cv_header_stdint_o" = "_" ; then
AX_CHECK_HEADER_STDINT_U(dnl,
sys/types.h inttypes.h sys/inttypes.h $inttype_headers,
ac_cv_stdint_result="(seen u_int32_t$and64 in $i)")
fi fi
dnl if there was no good C99 header file, do some typedef checks...
if test "_$ac_cv_header_stdint_x" = "_" ; then
AC_MSG_CHECKING([for stdint datatype model])
AC_MSG_RESULT([(..)])
AX_CHECK_DATA_MODEL
fi
if test "_$ac_cv_header_stdint_x" != "_" ; then
ac_cv_header_stdint="$ac_cv_header_stdint_x"
elif test "_$ac_cv_header_stdint_o" != "_" ; then
ac_cv_header_stdint="$ac_cv_header_stdint_o"
elif test "_$ac_cv_header_stdint_u" != "_" ; then
ac_cv_header_stdint="$ac_cv_header_stdint_u"
else
ac_cv_header_stdint="stddef.h"
fi
AC_MSG_CHECKING([for extra inttypes in chosen header])
AC_MSG_RESULT([($ac_cv_header_stdint)])
dnl see if int_least and int_fast types are present in _this_ header.
unset ac_cv_type_int_least32_t
unset ac_cv_type_int_fast32_t
AC_CHECK_TYPE(int_least32_t,,,[#include <$ac_cv_header_stdint>])
AC_CHECK_TYPE(int_fast32_t,,,[#include<$ac_cv_header_stdint>])
AC_CHECK_TYPE(intmax_t,,,[#include <$ac_cv_header_stdint>])
fi # shortcircut to system "stdint.h"
# ------------------ PREPARE VARIABLES ------------------------------
if test "$GCC" = "yes" ; then
ac_cv_stdint_message="using gnu compiler "`$CC --version | head -1`
else
ac_cv_stdint_message="using $CC"
fi
AC_MSG_RESULT([make use of $ac_cv_header_stdint in $ac_stdint_h dnl
$ac_cv_stdint_result])
dnl -----------------------------------------------------------------
# ----------------- DONE inttypes.h checks START header -------------
AC_CONFIG_COMMANDS([$ac_stdint_h],[
AC_MSG_NOTICE(creating $ac_stdint_h : $_ac_stdint_h)
ac_stdint=$tmp/_stdint.h
echo "#ifndef" $_ac_stdint_h >$ac_stdint
echo "#define" $_ac_stdint_h "1" >>$ac_stdint
echo "#ifndef" _GENERATED_STDINT_H >>$ac_stdint
echo "#define" _GENERATED_STDINT_H '"'$PACKAGE $VERSION'"' >>$ac_stdint
echo "/* generated $ac_cv_stdint_message */" >>$ac_stdint
if test "_$ac_cv_header_stdint_t" != "_" ; then
echo "#define _STDINT_HAVE_STDINT_H" "1" >>$ac_stdint
echo "#include <stdint.h>" >>$ac_stdint
echo "#endif" >>$ac_stdint
echo "#endif" >>$ac_stdint
else
cat >>$ac_stdint <<STDINT_EOF
/* ................... shortcircuit part ........................... */
#if defined HAVE_STDINT_H || defined _STDINT_HAVE_STDINT_H
#include <stdint.h>
#else
#include <stddef.h>
/* .................... configured part ............................ */
STDINT_EOF
echo "/* whether we have a C99 compatible stdint header file */" >>$ac_stdint
if test "_$ac_cv_header_stdint_x" != "_" ; then
ac_header="$ac_cv_header_stdint_x"
echo "#define _STDINT_HEADER_INTPTR" '"'"$ac_header"'"' >>$ac_stdint
else
echo "/* #undef _STDINT_HEADER_INTPTR */" >>$ac_stdint
fi
echo "/* whether we have a C96 compatible inttypes header file */" >>$ac_stdint
if test "_$ac_cv_header_stdint_o" != "_" ; then
ac_header="$ac_cv_header_stdint_o"
echo "#define _STDINT_HEADER_UINT32" '"'"$ac_header"'"' >>$ac_stdint
else
echo "/* #undef _STDINT_HEADER_UINT32 */" >>$ac_stdint
fi
echo "/* whether we have a BSD compatible inet types header */" >>$ac_stdint
if test "_$ac_cv_header_stdint_u" != "_" ; then
ac_header="$ac_cv_header_stdint_u"
echo "#define _STDINT_HEADER_U_INT32" '"'"$ac_header"'"' >>$ac_stdint
else
echo "/* #undef _STDINT_HEADER_U_INT32 */" >>$ac_stdint
fi
echo "" >>$ac_stdint
if test "_$ac_header" != "_" ; then if test "$ac_header" != "stddef.h" ; then
echo "#include <$ac_header>" >>$ac_stdint
echo "" >>$ac_stdint
fi fi
echo "/* which 64bit typedef has been found */" >>$ac_stdint
if test "$ac_cv_type_uint64_t" = "yes" ; then
echo "#define _STDINT_HAVE_UINT64_T" "1" >>$ac_stdint
else
echo "/* #undef _STDINT_HAVE_UINT64_T */" >>$ac_stdint
fi
if test "$ac_cv_type_u_int64_t" = "yes" ; then
echo "#define _STDINT_HAVE_U_INT64_T" "1" >>$ac_stdint
else
echo "/* #undef _STDINT_HAVE_U_INT64_T */" >>$ac_stdint
fi
echo "" >>$ac_stdint
echo "/* which type model has been detected */" >>$ac_stdint
if test "_$ac_cv_char_data_model" != "_" ; then
echo "#define _STDINT_CHAR_MODEL" "$ac_cv_char_data_model" >>$ac_stdint
echo "#define _STDINT_LONG_MODEL" "$ac_cv_long_data_model" >>$ac_stdint
else
echo "/* #undef _STDINT_CHAR_MODEL // skipped */" >>$ac_stdint
echo "/* #undef _STDINT_LONG_MODEL // skipped */" >>$ac_stdint
fi
echo "" >>$ac_stdint
echo "/* whether int_least types were detected */" >>$ac_stdint
if test "$ac_cv_type_int_least32_t" = "yes"; then
echo "#define _STDINT_HAVE_INT_LEAST32_T" "1" >>$ac_stdint
else
echo "/* #undef _STDINT_HAVE_INT_LEAST32_T */" >>$ac_stdint
fi
echo "/* whether int_fast types were detected */" >>$ac_stdint
if test "$ac_cv_type_int_fast32_t" = "yes"; then
echo "#define _STDINT_HAVE_INT_FAST32_T" "1" >>$ac_stdint
else
echo "/* #undef _STDINT_HAVE_INT_FAST32_T */" >>$ac_stdint
fi
echo "/* whether intmax_t type was detected */" >>$ac_stdint
if test "$ac_cv_type_intmax_t" = "yes"; then
echo "#define _STDINT_HAVE_INTMAX_T" "1" >>$ac_stdint
else
echo "/* #undef _STDINT_HAVE_INTMAX_T */" >>$ac_stdint
fi
echo "" >>$ac_stdint
cat >>$ac_stdint <<STDINT_EOF
/* .................... detections part ............................ */
/* whether we need to define bitspecific types from compiler base types */
#ifndef _STDINT_HEADER_INTPTR
#ifndef _STDINT_HEADER_UINT32
#ifndef _STDINT_HEADER_U_INT32
#define _STDINT_NEED_INT_MODEL_T
#else
#define _STDINT_HAVE_U_INT_TYPES
#endif
#endif
#endif
#ifdef _STDINT_HAVE_U_INT_TYPES
#undef _STDINT_NEED_INT_MODEL_T
#endif
#ifdef _STDINT_CHAR_MODEL
#if _STDINT_CHAR_MODEL+0 == 122 || _STDINT_CHAR_MODEL+0 == 124
#ifndef _STDINT_BYTE_MODEL
#define _STDINT_BYTE_MODEL 12
#endif
#endif
#endif
#ifndef _STDINT_HAVE_INT_LEAST32_T
#define _STDINT_NEED_INT_LEAST_T
#endif
#ifndef _STDINT_HAVE_INT_FAST32_T
#define _STDINT_NEED_INT_FAST_T
#endif
#ifndef _STDINT_HEADER_INTPTR
#define _STDINT_NEED_INTPTR_T
#ifndef _STDINT_HAVE_INTMAX_T
#define _STDINT_NEED_INTMAX_T
#endif
#endif
/* .................... definition part ............................ */
/* some system headers have good uint64_t */
#ifndef _HAVE_UINT64_T
#if defined _STDINT_HAVE_UINT64_T || defined HAVE_UINT64_T
#define _HAVE_UINT64_T
#elif defined _STDINT_HAVE_U_INT64_T || defined HAVE_U_INT64_T
#define _HAVE_UINT64_T
typedef u_int64_t uint64_t;
#endif
#endif
#ifndef _HAVE_UINT64_T
/* .. here are some common heuristics using compiler runtime specifics */
#if defined __STDC_VERSION__ && defined __STDC_VERSION__ >= 199901L
#define _HAVE_UINT64_T
#define _HAVE_LONGLONG_UINT64_T
typedef long long int64_t;
typedef unsigned long long uint64_t;
#elif !defined __STRICT_ANSI__
#if defined _MSC_VER || defined __WATCOMC__ || defined __BORLANDC__
#define _HAVE_UINT64_T
typedef __int64 int64_t;
typedef unsigned __int64 uint64_t;
#elif defined __GNUC__ || defined __MWERKS__ || defined __ELF__
/* note: all ELF-systems seem to have loff-support which needs 64-bit */
#if !defined _NO_LONGLONG
#define _HAVE_UINT64_T
#define _HAVE_LONGLONG_UINT64_T
typedef long long int64_t;
typedef unsigned long long uint64_t;
#endif
#elif defined __alpha || (defined __mips && defined _ABIN32)
#if !defined _NO_LONGLONG
typedef long int64_t;
typedef unsigned long uint64_t;
#endif
/* compiler/cpu type to define int64_t */
#endif
#endif
#endif
#if defined _STDINT_HAVE_U_INT_TYPES
/* int8_t int16_t int32_t defined by inet code, redeclare the u_intXX types */
typedef u_int8_t uint8_t;
typedef u_int16_t uint16_t;
typedef u_int32_t uint32_t;
/* glibc compatibility */
#ifndef __int8_t_defined
#define __int8_t_defined
#endif
#endif
#ifdef _STDINT_NEED_INT_MODEL_T
/* we must guess all the basic types. Apart from byte-adressable system, */
/* there a few 32-bit-only dsp-systems that we guard with BYTE_MODEL 8-} */
/* (btw, those nibble-addressable systems are way off, or so we assume) */
dnl /* have a look at "64bit and data size neutrality" at */
dnl /* http://unix.org/version2/whatsnew/login_64bit.html */
dnl /* (the shorthand "ILP" types always have a "P" part) */
#if defined _STDINT_BYTE_MODEL
#if _STDINT_LONG_MODEL+0 == 242
/* 2:4:2 = IP16 = a normal 16-bit system */
typedef unsigned char uint8_t;
typedef unsigned short uint16_t;
typedef unsigned long uint32_t;
#ifndef __int8_t_defined
#define __int8_t_defined
typedef char int8_t;
typedef short int16_t;
typedef long int32_t;
#endif
#elif _STDINT_LONG_MODEL+0 == 244 || _STDINT_LONG_MODEL == 444
/* 2:4:4 = LP32 = a 32-bit system derived from a 16-bit */
/* 4:4:4 = ILP32 = a normal 32-bit system */
typedef unsigned char uint8_t;
typedef unsigned short uint16_t;
typedef unsigned int uint32_t;
#ifndef __int8_t_defined
#define __int8_t_defined
typedef char int8_t;
typedef short int16_t;
typedef int int32_t;
#endif
#elif _STDINT_LONG_MODEL+0 == 484 || _STDINT_LONG_MODEL+0 == 488
/* 4:8:4 = IP32 = a 32-bit system prepared for 64-bit */
/* 4:8:8 = LP64 = a normal 64-bit system */
typedef unsigned char uint8_t;
typedef unsigned short uint16_t;
typedef unsigned int uint32_t;
#ifndef __int8_t_defined
#define __int8_t_defined
typedef char int8_t;
typedef short int16_t;
typedef int int32_t;
#endif
/* this system has a "long" of 64bit */
#ifndef _HAVE_UINT64_T
#define _HAVE_UINT64_T
typedef unsigned long uint64_t;
typedef long int64_t;
#endif
#elif _STDINT_LONG_MODEL+0 == 448
/* LLP64 a 64-bit system derived from a 32-bit system */
typedef unsigned char uint8_t;
typedef unsigned short uint16_t;
typedef unsigned int uint32_t;
#ifndef __int8_t_defined
#define __int8_t_defined
typedef char int8_t;
typedef short int16_t;
typedef int int32_t;
#endif
/* assuming the system has a "long long" */
#ifndef _HAVE_UINT64_T
#define _HAVE_UINT64_T
#define _HAVE_LONGLONG_UINT64_T
typedef unsigned long long uint64_t;
typedef long long int64_t;
#endif
#else
#define _STDINT_NO_INT32_T
#endif
#else
#define _STDINT_NO_INT8_T
#define _STDINT_NO_INT32_T
#endif
#endif
/*
* quote from SunOS-5.8 sys/inttypes.h:
* Use at your own risk. As of February 1996, the committee is squarely
* behind the fixed sized types; the "least" and "fast" types are still being
* discussed. The probability that the "fast" types may be removed before
* the standard is finalized is high enough that they are not currently
* implemented.
*/
#if defined _STDINT_NEED_INT_LEAST_T
typedef int8_t int_least8_t;
typedef int16_t int_least16_t;
typedef int32_t int_least32_t;
#ifdef _HAVE_UINT64_T
typedef int64_t int_least64_t;
#endif
typedef uint8_t uint_least8_t;
typedef uint16_t uint_least16_t;
typedef uint32_t uint_least32_t;
#ifdef _HAVE_UINT64_T
typedef uint64_t uint_least64_t;
#endif
/* least types */
#endif
#if defined _STDINT_NEED_INT_FAST_T
typedef int8_t int_fast8_t;
typedef int int_fast16_t;
typedef int32_t int_fast32_t;
#ifdef _HAVE_UINT64_T
typedef int64_t int_fast64_t;
#endif
typedef uint8_t uint_fast8_t;
typedef unsigned uint_fast16_t;
typedef uint32_t uint_fast32_t;
#ifdef _HAVE_UINT64_T
typedef uint64_t uint_fast64_t;
#endif
/* fast types */
#endif
#ifdef _STDINT_NEED_INTMAX_T
#ifdef _HAVE_UINT64_T
typedef int64_t intmax_t;
typedef uint64_t uintmax_t;
#else
typedef long intmax_t;
typedef unsigned long uintmax_t;
#endif
#endif
#ifdef _STDINT_NEED_INTPTR_T
#ifndef __intptr_t_defined
#define __intptr_t_defined
/* we encourage using "long" to store pointer values, never use "int" ! */
#if _STDINT_LONG_MODEL+0 == 242 || _STDINT_LONG_MODEL+0 == 484
typedef unsigned int uintptr_t;
typedef int intptr_t;
#elif _STDINT_LONG_MODEL+0 == 244 || _STDINT_LONG_MODEL+0 == 444
typedef unsigned long uintptr_t;
typedef long intptr_t;
#elif _STDINT_LONG_MODEL+0 == 448 && defined _HAVE_UINT64_T
typedef uint64_t uintptr_t;
typedef int64_t intptr_t;
#else /* matches typical system types ILP32 and LP64 - but not IP16 or LLP64 */
typedef unsigned long uintptr_t;
typedef long intptr_t;
#endif
#endif
#endif
/* The ISO C99 standard specifies that in C++ implementations these
should only be defined if explicitly requested. */
#if !defined __cplusplus || defined __STDC_CONSTANT_MACROS
#ifndef UINT32_C
/* Signed. */
# define INT8_C(c) c
# define INT16_C(c) c
# define INT32_C(c) c
# ifdef _HAVE_LONGLONG_UINT64_T
# define INT64_C(c) c ## L
# else
# define INT64_C(c) c ## LL
# endif
/* Unsigned. */
# define UINT8_C(c) c ## U
# define UINT16_C(c) c ## U
# define UINT32_C(c) c ## U
# ifdef _HAVE_LONGLONG_UINT64_T
# define UINT64_C(c) c ## UL
# else
# define UINT64_C(c) c ## ULL
# endif
/* Maximal type. */
# ifdef _HAVE_LONGLONG_UINT64_T
# define INTMAX_C(c) c ## L
# define UINTMAX_C(c) c ## UL
# else
# define INTMAX_C(c) c ## LL
# define UINTMAX_C(c) c ## ULL
# endif
/* literalnumbers */
#endif
#endif
/* These limits are merily those of a two complement byte-oriented system */
/* Minimum of signed integral types. */
# define INT8_MIN (-128)
# define INT16_MIN (-32767-1)
# define INT32_MIN (-2147483647-1)
# define INT64_MIN (-__INT64_C(9223372036854775807)-1)
/* Maximum of signed integral types. */
# define INT8_MAX (127)
# define INT16_MAX (32767)
# define INT32_MAX (2147483647)
# define INT64_MAX (__INT64_C(9223372036854775807))
/* Maximum of unsigned integral types. */
# define UINT8_MAX (255)
# define UINT16_MAX (65535)
# define UINT32_MAX (4294967295U)
# define UINT64_MAX (__UINT64_C(18446744073709551615))
/* Minimum of signed integral types having a minimum size. */
# define INT_LEAST8_MIN INT8_MIN
# define INT_LEAST16_MIN INT16_MIN
# define INT_LEAST32_MIN INT32_MIN
# define INT_LEAST64_MIN INT64_MIN
/* Maximum of signed integral types having a minimum size. */
# define INT_LEAST8_MAX INT8_MAX
# define INT_LEAST16_MAX INT16_MAX
# define INT_LEAST32_MAX INT32_MAX
# define INT_LEAST64_MAX INT64_MAX
/* Maximum of unsigned integral types having a minimum size. */
# define UINT_LEAST8_MAX UINT8_MAX
# define UINT_LEAST16_MAX UINT16_MAX
# define UINT_LEAST32_MAX UINT32_MAX
# define UINT_LEAST64_MAX UINT64_MAX
/* shortcircuit*/
#endif
/* once */
#endif
#endif
STDINT_EOF
fi
if cmp -s $ac_stdint_h $ac_stdint 2>/dev/null; then
AC_MSG_NOTICE([$ac_stdint_h is unchanged])
else
ac_dir=`AS_DIRNAME(["$ac_stdint_h"])`
AS_MKDIR_P(["$ac_dir"])
rm -f $ac_stdint_h
mv $ac_stdint $ac_stdint_h
fi
],[# variables for create stdint.h replacement
PACKAGE="$PACKAGE"
VERSION="$VERSION"
ac_stdint_h="$ac_stdint_h"
_ac_stdint_h=AS_TR_CPP(_$PACKAGE-$ac_stdint_h)
ac_cv_stdint_message="$ac_cv_stdint_message"
ac_cv_header_stdint_t="$ac_cv_header_stdint_t"
ac_cv_header_stdint_x="$ac_cv_header_stdint_x"
ac_cv_header_stdint_o="$ac_cv_header_stdint_o"
ac_cv_header_stdint_u="$ac_cv_header_stdint_u"
ac_cv_type_uint64_t="$ac_cv_type_uint64_t"
ac_cv_type_u_int64_t="$ac_cv_type_u_int64_t"
ac_cv_char_data_model="$ac_cv_char_data_model"
ac_cv_long_data_model="$ac_cv_long_data_model"
ac_cv_type_int_least32_t="$ac_cv_type_int_least32_t"
ac_cv_type_int_fast32_t="$ac_cv_type_int_fast32_t"
ac_cv_type_intmax_t="$ac_cv_type_intmax_t"
])
])

157
nestegg/m4/pkg.m4 Normal file
View File

@@ -0,0 +1,157 @@
# pkg.m4 - Macros to locate and utilise pkg-config. -*- Autoconf -*-
#
# Copyright © 2004 Scott James Remnant <scott@netsplit.com>.
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program 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
# General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
#
# As a special exception to the GNU General Public License, if you
# distribute this file as part of a program that contains a
# configuration script generated by Autoconf, you may include it under
# the same distribution terms that you use for the rest of that program.
# PKG_PROG_PKG_CONFIG([MIN-VERSION])
# ----------------------------------
AC_DEFUN([PKG_PROG_PKG_CONFIG],
[m4_pattern_forbid([^_?PKG_[A-Z_]+$])
m4_pattern_allow([^PKG_CONFIG(_PATH)?$])
AC_ARG_VAR([PKG_CONFIG], [path to pkg-config utility])dnl
if test "x$ac_cv_env_PKG_CONFIG_set" != "xset"; then
AC_PATH_TOOL([PKG_CONFIG], [pkg-config])
fi
if test -n "$PKG_CONFIG"; then
_pkg_min_version=m4_default([$1], [0.9.0])
AC_MSG_CHECKING([pkg-config is at least version $_pkg_min_version])
if $PKG_CONFIG --atleast-pkgconfig-version $_pkg_min_version; then
AC_MSG_RESULT([yes])
else
AC_MSG_RESULT([no])
PKG_CONFIG=""
fi
fi[]dnl
])# PKG_PROG_PKG_CONFIG
# PKG_CHECK_EXISTS(MODULES, [ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND])
#
# Check to see whether a particular set of modules exists. Similar
# to PKG_CHECK_MODULES(), but does not set variables or print errors.
#
#
# Similar to PKG_CHECK_MODULES, make sure that the first instance of
# this or PKG_CHECK_MODULES is called, or make sure to call
# PKG_CHECK_EXISTS manually
# --------------------------------------------------------------
AC_DEFUN([PKG_CHECK_EXISTS],
[AC_REQUIRE([PKG_PROG_PKG_CONFIG])dnl
if test -n "$PKG_CONFIG" && \
AC_RUN_LOG([$PKG_CONFIG --exists --print-errors "$1"]); then
m4_ifval([$2], [$2], [:])
m4_ifvaln([$3], [else
$3])dnl
fi])
# _PKG_CONFIG([VARIABLE], [COMMAND], [MODULES])
# ---------------------------------------------
m4_define([_PKG_CONFIG],
[if test -n "$PKG_CONFIG"; then
if test -n "$$1"; then
pkg_cv_[]$1="$$1"
else
PKG_CHECK_EXISTS([$3],
[pkg_cv_[]$1=`$PKG_CONFIG --[]$2 "$3" 2>/dev/null`],
[pkg_failed=yes])
fi
else
pkg_failed=untried
fi[]dnl
])# _PKG_CONFIG
# _PKG_SHORT_ERRORS_SUPPORTED
# -----------------------------
AC_DEFUN([_PKG_SHORT_ERRORS_SUPPORTED],
[AC_REQUIRE([PKG_PROG_PKG_CONFIG])
if $PKG_CONFIG --atleast-pkgconfig-version 0.20; then
_pkg_short_errors_supported=yes
else
_pkg_short_errors_supported=no
fi[]dnl
])# _PKG_SHORT_ERRORS_SUPPORTED
# PKG_CHECK_MODULES(VARIABLE-PREFIX, MODULES, [ACTION-IF-FOUND],
# [ACTION-IF-NOT-FOUND])
#
#
# Note that if there is a possibility the first call to
# PKG_CHECK_MODULES might not happen, you should be sure to include an
# explicit call to PKG_PROG_PKG_CONFIG in your configure.ac
#
#
# --------------------------------------------------------------
AC_DEFUN([PKG_CHECK_MODULES],
[AC_REQUIRE([PKG_PROG_PKG_CONFIG])dnl
AC_ARG_VAR([$1][_CFLAGS], [C compiler flags for $1, overriding pkg-config])dnl
AC_ARG_VAR([$1][_LIBS], [linker flags for $1, overriding pkg-config])dnl
pkg_failed=no
AC_MSG_CHECKING([for $1])
_PKG_CONFIG([$1][_CFLAGS], [cflags], [$2])
_PKG_CONFIG([$1][_LIBS], [libs], [$2])
m4_define([_PKG_TEXT], [Alternatively, you may set the environment variables $1[]_CFLAGS
and $1[]_LIBS to avoid the need to call pkg-config.
See the pkg-config man page for more details.])
if test $pkg_failed = yes; then
_PKG_SHORT_ERRORS_SUPPORTED
if test $_pkg_short_errors_supported = yes; then
$1[]_PKG_ERRORS=`$PKG_CONFIG --short-errors --errors-to-stdout --print-errors "$2"`
else
$1[]_PKG_ERRORS=`$PKG_CONFIG --errors-to-stdout --print-errors "$2"`
fi
# Put the nasty error message in config.log where it belongs
echo "$$1[]_PKG_ERRORS" >&AS_MESSAGE_LOG_FD
ifelse([$4], , [AC_MSG_ERROR(dnl
[Package requirements ($2) were not met:
$$1_PKG_ERRORS
Consider adjusting the PKG_CONFIG_PATH environment variable if you
installed software in a non-standard prefix.
_PKG_TEXT
])],
[AC_MSG_RESULT([no])
$4])
elif test $pkg_failed = untried; then
ifelse([$4], , [AC_MSG_FAILURE(dnl
[The pkg-config script could not be found or is too old. Make sure it
is in your PATH or set the PKG_CONFIG environment variable to the full
path to pkg-config.
_PKG_TEXT
To get pkg-config, see <http://pkg-config.freedesktop.org/>.])],
[$4])
else
$1[]_CFLAGS=$pkg_cv_[]$1[]_CFLAGS
$1[]_LIBS=$pkg_cv_[]$1[]_LIBS
AC_MSG_RESULT([yes])
ifelse([$3], , :, [$3])
fi[]dnl
])# PKG_CHECK_MODULES

13
nestegg/nestegg-uninstalled.pc.in Normal file
View File

@@ -0,0 +1,13 @@
# nestegg uninstalled pkg-config file
prefix=@prefix@
exec_prefix=@exec_prefix@
libdir=@libdir@
includedir=@includedir@
Name: nestegg
Description: WebM/Matroska demuxer
Version: @VERSION@
Conflicts:
Libs: -L${libdir} -lnestegg
Cflags: -I${includedir}

13
nestegg/nestegg.pc.in Normal file
View File

@@ -0,0 +1,13 @@
# nestegg installed pkg-config file
prefix=@prefix@
exec_prefix=@exec_prefix@
libdir=@libdir@
includedir=@includedir@
Name: nestegg
Description: WebM/Matroska demuxer
Version: @VERSION@
Conflicts:
Libs: -L${libdir} -lnestegg
Cflags: -I${includedir}

1938
nestegg/src/nestegg.c Normal file
View File

File diff suppressed because it is too large Load Diff

248
nestegg/test/test.c Normal file
View File

@@ -0,0 +1,248 @@
/*
* Copyright © 2010 Mozilla Foundation
*
* This program is made available under an ISC-style license. See the
* accompanying file LICENSE for details.
*/
#include <assert.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include "nestegg/nestegg.h"
#undef DEBUG
#define SEEK_TEST
static int
stdio_read(void * p, size_t length, void * fp)
{
size_t r;
r = fread(p, length, 1, fp);
if (r == 0 && feof(fp))
return 0;
return r == 0 ? -1 : 1;
}
static int
stdio_seek(int64_t offset, int whence, void * fp)
{
return fseek(fp, offset, whence);
}
static int64_t
stdio_tell(void * fp)
{
return ftell(fp);
}
static void
log_callback(nestegg * ctx, unsigned int severity, char const * fmt, ...)
{
va_list ap;
char const * sev = NULL;
#ifndef DEBUG
if (severity < NESTEGG_LOG_WARNING)
return;
#endif
switch (severity) {
case NESTEGG_LOG_DEBUG:
sev = "debug: ";
break;
case NESTEGG_LOG_WARNING:
sev = "warning: ";
break;
case NESTEGG_LOG_CRITICAL:
sev = "critical:";
break;
default:
sev = "unknown: ";
}
fprintf(stderr, "%p %s ", (void *) ctx, sev);
va_start(ap, fmt);
vfprintf(stderr, fmt, ap);
va_end(ap);
fprintf(stderr, "\n");
}
int
main(int argc, char * argv[])
{
FILE * fp;
int r, type;
nestegg * ctx;
nestegg_audio_params aparams;
nestegg_packet * pkt;
nestegg_video_params vparams;
size_t length, size;
uint64_t duration, tstamp, pkt_tstamp;
unsigned char * codec_data, * ptr;
unsigned int cnt, i, j, track, tracks, pkt_cnt, pkt_track;
unsigned int data_items = 0;
nestegg_io io = {
stdio_read,
stdio_seek,
stdio_tell,
NULL
};
if (argc != 2)
return EXIT_FAILURE;
fp = fopen(argv[1], "rb");
if (!fp)
return EXIT_FAILURE;
io.userdata = fp;
ctx = NULL;
r = nestegg_init(&ctx, io, log_callback);
if (r != 0)
return EXIT_FAILURE;
nestegg_track_count(ctx, &tracks);
nestegg_duration(ctx, &duration);
#ifdef DEBUG
fprintf(stderr, "media has %u tracks and duration %fs\n", tracks, duration / 1e9);
#endif
for (i = 0; i < tracks; ++i) {
type = nestegg_track_type(ctx, i);
#ifdef DEBUG
fprintf(stderr, "track %u: type: %d codec: %d", i,
type, nestegg_track_codec_id(ctx, i));
#endif
nestegg_track_codec_data_count(ctx, i, &data_items);
for (j = 0; j < data_items; ++j) {
nestegg_track_codec_data(ctx, i, j, &codec_data, &length);
#ifdef DEBUG
fprintf(stderr, " (%p, %u)", codec_data, (unsigned int) length);
#endif
}
if (type == NESTEGG_TRACK_VIDEO) {
nestegg_track_video_params(ctx, i, &vparams);
#ifdef DEBUG
fprintf(stderr, " video: %ux%u (d: %ux%u %ux%ux%ux%u)",
vparams.width, vparams.height,
vparams.display_width, vparams.display_height,
vparams.crop_top, vparams.crop_left, vparams.crop_bottom, vparams.crop_right);
#endif
} else if (type == NESTEGG_TRACK_AUDIO) {
nestegg_track_audio_params(ctx, i, &aparams);
#ifdef DEBUG
fprintf(stderr, " audio: %.2fhz %u bit %u channels",
aparams.rate, aparams.depth, aparams.channels);
#endif
}
#ifdef DEBUG
fprintf(stderr, "\n");
#endif
}
#ifdef SEEK_TEST
#ifdef DEBUG
fprintf(stderr, "seek to middle\n");
#endif
r = nestegg_track_seek(ctx, 0, duration / 2);
if (r == 0) {
#ifdef DEBUG
fprintf(stderr, "middle ");
#endif
r = nestegg_read_packet(ctx, &pkt);
if (r == 1) {
nestegg_packet_track(pkt, &track);
nestegg_packet_count(pkt, &cnt);
nestegg_packet_tstamp(pkt, &tstamp);
#ifdef DEBUG
fprintf(stderr, "* t %u pts %f frames %u\n", track, tstamp / 1e9, cnt);
#endif
nestegg_free_packet(pkt);
} else {
#ifdef DEBUG
fprintf(stderr, "middle seek failed\n");
#endif
}
}
#ifdef DEBUG
fprintf(stderr, "seek to ~end\n");
#endif
r = nestegg_track_seek(ctx, 0, duration - (duration / 10));
if (r == 0) {
#ifdef DEBUG
fprintf(stderr, "end ");
#endif
r = nestegg_read_packet(ctx, &pkt);
if (r == 1) {
nestegg_packet_track(pkt, &track);
nestegg_packet_count(pkt, &cnt);
nestegg_packet_tstamp(pkt, &tstamp);
#ifdef DEBUG
fprintf(stderr, "* t %u pts %f frames %u\n", track, tstamp / 1e9, cnt);
#endif
nestegg_free_packet(pkt);
} else {
#ifdef DEBUG
fprintf(stderr, "end seek failed\n");
#endif
}
}
#ifdef DEBUG
fprintf(stderr, "seek to ~start\n");
#endif
r = nestegg_track_seek(ctx, 0, duration / 10);
if (r == 0) {
#ifdef DEBUG
fprintf(stderr, "start ");
#endif
r = nestegg_read_packet(ctx, &pkt);
if (r == 1) {
nestegg_packet_track(pkt, &track);
nestegg_packet_count(pkt, &cnt);
nestegg_packet_tstamp(pkt, &tstamp);
#ifdef DEBUG
fprintf(stderr, "* t %u pts %f frames %u\n", track, tstamp / 1e9, cnt);
#endif
nestegg_free_packet(pkt);
} else {
#ifdef DEBUG
fprintf(stderr, "start seek failed\n");
#endif
}
}
#endif
while (nestegg_read_packet(ctx, &pkt) > 0) {
nestegg_packet_track(pkt, &pkt_track);
nestegg_packet_count(pkt, &pkt_cnt);
nestegg_packet_tstamp(pkt, &pkt_tstamp);
#ifdef DEBUG
fprintf(stderr, "t %u pts %f frames %u: ", pkt_track, pkt_tstamp / 1e9, pkt_cnt);
#endif
for (i = 0; i < pkt_cnt; ++i) {
nestegg_packet_data(pkt, i, &ptr, &size);
#ifdef DEBUG
fprintf(stderr, "%u ", (unsigned int) size);
#endif
}
#ifdef DEBUG
fprintf(stderr, "\n");
#endif
nestegg_free_packet(pkt);
}
nestegg_destroy(ctx);
fclose(fp);
return EXIT_SUCCESS;
}

210
release.sh
View File

@@ -1,210 +0,0 @@
#!/bin/sh
##
## Copyright (c) 2010 The WebM project authors. All Rights Reserved.
##
## Use of this source code is governed by a BSD-style license
## that can be found in the LICENSE file in the root of the source
## tree. An additional intellectual property rights grant can be found
## in the file PATENTS. All contributing project authors may
## be found in the AUTHORS file in the root of the source tree.
##
self=$0
for opt; do
case $opt in
--clean) clean=yes;;
-j*) jopt=$opt;;
*) echo "Unsupported option $opt"; exit 1;;
esac
done
TAB="$(printf '\t')"
cat > release.mk << EOF
%\$(BUILD_SFX).tar.bz2: %/.done
${TAB}@echo "\$(subst .tar.bz2,,\$@): tarball"
${TAB}@cd \$(dir \$<); tar -cf - \$(subst .tar.bz2,,\$@) | bzip2 > ../\$@
%\$(BUILD_SFX).zip: %/.done
${TAB}@echo "\$(subst .zip,,\$@): zip"
${TAB}@rm -f \$@; cd \$(dir \$<); zip -rq ../\$@ \$(subst .zip,,\$@)
logs/%\$(BUILD_SFX).log.bz2: %/.done
${TAB}@echo "\$(subst .log.bz2,,\$(notdir \$@)): tarlog"
${TAB}@mkdir -p logs
${TAB}@cat \$< | bzip2 > \$@
%/.done:
${TAB}@mkdir -p \$(dir \$@)
${TAB}@echo "\$(dir \$@): configure \$(CONFIG_OPTS) \$(EXTRA_PATH)"
${TAB}@cd \$(dir \$@); export PATH=\$\$PATH\$(EXTRA_PATH); ../\$(SRC_ROOT)/configure \$(CONFIG_OPTS) >makelog.txt 2>&1
${TAB}@echo "\$(dir \$@): make"
${TAB}@cd \$(dir \$@); PATH=\$\$PATH\$(EXTRA_PATH) \$(MAKE) >>makelog.txt 2>&1
${TAB}@echo "\$(dir \$@): test install"
${TAB}@cd \$(dir \$@); PATH=\$\$PATH\$(EXTRA_PATH) \$(MAKE) install >>makelog.txt 2>&1
${TAB}@cd \$(dir \$@)/dist/build; PATH=\$\$PATH\$(EXTRA_PATH) \$(MAKE) >>makelog.txt 2>&1
${TAB}@echo "\$(dir \$@): install"
${TAB}@cd \$(dir \$@); PATH=\$\$PATH\$(EXTRA_PATH) \$(MAKE) install DIST_DIR=\$(TGT) >>makelog.txt 2>&1
${TAB}@touch \$@
#include release-deps.mk
EOF
#[ -f release-deps.mk ] || \
# find ${self%/*} -name .git -prune -o -type f -print0 \
# | xargs -0 -n1 echo \
# | sed -e 's; ;\\ ;g' | awk '{print "$(TGT)/.done: "$0}' > release-deps.mk
build_config_list() {
for codec in $CODEC_LIST; do
for arch in $ARCH_LIST; do
if [ -n "$OS_LIST" ]; then
for os in $OS_LIST; do
CONFIGS="$CONFIGS vpx-${codec}-${arch}-${os}"
done
else
CONFIGS="$CONFIGS vpx-${codec}-${arch}"
fi
done
done
}
CODEC_LIST="vp8 vp8cx vp8dx"
case `uname` in
Linux*)
ARCH_LIST="x86 x86_64"
OS_LIST="linux"
build_config_list
ARCH_LIST="armv5te armv6 armv7"
OS_LIST="linux-gcc"
;;
CYGWIN*)
TAR_SFX=.zip
for vs in vs7 vs8; do
for arch in x86-win32 x86_64-win64; do
for msvcrt in md mt; do
case $vs,$arch in
vs7,x86_64-win64) continue ;;
esac
ARCH_LIST="$ARCH_LIST ${arch}${msvcrt}-${vs}"
done
done
done
;;
Darwin*)
ARCH_LIST="universal"
OS_LIST="darwin8 darwin9"
;;
sun_os*)
ARCH_LIST="x86 x86_64"
OS_LIST="solaris"
;;
esac
build_config_list
TAR_SFX=${TAR_SFX:-.tar.bz2}
ARM_TOOLCHAIN=/usr/local/google/csl-2009q3-67
for cfg in $CONFIGS; do
full_cfg=$cfg
cfg=${cfg#vpx-}
opts=
rm -f makelog.txt
case $cfg in
src-*) opts="$opts --enable-codec-srcs"
cfg=${cfg#src-}
;;
eval-*) opts="$opts --enable-eval-limit"
cfg=${cfg#src-}
;;
esac
case $cfg in
#
# Linux
#
*x86-linux)
opts="$opts --target=x86-linux-gcc" ;;
*x86_64-linux)
opts="$opts --target=x86_64-linux-gcc" ;;
*arm*-linux-gcc)
armv=${cfg##*armv}
armv=${armv%%-*}
opts="$opts --target=armv${armv}-linux-gcc" ;;
*arm*-linux-rvct)
armv=${cfg##*armv}
armv=${armv%%-*}
opts="$opts --target=armv${armv}-linux-rvct"
opts="$opts --libc=${ARM_TOOLCHAIN}/arm-none-linux-gnueabi/libc" ;;
#
# Windows
#
# need --enable-debug-libs for now until we're smarter about
# building the debug/release from the customer installed
# environment
*-x86-win32*-vs*)
opts="$opts --target=x86-win32-vs${cfg##*-vs} --enable-debug-libs";;
*-x86_64-win64*-vs8)
opts="$opts --target=x86_64-win64-vs8 --enable-debug-libs" ;;
#
# Darwin
#
*-universal-darwin*)
opts="$opts --target=universal-darwin${cfg##*-darwin}-gcc" ;;
#
# Solaris
#
*x86-solaris)
opts="$opts --target=x86-solaris-gcc" ;;
*x86_64-solaris)
opts="$opts --target=x86_64-solaris-gcc" ;;
esac
case $cfg in
*x86-linux | *x86-solaris) opts="$opts --enable-pic" ;;
esac
case $cfg in
*-win[36][24]mt*) opts="$opts --enable-static-msvcrt" ;;
*-win[36][24]md*) opts="$opts --disable-static-msvcrt" ;;
esac
opts="$opts --disable-codecs"
case $cfg in
vp8*) opts="$opts --enable-vp8" ;;
esac
case $cfg in
*cx-*) opts="${opts}-encoder" ;;
*dx-*) opts="${opts}-decoder" ;;
esac
opts="$opts --enable-postproc"
[ "x${clean}" = "xyes" ] \
&& rm -rf ${full_cfg}${BUILD_SFX}${TAR_SFX} \
&& rm -rf logs/${full_cfg}${BUILD_SFX}.log.bz2
TGT=${full_cfg}${BUILD_SFX}
BUILD_TARGETS="logs/${TGT}.log.bz2 ${TGT}${TAR_SFX}"
echo "${BUILD_TARGETS}: CONFIG_OPTS=$opts" >>release.mk
echo "${BUILD_TARGETS}: TGT=${TGT}" >>release.mk
case $cfg in
*-arm*-linux-*)
echo "${BUILD_TARGETS}: EXTRA_PATH=:${ARM_TOOLCHAIN}/bin/" >>release.mk ;;
*-vs7)
echo "${BUILD_TARGETS}: EXTRA_PATH=:/cygdrive/c/Program\ Files/Microsoft\ Visual\ Studio\ .NET\ 2003/Common7/IDE" >>release.mk ;;
*-vs8)
echo "${BUILD_TARGETS}: EXTRA_PATH=:/cygdrive/c/Program\ Files/Microsoft\ Visual\ Studio\ 8/Common7/IDE" >>release.mk ;;
esac
MAKE_TGTS="$MAKE_TGTS ${TGT}${TAR_SFX} logs/${TGT}.log.bz2"
done
${MAKE:-make} ${jopt:--j3} -f release.mk \
SRC_ROOT=${self%/*} BUILD_SFX=${BUILD_SFX} ${MAKE_TGTS}

2
solution.mk
View File

@@ -22,7 +22,7 @@ else
vpx.sln: $(wildcard *.vcproj)
@echo " [CREATE] $@"
$(SRC_PATH_BARE)/build/make/gen_msvs_sln.sh \
$(if $(filter %vpx.vcproj,$^),--dep=ivfdec:vpx) \
$(if $(filter %vpx.vcproj,$^),--dep=vpxdec:vpx) \
$(if $(filter %vpx.vcproj,$^),--dep=xma:vpx) \
--ver=$(CONFIG_VS_VERSION)\
--target=$(TOOLCHAIN)\

24
tools_common.c Normal file
View File

@@ -0,0 +1,24 @@
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include <stdio.h>
#include "tools_common.h"
#ifdef _WIN32
#include <io.h>
#include <fcntl.h>
#endif
FILE* set_binary_mode(FILE *stream)
{
(void)stream;
#ifdef _WIN32
_setmode(_fileno(stream), _O_BINARY);
#endif
return stream;
}

16
tools_common.h Normal file
View File

@@ -0,0 +1,16 @@
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#ifndef TOOLS_COMMON_H
#define TOOLS_COMMON_H
/* Sets a stdio stream into binary mode */
FILE* set_binary_mode(FILE *stream);
#endif

6
vp8/common/alloccommon.c
View File

@@ -56,7 +56,7 @@ int vp8_alloc_frame_buffers(VP8_COMMON *oci, int width, int height)
vp8_de_alloc_frame_buffers(oci);
// our internal buffers are always multiples of 16
/* our internal buffers are always multiples of 16 */
if ((width & 0xf) != 0)
width += 16 - (width & 0xf);
@@ -179,10 +179,10 @@ void vp8_create_common(VP8_COMMON *oci)
oci->clr_type = REG_YUV;
oci->clamp_type = RECON_CLAMP_REQUIRED;
// Initialise reference frame sign bias structure to defaults
/* Initialise reference frame sign bias structure to defaults */
vpx_memset(oci->ref_frame_sign_bias, 0, sizeof(oci->ref_frame_sign_bias));
// Default disable buffer to buffer copying
/* Default disable buffer to buffer copying */
oci->copy_buffer_to_gf = 0;
oci->copy_buffer_to_arf = 0;
}

136
vp8/common/arm/arm_systemdependent.c Normal file
View File

@@ -0,0 +1,136 @@
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_ports/config.h"
#include "vpx_ports/arm.h"
#include "g_common.h"
#include "pragmas.h"
#include "subpixel.h"
#include "loopfilter.h"
#include "recon.h"
#include "idct.h"
#include "onyxc_int.h"
extern void (*vp8_build_intra_predictors_mby_ptr)(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby_neon(MACROBLOCKD *x);
extern void (*vp8_build_intra_predictors_mby_s_ptr)(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby_s_neon(MACROBLOCKD *x);
void vp8_arch_arm_common_init(VP8_COMMON *ctx)
{
#if CONFIG_RUNTIME_CPU_DETECT
VP8_COMMON_RTCD *rtcd = &ctx->rtcd;
int flags = arm_cpu_caps();
int has_edsp = flags & HAS_EDSP;
int has_media = flags & HAS_MEDIA;
int has_neon = flags & HAS_NEON;
rtcd->flags = flags;
/* Override default functions with fastest ones for this CPU. */
#if HAVE_ARMV6
if (has_media)
{
rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_armv6;
rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_armv6;
rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_armv6;
rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_armv6;
rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_armv6;
rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_armv6;
rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_armv6;
rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_armv6;
rtcd->idct.idct1 = vp8_short_idct4x4llm_1_v6;
rtcd->idct.idct16 = vp8_short_idct4x4llm_v6_dual;
rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_v6;
rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_v6;
rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_armv6;
rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_armv6;
rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_armv6;
rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_armv6;
rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_armv6;
rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_armv6;
rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_armv6;
rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_armv6;
rtcd->recon.copy16x16 = vp8_copy_mem16x16_v6;
rtcd->recon.copy8x8 = vp8_copy_mem8x8_v6;
rtcd->recon.copy8x4 = vp8_copy_mem8x4_v6;
rtcd->recon.recon = vp8_recon_b_armv6;
rtcd->recon.recon2 = vp8_recon2b_armv6;
rtcd->recon.recon4 = vp8_recon4b_armv6;
}
#endif
#if HAVE_ARMV7
if (has_neon)
{
rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_neon;
rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_neon;
rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_neon;
rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_neon;
rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_neon;
rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_neon;
rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_neon;
rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_neon;
rtcd->idct.idct1 = vp8_short_idct4x4llm_1_neon;
rtcd->idct.idct16 = vp8_short_idct4x4llm_neon;
rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_neon;
rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_neon;
rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_neon;
rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_neon;
rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_neon;
rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_neon;
rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_neon;
rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_neon;
rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_neon;
rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_neon;
rtcd->recon.copy16x16 = vp8_copy_mem16x16_neon;
rtcd->recon.copy8x8 = vp8_copy_mem8x8_neon;
rtcd->recon.copy8x4 = vp8_copy_mem8x4_neon;
rtcd->recon.recon = vp8_recon_b_neon;
rtcd->recon.recon2 = vp8_recon2b_neon;
rtcd->recon.recon4 = vp8_recon4b_neon;
rtcd->recon.recon_mb = vp8_recon_mb_neon;
}
#endif
#endif
#if HAVE_ARMV6
#if CONFIG_RUNTIME_CPU_DETECT
if (has_media)
#endif
{
vp8_build_intra_predictors_mby_ptr = vp8_build_intra_predictors_mby;
vp8_build_intra_predictors_mby_s_ptr = vp8_build_intra_predictors_mby_s;
}
#endif
#if HAVE_ARMV7
#if CONFIG_RUNTIME_CPU_DETECT
if (has_neon)
#endif
{
vp8_build_intra_predictors_mby_ptr =
vp8_build_intra_predictors_mby_neon;
vp8_build_intra_predictors_mby_s_ptr =
vp8_build_intra_predictors_mby_s_neon;
}
#endif
}

59
vp8/common/arm/armv6/filter_v6.asm
View File

@@ -11,6 +11,7 @@
EXPORT |vp8_filter_block2d_first_pass_armv6|
EXPORT |vp8_filter_block2d_second_pass_armv6|
EXPORT |vp8_filter4_block2d_second_pass_armv6|
EXPORT |vp8_filter_block2d_first_pass_only_armv6|
EXPORT |vp8_filter_block2d_second_pass_only_armv6|
@@ -192,6 +193,64 @@
ENDP
;---------------------------------
; r0 short *src_ptr,
; r1 unsigned char *output_ptr,
; r2 unsigned int output_pitch,
; r3 unsigned int cnt,
; stack const short *vp8_filter
;---------------------------------
|vp8_filter4_block2d_second_pass_armv6| PROC
stmdb sp!, {r4 - r11, lr}
ldr r11, [sp, #36] ; vp8_filter address
mov r7, r3, lsl #16 ; height is top part of counter
ldr r4, [r11] ; load up packed filter coefficients
add lr, r1, r3 ; save final destination pointer
ldr r5, [r11, #4]
ldr r6, [r11, #8]
pkhbt r12, r5, r4 ; pack the filter differently
pkhbt r11, r6, r5
mov r4, #0x40 ; rounding factor (for smlad{x})
|height_loop_2nd_4|
ldrd r8, [r0, #-4] ; load the data
orr r7, r7, r3, lsr #1 ; loop counter
|width_loop_2nd_4|
ldr r10, [r0, #4]!
smladx r6, r9, r12, r4 ; apply filter
pkhbt r8, r9, r8
smlad r5, r8, r12, r4
pkhbt r8, r10, r9
smladx r6, r10, r11, r6
sub r7, r7, #1
smlad r5, r8, r11, r5
mov r8, r9 ; shift the data for the next loop
mov r9, r10
usat r6, #8, r6, asr #7 ; shift and clamp
usat r5, #8, r5, asr #7
strb r5, [r1], r2 ; the result is transposed back and stored
tst r7, #0xff
strb r6, [r1], r2
bne width_loop_2nd_4
subs r7, r7, #0x10000
add r0, r0, #16 ; update src for next loop
sub r1, lr, r7, lsr #16 ; update dst for next loop
bne height_loop_2nd_4
ldmia sp!, {r4 - r11, pc}
ENDP
;------------------------------------
; r0 unsigned char *src_ptr
; r1 unsigned char *output_ptr,

22
vp8/common/arm/bilinearfilter_arm.c
View File

@@ -49,7 +49,7 @@ extern void vp8_filter_block2d_bil_second_pass_armv6
const short *vp8_filter
);
/*
#if 0
void vp8_filter_block2d_bil_first_pass_6
(
unsigned char *src_ptr,
@@ -66,14 +66,14 @@ void vp8_filter_block2d_bil_first_pass_6
{
for ( j=0; j<output_width; j++ )
{
// Apply bilinear filter
/* Apply bilinear filter */
output_ptr[j] = ( ( (int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[1] * vp8_filter[1]) +
(VP8_FILTER_WEIGHT/2) ) >> VP8_FILTER_SHIFT;
src_ptr++;
}
// Next row...
/* Next row... */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_width;
}
@@ -96,7 +96,7 @@ void vp8_filter_block2d_bil_second_pass_6
{
for ( j=0; j<output_width; j++ )
{
// Apply filter
/* Apply filter */
Temp = ((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[output_width] * vp8_filter[1]) +
(VP8_FILTER_WEIGHT/2);
@@ -104,12 +104,12 @@ void vp8_filter_block2d_bil_second_pass_6
src_ptr++;
}
// Next row...
//src_ptr += src_pixels_per_line - output_width;
/* Next row... */
/*src_ptr += src_pixels_per_line - output_width;*/
output_ptr += output_pitch;
}
}
*/
#endif
void vp8_filter_block2d_bil_armv6
(
@@ -124,13 +124,13 @@ void vp8_filter_block2d_bil_armv6
)
{
unsigned short FData[36*16]; // Temp data bufffer used in filtering
unsigned short FData[36*16]; /* Temp data bufffer used in filtering */
// First filter 1-D horizontally...
// pixel_step = 1;
/* First filter 1-D horizontally... */
/* pixel_step = 1; */
vp8_filter_block2d_bil_first_pass_armv6(src_ptr, FData, src_pixels_per_line, Height + 1, Width, HFilter);
// then 1-D vertically...
/* then 1-D vertically... */
vp8_filter_block2d_bil_second_pass_armv6(FData, output_ptr, dst_pitch, Height, Width, VFilter);
}

115
vp8/common/arm/filter_arm.c
View File

@@ -20,13 +20,13 @@
DECLARE_ALIGNED(16, static const short, sub_pel_filters[8][6]) =
{
{ 0, 0, 128, 0, 0, 0 }, // note that 1/8 pel positions are just as per alpha -0.5 bicubic
{ 0, 0, 128, 0, 0, 0 }, /* note that 1/8 pel positions are just as per alpha -0.5 bicubic */
{ 0, -6, 123, 12, -1, 0 },
{ 2, -11, 108, 36, -8, 1 }, // New 1/4 pel 6 tap filter
{ 2, -11, 108, 36, -8, 1 }, /* New 1/4 pel 6 tap filter */
{ 0, -9, 93, 50, -6, 0 },
{ 3, -16, 77, 77, -16, 3 }, // New 1/2 pel 6 tap filter
{ 3, -16, 77, 77, -16, 3 }, /* New 1/2 pel 6 tap filter */
{ 0, -6, 50, 93, -9, 0 },
{ 1, -8, 36, 108, -11, 2 }, // New 1/4 pel 6 tap filter
{ 1, -8, 36, 108, -11, 2 }, /* New 1/4 pel 6 tap filter */
{ 0, -1, 12, 123, -6, 0 },
};
@@ -50,6 +50,15 @@ extern void vp8_filter_block2d_second_pass_armv6
const short *vp8_filter
);
extern void vp8_filter4_block2d_second_pass_armv6
(
short *src_ptr,
unsigned char *output_ptr,
unsigned int output_pitch,
unsigned int cnt,
const short *vp8_filter
);
extern void vp8_filter_block2d_first_pass_only_armv6
(
unsigned char *src_ptr,
@@ -84,39 +93,43 @@ void vp8_sixtap_predict_armv6
{
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED_ARRAY(4, short, FData, 12*4); // Temp data bufffer used in filtering
DECLARE_ALIGNED_ARRAY(4, short, FData, 12*4); /* Temp data bufffer used in filtering */
HFilter = sub_pel_filters[xoffset]; // 6 tap
VFilter = sub_pel_filters[yoffset]; // 6 tap
HFilter = sub_pel_filters[xoffset]; /* 6 tap */
VFilter = sub_pel_filters[yoffset]; /* 6 tap */
// Vfilter is null. First pass only
/* Vfilter is null. First pass only */
if (xoffset && !yoffset)
{
//vp8_filter_block2d_first_pass_armv6 ( src_ptr, FData+2, src_pixels_per_line, 4, 4, HFilter );
//vp8_filter_block2d_second_pass_armv6 ( FData+2, dst_ptr, dst_pitch, 4, VFilter );
/*vp8_filter_block2d_first_pass_armv6 ( src_ptr, FData+2, src_pixels_per_line, 4, 4, HFilter );
vp8_filter_block2d_second_pass_armv6 ( FData+2, dst_ptr, dst_pitch, 4, VFilter );*/
vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, HFilter);
}
// Hfilter is null. Second pass only
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset)
{
vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 4, dst_pitch, VFilter);
}
else
{
// Vfilter is a 4 tap filter
/* Vfilter is a 4 tap filter */
if (yoffset & 0x1)
{
vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 4, 7, HFilter);
// Vfilter is 6 tap filter
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
}
/* Vfilter is 6 tap filter */
else
{
vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 4, 9, HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 4, VFilter);
}
}
}
/*
#if 0
void vp8_sixtap_predict8x4_armv6
(
unsigned char *src_ptr,
@@ -129,33 +142,33 @@ void vp8_sixtap_predict8x4_armv6
{
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED_ARRAY(4, short, FData, 16*8); // Temp data bufffer used in filtering
DECLARE_ALIGNED_ARRAY(4, short, FData, 16*8); /* Temp data bufffer used in filtering */
HFilter = sub_pel_filters[xoffset]; // 6 tap
VFilter = sub_pel_filters[yoffset]; // 6 tap
HFilter = sub_pel_filters[xoffset]; /* 6 tap */
VFilter = sub_pel_filters[yoffset]; /* 6 tap */
// if (xoffset && !yoffset)
// {
// vp8_filter_block2d_first_pass_only_armv6 ( src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, HFilter );
// }
// Hfilter is null. Second pass only
// else if (!xoffset && yoffset)
// {
// vp8_filter_block2d_second_pass_only_armv6 ( src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, VFilter );
// }
// else
// {
// if (yoffset & 0x1)
// vp8_filter_block2d_first_pass_armv6 ( src_ptr-src_pixels_per_line, FData+1, src_pixels_per_line, 8, 7, HFilter );
// else
/*if (xoffset && !yoffset)
{
vp8_filter_block2d_first_pass_only_armv6 ( src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, HFilter );
}*/
/* Hfilter is null. Second pass only */
/*else if (!xoffset && yoffset)
{
vp8_filter_block2d_second_pass_only_armv6 ( src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, VFilter );
}
else
{
if (yoffset & 0x1)
vp8_filter_block2d_first_pass_armv6 ( src_ptr-src_pixels_per_line, FData+1, src_pixels_per_line, 8, 7, HFilter );
else*/
vp8_filter_block2d_first_pass_armv6 ( src_ptr-(2*src_pixels_per_line), FData, src_pixels_per_line, 8, 9, HFilter );
vp8_filter_block2d_second_pass_armv6 ( FData+2, dst_ptr, dst_pitch, 4, 8, VFilter );
// }
/*}*/
}
*/
#endif
void vp8_sixtap_predict8x8_armv6
(
@@ -169,16 +182,16 @@ void vp8_sixtap_predict8x8_armv6
{
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED_ARRAY(4, short, FData, 16*8); // Temp data bufffer used in filtering
DECLARE_ALIGNED_ARRAY(4, short, FData, 16*8); /* Temp data bufffer used in filtering */
HFilter = sub_pel_filters[xoffset]; // 6 tap
VFilter = sub_pel_filters[yoffset]; // 6 tap
HFilter = sub_pel_filters[xoffset]; /* 6 tap */
VFilter = sub_pel_filters[yoffset]; /* 6 tap */
if (xoffset && !yoffset)
{
vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, HFilter);
}
// Hfilter is null. Second pass only
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset)
{
vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 8, dst_pitch, VFilter);
@@ -186,11 +199,15 @@ void vp8_sixtap_predict8x8_armv6
else
{
if (yoffset & 0x1)
{
vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 8, 11, HFilter);
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
}
else
{
vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 8, 13, HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 8, VFilter);
}
}
}
@@ -207,16 +224,16 @@ void vp8_sixtap_predict16x16_armv6
{
const short *HFilter;
const short *VFilter;
DECLARE_ALIGNED_ARRAY(4, short, FData, 24*16); // Temp data bufffer used in filtering
DECLARE_ALIGNED_ARRAY(4, short, FData, 24*16); /* Temp data bufffer used in filtering */
HFilter = sub_pel_filters[xoffset]; // 6 tap
VFilter = sub_pel_filters[yoffset]; // 6 tap
HFilter = sub_pel_filters[xoffset]; /* 6 tap */
VFilter = sub_pel_filters[yoffset]; /* 6 tap */
if (xoffset && !yoffset)
{
vp8_filter_block2d_first_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, HFilter);
}
// Hfilter is null. Second pass only
/* Hfilter is null. Second pass only */
else if (!xoffset && yoffset)
{
vp8_filter_block2d_second_pass_only_armv6(src_ptr, dst_ptr, src_pixels_per_line, 16, dst_pitch, VFilter);
@@ -224,11 +241,15 @@ void vp8_sixtap_predict16x16_armv6
else
{
if (yoffset & 0x1)
{
vp8_filter_block2d_first_pass_armv6(src_ptr - src_pixels_per_line, FData + 1, src_pixels_per_line, 16, 19, HFilter);
vp8_filter4_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
}
else
{
vp8_filter_block2d_first_pass_armv6(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 16, 21, HFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
vp8_filter_block2d_second_pass_armv6(FData + 2, dst_ptr, dst_pitch, 16, VFilter);
}
}
}

4
vp8/common/arm/idct_arm.h
View File

@@ -19,6 +19,7 @@ extern prototype_idct_scalar_add(vp8_dc_only_idct_add_v6);
extern prototype_second_order(vp8_short_inv_walsh4x4_1_v6);
extern prototype_second_order(vp8_short_inv_walsh4x4_v6);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_idct_idct1
#define vp8_idct_idct1 vp8_short_idct4x4llm_1_v6
@@ -34,6 +35,7 @@ extern prototype_second_order(vp8_short_inv_walsh4x4_v6);
#undef vp8_idct_iwalsh16
#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_v6
#endif
#endif
#if HAVE_ARMV7
extern prototype_idct(vp8_short_idct4x4llm_1_neon);
@@ -42,6 +44,7 @@ extern prototype_idct_scalar_add(vp8_dc_only_idct_add_neon);
extern prototype_second_order(vp8_short_inv_walsh4x4_1_neon);
extern prototype_second_order(vp8_short_inv_walsh4x4_neon);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_idct_idct1
#define vp8_idct_idct1 vp8_short_idct4x4llm_1_neon
@@ -57,5 +60,6 @@ extern prototype_second_order(vp8_short_inv_walsh4x4_neon);
#undef vp8_idct_iwalsh16
#define vp8_idct_iwalsh16 vp8_short_inv_walsh4x4_neon
#endif
#endif
#endif

20
vp8/common/arm/loopfilter_arm.c
View File

@@ -35,8 +35,8 @@ extern loop_filter_uvfunction vp8_mbloop_filter_vertical_edge_uv_neon;
#if HAVE_ARMV6
//ARMV6 loopfilter functions
// Horizontal MB filtering
/*ARMV6 loopfilter functions*/
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -60,7 +60,7 @@ void vp8_loop_filter_mbhs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsi
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
}
// Vertical MB Filtering
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -84,7 +84,7 @@ void vp8_loop_filter_mbvs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsi
vp8_loop_filter_simple_vertical_edge_armv6(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
}
// Horizontal B Filtering
/* Horizontal B Filtering */
void vp8_loop_filter_bh_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -112,7 +112,7 @@ void vp8_loop_filter_bhs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsig
vp8_loop_filter_simple_horizontal_edge_armv6(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
}
// Vertical B Filtering
/* Vertical B Filtering */
void vp8_loop_filter_bv_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -142,8 +142,8 @@ void vp8_loop_filter_bvs_armv6(unsigned char *y_ptr, unsigned char *u_ptr, unsig
#endif
#if HAVE_ARMV7
// NEON loopfilter functions
// Horizontal MB filtering
/* NEON loopfilter functions */
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -164,7 +164,7 @@ void vp8_loop_filter_mbhs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsig
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
}
// Vertical MB Filtering
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -185,7 +185,7 @@ void vp8_loop_filter_mbvs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsig
vp8_loop_filter_simple_vertical_edge_neon(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
}
// Horizontal B Filtering
/* Horizontal B Filtering */
void vp8_loop_filter_bh_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -210,7 +210,7 @@ void vp8_loop_filter_bhs_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsign
vp8_loop_filter_simple_horizontal_edge_neon(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
}
// Vertical B Filtering
/* Vertical B Filtering */
void vp8_loop_filter_bv_neon(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{

4
vp8/common/arm/loopfilter_arm.h
View File

@@ -22,6 +22,7 @@ extern prototype_loopfilter_block(vp8_loop_filter_bvs_armv6);
extern prototype_loopfilter_block(vp8_loop_filter_mbhs_armv6);
extern prototype_loopfilter_block(vp8_loop_filter_bhs_armv6);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_lf_normal_mb_v
#define vp8_lf_normal_mb_v vp8_loop_filter_mbv_armv6
@@ -46,6 +47,7 @@ extern prototype_loopfilter_block(vp8_loop_filter_bhs_armv6);
#undef vp8_lf_simple_b_h
#define vp8_lf_simple_b_h vp8_loop_filter_bhs_armv6
#endif
#endif
#if HAVE_ARMV7
extern prototype_loopfilter_block(vp8_loop_filter_mbv_neon);
@@ -57,6 +59,7 @@ extern prototype_loopfilter_block(vp8_loop_filter_bvs_neon);
extern prototype_loopfilter_block(vp8_loop_filter_mbhs_neon);
extern prototype_loopfilter_block(vp8_loop_filter_bhs_neon);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_lf_normal_mb_v
#define vp8_lf_normal_mb_v vp8_loop_filter_mbv_neon
@@ -81,5 +84,6 @@ extern prototype_loopfilter_block(vp8_loop_filter_bhs_neon);
#undef vp8_lf_simple_b_h
#define vp8_lf_simple_b_h vp8_loop_filter_bhs_neon
#endif
#endif
#endif

409
vp8/common/arm/neon/loopfilter_neon.asm Normal file
View File

@@ -0,0 +1,409 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_loop_filter_horizontal_edge_y_neon|
EXPORT |vp8_loop_filter_horizontal_edge_uv_neon|
EXPORT |vp8_loop_filter_vertical_edge_y_neon|
EXPORT |vp8_loop_filter_vertical_edge_uv_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
; flimit, limit, and thresh should be positive numbers.
; All 16 elements in these variables are equal.
; void vp8_loop_filter_horizontal_edge_y_neon(unsigned char *src, int pitch,
; const signed char *flimit,
; const signed char *limit,
; const signed char *thresh,
; int count)
; r0 unsigned char *src
; r1 int pitch
; r2 const signed char *flimit
; r3 const signed char *limit
; sp const signed char *thresh,
; sp+4 int count (unused)
|vp8_loop_filter_horizontal_edge_y_neon| PROC
stmdb sp!, {lr}
vld1.s8 {d0[], d1[]}, [r2] ; flimit
vld1.s8 {d2[], d3[]}, [r3] ; limit
sub r2, r0, r1, lsl #2 ; move src pointer down by 4 lines
ldr r12, [sp, #4] ; load thresh pointer
vld1.u8 {q3}, [r2], r1 ; p3
vld1.u8 {q4}, [r2], r1 ; p2
vld1.u8 {q5}, [r2], r1 ; p1
vld1.u8 {q6}, [r2], r1 ; p0
vld1.u8 {q7}, [r2], r1 ; q0
vld1.u8 {q8}, [r2], r1 ; q1
vld1.u8 {q9}, [r2], r1 ; q2
vld1.u8 {q10}, [r2] ; q3
vld1.s8 {d4[], d5[]}, [r12] ; thresh
sub r0, r0, r1, lsl #1
bl vp8_loop_filter_neon
vst1.u8 {q5}, [r0], r1 ; store op1
vst1.u8 {q6}, [r0], r1 ; store op0
vst1.u8 {q7}, [r0], r1 ; store oq0
vst1.u8 {q8}, [r0], r1 ; store oq1
ldmia sp!, {pc}
ENDP ; |vp8_loop_filter_horizontal_edge_y_neon|
; void vp8_loop_filter_horizontal_edge_uv_neon(unsigned char *u, int pitch
; const signed char *flimit,
; const signed char *limit,
; const signed char *thresh,
; unsigned char *v)
; r0 unsigned char *u,
; r1 int pitch,
; r2 const signed char *flimit,
; r3 const signed char *limit,
; sp const signed char *thresh,
; sp+4 unsigned char *v
|vp8_loop_filter_horizontal_edge_uv_neon| PROC
stmdb sp!, {lr}
vld1.s8 {d0[], d1[]}, [r2] ; flimit
vld1.s8 {d2[], d3[]}, [r3] ; limit
ldr r2, [sp, #8] ; load v ptr
sub r3, r0, r1, lsl #2 ; move u pointer down by 4 lines
vld1.u8 {d6}, [r3], r1 ; p3
vld1.u8 {d8}, [r3], r1 ; p2
vld1.u8 {d10}, [r3], r1 ; p1
vld1.u8 {d12}, [r3], r1 ; p0
vld1.u8 {d14}, [r3], r1 ; q0
vld1.u8 {d16}, [r3], r1 ; q1
vld1.u8 {d18}, [r3], r1 ; q2
vld1.u8 {d20}, [r3] ; q3
ldr r3, [sp, #4] ; load thresh pointer
sub r12, r2, r1, lsl #2 ; move v pointer down by 4 lines
vld1.u8 {d7}, [r12], r1 ; p3
vld1.u8 {d9}, [r12], r1 ; p2
vld1.u8 {d11}, [r12], r1 ; p1
vld1.u8 {d13}, [r12], r1 ; p0
vld1.u8 {d15}, [r12], r1 ; q0
vld1.u8 {d17}, [r12], r1 ; q1
vld1.u8 {d19}, [r12], r1 ; q2
vld1.u8 {d21}, [r12] ; q3
vld1.s8 {d4[], d5[]}, [r3] ; thresh
bl vp8_loop_filter_neon
sub r0, r0, r1, lsl #1
sub r2, r2, r1, lsl #1
vst1.u8 {d10}, [r0], r1 ; store u op1
vst1.u8 {d11}, [r2], r1 ; store v op1
vst1.u8 {d12}, [r0], r1 ; store u op0
vst1.u8 {d13}, [r2], r1 ; store v op0
vst1.u8 {d14}, [r0], r1 ; store u oq0
vst1.u8 {d15}, [r2], r1 ; store v oq0
vst1.u8 {d16}, [r0] ; store u oq1
vst1.u8 {d17}, [r2] ; store v oq1
ldmia sp!, {pc}
ENDP ; |vp8_loop_filter_horizontal_edge_uv_neon|
; void vp8_loop_filter_vertical_edge_y_neon(unsigned char *src, int pitch,
; const signed char *flimit,
; const signed char *limit,
; const signed char *thresh,
; int count)
; r0 unsigned char *src,
; r1 int pitch,
; r2 const signed char *flimit,
; r3 const signed char *limit,
; sp const signed char *thresh,
; sp+4 int count (unused)
|vp8_loop_filter_vertical_edge_y_neon| PROC
stmdb sp!, {lr}
vld1.s8 {d0[], d1[]}, [r2] ; flimit
vld1.s8 {d2[], d3[]}, [r3] ; limit
sub r2, r0, #4 ; src ptr down by 4 columns
sub r0, r0, #2 ; dst ptr
ldr r12, [sp, #4] ; load thresh pointer
vld1.u8 {d6}, [r2], r1 ; load first 8-line src data
vld1.u8 {d8}, [r2], r1
vld1.u8 {d10}, [r2], r1
vld1.u8 {d12}, [r2], r1
vld1.u8 {d14}, [r2], r1
vld1.u8 {d16}, [r2], r1
vld1.u8 {d18}, [r2], r1
vld1.u8 {d20}, [r2], r1
vld1.s8 {d4[], d5[]}, [r12] ; thresh
vld1.u8 {d7}, [r2], r1 ; load second 8-line src data
vld1.u8 {d9}, [r2], r1
vld1.u8 {d11}, [r2], r1
vld1.u8 {d13}, [r2], r1
vld1.u8 {d15}, [r2], r1
vld1.u8 {d17}, [r2], r1
vld1.u8 {d19}, [r2], r1
vld1.u8 {d21}, [r2]
;transpose to 8x16 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
bl vp8_loop_filter_neon
vswp d12, d11
vswp d16, d13
vswp d14, d12
vswp d16, d15
;store op1, op0, oq0, oq1
vst4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r1
vst4.8 {d10[1], d11[1], d12[1], d13[1]}, [r0], r1
vst4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r1
vst4.8 {d10[3], d11[3], d12[3], d13[3]}, [r0], r1
vst4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r1
vst4.8 {d10[5], d11[5], d12[5], d13[5]}, [r0], r1
vst4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r1
vst4.8 {d10[7], d11[7], d12[7], d13[7]}, [r0], r1
vst4.8 {d14[0], d15[0], d16[0], d17[0]}, [r0], r1
vst4.8 {d14[1], d15[1], d16[1], d17[1]}, [r0], r1
vst4.8 {d14[2], d15[2], d16[2], d17[2]}, [r0], r1
vst4.8 {d14[3], d15[3], d16[3], d17[3]}, [r0], r1
vst4.8 {d14[4], d15[4], d16[4], d17[4]}, [r0], r1
vst4.8 {d14[5], d15[5], d16[5], d17[5]}, [r0], r1
vst4.8 {d14[6], d15[6], d16[6], d17[6]}, [r0], r1
vst4.8 {d14[7], d15[7], d16[7], d17[7]}, [r0]
ldmia sp!, {pc}
ENDP ; |vp8_loop_filter_vertical_edge_y_neon|
; void vp8_loop_filter_vertical_edge_uv_neon(unsigned char *u, int pitch
; const signed char *flimit,
; const signed char *limit,
; const signed char *thresh,
; unsigned char *v)
; r0 unsigned char *u,
; r1 int pitch,
; r2 const signed char *flimit,
; r3 const signed char *limit,
; sp const signed char *thresh,
; sp+4 unsigned char *v
|vp8_loop_filter_vertical_edge_uv_neon| PROC
stmdb sp!, {lr}
sub r12, r0, #4 ; move u pointer down by 4 columns
vld1.s8 {d0[], d1[]}, [r2] ; flimit
vld1.s8 {d2[], d3[]}, [r3] ; limit
ldr r2, [sp, #8] ; load v ptr
vld1.u8 {d6}, [r12], r1 ;load u data
vld1.u8 {d8}, [r12], r1
vld1.u8 {d10}, [r12], r1
vld1.u8 {d12}, [r12], r1
vld1.u8 {d14}, [r12], r1
vld1.u8 {d16}, [r12], r1
vld1.u8 {d18}, [r12], r1
vld1.u8 {d20}, [r12]
sub r3, r2, #4 ; move v pointer down by 4 columns
vld1.u8 {d7}, [r3], r1 ;load v data
vld1.u8 {d9}, [r3], r1
vld1.u8 {d11}, [r3], r1
vld1.u8 {d13}, [r3], r1
vld1.u8 {d15}, [r3], r1
vld1.u8 {d17}, [r3], r1
vld1.u8 {d19}, [r3], r1
vld1.u8 {d21}, [r3]
ldr r12, [sp, #4] ; load thresh pointer
;transpose to 8x16 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
vld1.s8 {d4[], d5[]}, [r12] ; thresh
bl vp8_loop_filter_neon
sub r0, r0, #2
sub r2, r2, #2
vswp d12, d11
vswp d16, d13
vswp d14, d12
vswp d16, d15
;store op1, op0, oq0, oq1
vst4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r1
vst4.8 {d14[0], d15[0], d16[0], d17[0]}, [r2], r1
vst4.8 {d10[1], d11[1], d12[1], d13[1]}, [r0], r1
vst4.8 {d14[1], d15[1], d16[1], d17[1]}, [r2], r1
vst4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r1
vst4.8 {d14[2], d15[2], d16[2], d17[2]}, [r2], r1
vst4.8 {d10[3], d11[3], d12[3], d13[3]}, [r0], r1
vst4.8 {d14[3], d15[3], d16[3], d17[3]}, [r2], r1
vst4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r1
vst4.8 {d14[4], d15[4], d16[4], d17[4]}, [r2], r1
vst4.8 {d10[5], d11[5], d12[5], d13[5]}, [r0], r1
vst4.8 {d14[5], d15[5], d16[5], d17[5]}, [r2], r1
vst4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r1
vst4.8 {d14[6], d15[6], d16[6], d17[6]}, [r2], r1
vst4.8 {d10[7], d11[7], d12[7], d13[7]}, [r0]
vst4.8 {d14[7], d15[7], d16[7], d17[7]}, [r2]
ldmia sp!, {pc}
ENDP ; |vp8_loop_filter_vertical_edge_uv_neon|
; void vp8_loop_filter_neon();
; This is a helper function for the loopfilters. The invidual functions do the
; necessary load, transpose (if necessary) and store.
; r0-r3 PRESERVE
; q0 flimit
; q1 limit
; q2 thresh
; q3 p3
; q4 p2
; q5 p1
; q6 p0
; q7 q0
; q8 q1
; q9 q2
; q10 q3
|vp8_loop_filter_neon| PROC
ldr r12, _lf_coeff_
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q4, q10, q9 ; abs(q3 - q2)
vabd.u8 q9, q6, q7 ; abs(p0 - q0)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q4
vmax.u8 q15, q11, q12
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vadd.u8 q0, q0, q0 ; flimit * 2
vadd.u8 q0, q0, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q2, q5, q8 ; a = abs(p1 - q1)
vqadd.u8 q9, q9, q9 ; b = abs(p0 - q0) * 2
vshr.u8 q2, q2, #1 ; a = a / 2
vqadd.u8 q9, q9, q2 ; a = b + a
vcge.u8 q9, q0, q9 ; (a > flimit * 2 + limit) * -1
vld1.u8 {q0}, [r12]!
; vp8_filter() function
; convert to signed
veor q7, q7, q0 ; qs0
veor q6, q6, q0 ; ps0
veor q5, q5, q0 ; ps1
veor q8, q8, q0 ; qs1
vld1.u8 {q10}, [r12]!
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q11, d15, d13
vmovl.u8 q4, d20
vqsub.s8 q1, q5, q8 ; vp8_filter = clamp(ps1-qs1)
vorr q14, q13, q14 ; vp8_hevmask
vmul.i16 q2, q2, q4 ; 3 * ( qs0 - ps0)
vmul.i16 q11, q11, q4
vand q1, q1, q14 ; vp8_filter &= hev
vand q15, q15, q9 ; vp8_filter_mask
vaddw.s8 q2, q2, d2
vaddw.s8 q11, q11, d3
vld1.u8 {q9}, [r12]!
; vp8_filter = clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d2, q2
vqmovn.s16 d3, q11
vand q1, q1, q15 ; vp8_filter &= mask
vqadd.s8 q2, q1, q10 ; Filter2 = clamp(vp8_filter+3)
vqadd.s8 q1, q1, q9 ; Filter1 = clamp(vp8_filter+4)
vshr.s8 q2, q2, #3 ; Filter2 >>= 3
vshr.s8 q1, q1, #3 ; Filter1 >>= 3
vqadd.s8 q11, q6, q2 ; u = clamp(ps0 + Filter2)
vqsub.s8 q10, q7, q1 ; u = clamp(qs0 - Filter1)
; outer tap adjustments: ++vp8_filter >> 1
vrshr.s8 q1, q1, #1
vbic q1, q1, q14 ; vp8_filter &= ~hev
vqadd.s8 q13, q5, q1 ; u = clamp(ps1 + vp8_filter)
vqsub.s8 q12, q8, q1 ; u = clamp(qs1 - vp8_filter)
veor q5, q13, q0 ; *op1 = u^0x80
veor q6, q11, q0 ; *op0 = u^0x80
veor q7, q10, q0 ; *oq0 = u^0x80
veor q8, q12, q0 ; *oq1 = u^0x80
bx lr
ENDP ; |vp8_loop_filter_horizontal_edge_y_neon|
AREA loopfilter_dat, DATA, READONLY
_lf_coeff_
DCD lf_coeff
lf_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x01010101, 0x01010101, 0x01010101, 0x01010101
END

178
vp8/common/arm/neon/loopfilterhorizontaledge_uv_neon.asm
View File

@@ -1,178 +0,0 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_loop_filter_horizontal_edge_uv_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
;are equal. So, in the code, only one load is needed
;for flimit. Same way applies to limit and thresh.
; r0 unsigned char *u,
; r1 int p, //pitch
; r2 const signed char *flimit,
; r3 const signed char *limit,
; stack(r4) const signed char *thresh,
; stack(r5) unsigned char *v
|vp8_loop_filter_horizontal_edge_uv_neon| PROC
sub r0, r0, r1, lsl #2 ; move u pointer down by 4 lines
vld1.s8 {d0[], d1[]}, [r2] ; flimit
ldr r2, [sp, #4] ; load v ptr
ldr r12, [sp, #0] ; load thresh pointer
sub r2, r2, r1, lsl #2 ; move v pointer down by 4 lines
vld1.u8 {d6}, [r0], r1 ; p3
vld1.u8 {d7}, [r2], r1 ; p3
vld1.u8 {d8}, [r0], r1 ; p2
vld1.u8 {d9}, [r2], r1 ; p2
vld1.u8 {d10}, [r0], r1 ; p1
vld1.u8 {d11}, [r2], r1 ; p1
vld1.u8 {d12}, [r0], r1 ; p0
vld1.u8 {d13}, [r2], r1 ; p0
vld1.u8 {d14}, [r0], r1 ; q0
vld1.u8 {d15}, [r2], r1 ; q0
vld1.u8 {d16}, [r0], r1 ; q1
vld1.u8 {d17}, [r2], r1 ; q1
vld1.u8 {d18}, [r0], r1 ; q2
vld1.u8 {d19}, [r2], r1 ; q2
vld1.u8 {d20}, [r0], r1 ; q3
vld1.u8 {d21}, [r2], r1 ; q3
vld1.s8 {d2[], d3[]}, [r3] ; limit
vld1.s8 {d4[], d5[]}, [r12] ; thresh
ldr r12, _lfhuv_coeff_
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q4, q10, q9 ; abs(q3 - q2)
vabd.u8 q9, q6, q7 ; abs(p0 - q0)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q4
vmax.u8 q15, q11, q12
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vadd.u8 q0, q0, q0 ; flimit * 2
vadd.u8 q0, q0, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15 ; (max > limit) * -1
vabd.u8 q2, q5, q8 ; abs(p1 - q1)
vqadd.u8 q9, q9, q9 ; abs(p0 - q0) * 2
vshr.u8 q2, q2, #1 ; abs(p1 - q1) / 2
vqadd.u8 q9, q9, q2 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
vcge.u8 q9, q0, q9 ; (abs(p0 - q0)*2 + abs(p1-q1)/2 > flimit*2 + limit)*-1
vld1.u8 {q0}, [r12]!
;vp8_filter() function
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
;;;;;;;;;;;;;;
vld1.u8 {q10}, [r12]!
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q11, d15, d13
vmovl.u8 q4, d20
vqsub.s8 q1, q5, q8 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vorr q14, q13, q14 ; q14: vp8_hevmask
vmul.i16 q2, q2, q4 ; 3 * ( qs0 - ps0)
vmul.i16 q11, q11, q4
vand q1, q1, q14 ; vp8_filter &= hev
vand q15, q15, q9 ; vp8_filter_mask
vaddw.s8 q2, q2, d2
vaddw.s8 q11, q11, d3
vld1.u8 {q9}, [r12]!
;
vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d3, q11
;;
vand q1, q1, q15 ; vp8_filter &= mask
vqadd.s8 q2, q1, q10 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
vqadd.s8 q1, q1, q9 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
vshr.s8 q2, q2, #3 ; Filter2 >>= 3
vshr.s8 q1, q1, #3 ; Filter1 >>= 3
;calculate output
vqadd.s8 q11, q6, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
vqsub.s8 q10, q7, q1 ; u = vp8_signed_char_clamp(qs0 - Filter1)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vrshr.s8 q1, q1, #1 ;round/shift: vp8_filter += 1; vp8_filter >>= 1
sub r0, r0, r1, lsl #2
sub r0, r0, r1, lsl #1
;
vbic q1, q1, q14 ; vp8_filter &= ~hev
sub r2, r2, r1, lsl #2
sub r2, r2, r1, lsl #1
;;
vqadd.s8 q13, q5, q1 ; u = vp8_signed_char_clamp(ps1 + vp8_filter)
vqsub.s8 q12, q8, q1 ; u = vp8_signed_char_clamp(qs1 - vp8_filter)
;
veor q5, q13, q0 ; *op1 = u^0x80
veor q6, q11, q0 ; *op0 = u^0x80
veor q7, q10, q0 ; *oq0 = u^0x80
veor q8, q12, q0 ; *oq1 = u^0x80
;
vst1.u8 {d10}, [r0], r1 ; store u op1
vst1.u8 {d11}, [r2], r1 ; store v op1
vst1.u8 {d12}, [r0], r1 ; store u op0
vst1.u8 {d13}, [r2], r1 ; store v op0
vst1.u8 {d14}, [r0], r1 ; store u oq0
vst1.u8 {d15}, [r2], r1 ; store v oq0
vst1.u8 {d16}, [r0], r1 ; store u oq1
vst1.u8 {d17}, [r2], r1 ; store v oq1
bx lr
ENDP ; |vp8_loop_filter_horizontal_edge_uv_neon|
;-----------------
AREA hloopfilteruv_dat, DATA, READWRITE ;read/write by default
;Data section with name data_area is specified. DCD reserves space in memory for 16 data.
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
_lfhuv_coeff_
DCD lfhuv_coeff
lfhuv_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x01010101, 0x01010101, 0x01010101, 0x01010101
END

161
vp8/common/arm/neon/loopfilterhorizontaledge_y_neon.asm
View File

@@ -1,161 +0,0 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_loop_filter_horizontal_edge_y_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
;are equal. So, in the code, only one load is needed
;for flimit. Same way applies to limit and thresh.
; r0 unsigned char *s,
; r1 int p, //pitch
; r2 const signed char *flimit,
; r3 const signed char *limit,
; stack(r4) const signed char *thresh,
; //stack(r5) int count --unused
|vp8_loop_filter_horizontal_edge_y_neon| PROC
sub r0, r0, r1, lsl #2 ; move src pointer down by 4 lines
ldr r12, [sp, #0] ; load thresh pointer
vld1.u8 {q3}, [r0], r1 ; p3
vld1.s8 {d0[], d1[]}, [r2] ; flimit
vld1.u8 {q4}, [r0], r1 ; p2
vld1.s8 {d2[], d3[]}, [r3] ; limit
vld1.u8 {q5}, [r0], r1 ; p1
vld1.s8 {d4[], d5[]}, [r12] ; thresh
vld1.u8 {q6}, [r0], r1 ; p0
ldr r12, _lfhy_coeff_
vld1.u8 {q7}, [r0], r1 ; q0
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vld1.u8 {q8}, [r0], r1 ; q1
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vld1.u8 {q9}, [r0], r1 ; q2
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vld1.u8 {q10}, [r0], r1 ; q3
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q4, q10, q9 ; abs(q3 - q2)
vabd.u8 q9, q6, q7 ; abs(p0 - q0)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q4
vmax.u8 q15, q11, q12
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vadd.u8 q0, q0, q0 ; flimit * 2
vadd.u8 q0, q0, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q2, q5, q8 ; abs(p1 - q1)
vqadd.u8 q9, q9, q9 ; abs(p0 - q0) * 2
vshr.u8 q2, q2, #1 ; abs(p1 - q1) / 2
vqadd.u8 q9, q9, q2 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
vcge.u8 q9, q0, q9 ; (abs(p0 - q0)*2 + abs(p1-q1)/2 > flimit*2 + limit)*-1
vld1.u8 {q0}, [r12]!
;vp8_filter() function
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
;;;;;;;;;;;;;;
vld1.u8 {q10}, [r12]!
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q11, d15, d13
vmovl.u8 q4, d20
vqsub.s8 q1, q5, q8 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vorr q14, q13, q14 ; q14: vp8_hevmask
vmul.i16 q2, q2, q4 ; 3 * ( qs0 - ps0)
vmul.i16 q11, q11, q4
vand q1, q1, q14 ; vp8_filter &= hev
vand q15, q15, q9 ; vp8_filter_mask
vaddw.s8 q2, q2, d2
vaddw.s8 q11, q11, d3
vld1.u8 {q9}, [r12]!
;
vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d3, q11
;;
vand q1, q1, q15 ; vp8_filter &= mask
vqadd.s8 q2, q1, q10 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
vqadd.s8 q1, q1, q9 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
vshr.s8 q2, q2, #3 ; Filter2 >>= 3
vshr.s8 q1, q1, #3 ; Filter1 >>= 3
;calculate output
vqadd.s8 q11, q6, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
vqsub.s8 q10, q7, q1 ; u = vp8_signed_char_clamp(qs0 - Filter1)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vrshr.s8 q1, q1, #1 ;round/shift: vp8_filter += 1; vp8_filter >>= 1
sub r0, r0, r1, lsl #2
sub r0, r0, r1, lsl #1
;
vbic q1, q1, q14 ; vp8_filter &= ~hev
;
add r2, r1, r0
vqadd.s8 q13, q5, q1 ; u = vp8_signed_char_clamp(ps1 + vp8_filter)
vqsub.s8 q12, q8, q1 ; u = vp8_signed_char_clamp(qs1 - vp8_filter)
add r3, r2, r1
veor q5, q13, q0 ; *op1 = u^0x80
veor q6, q11, q0 ; *op0 = u^0x80
veor q7, q10, q0 ; *oq0 = u^0x80
veor q8, q12, q0 ; *oq1 = u^0x80
add r12, r3, r1
vst1.u8 {q5}, [r0] ; store op1
vst1.u8 {q6}, [r2] ; store op0
vst1.u8 {q7}, [r3] ; store oq0
vst1.u8 {q8}, [r12] ; store oq1
bx lr
ENDP ; |vp8_loop_filter_horizontal_edge_y_neon|
;-----------------
AREA hloopfiltery_dat, DATA, READWRITE ;read/write by default
;Data section with name data_area is specified. DCD reserves space in memory for 16 data.
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
_lfhy_coeff_
DCD lfhy_coeff
lfhy_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x01010101, 0x01010101, 0x01010101, 0x01010101
END

203
vp8/common/arm/neon/loopfilterverticaledge_uv_neon.asm
View File

@@ -1,203 +0,0 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_loop_filter_vertical_edge_uv_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
;are equal. So, in the code, only one load is needed
;for flimit. Same way applies to limit and thresh.
; r0 unsigned char *u,
; r1 int p, //pitch
; r2 const signed char *flimit,
; r3 const signed char *limit,
; stack(r4) const signed char *thresh,
; stack(r5) unsigned char *v
|vp8_loop_filter_vertical_edge_uv_neon| PROC
sub r0, r0, #4 ; move u pointer down by 4 columns
vld1.s8 {d0[], d1[]}, [r2] ; flimit
ldr r2, [sp, #4] ; load v ptr
ldr r12, [sp, #0] ; load thresh pointer
sub r2, r2, #4 ; move v pointer down by 4 columns
vld1.u8 {d6}, [r0], r1 ;load u data
vld1.u8 {d7}, [r2], r1 ;load v data
vld1.u8 {d8}, [r0], r1
vld1.u8 {d9}, [r2], r1
vld1.u8 {d10}, [r0], r1
vld1.u8 {d11}, [r2], r1
vld1.u8 {d12}, [r0], r1
vld1.u8 {d13}, [r2], r1
vld1.u8 {d14}, [r0], r1
vld1.u8 {d15}, [r2], r1
vld1.u8 {d16}, [r0], r1
vld1.u8 {d17}, [r2], r1
vld1.u8 {d18}, [r0], r1
vld1.u8 {d19}, [r2], r1
vld1.u8 {d20}, [r0], r1
vld1.u8 {d21}, [r2], r1
;transpose to 8x16 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
vld1.s8 {d2[], d3[]}, [r3] ; limit
vld1.s8 {d4[], d5[]}, [r12] ; thresh
ldr r12, _vlfuv_coeff_
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q4, q10, q9 ; abs(q3 - q2)
vabd.u8 q9, q6, q7 ; abs(p0 - q0)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q4
vmax.u8 q15, q11, q12
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vadd.u8 q0, q0, q0 ; flimit * 2
vadd.u8 q0, q0, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q2, q5, q8 ; abs(p1 - q1)
vqadd.u8 q9, q9, q9 ; abs(p0 - q0) * 2
vshr.u8 q2, q2, #1 ; abs(p1 - q1) / 2
vqadd.u8 q9, q9, q2 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
vcge.u8 q9, q0, q9 ; (abs(p0 - q0)*2 + abs(p1-q1)/2 > flimit*2 + limit)*-1
vld1.u8 {q0}, [r12]!
;vp8_filter() function
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
;;;;;;;;;;;;;;
vld1.u8 {q10}, [r12]!
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q11, d15, d13
vmovl.u8 q4, d20
vqsub.s8 q1, q5, q8 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vorr q14, q13, q14 ; q14: vp8_hevmask
vmul.i16 q2, q2, q4 ; 3 * ( qs0 - ps0)
vmul.i16 q11, q11, q4
vand q1, q1, q14 ; vp8_filter &= hev
vand q15, q15, q9 ; vp8_filter_mask
vaddw.s8 q2, q2, d2
vaddw.s8 q11, q11, d3
vld1.u8 {q9}, [r12]!
;
vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d3, q11
;;
vand q1, q1, q15 ; vp8_filter &= mask
vqadd.s8 q2, q1, q10 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
vqadd.s8 q1, q1, q9 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
vshr.s8 q2, q2, #3 ; Filter2 >>= 3
vshr.s8 q1, q1, #3 ; Filter1 >>= 3
;calculate output
vqadd.s8 q11, q6, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
vqsub.s8 q10, q7, q1 ; u = vp8_signed_char_clamp(qs0 - Filter1)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vrshr.s8 q1, q1, #1 ;round/shift: vp8_filter += 1; vp8_filter >>= 1
sub r0, r0, r1, lsl #3
add r0, r0, #2
vbic q1, q1, q14 ; vp8_filter &= ~hev
sub r2, r2, r1, lsl #3
add r2, r2, #2
vqadd.s8 q13, q5, q1 ; u = vp8_signed_char_clamp(ps1 + vp8_filter)
vqsub.s8 q12, q8, q1 ; u = vp8_signed_char_clamp(qs1 - vp8_filter)
veor q7, q10, q0 ; *oq0 = u^0x80
veor q5, q13, q0 ; *op1 = u^0x80
veor q6, q11, q0 ; *op0 = u^0x80
veor q8, q12, q0 ; *oq1 = u^0x80
vswp d12, d11
vswp d16, d13
vswp d14, d12
vswp d16, d15
;store op1, op0, oq0, oq1
vst4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0], r1
vst4.8 {d14[0], d15[0], d16[0], d17[0]}, [r2], r1
vst4.8 {d10[1], d11[1], d12[1], d13[1]}, [r0], r1
vst4.8 {d14[1], d15[1], d16[1], d17[1]}, [r2], r1
vst4.8 {d10[2], d11[2], d12[2], d13[2]}, [r0], r1
vst4.8 {d14[2], d15[2], d16[2], d17[2]}, [r2], r1
vst4.8 {d10[3], d11[3], d12[3], d13[3]}, [r0], r1
vst4.8 {d14[3], d15[3], d16[3], d17[3]}, [r2], r1
vst4.8 {d10[4], d11[4], d12[4], d13[4]}, [r0], r1
vst4.8 {d14[4], d15[4], d16[4], d17[4]}, [r2], r1
vst4.8 {d10[5], d11[5], d12[5], d13[5]}, [r0], r1
vst4.8 {d14[5], d15[5], d16[5], d17[5]}, [r2], r1
vst4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0], r1
vst4.8 {d14[6], d15[6], d16[6], d17[6]}, [r2], r1
vst4.8 {d10[7], d11[7], d12[7], d13[7]}, [r0], r1
vst4.8 {d14[7], d15[7], d16[7], d17[7]}, [r2], r1
bx lr
ENDP ; |vp8_loop_filter_vertical_edge_uv_neon|
;-----------------
AREA vloopfilteruv_dat, DATA, READWRITE ;read/write by default
;Data section with name data_area is specified. DCD reserves space in memory for 16 data.
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
_vlfuv_coeff_
DCD vlfuv_coeff
vlfuv_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x01010101, 0x01010101, 0x01010101, 0x01010101
END

207
vp8/common/arm/neon/loopfilterverticaledge_y_neon.asm
View File

@@ -1,207 +0,0 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_loop_filter_vertical_edge_y_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
;are equal. So, in the code, only one load is needed
;for flimit. Same way applies to limit and thresh.
; r0 unsigned char *s,
; r1 int p, //pitch
; r2 const signed char *flimit,
; r3 const signed char *limit,
; stack(r4) const signed char *thresh,
; //stack(r5) int count --unused
|vp8_loop_filter_vertical_edge_y_neon| PROC
sub r0, r0, #4 ; move src pointer down by 4 columns
ldr r12, [sp, #0] ; load thresh pointer
vld1.u8 {d6}, [r0], r1 ; load first 8-line src data
vld1.s8 {d0[], d1[]}, [r2] ; flimit
vld1.u8 {d8}, [r0], r1
vld1.s8 {d2[], d3[]}, [r3] ; limit
vld1.u8 {d10}, [r0], r1
vld1.s8 {d4[], d5[]}, [r12] ; thresh
vld1.u8 {d12}, [r0], r1
ldr r12, _vlfy_coeff_
vld1.u8 {d14}, [r0], r1
vld1.u8 {d16}, [r0], r1
vld1.u8 {d18}, [r0], r1
vld1.u8 {d20}, [r0], r1
vld1.u8 {d7}, [r0], r1 ; load second 8-line src data
vld1.u8 {d9}, [r0], r1
vld1.u8 {d11}, [r0], r1
vld1.u8 {d13}, [r0], r1
vld1.u8 {d15}, [r0], r1
vld1.u8 {d17}, [r0], r1
vld1.u8 {d19}, [r0], r1
vld1.u8 {d21}, [r0], r1
;transpose to 8x16 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q4, q10, q9 ; abs(q3 - q2)
vabd.u8 q9, q6, q7 ; abs(p0 - q0)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q4
vmax.u8 q15, q11, q12
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vadd.u8 q0, q0, q0 ; flimit * 2
vadd.u8 q0, q0, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q2, q5, q8 ; abs(p1 - q1)
vqadd.u8 q9, q9, q9 ; abs(p0 - q0) * 2
vshr.u8 q2, q2, #1 ; abs(p1 - q1) / 2
vqadd.u8 q9, q9, q2 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
vcge.u8 q9, q0, q9 ; (abs(p0 - q0)*2 + abs(p1-q1)/2 > flimit*2 + limit)*-1
vld1.u8 {q0}, [r12]!
;vp8_filter() function
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
;;;;;;;;;;;;;;
vld1.u8 {q10}, [r12]!
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q11, d15, d13
vmovl.u8 q4, d20
vqsub.s8 q1, q5, q8 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vorr q14, q13, q14 ; q14: vp8_hevmask
vmul.i16 q2, q2, q4 ; 3 * ( qs0 - ps0)
vmul.i16 q11, q11, q4
vand q1, q1, q14 ; vp8_filter &= hev
vand q15, q15, q9 ; vp8_filter_mask
vaddw.s8 q2, q2, d2
vaddw.s8 q11, q11, d3
vld1.u8 {q9}, [r12]!
;
vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d3, q11
;;
vand q1, q1, q15 ; vp8_filter &= mask
vqadd.s8 q2, q1, q10 ; Filter2 = vp8_signed_char_clamp(vp8_filter+3)
vqadd.s8 q1, q1, q9 ; Filter1 = vp8_signed_char_clamp(vp8_filter+4)
vshr.s8 q2, q2, #3 ; Filter2 >>= 3
vshr.s8 q1, q1, #3 ; Filter1 >>= 3
;calculate output
vqadd.s8 q11, q6, q2 ; u = vp8_signed_char_clamp(ps0 + Filter2)
vqsub.s8 q10, q7, q1 ; u = vp8_signed_char_clamp(qs0 - Filter1)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vrshr.s8 q1, q1, #1 ;round/shift: vp8_filter += 1; vp8_filter >>= 1
sub r0, r0, r1, lsl #4
add r0, r0, #2
;
vbic q1, q1, q14 ; vp8_filter &= ~hev
add r2, r0, r1
;
vqadd.s8 q13, q5, q1 ; u = vp8_signed_char_clamp(ps1 + vp8_filter)
vqsub.s8 q12, q8, q1 ; u = vp8_signed_char_clamp(qs1 - vp8_filter)
veor q7, q10, q0 ; *oq0 = u^0x80
veor q5, q13, q0 ; *op1 = u^0x80
veor q6, q11, q0 ; *op0 = u^0x80
veor q8, q12, q0 ; *oq1 = u^0x80
add r3, r2, r1
;
vswp d12, d11
vswp d16, d13
add r12, r3, r1
vswp d14, d12
vswp d16, d15
;store op1, op0, oq0, oq1
vst4.8 {d10[0], d11[0], d12[0], d13[0]}, [r0]
vst4.8 {d10[1], d11[1], d12[1], d13[1]}, [r2]
vst4.8 {d10[2], d11[2], d12[2], d13[2]}, [r3]
vst4.8 {d10[3], d11[3], d12[3], d13[3]}, [r12], r1
add r0, r12, r1
vst4.8 {d10[4], d11[4], d12[4], d13[4]}, [r12]
vst4.8 {d10[5], d11[5], d12[5], d13[5]}, [r0], r1
add r2, r0, r1
vst4.8 {d10[6], d11[6], d12[6], d13[6]}, [r0]
vst4.8 {d10[7], d11[7], d12[7], d13[7]}, [r2], r1
add r3, r2, r1
vst4.8 {d14[0], d15[0], d16[0], d17[0]}, [r2]
vst4.8 {d14[1], d15[1], d16[1], d17[1]}, [r3], r1
add r12, r3, r1
vst4.8 {d14[2], d15[2], d16[2], d17[2]}, [r3]
vst4.8 {d14[3], d15[3], d16[3], d17[3]}, [r12], r1
add r0, r12, r1
vst4.8 {d14[4], d15[4], d16[4], d17[4]}, [r12]
vst4.8 {d14[5], d15[5], d16[5], d17[5]}, [r0], r1
add r2, r0, r1
vst4.8 {d14[6], d15[6], d16[6], d17[6]}, [r0]
vst4.8 {d14[7], d15[7], d16[7], d17[7]}, [r2]
bx lr
ENDP ; |vp8_loop_filter_vertical_edge_y_neon|
;-----------------
AREA vloopfiltery_dat, DATA, READWRITE ;read/write by default
;Data section with name data_area is specified. DCD reserves space in memory for 16 data.
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
_vlfy_coeff_
DCD vlfy_coeff
vlfy_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x01010101, 0x01010101, 0x01010101, 0x01010101
END

519
vp8/common/arm/neon/mbloopfilter_neon.asm Normal file
View File

@@ -0,0 +1,519 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_mbloop_filter_horizontal_edge_y_neon|
EXPORT |vp8_mbloop_filter_horizontal_edge_uv_neon|
EXPORT |vp8_mbloop_filter_vertical_edge_y_neon|
EXPORT |vp8_mbloop_filter_vertical_edge_uv_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
; flimit, limit, and thresh should be positive numbers.
; All 16 elements in these variables are equal.
; void vp8_mbloop_filter_horizontal_edge_y_neon(unsigned char *src, int pitch,
; const signed char *flimit,
; const signed char *limit,
; const signed char *thresh,
; int count)
; r0 unsigned char *src,
; r1 int pitch,
; r2 const signed char *flimit,
; r3 const signed char *limit,
; sp const signed char *thresh,
; sp+4 int count (unused)
|vp8_mbloop_filter_horizontal_edge_y_neon| PROC
stmdb sp!, {lr}
sub r0, r0, r1, lsl #2 ; move src pointer down by 4 lines
ldr r12, [sp, #4] ; load thresh pointer
vld1.u8 {q3}, [r0], r1 ; p3
vld1.s8 {d2[], d3[]}, [r3] ; limit
vld1.u8 {q4}, [r0], r1 ; p2
vld1.s8 {d4[], d5[]}, [r12] ; thresh
vld1.u8 {q5}, [r0], r1 ; p1
vld1.u8 {q6}, [r0], r1 ; p0
vld1.u8 {q7}, [r0], r1 ; q0
vld1.u8 {q8}, [r0], r1 ; q1
vld1.u8 {q9}, [r0], r1 ; q2
vld1.u8 {q10}, [r0], r1 ; q3
bl vp8_mbloop_filter_neon
sub r0, r0, r1, lsl #3
add r0, r0, r1
add r2, r0, r1
add r3, r2, r1
vst1.u8 {q4}, [r0] ; store op2
vst1.u8 {q5}, [r2] ; store op1
vst1.u8 {q6}, [r3], r1 ; store op0
add r12, r3, r1
vst1.u8 {q7}, [r3] ; store oq0
vst1.u8 {q8}, [r12], r1 ; store oq1
vst1.u8 {q9}, [r12] ; store oq2
ldmia sp!, {pc}
ENDP ; |vp8_mbloop_filter_horizontal_edge_y_neon|
; void vp8_mbloop_filter_horizontal_edge_uv_neon(unsigned char *u, int pitch,
; const signed char *flimit,
; const signed char *limit,
; const signed char *thresh,
; unsigned char *v)
; r0 unsigned char *u,
; r1 int pitch,
; r2 const signed char *flimit,
; r3 const signed char *limit,
; sp const signed char *thresh,
; sp+4 unsigned char *v
|vp8_mbloop_filter_horizontal_edge_uv_neon| PROC
stmdb sp!, {lr}
sub r0, r0, r1, lsl #2 ; move u pointer down by 4 lines
vld1.s8 {d2[], d3[]}, [r3] ; limit
ldr r3, [sp, #8] ; load v ptr
ldr r12, [sp, #4] ; load thresh pointer
sub r3, r3, r1, lsl #2 ; move v pointer down by 4 lines
vld1.u8 {d6}, [r0], r1 ; p3
vld1.u8 {d7}, [r3], r1 ; p3
vld1.u8 {d8}, [r0], r1 ; p2
vld1.u8 {d9}, [r3], r1 ; p2
vld1.u8 {d10}, [r0], r1 ; p1
vld1.u8 {d11}, [r3], r1 ; p1
vld1.u8 {d12}, [r0], r1 ; p0
vld1.u8 {d13}, [r3], r1 ; p0
vld1.u8 {d14}, [r0], r1 ; q0
vld1.u8 {d15}, [r3], r1 ; q0
vld1.u8 {d16}, [r0], r1 ; q1
vld1.u8 {d17}, [r3], r1 ; q1
vld1.u8 {d18}, [r0], r1 ; q2
vld1.u8 {d19}, [r3], r1 ; q2
vld1.u8 {d20}, [r0], r1 ; q3
vld1.u8 {d21}, [r3], r1 ; q3
vld1.s8 {d4[], d5[]}, [r12] ; thresh
bl vp8_mbloop_filter_neon
sub r0, r0, r1, lsl #3
sub r3, r3, r1, lsl #3
add r0, r0, r1
add r3, r3, r1
vst1.u8 {d8}, [r0], r1 ; store u op2
vst1.u8 {d9}, [r3], r1 ; store v op2
vst1.u8 {d10}, [r0], r1 ; store u op1
vst1.u8 {d11}, [r3], r1 ; store v op1
vst1.u8 {d12}, [r0], r1 ; store u op0
vst1.u8 {d13}, [r3], r1 ; store v op0
vst1.u8 {d14}, [r0], r1 ; store u oq0
vst1.u8 {d15}, [r3], r1 ; store v oq0
vst1.u8 {d16}, [r0], r1 ; store u oq1
vst1.u8 {d17}, [r3], r1 ; store v oq1
vst1.u8 {d18}, [r0], r1 ; store u oq2
vst1.u8 {d19}, [r3], r1 ; store v oq2
ldmia sp!, {pc}
ENDP ; |vp8_mbloop_filter_horizontal_edge_uv_neon|
; void vp8_mbloop_filter_vertical_edge_y_neon(unsigned char *src, int pitch,
; const signed char *flimit,
; const signed char *limit,
; const signed char *thresh,
; int count)
; r0 unsigned char *src,
; r1 int pitch,
; r2 const signed char *flimit,
; r3 const signed char *limit,
; sp const signed char *thresh,
; sp+4 int count (unused)
|vp8_mbloop_filter_vertical_edge_y_neon| PROC
stmdb sp!, {lr}
sub r0, r0, #4 ; move src pointer down by 4 columns
vld1.u8 {d6}, [r0], r1 ; load first 8-line src data
ldr r12, [sp, #4] ; load thresh pointer
vld1.u8 {d8}, [r0], r1
sub sp, sp, #32
vld1.u8 {d10}, [r0], r1
vld1.u8 {d12}, [r0], r1
vld1.u8 {d14}, [r0], r1
vld1.u8 {d16}, [r0], r1
vld1.u8 {d18}, [r0], r1
vld1.u8 {d20}, [r0], r1
vld1.u8 {d7}, [r0], r1 ; load second 8-line src data
vld1.u8 {d9}, [r0], r1
vld1.u8 {d11}, [r0], r1
vld1.u8 {d13}, [r0], r1
vld1.u8 {d15}, [r0], r1
vld1.u8 {d17}, [r0], r1
vld1.u8 {d19}, [r0], r1
vld1.u8 {d21}, [r0], r1
;transpose to 8x16 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
vld1.s8 {d4[], d5[]}, [r12] ; thresh
vld1.s8 {d2[], d3[]}, [r3] ; limit
mov r12, sp
vst1.u8 {q3}, [r12]!
vst1.u8 {q10}, [r12]!
bl vp8_mbloop_filter_neon
sub r0, r0, r1, lsl #4
add r2, r0, r1
add r3, r2, r1
vld1.u8 {q3}, [sp]!
vld1.u8 {q10}, [sp]!
;transpose to 16x8 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
add r12, r3, r1
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
;store op2, op1, op0, oq0, oq1, oq2
vst1.8 {d6}, [r0]
vst1.8 {d8}, [r2]
vst1.8 {d10}, [r3]
vst1.8 {d12}, [r12], r1
add r0, r12, r1
vst1.8 {d14}, [r12]
vst1.8 {d16}, [r0], r1
add r2, r0, r1
vst1.8 {d18}, [r0]
vst1.8 {d20}, [r2], r1
add r3, r2, r1
vst1.8 {d7}, [r2]
vst1.8 {d9}, [r3], r1
add r12, r3, r1
vst1.8 {d11}, [r3]
vst1.8 {d13}, [r12], r1
add r0, r12, r1
vst1.8 {d15}, [r12]
vst1.8 {d17}, [r0], r1
add r2, r0, r1
vst1.8 {d19}, [r0]
vst1.8 {d21}, [r2]
ldmia sp!, {pc}
ENDP ; |vp8_mbloop_filter_vertical_edge_y_neon|
; void vp8_mbloop_filter_vertical_edge_uv_neon(unsigned char *u, int pitch,
; const signed char *flimit,
; const signed char *limit,
; const signed char *thresh,
; unsigned char *v)
; r0 unsigned char *u,
; r1 int pitch,
; r2 const signed char *flimit,
; r3 const signed char *limit,
; sp const signed char *thresh,
; sp+4 unsigned char *v
|vp8_mbloop_filter_vertical_edge_uv_neon| PROC
stmdb sp!, {lr}
sub r0, r0, #4 ; move src pointer down by 4 columns
vld1.s8 {d2[], d3[]}, [r3] ; limit
ldr r3, [sp, #8] ; load v ptr
ldr r12, [sp, #4] ; load thresh pointer
sub r3, r3, #4 ; move v pointer down by 4 columns
vld1.u8 {d6}, [r0], r1 ;load u data
vld1.u8 {d7}, [r3], r1 ;load v data
vld1.u8 {d8}, [r0], r1
vld1.u8 {d9}, [r3], r1
vld1.u8 {d10}, [r0], r1
vld1.u8 {d11}, [r3], r1
vld1.u8 {d12}, [r0], r1
vld1.u8 {d13}, [r3], r1
vld1.u8 {d14}, [r0], r1
vld1.u8 {d15}, [r3], r1
vld1.u8 {d16}, [r0], r1
vld1.u8 {d17}, [r3], r1
vld1.u8 {d18}, [r0], r1
vld1.u8 {d19}, [r3], r1
vld1.u8 {d20}, [r0], r1
vld1.u8 {d21}, [r3], r1
;transpose to 8x16 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
sub sp, sp, #32
vld1.s8 {d4[], d5[]}, [r12] ; thresh
mov r12, sp
vst1.u8 {q3}, [r12]!
vst1.u8 {q10}, [r12]!
bl vp8_mbloop_filter_neon
sub r0, r0, r1, lsl #3
sub r3, r3, r1, lsl #3
vld1.u8 {q3}, [sp]!
vld1.u8 {q10}, [sp]!
;transpose to 16x8 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
;store op2, op1, op0, oq0, oq1, oq2
vst1.8 {d6}, [r0], r1
vst1.8 {d7}, [r3], r1
vst1.8 {d8}, [r0], r1
vst1.8 {d9}, [r3], r1
vst1.8 {d10}, [r0], r1
vst1.8 {d11}, [r3], r1
vst1.8 {d12}, [r0], r1
vst1.8 {d13}, [r3], r1
vst1.8 {d14}, [r0], r1
vst1.8 {d15}, [r3], r1
vst1.8 {d16}, [r0], r1
vst1.8 {d17}, [r3], r1
vst1.8 {d18}, [r0], r1
vst1.8 {d19}, [r3], r1
vst1.8 {d20}, [r0], r1
vst1.8 {d21}, [r3], r1
ldmia sp!, {pc}
ENDP ; |vp8_mbloop_filter_vertical_edge_uv_neon|
; void vp8_mbloop_filter_neon()
; This is a helper function for the macroblock loopfilters. The individual
; functions do the necessary load, transpose (if necessary), preserve (if
; necessary) and store.
; TODO:
; The vertical filter writes p3/q3 back out because two 4 element writes are
; much simpler than ordering and writing two 3 element sets (or three 2 elements
; sets, or whichever other combinations are possible).
; If we can preserve q3 and q10, the vertical filter will be able to avoid
; storing those values on the stack and reading them back after the filter.
; r0,r1 PRESERVE
; r2 flimit
; r3 PRESERVE
; q1 limit
; q2 thresh
; q3 p3
; q4 p2
; q5 p1
; q6 p0
; q7 q0
; q8 q1
; q9 q2
; q10 q3
|vp8_mbloop_filter_neon| PROC
ldr r12, _mblf_coeff_
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q0, q10, q9 ; abs(q3 - q2)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q0
vmax.u8 q15, q11, q12
vabd.u8 q12, q6, q7 ; abs(p0 - q0)
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh) * -1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh) * -1
vmax.u8 q15, q15, q3
vld1.s8 {d4[], d5[]}, [r2] ; flimit
vld1.u8 {q0}, [r12]!
vadd.u8 q2, q2, q2 ; flimit * 2
vadd.u8 q2, q2, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q1, q5, q8 ; a = abs(p1 - q1)
vqadd.u8 q12, q12, q12 ; b = abs(p0 - q0) * 2
vshr.u8 q1, q1, #1 ; a = a / 2
vqadd.u8 q12, q12, q1 ; a = b + a
vcge.u8 q12, q2, q12 ; (a > flimit * 2 + limit) * -1
; vp8_filter
; convert to signed
veor q7, q7, q0 ; qs0
veor q6, q6, q0 ; ps0
veor q5, q5, q0 ; ps1
veor q8, q8, q0 ; qs1
veor q4, q4, q0 ; ps2
veor q9, q9, q0 ; qs2
vorr q14, q13, q14 ; vp8_hevmask
vsubl.s8 q2, d14, d12 ; qs0 - ps0
vsubl.s8 q13, d15, d13
vqsub.s8 q1, q5, q8 ; vp8_filter = clamp(ps1-qs1)
vadd.s16 q10, q2, q2 ; 3 * (qs0 - ps0)
vadd.s16 q11, q13, q13
vand q15, q15, q12 ; vp8_filter_mask
vadd.s16 q2, q2, q10
vadd.s16 q13, q13, q11
vld1.u8 {q12}, [r12]! ; #3
vaddw.s8 q2, q2, d2 ; vp8_filter + 3 * ( qs0 - ps0)
vaddw.s8 q13, q13, d3
vld1.u8 {q11}, [r12]! ; #4
; vp8_filter = clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d2, q2
vqmovn.s16 d3, q13
vand q1, q1, q15 ; vp8_filter &= mask
vld1.u8 {q15}, [r12]! ; #63
;
vand q13, q1, q14 ; Filter2 &= hev
vld1.u8 {d7}, [r12]! ; #9
vqadd.s8 q2, q13, q11 ; Filter1 = clamp(Filter2+4)
vqadd.s8 q13, q13, q12 ; Filter2 = clamp(Filter2+3)
vld1.u8 {d6}, [r12]! ; #18
vshr.s8 q2, q2, #3 ; Filter1 >>= 3
vshr.s8 q13, q13, #3 ; Filter2 >>= 3
vmov q10, q15
vmov q12, q15
vqsub.s8 q7, q7, q2 ; qs0 = clamp(qs0 - Filter1)
vld1.u8 {d5}, [r12]! ; #27
vqadd.s8 q6, q6, q13 ; ps0 = clamp(ps0 + Filter2)
vbic q1, q1, q14 ; vp8_filter &= ~hev
; roughly 1/7th difference across boundary
; roughly 2/7th difference across boundary
; roughly 3/7th difference across boundary
vmov q11, q15
vmov q13, q15
vmov q14, q15
vmlal.s8 q10, d2, d7 ; Filter2 * 9
vmlal.s8 q11, d3, d7
vmlal.s8 q12, d2, d6 ; Filter2 * 18
vmlal.s8 q13, d3, d6
vmlal.s8 q14, d2, d5 ; Filter2 * 27
vmlal.s8 q15, d3, d5
vqshrn.s16 d20, q10, #7 ; u = clamp((63 + Filter2 * 9)>>7)
vqshrn.s16 d21, q11, #7
vqshrn.s16 d24, q12, #7 ; u = clamp((63 + Filter2 * 18)>>7)
vqshrn.s16 d25, q13, #7
vqshrn.s16 d28, q14, #7 ; u = clamp((63 + Filter2 * 27)>>7)
vqshrn.s16 d29, q15, #7
vqsub.s8 q11, q9, q10 ; s = clamp(qs2 - u)
vqadd.s8 q10, q4, q10 ; s = clamp(ps2 + u)
vqsub.s8 q13, q8, q12 ; s = clamp(qs1 - u)
vqadd.s8 q12, q5, q12 ; s = clamp(ps1 + u)
vqsub.s8 q15, q7, q14 ; s = clamp(qs0 - u)
vqadd.s8 q14, q6, q14 ; s = clamp(ps0 + u)
veor q9, q11, q0 ; *oq2 = s^0x80
veor q4, q10, q0 ; *op2 = s^0x80
veor q8, q13, q0 ; *oq1 = s^0x80
veor q5, q12, q0 ; *op2 = s^0x80
veor q7, q15, q0 ; *oq0 = s^0x80
veor q6, q14, q0 ; *op0 = s^0x80
bx lr
ENDP ; |vp8_mbloop_filter_neon|
AREA mbloopfilter_dat, DATA, READONLY
_mblf_coeff_
DCD mblf_coeff
mblf_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x003f003f, 0x003f003f, 0x003f003f, 0x003f003f
DCD 0x09090909, 0x09090909, 0x12121212, 0x12121212
DCD 0x1b1b1b1b, 0x1b1b1b1b
END

220
vp8/common/arm/neon/mbloopfilterhorizontaledge_uv_neon.asm
View File

@@ -1,220 +0,0 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_mbloop_filter_horizontal_edge_uv_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
;are equal. So, in the code, only one load is needed
;for flimit. Same way applies to limit and thresh.
; r0 unsigned char *u,
; r1 int p, //pitch
; r2 const signed char *flimit,
; r3 const signed char *limit,
; stack(r4) const signed char *thresh,
; stack(r5) unsigned char *v
|vp8_mbloop_filter_horizontal_edge_uv_neon| PROC
sub r0, r0, r1, lsl #2 ; move u pointer down by 4 lines
vld1.s8 {d2[], d3[]}, [r3] ; limit
ldr r3, [sp, #4] ; load v ptr
ldr r12, [sp, #0] ; load thresh pointer
sub r3, r3, r1, lsl #2 ; move v pointer down by 4 lines
vld1.u8 {d6}, [r0], r1 ; p3
vld1.u8 {d7}, [r3], r1 ; p3
vld1.u8 {d8}, [r0], r1 ; p2
vld1.u8 {d9}, [r3], r1 ; p2
vld1.u8 {d10}, [r0], r1 ; p1
vld1.u8 {d11}, [r3], r1 ; p1
vld1.u8 {d12}, [r0], r1 ; p0
vld1.u8 {d13}, [r3], r1 ; p0
vld1.u8 {d14}, [r0], r1 ; q0
vld1.u8 {d15}, [r3], r1 ; q0
vld1.u8 {d16}, [r0], r1 ; q1
vld1.u8 {d17}, [r3], r1 ; q1
vld1.u8 {d18}, [r0], r1 ; q2
vld1.u8 {d19}, [r3], r1 ; q2
vld1.u8 {d20}, [r0], r1 ; q3
vld1.u8 {d21}, [r3], r1 ; q3
vld1.s8 {d4[], d5[]}, [r12] ; thresh
ldr r12, _mbhlfuv_coeff_
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q0, q10, q9 ; abs(q3 - q2)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q0
vmax.u8 q15, q11, q12
vabd.u8 q12, q6, q7 ; abs(p0 - q0)
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vld1.s8 {d4[], d5[]}, [r2] ; flimit
vld1.u8 {q0}, [r12]!
vadd.u8 q2, q2, q2 ; flimit * 2
vadd.u8 q2, q2, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q1, q5, q8 ; abs(p1 - q1)
vqadd.u8 q12, q12, q12 ; abs(p0 - q0) * 2
vshr.u8 q1, q1, #1 ; abs(p1 - q1) / 2
vqadd.u8 q12, q12, q1 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
vcge.u8 q12, q2, q12 ; (abs(p0 - q0)*2 + abs(p1 - q1)/2 > flimit*2 + limit)*-1
;vp8_filter() function
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
veor q4, q4, q0 ; ps2: p2 offset to convert to a signed value
veor q9, q9, q0 ; qs2: q2 offset to convert to a signed value
;;;;;;;;;;;;;
vorr q14, q13, q14 ; q14: vp8_hevmask
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q13, d15, d13
vqsub.s8 q1, q5, q8 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vadd.s16 q10, q2, q2 ; 3 * ( qs0 - ps0)
vadd.s16 q11, q13, q13
vand q15, q15, q12 ; vp8_filter_mask
vadd.s16 q2, q2, q10
vadd.s16 q13, q13, q11
vld1.u8 {q12}, [r12]! ;#3
vaddw.s8 q2, q2, d2 ; vp8_filter + 3 * ( qs0 - ps0)
vaddw.s8 q13, q13, d3
vld1.u8 {q11}, [r12]! ;#4
vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d3, q13
;;;;;;;;;;;;;;
vand q1, q1, q15 ; vp8_filter &= mask
vld1.u8 {q15}, [r12]! ;#63
;
vand q13, q1, q14 ; Filter2: q13; Filter2 &= hev
vld1.u8 {d7}, [r12]! ;#9
vqadd.s8 q2, q13, q11 ; Filter1 = vp8_signed_char_clamp(Filter2+4)
vqadd.s8 q13, q13, q12 ; Filter2 = vp8_signed_char_clamp(Filter2+3)
vld1.u8 {d6}, [r12]! ;#18
sub r0, r0, r1, lsl #3
sub r3, r3, r1, lsl #3
vshr.s8 q2, q2, #3 ; Filter1 >>= 3
vshr.s8 q13, q13, #3 ; Filter2 >>= 3
vmov q10, q15
vmov q12, q15
vqsub.s8 q7, q7, q2 ; qs0 = vp8_signed_char_clamp(qs0 - Filter1)
vld1.u8 {d5}, [r12]! ;#27
add r0, r0, r1
add r3, r3, r1
vqadd.s8 q6, q6, q13 ; ps0 = vp8_signed_char_clamp(ps0 + Filter2)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vbic q1, q1, q14 ; Filter2: q1; vp8_filter &= ~hev; Filter2 = vp8_filter
; roughly 1/7th difference across boundary
; roughly 2/7th difference across boundary
; roughly 3/7th difference across boundary
vmov q11, q15
vmov q13, q15
vmov q14, q15
vmlal.s8 q10, d2, d7 ; Filter2 * 9
vmlal.s8 q11, d3, d7
vmlal.s8 q12, d2, d6 ; Filter2 * 18
vmlal.s8 q13, d3, d6
vmlal.s8 q14, d2, d5 ; Filter2 * 27
vmlal.s8 q15, d3, d5
vqshrn.s16 d20, q10, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
vqshrn.s16 d21, q11, #7
vqshrn.s16 d24, q12, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
vqshrn.s16 d25, q13, #7
vqshrn.s16 d28, q14, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
vqshrn.s16 d29, q15, #7
vqsub.s8 q11, q9, q10 ; s = vp8_signed_char_clamp(qs2 - u)
vqadd.s8 q10, q4, q10 ; s = vp8_signed_char_clamp(ps2 + u)
vqsub.s8 q13, q8, q12 ; s = vp8_signed_char_clamp(qs1 - u)
vqadd.s8 q12, q5, q12 ; s = vp8_signed_char_clamp(ps1 + u)
vqsub.s8 q15, q7, q14 ; s = vp8_signed_char_clamp(qs0 - u)
vqadd.s8 q14, q6, q14 ; s = vp8_signed_char_clamp(ps0 + u)
veor q9, q11, q0 ; *oq2 = s^0x80
veor q4, q10, q0 ; *op2 = s^0x80
veor q8, q13, q0 ; *oq1 = s^0x80
veor q5, q12, q0 ; *op2 = s^0x80
veor q7, q15, q0 ; *oq0 = s^0x80
veor q6, q14, q0 ; *op0 = s^0x80
vst1.u8 {d8}, [r0], r1 ; store u op2
vst1.u8 {d9}, [r3], r1 ; store v op2
vst1.u8 {d10}, [r0], r1 ; store u op1
vst1.u8 {d11}, [r3], r1 ; store v op1
vst1.u8 {d12}, [r0], r1 ; store u op0
vst1.u8 {d13}, [r3], r1 ; store v op0
vst1.u8 {d14}, [r0], r1 ; store u oq0
vst1.u8 {d15}, [r3], r1 ; store v oq0
vst1.u8 {d16}, [r0], r1 ; store u oq1
vst1.u8 {d17}, [r3], r1 ; store v oq1
vst1.u8 {d18}, [r0], r1 ; store u oq2
vst1.u8 {d19}, [r3], r1 ; store v oq2
bx lr
ENDP ; |vp8_mbloop_filter_horizontal_edge_uv_neon|
;-----------------
AREA mbhloopfilteruv_dat, DATA, READWRITE ;read/write by default
;Data section with name data_area is specified. DCD reserves space in memory for 16 data.
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
_mbhlfuv_coeff_
DCD mbhlfuv_coeff
mbhlfuv_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x003f003f, 0x003f003f, 0x003f003f, 0x003f003f
DCD 0x09090909, 0x09090909, 0x12121212, 0x12121212
DCD 0x1b1b1b1b, 0x1b1b1b1b
END

201
vp8/common/arm/neon/mbloopfilterhorizontaledge_y_neon.asm
View File

@@ -1,201 +0,0 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_mbloop_filter_horizontal_edge_y_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
;are equal. So, in the code, only one load is needed
;for flimit. Same way applies to limit and thresh.
; r0 unsigned char *s,
; r1 int p, //pitch
; r2 const signed char *flimit,
; r3 const signed char *limit,
; stack(r4) const signed char *thresh,
; //stack(r5) int count --unused
|vp8_mbloop_filter_horizontal_edge_y_neon| PROC
sub r0, r0, r1, lsl #2 ; move src pointer down by 4 lines
ldr r12, [sp, #0] ; load thresh pointer
vld1.u8 {q3}, [r0], r1 ; p3
vld1.s8 {d2[], d3[]}, [r3] ; limit
vld1.u8 {q4}, [r0], r1 ; p2
vld1.s8 {d4[], d5[]}, [r12] ; thresh
vld1.u8 {q5}, [r0], r1 ; p1
ldr r12, _mbhlfy_coeff_
vld1.u8 {q6}, [r0], r1 ; p0
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vld1.u8 {q7}, [r0], r1 ; q0
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vld1.u8 {q8}, [r0], r1 ; q1
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vld1.u8 {q9}, [r0], r1 ; q2
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vld1.u8 {q10}, [r0], r1 ; q3
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q0, q10, q9 ; abs(q3 - q2)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q0
vmax.u8 q15, q11, q12
vabd.u8 q12, q6, q7 ; abs(p0 - q0)
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vld1.s8 {d4[], d5[]}, [r2] ; flimit
vld1.u8 {q0}, [r12]!
vadd.u8 q2, q2, q2 ; flimit * 2
vadd.u8 q2, q2, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q1, q5, q8 ; abs(p1 - q1)
vqadd.u8 q12, q12, q12 ; abs(p0 - q0) * 2
vshr.u8 q1, q1, #1 ; abs(p1 - q1) / 2
vqadd.u8 q12, q12, q1 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
vcge.u8 q12, q2, q12 ; (abs(p0 - q0)*2 + abs(p1 - q1)/2 > flimit*2 + limit)*-1
;vp8_filter() function
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
veor q4, q4, q0 ; ps2: p2 offset to convert to a signed value
veor q9, q9, q0 ; qs2: q2 offset to convert to a signed value
;;;;;;;;;;;;;
vorr q14, q13, q14 ; q14: vp8_hevmask
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q13, d15, d13
vqsub.s8 q1, q5, q8 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vadd.s16 q10, q2, q2 ; 3 * ( qs0 - ps0)
vadd.s16 q11, q13, q13
vand q15, q15, q12 ; vp8_filter_mask
vadd.s16 q2, q2, q10
vadd.s16 q13, q13, q11
vld1.u8 {q12}, [r12]! ;#3
vaddw.s8 q2, q2, d2 ; vp8_filter + 3 * ( qs0 - ps0)
vaddw.s8 q13, q13, d3
vld1.u8 {q11}, [r12]! ;#4
vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d3, q13
;;;;;;;;;;;;;;
vand q1, q1, q15 ; vp8_filter &= mask
vld1.u8 {q15}, [r12]! ;#63
;
vand q13, q1, q14 ; Filter2: q13; Filter2 &= hev
vld1.u8 {d7}, [r12]! ;#9
sub r0, r0, r1, lsl #3
vqadd.s8 q2, q13, q11 ; Filter1 = vp8_signed_char_clamp(Filter2+4)
vqadd.s8 q13, q13, q12 ; Filter2 = vp8_signed_char_clamp(Filter2+3)
vld1.u8 {d6}, [r12]! ;#18
add r0, r0, r1
add r2, r0, r1
vshr.s8 q2, q2, #3 ; Filter1 >>= 3
vshr.s8 q13, q13, #3 ; Filter2 >>= 3
vmov q10, q15
vmov q12, q15
vqsub.s8 q7, q7, q2 ; qs0 = vp8_signed_char_clamp(qs0 - Filter1)
vld1.u8 {d5}, [r12]! ;#27
add r3, r2, r1
vqadd.s8 q6, q6, q13 ; ps0 = vp8_signed_char_clamp(ps0 + Filter2)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vbic q1, q1, q14 ; Filter2: q1; vp8_filter &= ~hev; Filter2 = vp8_filter
; roughly 1/7th difference across boundary
; roughly 2/7th difference across boundary
; roughly 3/7th difference across boundary
vmov q11, q15
vmov q13, q15
vmov q14, q15
vmlal.s8 q10, d2, d7 ; Filter2 * 9
vmlal.s8 q11, d3, d7
vmlal.s8 q12, d2, d6 ; Filter2 * 18
vmlal.s8 q13, d3, d6
vmlal.s8 q14, d2, d5 ; Filter2 * 27
vmlal.s8 q15, d3, d5
vqshrn.s16 d20, q10, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
vqshrn.s16 d21, q11, #7
vqshrn.s16 d24, q12, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
vqshrn.s16 d25, q13, #7
vqshrn.s16 d28, q14, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
vqshrn.s16 d29, q15, #7
vqsub.s8 q11, q9, q10 ; s = vp8_signed_char_clamp(qs2 - u)
vqadd.s8 q10, q4, q10 ; s = vp8_signed_char_clamp(ps2 + u)
vqsub.s8 q13, q8, q12 ; s = vp8_signed_char_clamp(qs1 - u)
vqadd.s8 q12, q5, q12 ; s = vp8_signed_char_clamp(ps1 + u)
vqsub.s8 q15, q7, q14 ; s = vp8_signed_char_clamp(qs0 - u)
vqadd.s8 q14, q6, q14 ; s = vp8_signed_char_clamp(ps0 + u)
veor q9, q11, q0 ; *oq2 = s^0x80
veor q4, q10, q0 ; *op2 = s^0x80
veor q5, q12, q0 ; *op2 = s^0x80
veor q6, q14, q0 ; *op0 = s^0x80
veor q8, q13, q0 ; *oq1 = s^0x80
veor q7, q15, q0 ; *oq0 = s^0x80
vst1.u8 {q4}, [r0] ; store op2
vst1.u8 {q5}, [r2] ; store op1
vst1.u8 {q6}, [r3], r1 ; store op0
add r12, r3, r1
vst1.u8 {q7}, [r3] ; store oq0
vst1.u8 {q8}, [r12], r1 ; store oq1
vst1.u8 {q9}, [r12] ; store oq2
bx lr
ENDP ; |vp8_mbloop_filter_horizontal_edge_y_neon|
;-----------------
AREA mbhloopfiltery_dat, DATA, READWRITE ;read/write by default
;Data section with name data_area is specified. DCD reserves space in memory for 16 data.
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
_mbhlfy_coeff_
DCD mbhlfy_coeff
mbhlfy_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x003f003f, 0x003f003f, 0x003f003f, 0x003f003f
DCD 0x09090909, 0x09090909, 0x12121212, 0x12121212
DCD 0x1b1b1b1b, 0x1b1b1b1b
END

262
vp8/common/arm/neon/mbloopfilterverticaledge_uv_neon.asm
View File

@@ -1,262 +0,0 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_mbloop_filter_vertical_edge_uv_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
;are equal. So, in the code, only one load is needed
;for flimit. Same way applies to limit and thresh.
; r0 unsigned char *u,
; r1 int p, //pitch
; r2 const signed char *flimit,
; r3 const signed char *limit,
; stack(r4) const signed char *thresh,
; stack(r5) unsigned char *v
|vp8_mbloop_filter_vertical_edge_uv_neon| PROC
sub r0, r0, #4 ; move src pointer down by 4 columns
vld1.s8 {d2[], d3[]}, [r3] ; limit
ldr r3, [sp, #4] ; load v ptr
ldr r12, [sp, #0] ; load thresh pointer
sub r3, r3, #4 ; move v pointer down by 4 columns
vld1.u8 {d6}, [r0], r1 ;load u data
vld1.u8 {d7}, [r3], r1 ;load v data
vld1.u8 {d8}, [r0], r1
vld1.u8 {d9}, [r3], r1
vld1.u8 {d10}, [r0], r1
vld1.u8 {d11}, [r3], r1
vld1.u8 {d12}, [r0], r1
vld1.u8 {d13}, [r3], r1
vld1.u8 {d14}, [r0], r1
vld1.u8 {d15}, [r3], r1
vld1.u8 {d16}, [r0], r1
vld1.u8 {d17}, [r3], r1
vld1.u8 {d18}, [r0], r1
vld1.u8 {d19}, [r3], r1
vld1.u8 {d20}, [r0], r1
vld1.u8 {d21}, [r3], r1
;transpose to 8x16 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
sub sp, sp, #32
vld1.s8 {d4[], d5[]}, [r12] ; thresh
vst1.u8 {q3}, [sp]!
ldr r12, _mbvlfuv_coeff_
vst1.u8 {q10}, [sp]!
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q0, q10, q9 ; abs(q3 - q2)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q0
vmax.u8 q15, q11, q12
vabd.u8 q12, q6, q7 ; abs(p0 - q0)
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vld1.s8 {d4[], d5[]}, [r2] ; flimit
vld1.u8 {q0}, [r12]!
vadd.u8 q2, q2, q2 ; flimit * 2
vadd.u8 q2, q2, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q1, q5, q8 ; abs(p1 - q1)
vqadd.u8 q12, q12, q12 ; abs(p0 - q0) * 2
vshr.u8 q1, q1, #1 ; abs(p1 - q1) / 2
vqadd.u8 q12, q12, q1 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
vcge.u8 q12, q2, q12 ; (abs(p0 - q0)*2 + abs(p1 - q1)/2 > flimit*2 + limit)*-1
;vp8_filter() function
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
veor q4, q4, q0 ; ps2: p2 offset to convert to a signed value
veor q9, q9, q0 ; qs2: q2 offset to convert to a signed value
;;;;;;;;;;;;;
vorr q14, q13, q14 ; q14: vp8_hevmask
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q13, d15, d13
vqsub.s8 q1, q5, q8 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vadd.s16 q10, q2, q2 ; 3 * ( qs0 - ps0)
vadd.s16 q11, q13, q13
vand q15, q15, q12 ; vp8_filter_mask
vadd.s16 q2, q2, q10
vadd.s16 q13, q13, q11
vld1.u8 {q12}, [r12]! ;#3
vaddw.s8 q2, q2, d2 ; vp8_filter + 3 * ( qs0 - ps0)
vaddw.s8 q13, q13, d3
vld1.u8 {q11}, [r12]! ;#4
vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d3, q13
;;;;;;;;;;;;;;
vand q1, q1, q15 ; vp8_filter &= mask
vld1.u8 {q15}, [r12]! ;#63
;
vand q13, q1, q14 ; Filter2: q13; Filter2 &= hev
vld1.u8 {d7}, [r12]! ;#9
;
vqadd.s8 q2, q13, q11 ; Filter1 = vp8_signed_char_clamp(Filter2+4)
vqadd.s8 q13, q13, q12 ; Filter2 = vp8_signed_char_clamp(Filter2+3)
vld1.u8 {d6}, [r12]! ;#18
sub r0, r0, r1, lsl #3
sub r3, r3, r1, lsl #3
vshr.s8 q2, q2, #3 ; Filter1 >>= 3
vshr.s8 q13, q13, #3 ; Filter2 >>= 3
vmov q10, q15
vmov q12, q15
vqsub.s8 q7, q7, q2 ; qs0 = vp8_signed_char_clamp(qs0 - Filter1)
vld1.u8 {d5}, [r12]! ;#27
sub sp, sp, #32
vqadd.s8 q6, q6, q13 ; ps0 = vp8_signed_char_clamp(ps0 + Filter2)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vbic q1, q1, q14 ; Filter2: q1; vp8_filter &= ~hev; Filter2 = vp8_filter
; roughly 1/7th difference across boundary
; roughly 2/7th difference across boundary
; roughly 3/7th difference across boundary
vmov q11, q15
vmov q13, q15
vmov q14, q15
vmlal.s8 q10, d2, d7 ; Filter2 * 9
vmlal.s8 q11, d3, d7
vmlal.s8 q12, d2, d6 ; Filter2 * 18
vmlal.s8 q13, d3, d6
vmlal.s8 q14, d2, d5 ; Filter2 * 27
vmlal.s8 q15, d3, d5
vqshrn.s16 d20, q10, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
vqshrn.s16 d21, q11, #7
vqshrn.s16 d24, q12, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
vqshrn.s16 d25, q13, #7
vqshrn.s16 d28, q14, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
vqshrn.s16 d29, q15, #7
vqsub.s8 q11, q9, q10 ; s = vp8_signed_char_clamp(qs2 - u)
vqadd.s8 q10, q4, q10 ; s = vp8_signed_char_clamp(ps2 + u)
vqsub.s8 q13, q8, q12 ; s = vp8_signed_char_clamp(qs1 - u)
vqadd.s8 q12, q5, q12 ; s = vp8_signed_char_clamp(ps1 + u)
vqsub.s8 q15, q7, q14 ; s = vp8_signed_char_clamp(qs0 - u)
vqadd.s8 q14, q6, q14 ; s = vp8_signed_char_clamp(ps0 + u)
veor q9, q11, q0 ; *oq2 = s^0x80
veor q4, q10, q0 ; *op2 = s^0x80
veor q8, q13, q0 ; *oq1 = s^0x80
veor q5, q12, q0 ; *op2 = s^0x80
veor q7, q15, q0 ; *oq0 = s^0x80
vld1.u8 {q3}, [sp]!
veor q6, q14, q0 ; *op0 = s^0x80
vld1.u8 {q10}, [sp]!
;transpose to 16x8 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
;store op2, op1, op0, oq0, oq1, oq2
vst1.8 {d6}, [r0], r1
vst1.8 {d7}, [r3], r1
vst1.8 {d8}, [r0], r1
vst1.8 {d9}, [r3], r1
vst1.8 {d10}, [r0], r1
vst1.8 {d11}, [r3], r1
vst1.8 {d12}, [r0], r1
vst1.8 {d13}, [r3], r1
vst1.8 {d14}, [r0], r1
vst1.8 {d15}, [r3], r1
vst1.8 {d16}, [r0], r1
vst1.8 {d17}, [r3], r1
vst1.8 {d18}, [r0], r1
vst1.8 {d19}, [r3], r1
vst1.8 {d20}, [r0], r1
vst1.8 {d21}, [r3], r1
bx lr
ENDP ; |vp8_mbloop_filter_vertical_edge_uv_neon|
;-----------------
AREA mbvloopfilteruv_dat, DATA, READWRITE ;read/write by default
;Data section with name data_area is specified. DCD reserves space in memory for 16 data.
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
_mbvlfuv_coeff_
DCD mbvlfuv_coeff
mbvlfuv_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x003f003f, 0x003f003f, 0x003f003f, 0x003f003f
DCD 0x09090909, 0x09090909, 0x12121212, 0x12121212
DCD 0x1b1b1b1b, 0x1b1b1b1b
END

267
vp8/common/arm/neon/mbloopfilterverticaledge_y_neon.asm
View File

@@ -1,267 +0,0 @@
;
; Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
; Use of this source code is governed by a BSD-style license
; that can be found in the LICENSE file in the root of the source
; tree. An additional intellectual property rights grant can be found
; in the file PATENTS. All contributing project authors may
; be found in the AUTHORS file in the root of the source tree.
;
EXPORT |vp8_mbloop_filter_vertical_edge_y_neon|
ARM
REQUIRE8
PRESERVE8
AREA ||.text||, CODE, READONLY, ALIGN=2
;Note: flimit, limit, and thresh shpuld be positive numbers. All 16 elements in flimit
;are equal. So, in the code, only one load is needed
;for flimit. Same way applies to limit and thresh.
; r0 unsigned char *s,
; r1 int p, //pitch
; r2 const signed char *flimit,
; r3 const signed char *limit,
; stack(r4) const signed char *thresh,
; //stack(r5) int count --unused
|vp8_mbloop_filter_vertical_edge_y_neon| PROC
sub r0, r0, #4 ; move src pointer down by 4 columns
vld1.u8 {d6}, [r0], r1 ; load first 8-line src data
ldr r12, [sp, #0] ; load thresh pointer
vld1.u8 {d8}, [r0], r1
sub sp, sp, #32
vld1.u8 {d10}, [r0], r1
vld1.u8 {d12}, [r0], r1
vld1.u8 {d14}, [r0], r1
vld1.u8 {d16}, [r0], r1
vld1.u8 {d18}, [r0], r1
vld1.u8 {d20}, [r0], r1
vld1.u8 {d7}, [r0], r1 ; load second 8-line src data
vld1.u8 {d9}, [r0], r1
vld1.u8 {d11}, [r0], r1
vld1.u8 {d13}, [r0], r1
vld1.u8 {d15}, [r0], r1
vld1.u8 {d17}, [r0], r1
vld1.u8 {d19}, [r0], r1
vld1.u8 {d21}, [r0], r1
;transpose to 8x16 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
vld1.s8 {d2[], d3[]}, [r3] ; limit
vst1.u8 {q3}, [sp]!
vld1.s8 {d4[], d5[]}, [r12] ; thresh
ldr r12, _mbvlfy_coeff_
vst1.u8 {q10}, [sp]!
; vp8_filter_mask
vabd.u8 q11, q3, q4 ; abs(p3 - p2)
vabd.u8 q12, q4, q5 ; abs(p2 - p1)
vabd.u8 q13, q5, q6 ; abs(p1 - p0)
vabd.u8 q14, q8, q7 ; abs(q1 - q0)
vabd.u8 q3, q9, q8 ; abs(q2 - q1)
vabd.u8 q0, q10, q9 ; abs(q3 - q2)
vmax.u8 q11, q11, q12
vmax.u8 q12, q13, q14
vmax.u8 q3, q3, q0
vmax.u8 q15, q11, q12
vabd.u8 q12, q6, q7 ; abs(p0 - q0)
; vp8_hevmask
vcgt.u8 q13, q13, q2 ; (abs(p1 - p0) > thresh)*-1
vcgt.u8 q14, q14, q2 ; (abs(q1 - q0) > thresh)*-1
vmax.u8 q15, q15, q3
vld1.s8 {d4[], d5[]}, [r2] ; flimit
vld1.u8 {q0}, [r12]!
vadd.u8 q2, q2, q2 ; flimit * 2
vadd.u8 q2, q2, q1 ; flimit * 2 + limit
vcge.u8 q15, q1, q15
vabd.u8 q1, q5, q8 ; abs(p1 - q1)
vqadd.u8 q12, q12, q12 ; abs(p0 - q0) * 2
vshr.u8 q1, q1, #1 ; abs(p1 - q1) / 2
vqadd.u8 q12, q12, q1 ; abs(p0 - q0) * 2 + abs(p1 - q1) / 2
vcge.u8 q12, q2, q12 ; (abs(p0 - q0)*2 + abs(p1 - q1)/2 > flimit*2 + limit)*-1
;vp8_filter() function
veor q7, q7, q0 ; qs0: q0 offset to convert to a signed value
veor q6, q6, q0 ; ps0: p0 offset to convert to a signed value
veor q5, q5, q0 ; ps1: p1 offset to convert to a signed value
veor q8, q8, q0 ; qs1: q1 offset to convert to a signed value
veor q4, q4, q0 ; ps2: p2 offset to convert to a signed value
veor q9, q9, q0 ; qs2: q2 offset to convert to a signed value
;;;;;;;;;;;;;
vorr q14, q13, q14 ; q14: vp8_hevmask
vsubl.s8 q2, d14, d12 ; ( qs0 - ps0)
vsubl.s8 q13, d15, d13
vqsub.s8 q1, q5, q8 ; vp8_filter = vp8_signed_char_clamp(ps1-qs1)
vadd.s16 q10, q2, q2 ; 3 * ( qs0 - ps0)
vadd.s16 q11, q13, q13
vand q15, q15, q12 ; vp8_filter_mask
vadd.s16 q2, q2, q10
vadd.s16 q13, q13, q11
vld1.u8 {q12}, [r12]! ;#3
vaddw.s8 q2, q2, d2 ; vp8_filter + 3 * ( qs0 - ps0)
vaddw.s8 q13, q13, d3
vld1.u8 {q11}, [r12]! ;#4
vqmovn.s16 d2, q2 ; vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * ( qs0 - ps0))
vqmovn.s16 d3, q13
;;;;;;;;;;;;;;
vand q1, q1, q15 ; vp8_filter &= mask
vld1.u8 {q15}, [r12]! ;#63
;
vand q13, q1, q14 ; Filter2: q13; Filter2 &= hev
vld1.u8 {d7}, [r12]! ;#9
;
vqadd.s8 q2, q13, q11 ; Filter1 = vp8_signed_char_clamp(Filter2+4)
vqadd.s8 q13, q13, q12 ; Filter2 = vp8_signed_char_clamp(Filter2+3)
vld1.u8 {d6}, [r12]! ;#18
sub r0, r0, r1, lsl #4
sub sp, sp, #32
add r2, r0, r1
vshr.s8 q2, q2, #3 ; Filter1 >>= 3
vshr.s8 q13, q13, #3 ; Filter2 >>= 3
vmov q10, q15
vmov q12, q15
vqsub.s8 q7, q7, q2 ; qs0 = vp8_signed_char_clamp(qs0 - Filter1)
vld1.u8 {d5}, [r12]! ;#27
add r3, r2, r1
vqadd.s8 q6, q6, q13 ; ps0 = vp8_signed_char_clamp(ps0 + Filter2)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
vbic q1, q1, q14 ; Filter2: q1; vp8_filter &= ~hev; Filter2 = vp8_filter
; roughly 1/7th difference across boundary
; roughly 2/7th difference across boundary
; roughly 3/7th difference across boundary
vmov q11, q15
vmov q13, q15
vmov q14, q15
vmlal.s8 q10, d2, d7 ; Filter2 * 9
vmlal.s8 q11, d3, d7
vmlal.s8 q12, d2, d6 ; Filter2 * 18
vmlal.s8 q13, d3, d6
vmlal.s8 q14, d2, d5 ; Filter2 * 27
vmlal.s8 q15, d3, d5
vqshrn.s16 d20, q10, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 9)>>7)
vqshrn.s16 d21, q11, #7
vqshrn.s16 d24, q12, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 18)>>7)
vqshrn.s16 d25, q13, #7
vqshrn.s16 d28, q14, #7 ; u = vp8_signed_char_clamp((63 + Filter2 * 27)>>7)
vqshrn.s16 d29, q15, #7
vqsub.s8 q11, q9, q10 ; s = vp8_signed_char_clamp(qs2 - u)
vqadd.s8 q10, q4, q10 ; s = vp8_signed_char_clamp(ps2 + u)
vqsub.s8 q13, q8, q12 ; s = vp8_signed_char_clamp(qs1 - u)
vqadd.s8 q12, q5, q12 ; s = vp8_signed_char_clamp(ps1 + u)
vqsub.s8 q15, q7, q14 ; s = vp8_signed_char_clamp(qs0 - u)
vqadd.s8 q14, q6, q14 ; s = vp8_signed_char_clamp(ps0 + u)
veor q9, q11, q0 ; *oq2 = s^0x80
veor q4, q10, q0 ; *op2 = s^0x80
veor q8, q13, q0 ; *oq1 = s^0x80
veor q5, q12, q0 ; *op2 = s^0x80
veor q7, q15, q0 ; *oq0 = s^0x80
vld1.u8 {q3}, [sp]!
veor q6, q14, q0 ; *op0 = s^0x80
vld1.u8 {q10}, [sp]!
;transpose to 16x8 matrix
vtrn.32 q3, q7
vtrn.32 q4, q8
vtrn.32 q5, q9
vtrn.32 q6, q10
add r12, r3, r1
vtrn.16 q3, q5
vtrn.16 q4, q6
vtrn.16 q7, q9
vtrn.16 q8, q10
vtrn.8 q3, q4
vtrn.8 q5, q6
vtrn.8 q7, q8
vtrn.8 q9, q10
;store op2, op1, op0, oq0, oq1, oq2
vst1.8 {d6}, [r0]
vst1.8 {d8}, [r2]
vst1.8 {d10}, [r3]
vst1.8 {d12}, [r12], r1
add r0, r12, r1
vst1.8 {d14}, [r12]
vst1.8 {d16}, [r0], r1
add r2, r0, r1
vst1.8 {d18}, [r0]
vst1.8 {d20}, [r2], r1
add r3, r2, r1
vst1.8 {d7}, [r2]
vst1.8 {d9}, [r3], r1
add r12, r3, r1
vst1.8 {d11}, [r3]
vst1.8 {d13}, [r12], r1
add r0, r12, r1
vst1.8 {d15}, [r12]
vst1.8 {d17}, [r0], r1
add r2, r0, r1
vst1.8 {d19}, [r0]
vst1.8 {d21}, [r2]
bx lr
ENDP ; |vp8_mbloop_filter_vertical_edge_y_neon|
;-----------------
AREA mbvloopfiltery_dat, DATA, READWRITE ;read/write by default
;Data section with name data_area is specified. DCD reserves space in memory for 16 data.
;One word each is reserved. Label filter_coeff can be used to access the data.
;Data address: filter_coeff, filter_coeff+4, filter_coeff+8 ...
_mbvlfy_coeff_
DCD mbvlfy_coeff
mbvlfy_coeff
DCD 0x80808080, 0x80808080, 0x80808080, 0x80808080
DCD 0x03030303, 0x03030303, 0x03030303, 0x03030303
DCD 0x04040404, 0x04040404, 0x04040404, 0x04040404
DCD 0x003f003f, 0x003f003f, 0x003f003f, 0x003f003f
DCD 0x09090909, 0x09090909, 0x12121212, 0x12121212
DCD 0x1b1b1b1b, 0x1b1b1b1b
END

29
vp8/common/arm/neon/recon_neon.c Normal file
View File

@@ -0,0 +1,29 @@
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_ports/config.h"
#include "recon.h"
#include "blockd.h"
extern void vp8_recon16x16mb_neon(unsigned char *pred_ptr, short *diff_ptr, unsigned char *dst_ptr, int ystride, unsigned char *udst_ptr, unsigned char *vdst_ptr);
void vp8_recon_mb_neon(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
unsigned char *pred_ptr = &x->predictor[0];
short *diff_ptr = &x->diff[0];
unsigned char *dst_ptr = x->dst.y_buffer;
unsigned char *udst_ptr = x->dst.u_buffer;
unsigned char *vdst_ptr = x->dst.v_buffer;
int ystride = x->dst.y_stride;
/*int uv_stride = x->dst.uv_stride;*/
vp8_recon16x16mb_neon(pred_ptr, diff_ptr, dst_ptr, ystride, udst_ptr, vdst_ptr);
}

109
vp8/common/arm/recon_arm.c
View File

@@ -1,109 +0,0 @@
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_ports/config.h"
#include "recon.h"
#include "blockd.h"
extern void vp8_recon16x16mb_neon(unsigned char *pred_ptr, short *diff_ptr, unsigned char *dst_ptr, int ystride, unsigned char *udst_ptr, unsigned char *vdst_ptr);
/*
void vp8_recon16x16mby(MACROBLOCKD *x)
{
int i;
for(i=0;i<16;i+=4)
{
//vp8_recon4b(&x->block[i]);
BLOCKD *b = &x->block[i];
vp8_recon4b (b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
}
*/
void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
BLOCKD *b = &x->block[0];
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[4];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[8];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[12];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
#if HAVE_ARMV7
void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
unsigned char *pred_ptr = &x->predictor[0];
short *diff_ptr = &x->diff[0];
unsigned char *dst_ptr = x->dst.y_buffer;
unsigned char *udst_ptr = x->dst.u_buffer;
unsigned char *vdst_ptr = x->dst.v_buffer;
int ystride = x->dst.y_stride;
//int uv_stride = x->dst.uv_stride;
vp8_recon16x16mb_neon(pred_ptr, diff_ptr, dst_ptr, ystride, udst_ptr, vdst_ptr);
}
#else
/*
void vp8_recon16x16mb(MACROBLOCKD *x)
{
int i;
for(i=0;i<16;i+=4)
{
// vp8_recon4b(&x->block[i]);
BLOCKD *b = &x->block[i];
vp8_recon4b (b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
for(i=16;i<24;i+=2)
{
// vp8_recon2b(&x->block[i]);
BLOCKD *b = &x->block[i];
vp8_recon2b (b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
}
*/
void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
BLOCKD *b = &x->block[0];
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
//b = &x->block[16];
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
#endif

9
vp8/common/arm/recon_arm.h
View File

@@ -21,6 +21,7 @@ extern prototype_copy_block(vp8_copy_mem8x8_v6);
extern prototype_copy_block(vp8_copy_mem8x4_v6);
extern prototype_copy_block(vp8_copy_mem16x16_v6);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_recon_recon
#define vp8_recon_recon vp8_recon_b_armv6
@@ -39,6 +40,7 @@ extern prototype_copy_block(vp8_copy_mem16x16_v6);
#undef vp8_recon_copy16x16
#define vp8_recon_copy16x16 vp8_copy_mem16x16_v6
#endif
#endif
#if HAVE_ARMV7
extern prototype_recon_block(vp8_recon_b_neon);
@@ -49,6 +51,9 @@ extern prototype_copy_block(vp8_copy_mem8x8_neon);
extern prototype_copy_block(vp8_copy_mem8x4_neon);
extern prototype_copy_block(vp8_copy_mem16x16_neon);
extern prototype_recon_macroblock(vp8_recon_mb_neon);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_recon_recon
#define vp8_recon_recon vp8_recon_b_neon
@@ -66,6 +71,10 @@ extern prototype_copy_block(vp8_copy_mem16x16_neon);
#undef vp8_recon_copy16x16
#define vp8_recon_copy16x16 vp8_copy_mem16x16_neon
#undef vp8_recon_recon_mb
#define vp8_recon_recon_mb vp8_recon_mb_neon
#endif
#endif
#endif

409
vp8/common/arm/reconintra4x4_arm.c
View File

@@ -1,409 +0,0 @@
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_ports/config.h"
#include "recon.h"
#include "vpx_mem/vpx_mem.h"
#include "reconintra.h"
void vp8_predict_intra4x4(BLOCKD *x,
int b_mode,
unsigned char *predictor)
{
int i, r, c;
unsigned char *Above = *(x->base_dst) + x->dst - x->dst_stride;
unsigned char Left[4];
unsigned char top_left = Above[-1];
Left[0] = (*(x->base_dst))[x->dst - 1];
Left[1] = (*(x->base_dst))[x->dst - 1 + x->dst_stride];
Left[2] = (*(x->base_dst))[x->dst - 1 + 2 * x->dst_stride];
Left[3] = (*(x->base_dst))[x->dst - 1 + 3 * x->dst_stride];
switch (b_mode)
{
case B_DC_PRED:
{
int expected_dc = 0;
for (i = 0; i < 4; i++)
{
expected_dc += Above[i];
expected_dc += Left[i];
}
expected_dc = (expected_dc + 4) >> 3;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
predictor[c] = expected_dc;
}
predictor += 16;
}
}
break;
case B_TM_PRED:
{
// prediction similar to true_motion prediction
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
int pred = Above[c] - top_left + Left[r];
if (pred < 0)
pred = 0;
if (pred > 255)
pred = 255;
predictor[c] = pred;
}
predictor += 16;
}
}
break;
case B_VE_PRED:
{
unsigned int ap[4];
ap[0] = (top_left + 2 * Above[0] + Above[1] + 2) >> 2;
ap[1] = (Above[0] + 2 * Above[1] + Above[2] + 2) >> 2;
ap[2] = (Above[1] + 2 * Above[2] + Above[3] + 2) >> 2;
ap[3] = (Above[2] + 2 * Above[3] + Above[4] + 2) >> 2;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
predictor[c] = ap[c];
}
predictor += 16;
}
}
break;
case B_HE_PRED:
{
unsigned int lp[4];
lp[0] = (top_left + 2 * Left[0] + Left[1] + 2) >> 2;
lp[1] = (Left[0] + 2 * Left[1] + Left[2] + 2) >> 2;
lp[2] = (Left[1] + 2 * Left[2] + Left[3] + 2) >> 2;
lp[3] = (Left[2] + 2 * Left[3] + Left[3] + 2) >> 2;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
predictor[c] = lp[r];
}
predictor += 16;
}
}
break;
case B_LD_PRED:
{
unsigned char *ptr = Above;
predictor[0 * 16 + 0] = (ptr[0] + ptr[1] * 2 + ptr[2] + 2) >> 2;
predictor[0 * 16 + 1] =
predictor[1 * 16 + 0] = (ptr[1] + ptr[2] * 2 + ptr[3] + 2) >> 2;
predictor[0 * 16 + 2] =
predictor[1 * 16 + 1] =
predictor[2 * 16 + 0] = (ptr[2] + ptr[3] * 2 + ptr[4] + 2) >> 2;
predictor[0 * 16 + 3] =
predictor[1 * 16 + 2] =
predictor[2 * 16 + 1] =
predictor[3 * 16 + 0] = (ptr[3] + ptr[4] * 2 + ptr[5] + 2) >> 2;
predictor[1 * 16 + 3] =
predictor[2 * 16 + 2] =
predictor[3 * 16 + 1] = (ptr[4] + ptr[5] * 2 + ptr[6] + 2) >> 2;
predictor[2 * 16 + 3] =
predictor[3 * 16 + 2] = (ptr[5] + ptr[6] * 2 + ptr[7] + 2) >> 2;
predictor[3 * 16 + 3] = (ptr[6] + ptr[7] * 2 + ptr[7] + 2) >> 2;
}
break;
case B_RD_PRED:
{
unsigned char pp[9];
pp[0] = Left[3];
pp[1] = Left[2];
pp[2] = Left[1];
pp[3] = Left[0];
pp[4] = top_left;
pp[5] = Above[0];
pp[6] = Above[1];
pp[7] = Above[2];
pp[8] = Above[3];
predictor[3 * 16 + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
predictor[3 * 16 + 1] =
predictor[2 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[3 * 16 + 2] =
predictor[2 * 16 + 1] =
predictor[1 * 16 + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
predictor[3 * 16 + 3] =
predictor[2 * 16 + 2] =
predictor[1 * 16 + 1] =
predictor[0 * 16 + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
predictor[2 * 16 + 3] =
predictor[1 * 16 + 2] =
predictor[0 * 16 + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
predictor[1 * 16 + 3] =
predictor[0 * 16 + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
predictor[0 * 16 + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
}
break;
case B_VR_PRED:
{
unsigned char pp[9];
pp[0] = Left[3];
pp[1] = Left[2];
pp[2] = Left[1];
pp[3] = Left[0];
pp[4] = top_left;
pp[5] = Above[0];
pp[6] = Above[1];
pp[7] = Above[2];
pp[8] = Above[3];
predictor[3 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[2 * 16 + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
predictor[3 * 16 + 1] =
predictor[1 * 16 + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
predictor[2 * 16 + 1] =
predictor[0 * 16 + 0] = (pp[4] + pp[5] + 1) >> 1;
predictor[3 * 16 + 2] =
predictor[1 * 16 + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
predictor[2 * 16 + 2] =
predictor[0 * 16 + 1] = (pp[5] + pp[6] + 1) >> 1;
predictor[3 * 16 + 3] =
predictor[1 * 16 + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
predictor[2 * 16 + 3] =
predictor[0 * 16 + 2] = (pp[6] + pp[7] + 1) >> 1;
predictor[1 * 16 + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
predictor[0 * 16 + 3] = (pp[7] + pp[8] + 1) >> 1;
}
break;
case B_VL_PRED:
{
unsigned char *pp = Above;
predictor[0 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
predictor[1 * 16 + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
predictor[2 * 16 + 0] =
predictor[0 * 16 + 1] = (pp[1] + pp[2] + 1) >> 1;
predictor[1 * 16 + 1] =
predictor[3 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[2 * 16 + 1] =
predictor[0 * 16 + 2] = (pp[2] + pp[3] + 1) >> 1;
predictor[3 * 16 + 1] =
predictor[1 * 16 + 2] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
predictor[0 * 16 + 3] =
predictor[2 * 16 + 2] = (pp[3] + pp[4] + 1) >> 1;
predictor[1 * 16 + 3] =
predictor[3 * 16 + 2] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
predictor[2 * 16 + 3] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
predictor[3 * 16 + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
}
break;
case B_HD_PRED:
{
unsigned char pp[9];
pp[0] = Left[3];
pp[1] = Left[2];
pp[2] = Left[1];
pp[3] = Left[0];
pp[4] = top_left;
pp[5] = Above[0];
pp[6] = Above[1];
pp[7] = Above[2];
pp[8] = Above[3];
predictor[3 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
predictor[3 * 16 + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
predictor[2 * 16 + 0] =
predictor[3 * 16 + 2] = (pp[1] + pp[2] + 1) >> 1;
predictor[2 * 16 + 1] =
predictor[3 * 16 + 3] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[2 * 16 + 2] =
predictor[1 * 16 + 0] = (pp[2] + pp[3] + 1) >> 1;
predictor[2 * 16 + 3] =
predictor[1 * 16 + 1] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
predictor[1 * 16 + 2] =
predictor[0 * 16 + 0] = (pp[3] + pp[4] + 1) >> 1;
predictor[1 * 16 + 3] =
predictor[0 * 16 + 1] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
predictor[0 * 16 + 2] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
predictor[0 * 16 + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
}
break;
case B_HU_PRED:
{
unsigned char *pp = Left;
predictor[0 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
predictor[0 * 16 + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
predictor[0 * 16 + 2] =
predictor[1 * 16 + 0] = (pp[1] + pp[2] + 1) >> 1;
predictor[0 * 16 + 3] =
predictor[1 * 16 + 1] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[1 * 16 + 2] =
predictor[2 * 16 + 0] = (pp[2] + pp[3] + 1) >> 1;
predictor[1 * 16 + 3] =
predictor[2 * 16 + 1] = (pp[2] + pp[3] * 2 + pp[3] + 2) >> 2;
predictor[2 * 16 + 2] =
predictor[2 * 16 + 3] =
predictor[3 * 16 + 0] =
predictor[3 * 16 + 1] =
predictor[3 * 16 + 2] =
predictor[3 * 16 + 3] = pp[3];
}
break;
}
}
// copy 4 bytes from the above right down so that the 4x4 prediction modes using pixels above and
// to the right prediction have filled in pixels to use.
void vp8_intra_prediction_down_copy(MACROBLOCKD *x)
{
unsigned char *above_right = *(x->block[0].base_dst) + x->block[0].dst - x->block[0].dst_stride + 16;
unsigned int *src_ptr = (unsigned int *)above_right;
unsigned int *dst_ptr0 = (unsigned int *)(above_right + 4 * x->block[0].dst_stride);
unsigned int *dst_ptr1 = (unsigned int *)(above_right + 8 * x->block[0].dst_stride);
unsigned int *dst_ptr2 = (unsigned int *)(above_right + 12 * x->block[0].dst_stride);
*dst_ptr0 = *src_ptr;
*dst_ptr1 = *src_ptr;
*dst_ptr2 = *src_ptr;
}
/*
void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
int i;
vp8_intra_prediction_down_copy(x);
for(i=0;i<16;i++)
{
BLOCKD *b = &x->block[i];
vp8_predict_intra4x4(b, x->block[i].bmi.mode,x->block[i].predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
vp8_recon_intra_mbuv(x);
}
*/
void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
int i;
BLOCKD *b = &x->block[0];
vp8_intra_prediction_down_copy(x);
{
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
vp8_recon_intra_mbuv(rtcd, x);
}

4
vp8/common/arm/subpixel_arm.h
View File

@@ -22,6 +22,7 @@ extern prototype_subpixel_predict(vp8_bilinear_predict8x8_armv6);
extern prototype_subpixel_predict(vp8_bilinear_predict8x4_armv6);
extern prototype_subpixel_predict(vp8_bilinear_predict4x4_armv6);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_subpix_sixtap16x16
#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_armv6
@@ -46,6 +47,7 @@ extern prototype_subpixel_predict(vp8_bilinear_predict4x4_armv6);
#undef vp8_subpix_bilinear4x4
#define vp8_subpix_bilinear4x4 vp8_bilinear_predict4x4_armv6
#endif
#endif
#if HAVE_ARMV7
extern prototype_subpixel_predict(vp8_sixtap_predict16x16_neon);
@@ -57,6 +59,7 @@ extern prototype_subpixel_predict(vp8_bilinear_predict8x8_neon);
extern prototype_subpixel_predict(vp8_bilinear_predict8x4_neon);
extern prototype_subpixel_predict(vp8_bilinear_predict4x4_neon);
#if !CONFIG_RUNTIME_CPU_DETECT
#undef vp8_subpix_sixtap16x16
#define vp8_subpix_sixtap16x16 vp8_sixtap_predict16x16_neon
@@ -81,5 +84,6 @@ extern prototype_subpixel_predict(vp8_bilinear_predict4x4_neon);
#undef vp8_subpix_bilinear4x4
#define vp8_subpix_bilinear4x4 vp8_bilinear_predict4x4_neon
#endif
#endif
#endif

149
vp8/common/arm/systemdependent.c
View File

@@ -1,149 +0,0 @@
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
*/
#include "vpx_ports/config.h"
#include "g_common.h"
#include "pragmas.h"
#include "subpixel.h"
#include "loopfilter.h"
#include "recon.h"
#include "idct.h"
#include "onyxc_int.h"
void (*vp8_build_intra_predictors_mby_ptr)(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby_neon(MACROBLOCKD *x);
void (*vp8_build_intra_predictors_mby_s_ptr)(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby_s_neon(MACROBLOCKD *x);
void vp8_machine_specific_config(VP8_COMMON *ctx)
{
#if CONFIG_RUNTIME_CPU_DETECT
VP8_COMMON_RTCD *rtcd = &ctx->rtcd;
#if HAVE_ARMV7
rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_neon;
rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_neon;
rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_neon;
rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_neon;
rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_neon;
rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_neon;
rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_neon;
rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_neon;
rtcd->idct.idct1 = vp8_short_idct4x4llm_1_neon;
rtcd->idct.idct16 = vp8_short_idct4x4llm_neon;
rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_neon;
rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_neon;
rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_neon;
rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_neon;
rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_neon;
rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_neon;
rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_neon;
rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_neon;
rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_neon;
rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_neon;
rtcd->recon.copy16x16 = vp8_copy_mem16x16_neon;
rtcd->recon.copy8x8 = vp8_copy_mem8x8_neon;
rtcd->recon.copy8x4 = vp8_copy_mem8x4_neon;
rtcd->recon.recon = vp8_recon_b_neon;
rtcd->recon.recon2 = vp8_recon2b_neon;
rtcd->recon.recon4 = vp8_recon4b_neon;
#elif HAVE_ARMV6
rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_armv6;
rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_armv6;
rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_armv6;
rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_armv6;
rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_armv6;
rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_armv6;
rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_armv6;
rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_armv6;
rtcd->idct.idct1 = vp8_short_idct4x4llm_1_v6;
rtcd->idct.idct16 = vp8_short_idct4x4llm_v6_dual;
rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_armv6;
rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_armv6;
rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_armv6;
rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_armv6;
rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_armv6;
rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_armv6;
rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_armv6;
rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_armv6;
rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_armv6;
rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_armv6;
rtcd->recon.copy16x16 = vp8_copy_mem16x16_v6;
rtcd->recon.copy8x8 = vp8_copy_mem8x8_v6;
rtcd->recon.copy8x4 = vp8_copy_mem8x4_v6;
rtcd->recon.recon = vp8_recon_b_armv6;
rtcd->recon.recon2 = vp8_recon2b_armv6;
rtcd->recon.recon4 = vp8_recon4b_armv6;
#else
//pure c
rtcd->idct.idct1 = vp8_short_idct4x4llm_1_c;
rtcd->idct.idct16 = vp8_short_idct4x4llm_c;
rtcd->idct.idct1_scalar = vp8_dc_only_idct_c;
rtcd->idct.iwalsh1 = vp8_short_inv_walsh4x4_1_c;
rtcd->idct.iwalsh16 = vp8_short_inv_walsh4x4_c;
rtcd->recon.copy16x16 = vp8_copy_mem16x16_c;
rtcd->recon.copy8x8 = vp8_copy_mem8x8_c;
rtcd->recon.copy8x4 = vp8_copy_mem8x4_c;
rtcd->recon.recon = vp8_recon_b_c;
rtcd->recon.recon2 = vp8_recon2b_c;
rtcd->recon.recon4 = vp8_recon4b_c;
rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_c;
rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_c;
rtcd->subpix.sixtap8x4 = vp8_sixtap_predict8x4_c;
rtcd->subpix.sixtap4x4 = vp8_sixtap_predict_c;
rtcd->subpix.bilinear16x16 = vp8_bilinear_predict16x16_c;
rtcd->subpix.bilinear8x8 = vp8_bilinear_predict8x8_c;
rtcd->subpix.bilinear8x4 = vp8_bilinear_predict8x4_c;
rtcd->subpix.bilinear4x4 = vp8_bilinear_predict4x4_c;
rtcd->loopfilter.normal_mb_v = vp8_loop_filter_mbv_c;
rtcd->loopfilter.normal_b_v = vp8_loop_filter_bv_c;
rtcd->loopfilter.normal_mb_h = vp8_loop_filter_mbh_c;
rtcd->loopfilter.normal_b_h = vp8_loop_filter_bh_c;
rtcd->loopfilter.simple_mb_v = vp8_loop_filter_mbvs_c;
rtcd->loopfilter.simple_b_v = vp8_loop_filter_bvs_c;
rtcd->loopfilter.simple_mb_h = vp8_loop_filter_mbhs_c;
rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_c;
#endif
#if CONFIG_POSTPROC || (CONFIG_VP8_ENCODER && CONFIG_PSNR)
rtcd->postproc.down = vp8_mbpost_proc_down_c;
rtcd->postproc.across = vp8_mbpost_proc_across_ip_c;
rtcd->postproc.downacross = vp8_post_proc_down_and_across_c;
rtcd->postproc.addnoise = vp8_plane_add_noise_c;
#endif
#endif
#if HAVE_ARMV7
vp8_build_intra_predictors_mby_ptr = vp8_build_intra_predictors_mby_neon;
vp8_build_intra_predictors_mby_s_ptr = vp8_build_intra_predictors_mby_s_neon;
#elif HAVE_ARMV6
vp8_build_intra_predictors_mby_ptr = vp8_build_intra_predictors_mby;
vp8_build_intra_predictors_mby_s_ptr = vp8_build_intra_predictors_mby_s;
#else
vp8_build_intra_predictors_mby_ptr = vp8_build_intra_predictors_mby;
vp8_build_intra_predictors_mby_s_ptr = vp8_build_intra_predictors_mby_s;
#endif
}

82
vp8/common/blockd.h
View File

@@ -24,7 +24,7 @@ void vpx_log(const char *format, ...);
#define TRUE 1
#define FALSE 0
//#define DCPRED 1
/*#define DCPRED 1*/
#define DCPREDSIMTHRESH 0
#define DCPREDCNTTHRESH 3
@@ -39,7 +39,7 @@ void vpx_log(const char *format, ...);
#define MAX_REF_LF_DELTAS 4
#define MAX_MODE_LF_DELTAS 4
// Segment Feature Masks
/* Segment Feature Masks */
#define SEGMENT_DELTADATA 0
#define SEGMENT_ABSDATA 1
@@ -75,11 +75,11 @@ typedef enum
typedef enum
{
DC_PRED, // average of above and left pixels
V_PRED, // vertical prediction
H_PRED, // horizontal prediction
TM_PRED, // Truemotion prediction
B_PRED, // block based prediction, each block has its own prediction mode
DC_PRED, /* average of above and left pixels */
V_PRED, /* vertical prediction */
H_PRED, /* horizontal prediction */
TM_PRED, /* Truemotion prediction */
B_PRED, /* block based prediction, each block has its own prediction mode */
NEARESTMV,
NEARMV,
@@ -90,16 +90,16 @@ typedef enum
MB_MODE_COUNT
} MB_PREDICTION_MODE;
// Macroblock level features
/* Macroblock level features */
typedef enum
{
MB_LVL_ALT_Q = 0, // Use alternate Quantizer ....
MB_LVL_ALT_LF = 1, // Use alternate loop filter value...
MB_LVL_MAX = 2, // Number of MB level features supported
MB_LVL_ALT_Q = 0, /* Use alternate Quantizer .... */
MB_LVL_ALT_LF = 1, /* Use alternate loop filter value... */
MB_LVL_MAX = 2 /* Number of MB level features supported */
} MB_LVL_FEATURES;
// Segment Feature Masks
/* Segment Feature Masks */
#define SEGMENT_ALTQ 0x01
#define SEGMENT_ALT_LF 0x02
@@ -110,11 +110,11 @@ typedef enum
typedef enum
{
B_DC_PRED, // average of above and left pixels
B_DC_PRED, /* average of above and left pixels */
B_TM_PRED,
B_VE_PRED, // vertical prediction
B_HE_PRED, // horizontal prediction
B_VE_PRED, /* vertical prediction */
B_HE_PRED, /* horizontal prediction */
B_LD_PRED,
B_RD_PRED,
@@ -169,14 +169,14 @@ typedef struct
MV as_mv;
} mv;
char partitioning;
unsigned char mb_skip_coeff; //does this mb has coefficients at all, 1=no coefficients, 0=need decode tokens
unsigned char partitioning;
unsigned char mb_skip_coeff; /* does this mb has coefficients at all, 1=no coefficients, 0=need decode tokens */
unsigned char dc_diff;
unsigned char need_to_clamp_mvs;
unsigned char segment_id; // Which set of segmentation parameters should be used for this MB
unsigned char segment_id; /* Which set of segmentation parameters should be used for this MB */
unsigned char force_no_skip; //encoder only
unsigned char force_no_skip; /* encoder only */
} MB_MODE_INFO;
@@ -195,9 +195,9 @@ typedef struct
short *diff;
short *reference;
short(*dequant)[4];
short *dequant;
// 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries
/* 16 Y blocks, 4 U blocks, 4 V blocks each with 16 entries */
unsigned char **base_pre;
int pre;
int pre_stride;
@@ -214,17 +214,17 @@ typedef struct
typedef struct
{
DECLARE_ALIGNED(16, short, diff[400]); // from idct diff
DECLARE_ALIGNED(16, short, diff[400]); /* from idct diff */
DECLARE_ALIGNED(16, unsigned char, predictor[384]);
//not used DECLARE_ALIGNED(16, short, reference[384]);
/* not used DECLARE_ALIGNED(16, short, reference[384]); */
DECLARE_ALIGNED(16, short, qcoeff[400]);
DECLARE_ALIGNED(16, short, dqcoeff[400]);
DECLARE_ALIGNED(16, char, eobs[25]);
// 16 Y blocks, 4 U, 4 V, 1 DC 2nd order block, each with 16 entries.
/* 16 Y blocks, 4 U, 4 V, 1 DC 2nd order block, each with 16 entries. */
BLOCKD block[25];
YV12_BUFFER_CONFIG pre; // Filtered copy of previous frame reconstruction
YV12_BUFFER_CONFIG pre; /* Filtered copy of previous frame reconstruction */
YV12_BUFFER_CONFIG dst;
MODE_INFO *mode_info_context;
@@ -235,39 +235,39 @@ typedef struct
int up_available;
int left_available;
// Y,U,V,Y2
/* Y,U,V,Y2 */
ENTROPY_CONTEXT_PLANES *above_context;
ENTROPY_CONTEXT_PLANES *left_context;
// 0 indicates segmentation at MB level is not enabled. Otherwise the individual bits indicate which features are active.
/* 0 indicates segmentation at MB level is not enabled. Otherwise the individual bits indicate which features are active. */
unsigned char segmentation_enabled;
// 0 (do not update) 1 (update) the macroblock segmentation map.
/* 0 (do not update) 1 (update) the macroblock segmentation map. */
unsigned char update_mb_segmentation_map;
// 0 (do not update) 1 (update) the macroblock segmentation feature data.
/* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
unsigned char update_mb_segmentation_data;
// 0 (do not update) 1 (update) the macroblock segmentation feature data.
/* 0 (do not update) 1 (update) the macroblock segmentation feature data. */
unsigned char mb_segement_abs_delta;
// Per frame flags that define which MB level features (such as quantizer or loop filter level)
// are enabled and when enabled the proabilities used to decode the per MB flags in MB_MODE_INFO
vp8_prob mb_segment_tree_probs[MB_FEATURE_TREE_PROBS]; // Probability Tree used to code Segment number
/* Per frame flags that define which MB level features (such as quantizer or loop filter level) */
/* are enabled and when enabled the proabilities used to decode the per MB flags in MB_MODE_INFO */
vp8_prob mb_segment_tree_probs[MB_FEATURE_TREE_PROBS]; /* Probability Tree used to code Segment number */
signed char segment_feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS]; // Segment parameters
signed char segment_feature_data[MB_LVL_MAX][MAX_MB_SEGMENTS]; /* Segment parameters */
// mode_based Loop filter adjustment
/* mode_based Loop filter adjustment */
unsigned char mode_ref_lf_delta_enabled;
unsigned char mode_ref_lf_delta_update;
// Delta values have the range +/- MAX_LOOP_FILTER
//char ref_lf_deltas[MAX_REF_LF_DELTAS]; // 0 = Intra, Last, GF, ARF
//char mode_lf_deltas[MAX_MODE_LF_DELTAS]; // 0 = BPRED, ZERO_MV, MV, SPLIT
signed char ref_lf_deltas[MAX_REF_LF_DELTAS]; // 0 = Intra, Last, GF, ARF
signed char mode_lf_deltas[MAX_MODE_LF_DELTAS]; // 0 = BPRED, ZERO_MV, MV, SPLIT
/* Delta values have the range +/- MAX_LOOP_FILTER */
signed char last_ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
signed char ref_lf_deltas[MAX_REF_LF_DELTAS]; /* 0 = Intra, Last, GF, ARF */
signed char last_mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
signed char mode_lf_deltas[MAX_MODE_LF_DELTAS]; /* 0 = BPRED, ZERO_MV, MV, SPLIT */
// Distance of MB away from frame edges
/* Distance of MB away from frame edges */
int mb_to_left_edge;
int mb_to_right_edge;
int mb_to_top_edge;

10
vp8/common/debugmodes.c
View File

@@ -21,7 +21,7 @@ void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols, int f
int mb_index = 0;
FILE *mvs = fopen("mvs.stt", "a");
// print out the macroblock Y modes
/* print out the macroblock Y modes */
mb_index = 0;
fprintf(mvs, "Mb Modes for Frame %d\n", frame);
@@ -60,7 +60,7 @@ void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols, int f
fprintf(mvs, "\n");
// print out the macroblock UV modes
/* print out the macroblock UV modes */
mb_index = 0;
fprintf(mvs, "UV Modes for Frame %d\n", frame);
@@ -80,7 +80,7 @@ void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols, int f
fprintf(mvs, "\n");
// print out the block modes
/* print out the block modes */
mb_index = 0;
fprintf(mvs, "Mbs for Frame %d\n", frame);
{
@@ -108,7 +108,7 @@ void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols, int f
}
fprintf(mvs, "\n");
// print out the macroblock mvs
/* print out the macroblock mvs */
mb_index = 0;
fprintf(mvs, "MVs for Frame %d\n", frame);
@@ -128,7 +128,7 @@ void vp8_print_modes_and_motion_vectors(MODE_INFO *mi, int rows, int cols, int f
fprintf(mvs, "\n");
// print out the block modes
/* print out the block modes */
mb_index = 0;
fprintf(mvs, "MVs for Frame %d\n", frame);
{

72
vp8/common/defaultcoefcounts.h
View File

@@ -15,204 +15,204 @@ static const unsigned int default_coef_counts [BLOCK_TYPES] [COEF_BANDS] [PREV_C
{
{
// Block Type ( 0 )
/* Block Type ( 0 ) */
{
// Coeff Band ( 0 )
/* Coeff Band ( 0 ) */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
},
{
// Coeff Band ( 1 )
/* Coeff Band ( 1 ) */
{30190, 26544, 225, 24, 4, 0, 0, 0, 0, 0, 0, 4171593,},
{26846, 25157, 1241, 130, 26, 6, 1, 0, 0, 0, 0, 149987,},
{10484, 9538, 1006, 160, 36, 18, 0, 0, 0, 0, 0, 15104,},
},
{
// Coeff Band ( 2 )
/* Coeff Band ( 2 ) */
{25842, 40456, 1126, 83, 11, 2, 0, 0, 0, 0, 0, 0,},
{9338, 8010, 512, 73, 7, 3, 2, 0, 0, 0, 0, 43294,},
{1047, 751, 149, 31, 13, 6, 1, 0, 0, 0, 0, 879,},
},
{
// Coeff Band ( 3 )
/* Coeff Band ( 3 ) */
{26136, 9826, 252, 13, 0, 0, 0, 0, 0, 0, 0, 0,},
{8134, 5574, 191, 14, 2, 0, 0, 0, 0, 0, 0, 35302,},
{ 605, 677, 116, 9, 1, 0, 0, 0, 0, 0, 0, 611,},
},
{
// Coeff Band ( 4 )
/* Coeff Band ( 4 ) */
{10263, 15463, 283, 17, 0, 0, 0, 0, 0, 0, 0, 0,},
{2773, 2191, 128, 9, 2, 2, 0, 0, 0, 0, 0, 10073,},
{ 134, 125, 32, 4, 0, 2, 0, 0, 0, 0, 0, 50,},
},
{
// Coeff Band ( 5 )
/* Coeff Band ( 5 ) */
{10483, 2663, 23, 1, 0, 0, 0, 0, 0, 0, 0, 0,},
{2137, 1251, 27, 1, 1, 0, 0, 0, 0, 0, 0, 14362,},
{ 116, 156, 14, 2, 1, 0, 0, 0, 0, 0, 0, 190,},
},
{
// Coeff Band ( 6 )
/* Coeff Band ( 6 ) */
{40977, 27614, 412, 28, 0, 0, 0, 0, 0, 0, 0, 0,},
{6113, 5213, 261, 22, 3, 0, 0, 0, 0, 0, 0, 26164,},
{ 382, 312, 50, 14, 2, 0, 0, 0, 0, 0, 0, 345,},
},
{
// Coeff Band ( 7 )
/* Coeff Band ( 7 ) */
{ 0, 26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 0, 13, 0, 0, 0, 0, 0, 0, 0, 0, 0, 319,},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8,},
},
},
{
// Block Type ( 1 )
/* Block Type ( 1 ) */
{
// Coeff Band ( 0 )
/* Coeff Band ( 0 ) */
{3268, 19382, 1043, 250, 93, 82, 49, 26, 17, 8, 25, 82289,},
{8758, 32110, 5436, 1832, 827, 668, 420, 153, 24, 0, 3, 52914,},
{9337, 23725, 8487, 3954, 2107, 1836, 1069, 399, 59, 0, 0, 18620,},
},
{
// Coeff Band ( 1 )
/* Coeff Band ( 1 ) */
{12419, 8420, 452, 62, 9, 1, 0, 0, 0, 0, 0, 0,},
{11715, 8705, 693, 92, 15, 7, 2, 0, 0, 0, 0, 53988,},
{7603, 8585, 2306, 778, 270, 145, 39, 5, 0, 0, 0, 9136,},
},
{
// Coeff Band ( 2 )
/* Coeff Band ( 2 ) */
{15938, 14335, 1207, 184, 55, 13, 4, 1, 0, 0, 0, 0,},
{7415, 6829, 1138, 244, 71, 26, 7, 0, 0, 0, 0, 9980,},
{1580, 1824, 655, 241, 89, 46, 10, 2, 0, 0, 0, 429,},
},
{
// Coeff Band ( 3 )
/* Coeff Band ( 3 ) */
{19453, 5260, 201, 19, 0, 0, 0, 0, 0, 0, 0, 0,},
{9173, 3758, 213, 22, 1, 1, 0, 0, 0, 0, 0, 9820,},
{1689, 1277, 276, 51, 17, 4, 0, 0, 0, 0, 0, 679,},
},
{
// Coeff Band ( 4 )
/* Coeff Band ( 4 ) */
{12076, 10667, 620, 85, 19, 9, 5, 0, 0, 0, 0, 0,},
{4665, 3625, 423, 55, 19, 9, 0, 0, 0, 0, 0, 5127,},
{ 415, 440, 143, 34, 20, 7, 2, 0, 0, 0, 0, 101,},
},
{
// Coeff Band ( 5 )
/* Coeff Band ( 5 ) */
{12183, 4846, 115, 11, 1, 0, 0, 0, 0, 0, 0, 0,},
{4226, 3149, 177, 21, 2, 0, 0, 0, 0, 0, 0, 7157,},
{ 375, 621, 189, 51, 11, 4, 1, 0, 0, 0, 0, 198,},
},
{
// Coeff Band ( 6 )
/* Coeff Band ( 6 ) */
{61658, 37743, 1203, 94, 10, 3, 0, 0, 0, 0, 0, 0,},
{15514, 11563, 903, 111, 14, 5, 0, 0, 0, 0, 0, 25195,},
{ 929, 1077, 291, 78, 14, 7, 1, 0, 0, 0, 0, 507,},
},
{
// Coeff Band ( 7 )
/* Coeff Band ( 7 ) */
{ 0, 990, 15, 3, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 0, 412, 13, 0, 0, 0, 0, 0, 0, 0, 0, 1641,},
{ 0, 18, 7, 1, 0, 0, 0, 0, 0, 0, 0, 30,},
},
},
{
// Block Type ( 2 )
/* Block Type ( 2 ) */
{
// Coeff Band ( 0 )
/* Coeff Band ( 0 ) */
{ 953, 24519, 628, 120, 28, 12, 4, 0, 0, 0, 0, 2248798,},
{1525, 25654, 2647, 617, 239, 143, 42, 5, 0, 0, 0, 66837,},
{1180, 11011, 3001, 1237, 532, 448, 239, 54, 5, 0, 0, 7122,},
},
{
// Coeff Band ( 1 )
/* Coeff Band ( 1 ) */
{1356, 2220, 67, 10, 4, 1, 0, 0, 0, 0, 0, 0,},
{1450, 2544, 102, 18, 4, 3, 0, 0, 0, 0, 0, 57063,},
{1182, 2110, 470, 130, 41, 21, 0, 0, 0, 0, 0, 6047,},
},
{
// Coeff Band ( 2 )
/* Coeff Band ( 2 ) */
{ 370, 3378, 200, 30, 5, 4, 1, 0, 0, 0, 0, 0,},
{ 293, 1006, 131, 29, 11, 0, 0, 0, 0, 0, 0, 5404,},
{ 114, 387, 98, 23, 4, 8, 1, 0, 0, 0, 0, 236,},
},
{
// Coeff Band ( 3 )
/* Coeff Band ( 3 ) */
{ 579, 194, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 395, 213, 5, 1, 0, 0, 0, 0, 0, 0, 0, 4157,},
{ 119, 122, 4, 0, 0, 0, 0, 0, 0, 0, 0, 300,},
},
{
// Coeff Band ( 4 )
/* Coeff Band ( 4 ) */
{ 38, 557, 19, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 21, 114, 12, 1, 0, 0, 0, 0, 0, 0, 0, 427,},
{ 0, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7,},
},
{
// Coeff Band ( 5 )
/* Coeff Band ( 5 ) */
{ 52, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 18, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 652,},
{ 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 30,},
},
{
// Coeff Band ( 6 )
/* Coeff Band ( 6 ) */
{ 640, 569, 10, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 25, 77, 2, 0, 0, 0, 0, 0, 0, 0, 0, 517,},
{ 4, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 3,},
},
{
// Coeff Band ( 7 )
/* Coeff Band ( 7 ) */
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
},
},
{
// Block Type ( 3 )
/* Block Type ( 3 ) */
{
// Coeff Band ( 0 )
/* Coeff Band ( 0 ) */
{2506, 20161, 2707, 767, 261, 178, 107, 30, 14, 3, 0, 100694,},
{8806, 36478, 8817, 3268, 1280, 850, 401, 114, 42, 0, 0, 58572,},
{11003, 27214, 11798, 5716, 2482, 2072, 1048, 175, 32, 0, 0, 19284,},
},
{
// Coeff Band ( 1 )
/* Coeff Band ( 1 ) */
{9738, 11313, 959, 205, 70, 18, 11, 1, 0, 0, 0, 0,},
{12628, 15085, 1507, 273, 52, 19, 9, 0, 0, 0, 0, 54280,},
{10701, 15846, 5561, 1926, 813, 570, 249, 36, 0, 0, 0, 6460,},
},
{
// Coeff Band ( 2 )
/* Coeff Band ( 2 ) */
{6781, 22539, 2784, 634, 182, 123, 20, 4, 0, 0, 0, 0,},
{6263, 11544, 2649, 790, 259, 168, 27, 5, 0, 0, 0, 20539,},
{3109, 4075, 2031, 896, 457, 386, 158, 29, 0, 0, 0, 1138,},
},
{
// Coeff Band ( 3 )
/* Coeff Band ( 3 ) */
{11515, 4079, 465, 73, 5, 14, 2, 0, 0, 0, 0, 0,},
{9361, 5834, 650, 96, 24, 8, 4, 0, 0, 0, 0, 22181,},
{4343, 3974, 1360, 415, 132, 96, 14, 1, 0, 0, 0, 1267,},
},
{
// Coeff Band ( 4 )
/* Coeff Band ( 4 ) */
{4787, 9297, 823, 168, 44, 12, 4, 0, 0, 0, 0, 0,},
{3619, 4472, 719, 198, 60, 31, 3, 0, 0, 0, 0, 8401,},
{1157, 1175, 483, 182, 88, 31, 8, 0, 0, 0, 0, 268,},
},
{
// Coeff Band ( 5 )
/* Coeff Band ( 5 ) */
{8299, 1226, 32, 5, 1, 0, 0, 0, 0, 0, 0, 0,},
{3502, 1568, 57, 4, 1, 1, 0, 0, 0, 0, 0, 9811,},
{1055, 1070, 166, 29, 6, 1, 0, 0, 0, 0, 0, 527,},
},
{
// Coeff Band ( 6 )
/* Coeff Band ( 6 ) */
{27414, 27927, 1989, 347, 69, 26, 0, 0, 0, 0, 0, 0,},
{5876, 10074, 1574, 341, 91, 24, 4, 0, 0, 0, 0, 21954,},
{1571, 2171, 778, 324, 124, 65, 16, 0, 0, 0, 0, 979,},
},
{
// Coeff Band ( 7 )
/* Coeff Band ( 7 ) */
{ 0, 29, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,},
{ 0, 23, 0, 0, 0, 0, 0, 0, 0, 0, 0, 459,},
{ 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 13,},

26
vp8/common/entropy.h
View File

@@ -17,18 +17,18 @@
/* Coefficient token alphabet */
#define ZERO_TOKEN 0 //0 Extra Bits 0+0
#define ONE_TOKEN 1 //1 Extra Bits 0+1
#define TWO_TOKEN 2 //2 Extra Bits 0+1
#define THREE_TOKEN 3 //3 Extra Bits 0+1
#define FOUR_TOKEN 4 //4 Extra Bits 0+1
#define DCT_VAL_CATEGORY1 5 //5-6 Extra Bits 1+1
#define DCT_VAL_CATEGORY2 6 //7-10 Extra Bits 2+1
#define DCT_VAL_CATEGORY3 7 //11-26 Extra Bits 4+1
#define DCT_VAL_CATEGORY4 8 //11-26 Extra Bits 5+1
#define DCT_VAL_CATEGORY5 9 //27-58 Extra Bits 5+1
#define DCT_VAL_CATEGORY6 10 //59+ Extra Bits 11+1
#define DCT_EOB_TOKEN 11 //EOB Extra Bits 0+0
#define ZERO_TOKEN 0 /* 0 Extra Bits 0+0 */
#define ONE_TOKEN 1 /* 1 Extra Bits 0+1 */
#define TWO_TOKEN 2 /* 2 Extra Bits 0+1 */
#define THREE_TOKEN 3 /* 3 Extra Bits 0+1 */
#define FOUR_TOKEN 4 /* 4 Extra Bits 0+1 */
#define DCT_VAL_CATEGORY1 5 /* 5-6 Extra Bits 1+1 */
#define DCT_VAL_CATEGORY2 6 /* 7-10 Extra Bits 2+1 */
#define DCT_VAL_CATEGORY3 7 /* 11-26 Extra Bits 4+1 */
#define DCT_VAL_CATEGORY4 8 /* 11-26 Extra Bits 5+1 */
#define DCT_VAL_CATEGORY5 9 /* 27-58 Extra Bits 5+1 */
#define DCT_VAL_CATEGORY6 10 /* 59+ Extra Bits 11+1 */
#define DCT_EOB_TOKEN 11 /* EOB Extra Bits 0+0 */
#define vp8_coef_tokens 12
#define MAX_ENTROPY_TOKENS vp8_coef_tokens
@@ -83,7 +83,7 @@ extern DECLARE_ALIGNED(16, const unsigned char, vp8_coef_bands[16]);
coefficient band (and since zigzag positions 0, 1, and 2 are in
distinct bands). */
/*# define DC_TOKEN_CONTEXTS 3 // 00, 0!0, !0!0 */
/*# define DC_TOKEN_CONTEXTS 3*/ /* 00, 0!0, !0!0 */
# define PREV_COEF_CONTEXTS 3
extern DECLARE_ALIGNED(16, const unsigned char, vp8_prev_token_class[vp8_coef_tokens]);

16
vp8/common/entropymv.c
View File

@@ -29,21 +29,21 @@ const MV_CONTEXT vp8_mv_update_probs[2] =
const MV_CONTEXT vp8_default_mv_context[2] =
{
{{
// row
162, // is short
128, // sign
225, 146, 172, 147, 214, 39, 156, // short tree
128, 129, 132, 75, 145, 178, 206, 239, 254, 254 // long bits
/* row */
162, /* is short */
128, /* sign */
225, 146, 172, 147, 214, 39, 156, /* short tree */
128, 129, 132, 75, 145, 178, 206, 239, 254, 254 /* long bits */
}},
{{
// same for column
164, // is short
/* same for column */
164, /* is short */
128,
204, 170, 119, 235, 140, 230, 228,
128, 130, 130, 74, 148, 180, 203, 236, 254, 254 // long bits
128, 130, 130, 74, 148, 180, 203, 236, 254, 254 /* long bits */
}}
};

31
vp8/common/extend.c
View File

@@ -15,14 +15,14 @@
static void extend_plane_borders
(
unsigned char *s, // source
int sp, // pitch
int h, // height
int w, // width
int et, // extend top border
int el, // extend left border
int eb, // extend bottom border
int er // extend right border
unsigned char *s, /* source */
int sp, /* pitch */
int h, /* height */
int w, /* width */
int et, /* extend top border */
int el, /* extend left border */
int eb, /* extend bottom border */
int er /* extend right border */
)
{
@@ -31,7 +31,7 @@ static void extend_plane_borders
unsigned char *dest_ptr1, *dest_ptr2;
int linesize;
// copy the left and right most columns out
/* copy the left and right most columns out */
src_ptr1 = s;
src_ptr2 = s + w - 1;
dest_ptr1 = s - el;
@@ -39,8 +39,9 @@ static void extend_plane_borders
for (i = 0; i < h - 0 + 1; i++)
{
// Some linkers will complain if we call vpx_memset with el set to a
// constant 0.
/* Some linkers will complain if we call vpx_memset with el set to a
* constant 0.
*/
if (el)
vpx_memset(dest_ptr1, src_ptr1[0], el);
vpx_memset(dest_ptr2, src_ptr2[0], er);
@@ -50,7 +51,7 @@ static void extend_plane_borders
dest_ptr2 += sp;
}
// Now copy the top and bottom source lines into each line of the respective borders
/* Now copy the top and bottom source lines into each line of the respective borders */
src_ptr1 = s - el;
src_ptr2 = s + sp * (h - 1) - el;
dest_ptr1 = s + sp * (-et) - el;
@@ -76,12 +77,12 @@ void vp8_extend_to_multiple_of16(YV12_BUFFER_CONFIG *ybf, int width, int height)
int er = 0xf & (16 - (width & 0xf));
int eb = 0xf & (16 - (height & 0xf));
// check for non multiples of 16
/* check for non multiples of 16 */
if (er != 0 || eb != 0)
{
extend_plane_borders(ybf->y_buffer, ybf->y_stride, height, width, 0, 0, eb, er);
//adjust for uv
/* adjust for uv */
height = (height + 1) >> 1;
width = (width + 1) >> 1;
er = 0x7 & (8 - (width & 0x7));
@@ -95,7 +96,7 @@ void vp8_extend_to_multiple_of16(YV12_BUFFER_CONFIG *ybf, int width, int height)
}
}
// note the extension is only for the last row, for intra prediction purpose
/* note the extension is only for the last row, for intra prediction purpose */
void vp8_extend_mb_row(YV12_BUFFER_CONFIG *ybf, unsigned char *YPtr, unsigned char *UPtr, unsigned char *VPtr)
{
int i;

76
vp8/common/filter_c.c
View File

@@ -32,13 +32,13 @@ static const int bilinear_filters[8][2] =
static const short sub_pel_filters[8][6] =
{
{ 0, 0, 128, 0, 0, 0 }, // note that 1/8 pel positions are just as per alpha -0.5 bicubic
{ 0, 0, 128, 0, 0, 0 }, /* note that 1/8 pel positions are just as per alpha -0.5 bicubic */
{ 0, -6, 123, 12, -1, 0 },
{ 2, -11, 108, 36, -8, 1 }, // New 1/4 pel 6 tap filter
{ 2, -11, 108, 36, -8, 1 }, /* New 1/4 pel 6 tap filter */
{ 0, -9, 93, 50, -6, 0 },
{ 3, -16, 77, 77, -16, 3 }, // New 1/2 pel 6 tap filter
{ 3, -16, 77, 77, -16, 3 }, /* New 1/2 pel 6 tap filter */
{ 0, -6, 50, 93, -9, 0 },
{ 1, -8, 36, 108, -11, 2 }, // New 1/4 pel 6 tap filter
{ 1, -8, 36, 108, -11, 2 }, /* New 1/4 pel 6 tap filter */
{ 0, -1, 12, 123, -6, 0 },
@@ -69,9 +69,9 @@ void vp8_filter_block2d_first_pass
((int)src_ptr[pixel_step] * vp8_filter[3]) +
((int)src_ptr[2*pixel_step] * vp8_filter[4]) +
((int)src_ptr[3*pixel_step] * vp8_filter[5]) +
(VP8_FILTER_WEIGHT >> 1); // Rounding
(VP8_FILTER_WEIGHT >> 1); /* Rounding */
// Normalize back to 0-255
/* Normalize back to 0-255 */
Temp = Temp >> VP8_FILTER_SHIFT;
if (Temp < 0)
@@ -83,7 +83,7 @@ void vp8_filter_block2d_first_pass
src_ptr++;
}
// Next row...
/* Next row... */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_width;
}
@@ -108,16 +108,16 @@ void vp8_filter_block2d_second_pass
{
for (j = 0; j < output_width; j++)
{
// Apply filter
/* Apply filter */
Temp = ((int)src_ptr[-2 * (int)pixel_step] * vp8_filter[0]) +
((int)src_ptr[-1 * (int)pixel_step] * vp8_filter[1]) +
((int)src_ptr[0] * vp8_filter[2]) +
((int)src_ptr[pixel_step] * vp8_filter[3]) +
((int)src_ptr[2*pixel_step] * vp8_filter[4]) +
((int)src_ptr[3*pixel_step] * vp8_filter[5]) +
(VP8_FILTER_WEIGHT >> 1); // Rounding
(VP8_FILTER_WEIGHT >> 1); /* Rounding */
// Normalize back to 0-255
/* Normalize back to 0-255 */
Temp = Temp >> VP8_FILTER_SHIFT;
if (Temp < 0)
@@ -129,7 +129,7 @@ void vp8_filter_block2d_second_pass
src_ptr++;
}
// Start next row
/* Start next row */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_pitch;
}
@@ -146,12 +146,12 @@ void vp8_filter_block2d
const short *VFilter
)
{
int FData[9*4]; // Temp data bufffer used in filtering
int FData[9*4]; /* Temp data bufffer used in filtering */
// First filter 1-D horizontally...
/* First filter 1-D horizontally... */
vp8_filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 9, 4, HFilter);
// then filter verticaly...
/* then filter verticaly... */
vp8_filter_block2d_second_pass(FData + 8, output_ptr, output_pitch, 4, 4, 4, 4, VFilter);
}
@@ -195,8 +195,8 @@ void vp8_sixtap_predict_c
const short *HFilter;
const short *VFilter;
HFilter = sub_pel_filters[xoffset]; // 6 tap
VFilter = sub_pel_filters[yoffset]; // 6 tap
HFilter = sub_pel_filters[xoffset]; /* 6 tap */
VFilter = sub_pel_filters[yoffset]; /* 6 tap */
vp8_filter_block2d(src_ptr, dst_ptr, src_pixels_per_line, dst_pitch, HFilter, VFilter);
}
@@ -212,16 +212,16 @@ void vp8_sixtap_predict8x8_c
{
const short *HFilter;
const short *VFilter;
int FData[13*16]; // Temp data bufffer used in filtering
int FData[13*16]; /* Temp data bufffer used in filtering */
HFilter = sub_pel_filters[xoffset]; // 6 tap
VFilter = sub_pel_filters[yoffset]; // 6 tap
HFilter = sub_pel_filters[xoffset]; /* 6 tap */
VFilter = sub_pel_filters[yoffset]; /* 6 tap */
// First filter 1-D horizontally...
/* First filter 1-D horizontally... */
vp8_filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 13, 8, HFilter);
// then filter verticaly...
/* then filter verticaly... */
vp8_filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 8, 8, VFilter);
}
@@ -238,16 +238,16 @@ void vp8_sixtap_predict8x4_c
{
const short *HFilter;
const short *VFilter;
int FData[13*16]; // Temp data bufffer used in filtering
int FData[13*16]; /* Temp data bufffer used in filtering */
HFilter = sub_pel_filters[xoffset]; // 6 tap
VFilter = sub_pel_filters[yoffset]; // 6 tap
HFilter = sub_pel_filters[xoffset]; /* 6 tap */
VFilter = sub_pel_filters[yoffset]; /* 6 tap */
// First filter 1-D horizontally...
/* First filter 1-D horizontally... */
vp8_filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 9, 8, HFilter);
// then filter verticaly...
/* then filter verticaly... */
vp8_filter_block2d_second_pass(FData + 16, dst_ptr, dst_pitch, 8, 8, 4, 8, VFilter);
}
@@ -264,16 +264,16 @@ void vp8_sixtap_predict16x16_c
{
const short *HFilter;
const short *VFilter;
int FData[21*24]; // Temp data bufffer used in filtering
int FData[21*24]; /* Temp data bufffer used in filtering */
HFilter = sub_pel_filters[xoffset]; // 6 tap
VFilter = sub_pel_filters[yoffset]; // 6 tap
HFilter = sub_pel_filters[xoffset]; /* 6 tap */
VFilter = sub_pel_filters[yoffset]; /* 6 tap */
// First filter 1-D horizontally...
/* First filter 1-D horizontally... */
vp8_filter_block2d_first_pass(src_ptr - (2 * src_pixels_per_line), FData, src_pixels_per_line, 1, 21, 16, HFilter);
// then filter verticaly...
/* then filter verticaly... */
vp8_filter_block2d_second_pass(FData + 32, dst_ptr, dst_pitch, 16, 16, 16, 16, VFilter);
}
@@ -324,14 +324,14 @@ void vp8_filter_block2d_bil_first_pass
{
for (j = 0; j < output_width; j++)
{
// Apply bilinear filter
/* Apply bilinear filter */
output_ptr[j] = (((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[pixel_step] * vp8_filter[1]) +
(VP8_FILTER_WEIGHT / 2)) >> VP8_FILTER_SHIFT;
src_ptr++;
}
// Next row...
/* Next row... */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_width;
}
@@ -384,7 +384,7 @@ void vp8_filter_block2d_bil_second_pass
{
for (j = 0; j < output_width; j++)
{
// Apply filter
/* Apply filter */
Temp = ((int)src_ptr[0] * vp8_filter[0]) +
((int)src_ptr[pixel_step] * vp8_filter[1]) +
(VP8_FILTER_WEIGHT / 2);
@@ -392,7 +392,7 @@ void vp8_filter_block2d_bil_second_pass
src_ptr++;
}
// Next row...
/* Next row... */
src_ptr += src_pixels_per_line - output_width;
output_ptr += output_pitch;
}
@@ -432,12 +432,12 @@ void vp8_filter_block2d_bil
)
{
unsigned short FData[17*16]; // Temp data bufffer used in filtering
unsigned short FData[17*16]; /* Temp data bufffer used in filtering */
// First filter 1-D horizontally...
/* First filter 1-D horizontally... */
vp8_filter_block2d_bil_first_pass(src_ptr, FData, src_pixels_per_line, 1, Height + 1, Width, HFilter);
// then 1-D vertically...
/* then 1-D vertically... */
vp8_filter_block2d_bil_second_pass(FData, output_ptr, dst_pitch, Width, Width, Height, Width, VFilter);
}

4
vp8/common/findnearmv.c
View File

@@ -168,7 +168,7 @@ void vp8_find_near_mvs
vp8_clamp_mv(nearest, xd);
vp8_clamp_mv(nearby, xd);
vp8_clamp_mv(best_mv, xd); //TODO: move this up before the copy
vp8_clamp_mv(best_mv, xd); /*TODO: move this up before the copy*/
}
vp8_prob *vp8_mv_ref_probs(
@@ -179,7 +179,7 @@ vp8_prob *vp8_mv_ref_probs(
p[1] = vp8_mode_contexts [near_mv_ref_ct[1]] [1];
p[2] = vp8_mode_contexts [near_mv_ref_ct[2]] [2];
p[3] = vp8_mode_contexts [near_mv_ref_ct[3]] [3];
//p[3] = vp8_mode_contexts [near_mv_ref_ct[1] + near_mv_ref_ct[2] + near_mv_ref_ct[3]] [3];
/*p[3] = vp8_mode_contexts [near_mv_ref_ct[1] + near_mv_ref_ct[2] + near_mv_ref_ct[3]] [3];*/
return p;
}

24
vp8/common/generic/systemdependent.c
View File

@@ -18,6 +18,7 @@
#include "onyxc_int.h"
extern void vp8_arch_x86_common_init(VP8_COMMON *ctx);
extern void vp8_arch_arm_common_init(VP8_COMMON *ctx);
void (*vp8_build_intra_predictors_mby_ptr)(MACROBLOCKD *x);
extern void vp8_build_intra_predictors_mby(MACROBLOCKD *x);
@@ -39,9 +40,11 @@ void vp8_machine_specific_config(VP8_COMMON *ctx)
rtcd->recon.copy16x16 = vp8_copy_mem16x16_c;
rtcd->recon.copy8x8 = vp8_copy_mem8x8_c;
rtcd->recon.copy8x4 = vp8_copy_mem8x4_c;
rtcd->recon.recon = vp8_recon_b_c;
rtcd->recon.recon = vp8_recon_b_c;
rtcd->recon.recon2 = vp8_recon2b_c;
rtcd->recon.recon4 = vp8_recon4b_c;
rtcd->recon.recon4 = vp8_recon4b_c;
rtcd->recon.recon_mb = vp8_recon_mb_c;
rtcd->recon.recon_mby = vp8_recon_mby_c;
rtcd->subpix.sixtap16x16 = vp8_sixtap_predict16x16_c;
rtcd->subpix.sixtap8x8 = vp8_sixtap_predict8x8_c;
@@ -62,14 +65,17 @@ void vp8_machine_specific_config(VP8_COMMON *ctx)
rtcd->loopfilter.simple_b_h = vp8_loop_filter_bhs_c;
#if CONFIG_POSTPROC || (CONFIG_VP8_ENCODER && CONFIG_PSNR)
rtcd->postproc.down = vp8_mbpost_proc_down_c;
rtcd->postproc.across = vp8_mbpost_proc_across_ip_c;
rtcd->postproc.downacross = vp8_post_proc_down_and_across_c;
rtcd->postproc.addnoise = vp8_plane_add_noise_c;
rtcd->postproc.down = vp8_mbpost_proc_down_c;
rtcd->postproc.across = vp8_mbpost_proc_across_ip_c;
rtcd->postproc.downacross = vp8_post_proc_down_and_across_c;
rtcd->postproc.addnoise = vp8_plane_add_noise_c;
rtcd->postproc.blend_mb_inner = vp8_blend_mb_inner_c;
rtcd->postproc.blend_mb_outer = vp8_blend_mb_outer_c;
rtcd->postproc.blend_b = vp8_blend_b_c;
#endif
#endif
// Pure C:
/* Pure C: */
vp8_build_intra_predictors_mby_ptr = vp8_build_intra_predictors_mby;
vp8_build_intra_predictors_mby_s_ptr = vp8_build_intra_predictors_mby_s;
@@ -77,4 +83,8 @@ void vp8_machine_specific_config(VP8_COMMON *ctx)
vp8_arch_x86_common_init(ctx);
#endif
#if ARCH_ARM
vp8_arch_arm_common_init(ctx);
#endif
}

4
vp8/common/invtrans.c
View File

@@ -38,7 +38,7 @@ void vp8_inverse_transform_mby(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *
{
int i;
// do 2nd order transform on the dc block
/* do 2nd order transform on the dc block */
IDCT_INVOKE(rtcd, iwalsh16)(x->block[24].dqcoeff, x->block[24].diff);
recon_dcblock(x);
@@ -68,7 +68,7 @@ void vp8_inverse_transform_mb(const vp8_idct_rtcd_vtable_t *rtcd, MACROBLOCKD *x
if (x->mode_info_context->mbmi.mode != B_PRED &&
x->mode_info_context->mbmi.mode != SPLITMV)
{
// do 2nd order transform on the dc block
/* do 2nd order transform on the dc block */
IDCT_INVOKE(rtcd, iwalsh16)(&x->block[24].dqcoeff[0], x->block[24].diff);
recon_dcblock(x);

132
vp8/common/loopfilter.c
View File

@@ -23,7 +23,7 @@ prototype_loopfilter(vp8_mbloop_filter_vertical_edge_c);
prototype_loopfilter(vp8_loop_filter_simple_horizontal_edge_c);
prototype_loopfilter(vp8_loop_filter_simple_vertical_edge_c);
// Horizontal MB filtering
/* Horizontal MB filtering */
void vp8_loop_filter_mbh_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -47,7 +47,7 @@ void vp8_loop_filter_mbhs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned
vp8_loop_filter_simple_horizontal_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
}
// Vertical MB Filtering
/* Vertical MB Filtering */
void vp8_loop_filter_mbv_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -71,7 +71,7 @@ void vp8_loop_filter_mbvs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned
vp8_loop_filter_simple_vertical_edge_c(y_ptr, y_stride, lfi->mbflim, lfi->lim, lfi->mbthr, 2);
}
// Horizontal B Filtering
/* Horizontal B Filtering */
void vp8_loop_filter_bh_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -99,7 +99,7 @@ void vp8_loop_filter_bhs_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned
vp8_loop_filter_simple_horizontal_edge_c(y_ptr + 12 * y_stride, y_stride, lfi->flim, lfi->lim, lfi->thr, 2);
}
// Vertical B Filtering
/* Vertical B Filtering */
void vp8_loop_filter_bv_c(unsigned char *y_ptr, unsigned char *u_ptr, unsigned char *v_ptr,
int y_stride, int uv_stride, loop_filter_info *lfi, int simpler_lpf)
{
@@ -140,7 +140,7 @@ void vp8_init_loop_filter(VP8_COMMON *cm)
const int yhedge_boost = 2;
const int uvhedge_boost = 2;
// For each possible value for the loop filter fill out a "loop_filter_info" entry.
/* For each possible value for the loop filter fill out a "loop_filter_info" entry. */
for (i = 0; i <= MAX_LOOP_FILTER; i++)
{
int filt_lvl = i;
@@ -166,7 +166,7 @@ void vp8_init_loop_filter(VP8_COMMON *cm)
HEVThresh = 0;
}
// Set loop filter paramaeters that control sharpness.
/* Set loop filter paramaeters that control sharpness. */
block_inside_limit = filt_lvl >> (sharpness_lvl > 0);
block_inside_limit = block_inside_limit >> (sharpness_lvl > 4);
@@ -195,7 +195,7 @@ void vp8_init_loop_filter(VP8_COMMON *cm)
}
// Set up the function pointers depending on the type of loop filtering selected
/* Set up the function pointers depending on the type of loop filtering selected */
if (lft == NORMAL_LOOPFILTER)
{
cm->lf_mbv = LF_INVOKE(&cm->rtcd.loopfilter, normal_mb_v);
@@ -212,14 +212,15 @@ void vp8_init_loop_filter(VP8_COMMON *cm)
}
}
// Put vp8_init_loop_filter() in vp8dx_create_decompressor(). Only call vp8_frame_init_loop_filter() while decoding
// each frame. Check last_frame_type to skip the function most of times.
/* Put vp8_init_loop_filter() in vp8dx_create_decompressor(). Only call vp8_frame_init_loop_filter() while decoding
* each frame. Check last_frame_type to skip the function most of times.
*/
void vp8_frame_init_loop_filter(loop_filter_info *lfi, int frame_type)
{
int HEVThresh;
int i, j;
// For each possible value for the loop filter fill out a "loop_filter_info" entry.
/* For each possible value for the loop filter fill out a "loop_filter_info" entry. */
for (i = 0; i <= MAX_LOOP_FILTER; i++)
{
int filt_lvl = i;
@@ -247,15 +248,15 @@ void vp8_frame_init_loop_filter(loop_filter_info *lfi, int frame_type)
for (j = 0; j < 16; j++)
{
//lfi[i].lim[j] = block_inside_limit;
//lfi[i].mbflim[j] = filt_lvl+yhedge_boost;
/*lfi[i].lim[j] = block_inside_limit;
lfi[i].mbflim[j] = filt_lvl+yhedge_boost;*/
lfi[i].mbthr[j] = HEVThresh;
//lfi[i].flim[j] = filt_lvl;
/*lfi[i].flim[j] = filt_lvl;*/
lfi[i].thr[j] = HEVThresh;
//lfi[i].uvlim[j] = block_inside_limit;
//lfi[i].uvmbflim[j] = filt_lvl+uvhedge_boost;
/*lfi[i].uvlim[j] = block_inside_limit;
lfi[i].uvmbflim[j] = filt_lvl+uvhedge_boost;*/
lfi[i].uvmbthr[j] = HEVThresh;
//lfi[i].uvflim[j] = filt_lvl;
/*lfi[i].uvflim[j] = filt_lvl;*/
lfi[i].uvthr[j] = HEVThresh;
}
}
@@ -268,32 +269,32 @@ void vp8_adjust_mb_lf_value(MACROBLOCKD *mbd, int *filter_level)
if (mbd->mode_ref_lf_delta_enabled)
{
// Aplly delta for reference frame
/* Apply delta for reference frame */
*filter_level += mbd->ref_lf_deltas[mbmi->ref_frame];
// Apply delta for mode
/* Apply delta for mode */
if (mbmi->ref_frame == INTRA_FRAME)
{
// Only the split mode BPRED has a further special case
/* Only the split mode BPRED has a further special case */
if (mbmi->mode == B_PRED)
*filter_level += mbd->mode_lf_deltas[0];
}
else
{
// Zero motion mode
/* Zero motion mode */
if (mbmi->mode == ZEROMV)
*filter_level += mbd->mode_lf_deltas[1];
// Split MB motion mode
/* Split MB motion mode */
else if (mbmi->mode == SPLITMV)
*filter_level += mbd->mode_lf_deltas[3];
// All other inter motion modes (Nearest, Near, New)
/* All other inter motion modes (Nearest, Near, New) */
else
*filter_level += mbd->mode_lf_deltas[2];
}
// Range check
/* Range check */
if (*filter_level > MAX_LOOP_FILTER)
*filter_level = MAX_LOOP_FILTER;
else if (*filter_level < 0)
@@ -311,7 +312,7 @@ void vp8_loop_filter_frame
{
YV12_BUFFER_CONFIG *post = cm->frame_to_show;
loop_filter_info *lfi = cm->lf_info;
int frame_type = cm->frame_type;
FRAME_TYPE frame_type = cm->frame_type;
int mb_row;
int mb_col;
@@ -324,21 +325,21 @@ void vp8_loop_filter_frame
int i;
unsigned char *y_ptr, *u_ptr, *v_ptr;
mbd->mode_info_context = cm->mi; // Point at base of Mb MODE_INFO list
mbd->mode_info_context = cm->mi; /* Point at base of Mb MODE_INFO list */
// Note the baseline filter values for each segment
/* Note the baseline filter values for each segment */
if (alt_flt_enabled)
{
for (i = 0; i < MAX_MB_SEGMENTS; i++)
{
// Abs value
/* Abs value */
if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i];
// Delta Value
/* Delta Value */
else
{
baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i];
baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; // Clamp to valid range
baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; /* Clamp to valid range */
}
}
}
@@ -348,18 +349,18 @@ void vp8_loop_filter_frame
baseline_filter_level[i] = default_filt_lvl;
}
// Initialize the loop filter for this frame.
/* Initialize the loop filter for this frame. */
if ((cm->last_filter_type != cm->filter_type) || (cm->last_sharpness_level != cm->sharpness_level))
vp8_init_loop_filter(cm);
else if (frame_type != cm->last_frame_type)
vp8_frame_init_loop_filter(lfi, frame_type);
// Set up the buffer pointers
/* Set up the buffer pointers */
y_ptr = post->y_buffer;
u_ptr = post->u_buffer;
v_ptr = post->v_buffer;
// vp8_filter each macro block
/* vp8_filter each macro block */
for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
{
for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
@@ -368,9 +369,10 @@ void vp8_loop_filter_frame
filter_level = baseline_filter_level[Segment];
// Distance of Mb to the various image edges.
// These specified to 8th pel as they are always compared to values that are in 1/8th pel units
// Apply any context driven MB level adjustment
/* Distance of Mb to the various image edges.
* These specified to 8th pel as they are always compared to values that are in 1/8th pel units
* Apply any context driven MB level adjustment
*/
vp8_adjust_mb_lf_value(mbd, &filter_level);
if (filter_level)
@@ -381,7 +383,7 @@ void vp8_loop_filter_frame
if (mbd->mode_info_context->mbmi.dc_diff > 0)
cm->lf_bv(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
// don't apply across umv border
/* don't apply across umv border */
if (mb_row > 0)
cm->lf_mbh(y_ptr, u_ptr, v_ptr, post->y_stride, post->uv_stride, &lfi[filter_level], cm->simpler_lpf);
@@ -393,14 +395,14 @@ void vp8_loop_filter_frame
u_ptr += 8;
v_ptr += 8;
mbd->mode_info_context++; // step to next MB
mbd->mode_info_context++; /* step to next MB */
}
y_ptr += post->y_stride * 16 - post->y_width;
u_ptr += post->uv_stride * 8 - post->uv_width;
v_ptr += post->uv_stride * 8 - post->uv_width;
mbd->mode_info_context++; // Skip border mb
mbd->mode_info_context++; /* Skip border mb */
}
}
@@ -424,26 +426,26 @@ void vp8_loop_filter_frame_yonly
int baseline_filter_level[MAX_MB_SEGMENTS];
int filter_level;
int alt_flt_enabled = mbd->segmentation_enabled;
int frame_type = cm->frame_type;
FRAME_TYPE frame_type = cm->frame_type;
(void) sharpness_lvl;
//MODE_INFO * this_mb_mode_info = cm->mi; // Point at base of Mb MODE_INFO list
mbd->mode_info_context = cm->mi; // Point at base of Mb MODE_INFO list
/*MODE_INFO * this_mb_mode_info = cm->mi;*/ /* Point at base of Mb MODE_INFO list */
mbd->mode_info_context = cm->mi; /* Point at base of Mb MODE_INFO list */
// Note the baseline filter values for each segment
/* Note the baseline filter values for each segment */
if (alt_flt_enabled)
{
for (i = 0; i < MAX_MB_SEGMENTS; i++)
{
// Abs value
/* Abs value */
if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i];
// Delta Value
/* Delta Value */
else
{
baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i];
baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; // Clamp to valid range
baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; /* Clamp to valid range */
}
}
}
@@ -453,16 +455,16 @@ void vp8_loop_filter_frame_yonly
baseline_filter_level[i] = default_filt_lvl;
}
// Initialize the loop filter for this frame.
/* Initialize the loop filter for this frame. */
if ((cm->last_filter_type != cm->filter_type) || (cm->last_sharpness_level != cm->sharpness_level))
vp8_init_loop_filter(cm);
else if (frame_type != cm->last_frame_type)
vp8_frame_init_loop_filter(lfi, frame_type);
// Set up the buffer pointers
/* Set up the buffer pointers */
y_ptr = post->y_buffer;
// vp8_filter each macro block
/* vp8_filter each macro block */
for (mb_row = 0; mb_row < cm->mb_rows; mb_row++)
{
for (mb_col = 0; mb_col < cm->mb_cols; mb_col++)
@@ -470,7 +472,7 @@ void vp8_loop_filter_frame_yonly
int Segment = (alt_flt_enabled) ? mbd->mode_info_context->mbmi.segment_id : 0;
filter_level = baseline_filter_level[Segment];
// Apply any context driven MB level adjustment
/* Apply any context driven MB level adjustment */
vp8_adjust_mb_lf_value(mbd, &filter_level);
if (filter_level)
@@ -481,7 +483,7 @@ void vp8_loop_filter_frame_yonly
if (mbd->mode_info_context->mbmi.dc_diff > 0)
cm->lf_bv(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
// don't apply across umv border
/* don't apply across umv border */
if (mb_row > 0)
cm->lf_mbh(y_ptr, 0, 0, post->y_stride, 0, &lfi[filter_level], 0);
@@ -490,12 +492,12 @@ void vp8_loop_filter_frame_yonly
}
y_ptr += 16;
mbd->mode_info_context ++; // step to next MB
mbd->mode_info_context ++; /* step to next MB */
}
y_ptr += post->y_stride * 16 - post->y_width;
mbd->mode_info_context ++; // Skip border mb
mbd->mode_info_context ++; /* Skip border mb */
}
}
@@ -516,7 +518,7 @@ void vp8_loop_filter_partial_frame
unsigned char *y_ptr;
int mb_row;
int mb_col;
//int mb_rows = post->y_height >> 4;
/*int mb_rows = post->y_height >> 4;*/
int mb_cols = post->y_width >> 4;
int linestocopy;
@@ -525,12 +527,12 @@ void vp8_loop_filter_partial_frame
int baseline_filter_level[MAX_MB_SEGMENTS];
int filter_level;
int alt_flt_enabled = mbd->segmentation_enabled;
int frame_type = cm->frame_type;
FRAME_TYPE frame_type = cm->frame_type;
(void) sharpness_lvl;
//MODE_INFO * this_mb_mode_info = cm->mi + (post->y_height>>5) * (mb_cols + 1); // Point at base of Mb MODE_INFO list
mbd->mode_info_context = cm->mi + (post->y_height >> 5) * (mb_cols + 1); // Point at base of Mb MODE_INFO list
/*MODE_INFO * this_mb_mode_info = cm->mi + (post->y_height>>5) * (mb_cols + 1);*/ /* Point at base of Mb MODE_INFO list */
mbd->mode_info_context = cm->mi + (post->y_height >> 5) * (mb_cols + 1); /* Point at base of Mb MODE_INFO list */
linestocopy = (post->y_height >> (4 + Fraction));
@@ -539,19 +541,19 @@ void vp8_loop_filter_partial_frame
linestocopy <<= 4;
// Note the baseline filter values for each segment
/* Note the baseline filter values for each segment */
if (alt_flt_enabled)
{
for (i = 0; i < MAX_MB_SEGMENTS; i++)
{
// Abs value
/* Abs value */
if (mbd->mb_segement_abs_delta == SEGMENT_ABSDATA)
baseline_filter_level[i] = mbd->segment_feature_data[MB_LVL_ALT_LF][i];
// Delta Value
/* Delta Value */
else
{
baseline_filter_level[i] = default_filt_lvl + mbd->segment_feature_data[MB_LVL_ALT_LF][i];
baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; // Clamp to valid range
baseline_filter_level[i] = (baseline_filter_level[i] >= 0) ? ((baseline_filter_level[i] <= MAX_LOOP_FILTER) ? baseline_filter_level[i] : MAX_LOOP_FILTER) : 0; /* Clamp to valid range */
}
}
}
@@ -561,16 +563,16 @@ void vp8_loop_filter_partial_frame
baseline_filter_level[i] = default_filt_lvl;
}
// Initialize the loop filter for this frame.
/* Initialize the loop filter for this frame. */
if ((cm->last_filter_type != cm->filter_type) || (cm->last_sharpness_level != cm->sharpness_level))
vp8_init_loop_filter(cm);
else if (frame_type != cm->last_frame_type)
vp8_frame_init_loop_filter(lfi, frame_type);
// Set up the buffer pointers
/* Set up the buffer pointers */
y_ptr = post->y_buffer + (post->y_height >> 5) * 16 * post->y_stride;
// vp8_filter each macro block
/* vp8_filter each macro block */
for (mb_row = 0; mb_row<(linestocopy >> 4); mb_row++)
{
for (mb_col = 0; mb_col < mb_cols; mb_col++)
@@ -593,10 +595,10 @@ void vp8_loop_filter_partial_frame
}
y_ptr += 16;
mbd->mode_info_context += 1; // step to next MB
mbd->mode_info_context += 1; /* step to next MB */
}
y_ptr += post->y_stride * 16 - post->y_width;
mbd->mode_info_context += 1; // Skip border mb
mbd->mode_info_context += 1; /* Skip border mb */
}
}

12
vp8/common/loopfilter.h
View File

@@ -22,10 +22,10 @@ typedef enum
SIMPLE_LOOPFILTER = 1
} LOOPFILTERTYPE;
// FRK
// Need to align this structure so when it is declared and
// passed it can be loaded into vector registers.
// FRK
/* FRK
* Need to align this structure so when it is declared and
* passed it can be loaded into vector registers.
*/
typedef struct
{
DECLARE_ALIGNED(16, signed char, lim[16]);
@@ -119,8 +119,8 @@ typedef struct
typedef void loop_filter_uvfunction
(
unsigned char *u, // source pointer
int p, // pitch
unsigned char *u, /* source pointer */
int p, /* pitch */
const signed char *flimit,
const signed char *limit,
const signed char *thresh,

69
vp8/common/loopfilter_filters.c
View File

@@ -23,7 +23,7 @@ static __inline signed char vp8_signed_char_clamp(int t)
}
// should we apply any filter at all ( 11111111 yes, 00000000 no)
/* should we apply any filter at all ( 11111111 yes, 00000000 no) */
static __inline signed char vp8_filter_mask(signed char limit, signed char flimit,
uc p3, uc p2, uc p1, uc p0, uc q0, uc q1, uc q2, uc q3)
{
@@ -39,7 +39,7 @@ static __inline signed char vp8_filter_mask(signed char limit, signed char flimi
return mask;
}
// is there high variance internal edge ( 11111111 yes, 00000000 no)
/* is there high variance internal edge ( 11111111 yes, 00000000 no) */
static __inline signed char vp8_hevmask(signed char thresh, uc p1, uc p0, uc q0, uc q1)
{
signed char hev = 0;
@@ -61,17 +61,18 @@ static __inline void vp8_filter(signed char mask, signed char hev, uc *op1, uc *
qs0 = (signed char) * oq0 ^ 0x80;
qs1 = (signed char) * oq1 ^ 0x80;
// add outer taps if we have high edge variance
/* add outer taps if we have high edge variance */
vp8_filter = vp8_signed_char_clamp(ps1 - qs1);
vp8_filter &= hev;
// inner taps
/* inner taps */
vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0));
vp8_filter &= mask;
// save bottom 3 bits so that we round one side +4 and the other +3
// if it equals 4 we'll set to adjust by -1 to account for the fact
// we'd round 3 the other way
/* save bottom 3 bits so that we round one side +4 and the other +3
* if it equals 4 we'll set to adjust by -1 to account for the fact
* we'd round 3 the other way
*/
Filter1 = vp8_signed_char_clamp(vp8_filter + 4);
Filter2 = vp8_signed_char_clamp(vp8_filter + 3);
Filter1 >>= 3;
@@ -82,7 +83,7 @@ static __inline void vp8_filter(signed char mask, signed char hev, uc *op1, uc *
*op0 = u ^ 0x80;
vp8_filter = Filter1;
// outer tap adjustments
/* outer tap adjustments */
vp8_filter += 1;
vp8_filter >>= 1;
vp8_filter &= ~hev;
@@ -96,19 +97,20 @@ static __inline void vp8_filter(signed char mask, signed char hev, uc *op1, uc *
void vp8_loop_filter_horizontal_edge_c
(
unsigned char *s,
int p, //pitch
int p, /* pitch */
const signed char *flimit,
const signed char *limit,
const signed char *thresh,
int count
)
{
int hev = 0; // high edge variance
int hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
// loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions.
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do
{
mask = vp8_filter_mask(limit[i], flimit[i],
@@ -134,12 +136,13 @@ void vp8_loop_filter_vertical_edge_c
int count
)
{
int hev = 0; // high edge variance
int hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
// loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions.
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do
{
mask = vp8_filter_mask(limit[i], flimit[i],
@@ -166,7 +169,7 @@ static __inline void vp8_mbfilter(signed char mask, signed char hev,
signed char qs1 = (signed char) * oq1 ^ 0x80;
signed char qs2 = (signed char) * oq2 ^ 0x80;
// add outer taps if we have high edge variance
/* add outer taps if we have high edge variance */
vp8_filter = vp8_signed_char_clamp(ps1 - qs1);
vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (qs0 - ps0));
vp8_filter &= mask;
@@ -174,7 +177,7 @@ static __inline void vp8_mbfilter(signed char mask, signed char hev,
Filter2 = vp8_filter;
Filter2 &= hev;
// save bottom 3 bits so that we round one side +4 and the other +3
/* save bottom 3 bits so that we round one side +4 and the other +3 */
Filter1 = vp8_signed_char_clamp(Filter2 + 4);
Filter2 = vp8_signed_char_clamp(Filter2 + 3);
Filter1 >>= 3;
@@ -183,25 +186,25 @@ static __inline void vp8_mbfilter(signed char mask, signed char hev,
ps0 = vp8_signed_char_clamp(ps0 + Filter2);
// only apply wider filter if not high edge variance
/* only apply wider filter if not high edge variance */
vp8_filter &= ~hev;
Filter2 = vp8_filter;
// roughly 3/7th difference across boundary
/* roughly 3/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 27) >> 7);
s = vp8_signed_char_clamp(qs0 - u);
*oq0 = s ^ 0x80;
s = vp8_signed_char_clamp(ps0 + u);
*op0 = s ^ 0x80;
// roughly 2/7th difference across boundary
/* roughly 2/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 18) >> 7);
s = vp8_signed_char_clamp(qs1 - u);
*oq1 = s ^ 0x80;
s = vp8_signed_char_clamp(ps1 + u);
*op1 = s ^ 0x80;
// roughly 1/7th difference across boundary
/* roughly 1/7th difference across boundary */
u = vp8_signed_char_clamp((63 + Filter2 * 9) >> 7);
s = vp8_signed_char_clamp(qs2 - u);
*oq2 = s ^ 0x80;
@@ -219,12 +222,13 @@ void vp8_mbloop_filter_horizontal_edge_c
int count
)
{
signed char hev = 0; // high edge variance
signed char hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
// loop filter designed to work using chars so that we can make maximum use
// of 8 bit simd instructions.
/* loop filter designed to work using chars so that we can make maximum use
* of 8 bit simd instructions.
*/
do
{
@@ -253,7 +257,7 @@ void vp8_mbloop_filter_vertical_edge_c
int count
)
{
signed char hev = 0; // high edge variance
signed char hev = 0; /* high edge variance */
signed char mask = 0;
int i = 0;
@@ -273,12 +277,13 @@ void vp8_mbloop_filter_vertical_edge_c
}
// should we apply any filter at all ( 11111111 yes, 00000000 no)
/* should we apply any filter at all ( 11111111 yes, 00000000 no) */
static __inline signed char vp8_simple_filter_mask(signed char limit, signed char flimit, uc p1, uc p0, uc q0, uc q1)
{
// Why does this cause problems for win32?
// error C2143: syntax error : missing ';' before 'type'
// (void) limit;
/* Why does this cause problems for win32?
* error C2143: syntax error : missing ';' before 'type'
* (void) limit;
*/
signed char mask = (abs(p0 - q0) * 2 + abs(p1 - q1) / 2 <= flimit * 2 + limit) * -1;
return mask;
}
@@ -296,7 +301,7 @@ static __inline void vp8_simple_filter(signed char mask, uc *op1, uc *op0, uc *o
vp8_filter = vp8_signed_char_clamp(vp8_filter + 3 * (q0 - p0));
vp8_filter &= mask;
// save bottom 3 bits so that we round one side +4 and the other +3
/* save bottom 3 bits so that we round one side +4 and the other +3 */
Filter1 = vp8_signed_char_clamp(vp8_filter + 4);
Filter1 >>= 3;
u = vp8_signed_char_clamp(q0 - Filter1);
@@ -324,7 +329,7 @@ void vp8_loop_filter_simple_horizontal_edge_c
do
{
//mask = vp8_simple_filter_mask( limit[i], flimit[i],s[-1*p],s[0*p]);
/*mask = vp8_simple_filter_mask( limit[i], flimit[i],s[-1*p],s[0*p]);*/
mask = vp8_simple_filter_mask(limit[i], flimit[i], s[-2*p], s[-1*p], s[0*p], s[1*p]);
vp8_simple_filter(mask, s - 2 * p, s - 1 * p, s, s + 1 * p);
++s;
@@ -348,7 +353,7 @@ void vp8_loop_filter_simple_vertical_edge_c
do
{
//mask = vp8_simple_filter_mask( limit[i], flimit[i],s[-1],s[0]);
/*mask = vp8_simple_filter_mask( limit[i], flimit[i],s[-1],s[0]);*/
mask = vp8_simple_filter_mask(limit[i], flimit[i], s[-2], s[-1], s[0], s[1]);
vp8_simple_filter(mask, s - 2, s - 1, s, s + 1);
s += p;

8
vp8/common/mbpitch.c
View File

@@ -14,7 +14,7 @@
typedef enum
{
PRED = 0,
DEST = 1,
DEST = 1
} BLOCKSET;
void vp8_setup_block
@@ -62,13 +62,13 @@ void vp8_setup_macroblock(MACROBLOCKD *x, BLOCKSET bs)
v = &x->pre.v_buffer;
}
for (block = 0; block < 16; block++) // y blocks
for (block = 0; block < 16; block++) /* y blocks */
{
vp8_setup_block(&x->block[block], x->dst.y_stride, y, x->dst.y_stride,
(block >> 2) * 4 * x->dst.y_stride + (block & 3) * 4, bs);
}
for (block = 16; block < 20; block++) // U and V blocks
for (block = 16; block < 20; block++) /* U and V blocks */
{
vp8_setup_block(&x->block[block], x->dst.uv_stride, u, x->dst.uv_stride,
((block - 16) >> 1) * 4 * x->dst.uv_stride + (block & 1) * 4, bs);
@@ -123,7 +123,7 @@ void vp8_setup_block_dptrs(MACROBLOCKD *x)
void vp8_build_block_doffsets(MACROBLOCKD *x)
{
// handle the destination pitch features
/* handle the destination pitch features */
vp8_setup_macroblock(x, DEST);
vp8_setup_macroblock(x, PRED);
}

12
vp8/common/modecont.c
View File

@@ -14,27 +14,27 @@
const int vp8_mode_contexts[6][4] =
{
{
// 0
/* 0 */
7, 1, 1, 143,
},
{
// 1
/* 1 */
14, 18, 14, 107,
},
{
// 2
/* 2 */
135, 64, 57, 68,
},
{
// 3
/* 3 */
60, 56, 128, 65,
},
{
// 4
/* 4 */
159, 134, 128, 34,
},
{
// 5
/* 5 */
234, 188, 128, 28,
},
};

220
vp8/common/modecontext.c
View File

@@ -14,133 +14,133 @@
const unsigned int vp8_kf_default_bmode_counts [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES] =
{
{
//Above Mode : 0
{ 43438, 2195, 470, 316, 615, 171, 217, 412, 124, 160, }, // left_mode 0
{ 5722, 2751, 296, 291, 81, 68, 80, 101, 100, 170, }, // left_mode 1
{ 1629, 201, 307, 25, 47, 16, 34, 72, 19, 28, }, // left_mode 2
{ 332, 266, 36, 500, 20, 65, 23, 14, 154, 106, }, // left_mode 3
{ 450, 97, 10, 24, 117, 10, 2, 12, 8, 71, }, // left_mode 4
{ 384, 49, 29, 44, 12, 162, 51, 5, 87, 42, }, // left_mode 5
{ 495, 53, 157, 27, 14, 57, 180, 17, 17, 34, }, // left_mode 6
{ 695, 64, 62, 9, 27, 5, 3, 147, 10, 26, }, // left_mode 7
{ 230, 54, 20, 124, 16, 125, 29, 12, 283, 37, }, // left_mode 8
{ 260, 87, 21, 120, 32, 16, 33, 16, 33, 203, }, // left_mode 9
/*Above Mode : 0*/
{ 43438, 2195, 470, 316, 615, 171, 217, 412, 124, 160, }, /* left_mode 0 */
{ 5722, 2751, 296, 291, 81, 68, 80, 101, 100, 170, }, /* left_mode 1 */
{ 1629, 201, 307, 25, 47, 16, 34, 72, 19, 28, }, /* left_mode 2 */
{ 332, 266, 36, 500, 20, 65, 23, 14, 154, 106, }, /* left_mode 3 */
{ 450, 97, 10, 24, 117, 10, 2, 12, 8, 71, }, /* left_mode 4 */
{ 384, 49, 29, 44, 12, 162, 51, 5, 87, 42, }, /* left_mode 5 */
{ 495, 53, 157, 27, 14, 57, 180, 17, 17, 34, }, /* left_mode 6 */
{ 695, 64, 62, 9, 27, 5, 3, 147, 10, 26, }, /* left_mode 7 */
{ 230, 54, 20, 124, 16, 125, 29, 12, 283, 37, }, /* left_mode 8 */
{ 260, 87, 21, 120, 32, 16, 33, 16, 33, 203, }, /* left_mode 9 */
},
{
//Above Mode : 1
{ 3934, 2573, 355, 137, 128, 87, 133, 117, 37, 27, }, // left_mode 0
{ 1036, 1929, 278, 135, 27, 37, 48, 55, 41, 91, }, // left_mode 1
{ 223, 256, 253, 15, 13, 9, 28, 64, 3, 3, }, // left_mode 2
{ 120, 129, 17, 316, 15, 11, 9, 4, 53, 74, }, // left_mode 3
{ 129, 58, 6, 11, 38, 2, 0, 5, 2, 67, }, // left_mode 4
{ 53, 22, 11, 16, 8, 26, 14, 3, 19, 12, }, // left_mode 5
{ 59, 26, 61, 11, 4, 9, 35, 13, 8, 8, }, // left_mode 6
{ 101, 52, 40, 8, 5, 2, 8, 59, 2, 20, }, // left_mode 7
{ 48, 34, 10, 52, 8, 15, 6, 6, 63, 20, }, // left_mode 8
{ 96, 48, 22, 63, 11, 14, 5, 8, 9, 96, }, // left_mode 9
/*Above Mode : 1*/
{ 3934, 2573, 355, 137, 128, 87, 133, 117, 37, 27, }, /* left_mode 0 */
{ 1036, 1929, 278, 135, 27, 37, 48, 55, 41, 91, }, /* left_mode 1 */
{ 223, 256, 253, 15, 13, 9, 28, 64, 3, 3, }, /* left_mode 2 */
{ 120, 129, 17, 316, 15, 11, 9, 4, 53, 74, }, /* left_mode 3 */
{ 129, 58, 6, 11, 38, 2, 0, 5, 2, 67, }, /* left_mode 4 */
{ 53, 22, 11, 16, 8, 26, 14, 3, 19, 12, }, /* left_mode 5 */
{ 59, 26, 61, 11, 4, 9, 35, 13, 8, 8, }, /* left_mode 6 */
{ 101, 52, 40, 8, 5, 2, 8, 59, 2, 20, }, /* left_mode 7 */
{ 48, 34, 10, 52, 8, 15, 6, 6, 63, 20, }, /* left_mode 8 */
{ 96, 48, 22, 63, 11, 14, 5, 8, 9, 96, }, /* left_mode 9 */
},
{
//Above Mode : 2
{ 709, 461, 506, 36, 27, 33, 151, 98, 24, 6, }, // left_mode 0
{ 201, 375, 442, 27, 13, 8, 46, 58, 6, 19, }, // left_mode 1
{ 122, 140, 417, 4, 13, 3, 33, 59, 4, 2, }, // left_mode 2
{ 36, 17, 22, 16, 6, 8, 12, 17, 9, 21, }, // left_mode 3
{ 51, 15, 7, 1, 14, 0, 4, 5, 3, 22, }, // left_mode 4
{ 18, 11, 30, 9, 7, 20, 11, 5, 2, 6, }, // left_mode 5
{ 38, 21, 103, 9, 4, 12, 79, 13, 2, 5, }, // left_mode 6
{ 64, 17, 66, 2, 12, 4, 2, 65, 4, 5, }, // left_mode 7
{ 14, 7, 7, 16, 3, 11, 4, 13, 15, 16, }, // left_mode 8
{ 36, 8, 32, 9, 9, 4, 14, 7, 6, 24, }, // left_mode 9
/*Above Mode : 2*/
{ 709, 461, 506, 36, 27, 33, 151, 98, 24, 6, }, /* left_mode 0 */
{ 201, 375, 442, 27, 13, 8, 46, 58, 6, 19, }, /* left_mode 1 */
{ 122, 140, 417, 4, 13, 3, 33, 59, 4, 2, }, /* left_mode 2 */
{ 36, 17, 22, 16, 6, 8, 12, 17, 9, 21, }, /* left_mode 3 */
{ 51, 15, 7, 1, 14, 0, 4, 5, 3, 22, }, /* left_mode 4 */
{ 18, 11, 30, 9, 7, 20, 11, 5, 2, 6, }, /* left_mode 5 */
{ 38, 21, 103, 9, 4, 12, 79, 13, 2, 5, }, /* left_mode 6 */
{ 64, 17, 66, 2, 12, 4, 2, 65, 4, 5, }, /* left_mode 7 */
{ 14, 7, 7, 16, 3, 11, 4, 13, 15, 16, }, /* left_mode 8 */
{ 36, 8, 32, 9, 9, 4, 14, 7, 6, 24, }, /* left_mode 9 */
},
{
//Above Mode : 3
{ 1340, 173, 36, 119, 30, 10, 13, 10, 20, 26, }, // left_mode 0
{ 156, 293, 26, 108, 5, 16, 2, 4, 23, 30, }, // left_mode 1
{ 60, 34, 13, 7, 3, 3, 0, 8, 4, 5, }, // left_mode 2
{ 72, 64, 1, 235, 3, 9, 2, 7, 28, 38, }, // left_mode 3
{ 29, 14, 1, 3, 5, 0, 2, 2, 5, 13, }, // left_mode 4
{ 22, 7, 4, 11, 2, 5, 1, 2, 6, 4, }, // left_mode 5
{ 18, 14, 5, 6, 4, 3, 14, 0, 9, 2, }, // left_mode 6
{ 41, 10, 7, 1, 2, 0, 0, 10, 2, 1, }, // left_mode 7
{ 23, 19, 2, 33, 1, 5, 2, 0, 51, 8, }, // left_mode 8
{ 33, 26, 7, 53, 3, 9, 3, 3, 9, 19, }, // left_mode 9
/*Above Mode : 3*/
{ 1340, 173, 36, 119, 30, 10, 13, 10, 20, 26, }, /* left_mode 0 */
{ 156, 293, 26, 108, 5, 16, 2, 4, 23, 30, }, /* left_mode 1 */
{ 60, 34, 13, 7, 3, 3, 0, 8, 4, 5, }, /* left_mode 2 */
{ 72, 64, 1, 235, 3, 9, 2, 7, 28, 38, }, /* left_mode 3 */
{ 29, 14, 1, 3, 5, 0, 2, 2, 5, 13, }, /* left_mode 4 */
{ 22, 7, 4, 11, 2, 5, 1, 2, 6, 4, }, /* left_mode 5 */
{ 18, 14, 5, 6, 4, 3, 14, 0, 9, 2, }, /* left_mode 6 */
{ 41, 10, 7, 1, 2, 0, 0, 10, 2, 1, }, /* left_mode 7 */
{ 23, 19, 2, 33, 1, 5, 2, 0, 51, 8, }, /* left_mode 8 */
{ 33, 26, 7, 53, 3, 9, 3, 3, 9, 19, }, /* left_mode 9 */
},
{
//Above Mode : 4
{ 410, 165, 43, 31, 66, 15, 30, 54, 8, 17, }, // left_mode 0
{ 115, 64, 27, 18, 30, 7, 11, 15, 4, 19, }, // left_mode 1
{ 31, 23, 25, 1, 7, 2, 2, 10, 0, 5, }, // left_mode 2
{ 17, 4, 1, 6, 8, 2, 7, 5, 5, 21, }, // left_mode 3
{ 120, 12, 1, 2, 83, 3, 0, 4, 1, 40, }, // left_mode 4
{ 4, 3, 1, 2, 1, 2, 5, 0, 3, 6, }, // left_mode 5
{ 10, 2, 13, 6, 6, 6, 8, 2, 4, 5, }, // left_mode 6
{ 58, 10, 5, 1, 28, 1, 1, 33, 1, 9, }, // left_mode 7
{ 8, 2, 1, 4, 2, 5, 1, 1, 2, 10, }, // left_mode 8
{ 76, 7, 5, 7, 18, 2, 2, 0, 5, 45, }, // left_mode 9
/*Above Mode : 4*/
{ 410, 165, 43, 31, 66, 15, 30, 54, 8, 17, }, /* left_mode 0 */
{ 115, 64, 27, 18, 30, 7, 11, 15, 4, 19, }, /* left_mode 1 */
{ 31, 23, 25, 1, 7, 2, 2, 10, 0, 5, }, /* left_mode 2 */
{ 17, 4, 1, 6, 8, 2, 7, 5, 5, 21, }, /* left_mode 3 */
{ 120, 12, 1, 2, 83, 3, 0, 4, 1, 40, }, /* left_mode 4 */
{ 4, 3, 1, 2, 1, 2, 5, 0, 3, 6, }, /* left_mode 5 */
{ 10, 2, 13, 6, 6, 6, 8, 2, 4, 5, }, /* left_mode 6 */
{ 58, 10, 5, 1, 28, 1, 1, 33, 1, 9, }, /* left_mode 7 */
{ 8, 2, 1, 4, 2, 5, 1, 1, 2, 10, }, /* left_mode 8 */
{ 76, 7, 5, 7, 18, 2, 2, 0, 5, 45, }, /* left_mode 9 */
},
{
//Above Mode : 5
{ 444, 46, 47, 20, 14, 110, 60, 14, 60, 7, }, // left_mode 0
{ 59, 57, 25, 18, 3, 17, 21, 6, 14, 6, }, // left_mode 1
{ 24, 17, 20, 6, 4, 13, 7, 2, 3, 2, }, // left_mode 2
{ 13, 11, 5, 14, 4, 9, 2, 4, 15, 7, }, // left_mode 3
{ 8, 5, 2, 1, 4, 0, 1, 1, 2, 12, }, // left_mode 4
{ 19, 5, 5, 7, 4, 40, 6, 3, 10, 4, }, // left_mode 5
{ 16, 5, 9, 1, 1, 16, 26, 2, 10, 4, }, // left_mode 6
{ 11, 4, 8, 1, 1, 4, 4, 5, 4, 1, }, // left_mode 7
{ 15, 1, 3, 7, 3, 21, 7, 1, 34, 5, }, // left_mode 8
{ 18, 5, 1, 3, 4, 3, 7, 1, 2, 9, }, // left_mode 9
/*Above Mode : 5*/
{ 444, 46, 47, 20, 14, 110, 60, 14, 60, 7, }, /* left_mode 0 */
{ 59, 57, 25, 18, 3, 17, 21, 6, 14, 6, }, /* left_mode 1 */
{ 24, 17, 20, 6, 4, 13, 7, 2, 3, 2, }, /* left_mode 2 */
{ 13, 11, 5, 14, 4, 9, 2, 4, 15, 7, }, /* left_mode 3 */
{ 8, 5, 2, 1, 4, 0, 1, 1, 2, 12, }, /* left_mode 4 */
{ 19, 5, 5, 7, 4, 40, 6, 3, 10, 4, }, /* left_mode 5 */
{ 16, 5, 9, 1, 1, 16, 26, 2, 10, 4, }, /* left_mode 6 */
{ 11, 4, 8, 1, 1, 4, 4, 5, 4, 1, }, /* left_mode 7 */
{ 15, 1, 3, 7, 3, 21, 7, 1, 34, 5, }, /* left_mode 8 */
{ 18, 5, 1, 3, 4, 3, 7, 1, 2, 9, }, /* left_mode 9 */
},
{
//Above Mode : 6
{ 476, 149, 94, 13, 14, 77, 291, 27, 23, 3, }, // left_mode 0
{ 79, 83, 42, 14, 2, 12, 63, 2, 4, 14, }, // left_mode 1
{ 43, 36, 55, 1, 3, 8, 42, 11, 5, 1, }, // left_mode 2
{ 9, 9, 6, 16, 1, 5, 6, 3, 11, 10, }, // left_mode 3
{ 10, 3, 1, 3, 10, 1, 0, 1, 1, 4, }, // left_mode 4
{ 14, 6, 15, 5, 1, 20, 25, 2, 5, 0, }, // left_mode 5
{ 28, 7, 51, 1, 0, 8, 127, 6, 2, 5, }, // left_mode 6
{ 13, 3, 3, 2, 3, 1, 2, 8, 1, 2, }, // left_mode 7
{ 10, 3, 3, 3, 3, 8, 2, 2, 9, 3, }, // left_mode 8
{ 13, 7, 11, 4, 0, 4, 6, 2, 5, 8, }, // left_mode 9
/*Above Mode : 6*/
{ 476, 149, 94, 13, 14, 77, 291, 27, 23, 3, }, /* left_mode 0 */
{ 79, 83, 42, 14, 2, 12, 63, 2, 4, 14, }, /* left_mode 1 */
{ 43, 36, 55, 1, 3, 8, 42, 11, 5, 1, }, /* left_mode 2 */
{ 9, 9, 6, 16, 1, 5, 6, 3, 11, 10, }, /* left_mode 3 */
{ 10, 3, 1, 3, 10, 1, 0, 1, 1, 4, }, /* left_mode 4 */
{ 14, 6, 15, 5, 1, 20, 25, 2, 5, 0, }, /* left_mode 5 */
{ 28, 7, 51, 1, 0, 8, 127, 6, 2, 5, }, /* left_mode 6 */
{ 13, 3, 3, 2, 3, 1, 2, 8, 1, 2, }, /* left_mode 7 */
{ 10, 3, 3, 3, 3, 8, 2, 2, 9, 3, }, /* left_mode 8 */
{ 13, 7, 11, 4, 0, 4, 6, 2, 5, 8, }, /* left_mode 9 */
},
{
//Above Mode : 7
{ 376, 135, 119, 6, 32, 8, 31, 224, 9, 3, }, // left_mode 0
{ 93, 60, 54, 6, 13, 7, 8, 92, 2, 12, }, // left_mode 1
{ 74, 36, 84, 0, 3, 2, 9, 67, 2, 1, }, // left_mode 2
{ 19, 4, 4, 8, 8, 2, 4, 7, 6, 16, }, // left_mode 3
{ 51, 7, 4, 1, 77, 3, 0, 14, 1, 15, }, // left_mode 4
{ 7, 7, 5, 7, 4, 7, 4, 5, 0, 3, }, // left_mode 5
{ 18, 2, 19, 2, 2, 4, 12, 11, 1, 2, }, // left_mode 6
{ 129, 6, 27, 1, 21, 3, 0, 189, 0, 6, }, // left_mode 7
{ 9, 1, 2, 8, 3, 7, 0, 5, 3, 3, }, // left_mode 8
{ 20, 4, 5, 10, 4, 2, 7, 17, 3, 16, }, // left_mode 9
/*Above Mode : 7*/
{ 376, 135, 119, 6, 32, 8, 31, 224, 9, 3, }, /* left_mode 0 */
{ 93, 60, 54, 6, 13, 7, 8, 92, 2, 12, }, /* left_mode 1 */
{ 74, 36, 84, 0, 3, 2, 9, 67, 2, 1, }, /* left_mode 2 */
{ 19, 4, 4, 8, 8, 2, 4, 7, 6, 16, }, /* left_mode 3 */
{ 51, 7, 4, 1, 77, 3, 0, 14, 1, 15, }, /* left_mode 4 */
{ 7, 7, 5, 7, 4, 7, 4, 5, 0, 3, }, /* left_mode 5 */
{ 18, 2, 19, 2, 2, 4, 12, 11, 1, 2, }, /* left_mode 6 */
{ 129, 6, 27, 1, 21, 3, 0, 189, 0, 6, }, /* left_mode 7 */
{ 9, 1, 2, 8, 3, 7, 0, 5, 3, 3, }, /* left_mode 8 */
{ 20, 4, 5, 10, 4, 2, 7, 17, 3, 16, }, /* left_mode 9 */
},
{
//Above Mode : 8
{ 617, 68, 34, 79, 11, 27, 25, 14, 75, 13, }, // left_mode 0
{ 51, 82, 21, 26, 6, 12, 13, 1, 26, 16, }, // left_mode 1
{ 29, 9, 12, 11, 3, 7, 1, 10, 2, 2, }, // left_mode 2
{ 17, 19, 11, 74, 4, 3, 2, 0, 58, 13, }, // left_mode 3
{ 10, 1, 1, 3, 4, 1, 0, 2, 1, 8, }, // left_mode 4
{ 14, 4, 5, 5, 1, 13, 2, 0, 27, 8, }, // left_mode 5
{ 10, 3, 5, 4, 1, 7, 6, 4, 5, 1, }, // left_mode 6
{ 10, 2, 6, 2, 1, 1, 1, 4, 2, 1, }, // left_mode 7
{ 14, 8, 5, 23, 2, 12, 6, 2, 117, 5, }, // left_mode 8
{ 9, 6, 2, 19, 1, 6, 3, 2, 9, 9, }, // left_mode 9
/*Above Mode : 8*/
{ 617, 68, 34, 79, 11, 27, 25, 14, 75, 13, }, /* left_mode 0 */
{ 51, 82, 21, 26, 6, 12, 13, 1, 26, 16, }, /* left_mode 1 */
{ 29, 9, 12, 11, 3, 7, 1, 10, 2, 2, }, /* left_mode 2 */
{ 17, 19, 11, 74, 4, 3, 2, 0, 58, 13, }, /* left_mode 3 */
{ 10, 1, 1, 3, 4, 1, 0, 2, 1, 8, }, /* left_mode 4 */
{ 14, 4, 5, 5, 1, 13, 2, 0, 27, 8, }, /* left_mode 5 */
{ 10, 3, 5, 4, 1, 7, 6, 4, 5, 1, }, /* left_mode 6 */
{ 10, 2, 6, 2, 1, 1, 1, 4, 2, 1, }, /* left_mode 7 */
{ 14, 8, 5, 23, 2, 12, 6, 2, 117, 5, }, /* left_mode 8 */
{ 9, 6, 2, 19, 1, 6, 3, 2, 9, 9, }, /* left_mode 9 */
},
{
//Above Mode : 9
{ 680, 73, 22, 38, 42, 5, 11, 9, 6, 28, }, // left_mode 0
{ 113, 112, 21, 22, 10, 2, 8, 4, 6, 42, }, // left_mode 1
{ 44, 20, 24, 6, 5, 4, 3, 3, 1, 2, }, // left_mode 2
{ 40, 23, 7, 71, 5, 2, 4, 1, 7, 22, }, // left_mode 3
{ 85, 9, 4, 4, 17, 2, 0, 3, 2, 23, }, // left_mode 4
{ 13, 4, 2, 6, 1, 7, 0, 1, 7, 6, }, // left_mode 5
{ 26, 6, 8, 3, 2, 3, 8, 1, 5, 4, }, // left_mode 6
{ 54, 8, 9, 6, 7, 0, 1, 11, 1, 3, }, // left_mode 7
{ 9, 10, 4, 13, 2, 5, 4, 2, 14, 8, }, // left_mode 8
{ 92, 9, 5, 19, 15, 3, 3, 1, 6, 58, }, // left_mode 9
/*Above Mode : 9*/
{ 680, 73, 22, 38, 42, 5, 11, 9, 6, 28, }, /* left_mode 0 */
{ 113, 112, 21, 22, 10, 2, 8, 4, 6, 42, }, /* left_mode 1 */
{ 44, 20, 24, 6, 5, 4, 3, 3, 1, 2, }, /* left_mode 2 */
{ 40, 23, 7, 71, 5, 2, 4, 1, 7, 22, }, /* left_mode 3 */
{ 85, 9, 4, 4, 17, 2, 0, 3, 2, 23, }, /* left_mode 4 */
{ 13, 4, 2, 6, 1, 7, 0, 1, 7, 6, }, /* left_mode 5 */
{ 26, 6, 8, 3, 2, 3, 8, 1, 5, 4, }, /* left_mode 6 */
{ 54, 8, 9, 6, 7, 0, 1, 11, 1, 3, }, /* left_mode 7 */
{ 9, 10, 4, 13, 2, 5, 4, 2, 14, 8, }, /* left_mode 8 */
{ 92, 9, 5, 19, 15, 3, 3, 1, 6, 58, }, /* left_mode 9 */
},
};

45
vp8/common/onyxc_int.h
View File

@@ -21,9 +21,9 @@
#include "recon.h"
#include "postproc.h"
//#ifdef PACKET_TESTING
/*#ifdef PACKET_TESTING*/
#include "header.h"
//#endif
/*#endif*/
/* Create/destroy static data structures. */
@@ -43,7 +43,7 @@ typedef struct frame_contexts
vp8_prob sub_mv_ref_prob [VP8_SUBMVREFS-1];
vp8_prob coef_probs [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1];
MV_CONTEXT mvc[2];
MV_CONTEXT pre_mvc[2]; //not to caculate the mvcost for the frame if mvc doesn't change.
MV_CONTEXT pre_mvc[2]; /* not to caculate the mvcost for the frame if mvc doesn't change. */
} FRAME_CONTEXT;
typedef enum
@@ -74,6 +74,7 @@ typedef struct VP8_COMMON_RTCD
vp8_subpix_rtcd_vtable_t subpix;
vp8_loopfilter_rtcd_vtable_t loopfilter;
vp8_postproc_rtcd_vtable_t postproc;
int flags;
#else
int unused;
#endif
@@ -83,9 +84,9 @@ typedef struct VP8Common
{
struct vpx_internal_error_info error;
DECLARE_ALIGNED(16, short, Y1dequant[QINDEX_RANGE][4][4]);
DECLARE_ALIGNED(16, short, Y2dequant[QINDEX_RANGE][4][4]);
DECLARE_ALIGNED(16, short, UVdequant[QINDEX_RANGE][4][4]);
DECLARE_ALIGNED(16, short, Y1dequant[QINDEX_RANGE][16]);
DECLARE_ALIGNED(16, short, Y2dequant[QINDEX_RANGE][16]);
DECLARE_ALIGNED(16, short, UVdequant[QINDEX_RANGE][16]);
int Width;
int Height;
@@ -104,7 +105,7 @@ typedef struct VP8Common
YV12_BUFFER_CONFIG post_proc_buffer;
YV12_BUFFER_CONFIG temp_scale_frame;
FRAME_TYPE last_frame_type; //Add to check if vp8_frame_init_loop_filter() can be skipped.
FRAME_TYPE last_frame_type; /* Add to check if vp8_frame_init_loop_filter() can be skipped. */
FRAME_TYPE frame_type;
int show_frame;
@@ -115,7 +116,7 @@ typedef struct VP8Common
int mb_cols;
int mode_info_stride;
// prfile settings
/* profile settings */
int experimental;
int mb_no_coeff_skip;
int no_lpf;
@@ -124,7 +125,7 @@ typedef struct VP8Common
int full_pixel;
int base_qindex;
int last_kf_gf_q; // Q used on the last GF or KF
int last_kf_gf_q; /* Q used on the last GF or KF */
int y1dc_delta_q;
int y2dc_delta_q;
@@ -154,31 +155,31 @@ typedef struct VP8Common
int last_sharpness_level;
int sharpness_level;
int refresh_last_frame; // Two state 0 = NO, 1 = YES
int refresh_golden_frame; // Two state 0 = NO, 1 = YES
int refresh_alt_ref_frame; // Two state 0 = NO, 1 = YES
int refresh_last_frame; /* Two state 0 = NO, 1 = YES */
int refresh_golden_frame; /* Two state 0 = NO, 1 = YES */
int refresh_alt_ref_frame; /* Two state 0 = NO, 1 = YES */
int copy_buffer_to_gf; // 0 none, 1 Last to GF, 2 ARF to GF
int copy_buffer_to_arf; // 0 none, 1 Last to ARF, 2 GF to ARF
int copy_buffer_to_gf; /* 0 none, 1 Last to GF, 2 ARF to GF */
int copy_buffer_to_arf; /* 0 none, 1 Last to ARF, 2 GF to ARF */
int refresh_entropy_probs; // Two state 0 = NO, 1 = YES
int refresh_entropy_probs; /* Two state 0 = NO, 1 = YES */
int ref_frame_sign_bias[MAX_REF_FRAMES]; // Two state 0, 1
int ref_frame_sign_bias[MAX_REF_FRAMES]; /* Two state 0, 1 */
// Y,U,V,Y2
ENTROPY_CONTEXT_PLANES *above_context; // row of context for each plane
ENTROPY_CONTEXT_PLANES left_context; // (up to) 4 contexts ""
/* Y,U,V,Y2 */
ENTROPY_CONTEXT_PLANES *above_context; /* row of context for each plane */
ENTROPY_CONTEXT_PLANES left_context; /* (up to) 4 contexts "" */
// keyframe block modes are predicted by their above, left neighbors
/* keyframe block modes are predicted by their above, left neighbors */
vp8_prob kf_bmode_prob [VP8_BINTRAMODES] [VP8_BINTRAMODES] [VP8_BINTRAMODES-1];
vp8_prob kf_ymode_prob [VP8_YMODES-1]; /* keyframe "" */
vp8_prob kf_uv_mode_prob [VP8_UV_MODES-1];
FRAME_CONTEXT lfc; // last frame entropy
FRAME_CONTEXT fc; // this frame entropy
FRAME_CONTEXT lfc; /* last frame entropy */
FRAME_CONTEXT fc; /* this frame entropy */
unsigned int current_video_frame;

520
vp8/common/postproc.c
View File

@@ -19,7 +19,53 @@
#include <math.h>
#include <stdlib.h>
#include <stdio.h>
// global constants
#define RGB_TO_YUV(t) \
( (0.257*(float)(t>>16)) + (0.504*(float)(t>>8&0xff)) + (0.098*(float)(t&0xff)) + 16), \
(-(0.148*(float)(t>>16)) - (0.291*(float)(t>>8&0xff)) + (0.439*(float)(t&0xff)) + 128), \
( (0.439*(float)(t>>16)) - (0.368*(float)(t>>8&0xff)) - (0.071*(float)(t&0xff)) + 128)
/* global constants */
static const unsigned char MB_PREDICTION_MODE_colors[MB_MODE_COUNT][3] =
{
{ RGB_TO_YUV(0x98FB98) }, /* PaleGreen */
{ RGB_TO_YUV(0x00FF00) }, /* Green */
{ RGB_TO_YUV(0xADFF2F) }, /* GreenYellow */
{ RGB_TO_YUV(0x228B22) }, /* ForestGreen */
{ RGB_TO_YUV(0x006400) }, /* DarkGreen */
{ RGB_TO_YUV(0x98F5FF) }, /* Cadet Blue */
{ RGB_TO_YUV(0x6CA6CD) }, /* Sky Blue */
{ RGB_TO_YUV(0x00008B) }, /* Dark blue */
{ RGB_TO_YUV(0x551A8B) }, /* Purple */
{ RGB_TO_YUV(0xFF0000) } /* Red */
};
static const unsigned char B_PREDICTION_MODE_colors[B_MODE_COUNT][3] =
{
{ RGB_TO_YUV(0x6633ff) }, /* Purple */
{ RGB_TO_YUV(0xcc33ff) }, /* Magenta */
{ RGB_TO_YUV(0xff33cc) }, /* Pink */
{ RGB_TO_YUV(0xff3366) }, /* Coral */
{ RGB_TO_YUV(0x3366ff) }, /* Blue */
{ RGB_TO_YUV(0xed00f5) }, /* Dark Blue */
{ RGB_TO_YUV(0x2e00b8) }, /* Dark Purple */
{ RGB_TO_YUV(0xff6633) }, /* Orange */
{ RGB_TO_YUV(0x33ccff) }, /* Light Blue */
{ RGB_TO_YUV(0x8ab800) }, /* Green */
{ RGB_TO_YUV(0xffcc33) }, /* Light Orange */
{ RGB_TO_YUV(0x33ffcc) }, /* Aqua */
{ RGB_TO_YUV(0x66ff33) }, /* Light Green */
{ RGB_TO_YUV(0xccff33) }, /* Yellow */
};
static const unsigned char MV_REFERENCE_FRAME_colors[MB_MODE_COUNT][3] =
{
{ RGB_TO_YUV(0x00ff00) }, /* Blue */
{ RGB_TO_YUV(0x0000ff) }, /* Green */
{ RGB_TO_YUV(0xffff00) }, /* Yellow */
{ RGB_TO_YUV(0xff0000) }, /* Red */
};
static const short kernel5[] =
{
@@ -76,7 +122,7 @@ const short vp8_rv[] =
extern void vp8_blit_text(const char *msg, unsigned char *address, const int pitch);
extern void vp8_blit_line(int x0, int x1, int y0, int y1, unsigned char *image, const int pitch);
/***********************************************************************************************************
*/
void vp8_post_proc_down_and_across_c
@@ -101,7 +147,7 @@ void vp8_post_proc_down_and_across_c
for (row = 0; row < rows; row++)
{
// post_proc_down for one row
/* post_proc_down for one row */
p_src = src_ptr;
p_dst = dst_ptr;
@@ -124,7 +170,7 @@ void vp8_post_proc_down_and_across_c
p_dst[col] = v;
}
// now post_proc_across
/* now post_proc_across */
p_src = dst_ptr;
p_dst = dst_ptr;
@@ -153,12 +199,12 @@ void vp8_post_proc_down_and_across_c
p_dst[col-2] = d[(col-2)&7];
}
//handle the last two pixels
/* handle the last two pixels */
p_dst[col-2] = d[(col-2)&7];
p_dst[col-1] = d[(col-1)&7];
//next row
/* next row */
src_ptr += pitch;
dst_ptr += pitch;
}
@@ -351,9 +397,9 @@ static void fillrd(struct postproc_state *state, int q, int a)
sigma = ai + .5 + .6 * (63 - qi) / 63.0;
// set up a lookup table of 256 entries that matches
// a gaussian distribution with sigma determined by q.
//
/* set up a lookup table of 256 entries that matches
* a gaussian distribution with sigma determined by q.
*/
{
double i;
int next, j;
@@ -444,6 +490,187 @@ void vp8_plane_add_noise_c(unsigned char *Start, char *noise,
}
}
/* Blend the macro block with a solid colored square. Leave the
* edges unblended to give distinction to macro blocks in areas
* filled with the same color block.
*/
void vp8_blend_mb_inner_c (unsigned char *y, unsigned char *u, unsigned char *v,
int y1, int u1, int v1, int alpha, int stride)
{
int i, j;
int y1_const = y1*((1<<16)-alpha);
int u1_const = u1*((1<<16)-alpha);
int v1_const = v1*((1<<16)-alpha);
y += 2*stride + 2;
for (i = 0; i < 12; i++)
{
for (j = 0; j < 12; j++)
{
y[j] = (y[j]*alpha + y1_const)>>16;
}
y += stride;
}
stride >>= 1;
u += stride + 1;
v += stride + 1;
for (i = 0; i < 6; i++)
{
for (j = 0; j < 6; j++)
{
u[j] = (u[j]*alpha + u1_const)>>16;
v[j] = (v[j]*alpha + v1_const)>>16;
}
u += stride;
v += stride;
}
}
/* Blend only the edge of the macro block. Leave center
* unblended to allow for other visualizations to be layered.
*/
void vp8_blend_mb_outer_c (unsigned char *y, unsigned char *u, unsigned char *v,
int y1, int u1, int v1, int alpha, int stride)
{
int i, j;
int y1_const = y1*((1<<16)-alpha);
int u1_const = u1*((1<<16)-alpha);
int v1_const = v1*((1<<16)-alpha);
for (i = 0; i < 2; i++)
{
for (j = 0; j < 16; j++)
{
y[j] = (y[j]*alpha + y1_const)>>16;
}
y += stride;
}
for (i = 0; i < 12; i++)
{
y[0] = (y[0]*alpha + y1_const)>>16;
y[1] = (y[1]*alpha + y1_const)>>16;
y[14] = (y[14]*alpha + y1_const)>>16;
y[15] = (y[15]*alpha + y1_const)>>16;
y += stride;
}
for (i = 0; i < 2; i++)
{
for (j = 0; j < 16; j++)
{
y[j] = (y[j]*alpha + y1_const)>>16;
}
y += stride;
}
stride >>= 1;
for (j = 0; j < 8; j++)
{
u[j] = (u[j]*alpha + u1_const)>>16;
v[j] = (v[j]*alpha + v1_const)>>16;
}
u += stride;
v += stride;
for (i = 0; i < 6; i++)
{
u[0] = (u[0]*alpha + u1_const)>>16;
v[0] = (v[0]*alpha + v1_const)>>16;
u[7] = (u[7]*alpha + u1_const)>>16;
v[7] = (v[7]*alpha + v1_const)>>16;
u += stride;
v += stride;
}
for (j = 0; j < 8; j++)
{
u[j] = (u[j]*alpha + u1_const)>>16;
v[j] = (v[j]*alpha + v1_const)>>16;
}
}
void vp8_blend_b_c (unsigned char *y, unsigned char *u, unsigned char *v,
int y1, int u1, int v1, int alpha, int stride)
{
int i, j;
int y1_const = y1*((1<<16)-alpha);
int u1_const = u1*((1<<16)-alpha);
int v1_const = v1*((1<<16)-alpha);
for (i = 0; i < 4; i++)
{
for (j = 0; j < 4; j++)
{
y[j] = (y[j]*alpha + y1_const)>>16;
}
y += stride;
}
stride >>= 1;
for (i = 0; i < 2; i++)
{
for (j = 0; j < 2; j++)
{
u[j] = (u[j]*alpha + u1_const)>>16;
v[j] = (v[j]*alpha + v1_const)>>16;
}
u += stride;
v += stride;
}
}
static void constrain_line (int x0, int *x1, int y0, int *y1, int width, int height)
{
int dx;
int dy;
if (*x1 > width)
{
dx = *x1 - x0;
dy = *y1 - y0;
*x1 = width;
if (dx)
*y1 = ((width-x0)*dy)/dx + y0;
}
if (*x1 < 0)
{
dx = *x1 - x0;
dy = *y1 - y0;
*x1 = 0;
if (dx)
*y1 = ((0-x0)*dy)/dx + y0;
}
if (*y1 > height)
{
dx = *x1 - x0;
dy = *y1 - y0;
*y1 = height;
if (dy)
*x1 = ((height-y0)*dx)/dy + x0;
}
if (*y1 < 0)
{
dx = *x1 - x0;
dy = *y1 - y0;
*y1 = 0;
if (dy)
*x1 = ((0-y0)*dx)/dy + x0;
}
}
#if CONFIG_RUNTIME_CPU_DETECT
#define RTCD_VTABLE(oci) (&(oci)->rtcd.postproc)
#else
@@ -465,7 +692,7 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
{
*dest = *oci->frame_to_show;
// handle problem with extending borders
/* handle problem with extending borders */
dest->y_width = oci->Width;
dest->y_height = oci->Height;
dest->uv_height = dest->y_height / 2;
@@ -521,7 +748,8 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
oci->mb_cols, oci->mb_rows);
vp8_blit_text(message, oci->post_proc_buffer.y_buffer, oci->post_proc_buffer.y_stride);
}
else if (flags & VP8D_DEBUG_LEVEL2)
if (flags & VP8D_DEBUG_LEVEL2)
{
int i, j;
unsigned char *y_ptr;
@@ -533,7 +761,7 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
y_ptr = post->y_buffer + 4 * post->y_stride + 4;
// vp8_filter each macro block
/* vp8_filter each macro block */
for (i = 0; i < mb_rows; i++)
{
for (j = 0; j < mb_cols; j++)
@@ -547,12 +775,13 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
y_ptr += 16;
}
mb_index ++; //border
mb_index ++; /* border */
y_ptr += post->y_stride * 16 - post->y_width;
}
}
else if (flags & VP8D_DEBUG_LEVEL3)
if (flags & VP8D_DEBUG_LEVEL3)
{
int i, j;
unsigned char *y_ptr;
@@ -564,7 +793,7 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
y_ptr = post->y_buffer + 4 * post->y_stride + 4;
// vp8_filter each macro block
/* vp8_filter each macro block */
for (i = 0; i < mb_rows; i++)
{
for (j = 0; j < mb_cols; j++)
@@ -581,12 +810,13 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
y_ptr += 16;
}
mb_index ++; //border
mb_index ++; /* border */
y_ptr += post->y_stride * 16 - post->y_width;
}
}
else if (flags & VP8D_DEBUG_LEVEL4)
if (flags & VP8D_DEBUG_LEVEL4)
{
sprintf(message, "Bitrate: %10.2f frame_rate: %10.2f ", oci->bitrate, oci->framerate);
vp8_blit_text(message, oci->post_proc_buffer.y_buffer, oci->post_proc_buffer.y_stride);
@@ -601,7 +831,7 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
y_ptr = post->y_buffer + 4 * post->y_stride + 4;
// vp8_filter each macro block
/* vp8_filter each macro block */
for (i = 0; i < mb_rows; i++)
{
for (j = 0; j < mb_cols; j++)
@@ -614,7 +844,7 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
y_ptr += 16;
}
mb_index ++; //border
mb_index ++; /* border */
y_ptr += post->y_stride * 16 - post->y_width;
}
@@ -623,11 +853,261 @@ int vp8_post_proc_frame(VP8_COMMON *oci, YV12_BUFFER_CONFIG *dest, int deblock_l
}
/* Draw motion vectors */
if (flags & VP8D_DEBUG_DRAW_MV)
{
YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
int width = post->y_width;
int height = post->y_height;
int mb_cols = width >> 4;
unsigned char *y_buffer = oci->post_proc_buffer.y_buffer;
int y_stride = oci->post_proc_buffer.y_stride;
MODE_INFO *mi = oci->mi;
int x0, y0;
for (y0 = 0; y0 < height; y0 += 16)
{
for (x0 = 0; x0 < width; x0 += 16)
{
int x1, y1;
if (mi->mbmi.mode == SPLITMV)
{
switch (mi->mbmi.partitioning)
{
case 0 : /* mv_top_bottom */
{
B_MODE_INFO *bmi = &mi->bmi[0];
MV *mv = &bmi->mv.as_mv;
x1 = x0 + 8 + (mv->col >> 3);
y1 = y0 + 4 + (mv->row >> 3);
constrain_line (x0+8, &x1, y0+4, &y1, width, height);
vp8_blit_line (x0+8, x1, y0+4, y1, y_buffer, y_stride);
bmi = &mi->bmi[8];
x1 = x0 + 8 + (mv->col >> 3);
y1 = y0 +12 + (mv->row >> 3);
constrain_line (x0+8, &x1, y0+12, &y1, width, height);
vp8_blit_line (x0+8, x1, y0+12, y1, y_buffer, y_stride);
break;
}
case 1 : /* mv_left_right */
{
B_MODE_INFO *bmi = &mi->bmi[0];
MV *mv = &bmi->mv.as_mv;
x1 = x0 + 4 + (mv->col >> 3);
y1 = y0 + 8 + (mv->row >> 3);
constrain_line (x0+4, &x1, y0+8, &y1, width, height);
vp8_blit_line (x0+4, x1, y0+8, y1, y_buffer, y_stride);
bmi = &mi->bmi[2];
x1 = x0 +12 + (mv->col >> 3);
y1 = y0 + 8 + (mv->row >> 3);
constrain_line (x0+12, &x1, y0+8, &y1, width, height);
vp8_blit_line (x0+12, x1, y0+8, y1, y_buffer, y_stride);
break;
}
case 2 : /* mv_quarters */
{
B_MODE_INFO *bmi = &mi->bmi[0];
MV *mv = &bmi->mv.as_mv;
x1 = x0 + 4 + (mv->col >> 3);
y1 = y0 + 4 + (mv->row >> 3);
constrain_line (x0+4, &x1, y0+4, &y1, width, height);
vp8_blit_line (x0+4, x1, y0+4, y1, y_buffer, y_stride);
bmi = &mi->bmi[2];
x1 = x0 +12 + (mv->col >> 3);
y1 = y0 + 4 + (mv->row >> 3);
constrain_line (x0+12, &x1, y0+4, &y1, width, height);
vp8_blit_line (x0+12, x1, y0+4, y1, y_buffer, y_stride);
bmi = &mi->bmi[8];
x1 = x0 + 4 + (mv->col >> 3);
y1 = y0 +12 + (mv->row >> 3);
constrain_line (x0+4, &x1, y0+12, &y1, width, height);
vp8_blit_line (x0+4, x1, y0+12, y1, y_buffer, y_stride);
bmi = &mi->bmi[10];
x1 = x0 +12 + (mv->col >> 3);
y1 = y0 +12 + (mv->row >> 3);
constrain_line (x0+12, &x1, y0+12, &y1, width, height);
vp8_blit_line (x0+12, x1, y0+12, y1, y_buffer, y_stride);
break;
}
default :
{
B_MODE_INFO *bmi = mi->bmi;
int bx0, by0;
for (by0 = y0; by0 < (y0+16); by0 += 4)
{
for (bx0 = x0; bx0 < (x0+16); bx0 += 4)
{
MV *mv = &bmi->mv.as_mv;
x1 = bx0 + 2 + (mv->col >> 3);
y1 = by0 + 2 + (mv->row >> 3);
constrain_line (bx0+2, &x1, by0+2, &y1, width, height);
vp8_blit_line (bx0+2, x1, by0+2, y1, y_buffer, y_stride);
bmi++;
}
}
}
}
}
else if (mi->mbmi.mode >= NEARESTMV)
{
MV *mv = &mi->mbmi.mv.as_mv;
const int lx0 = x0 + 8;
const int ly0 = y0 + 8;
x1 = lx0 + (mv->col >> 3);
y1 = ly0 + (mv->row >> 3);
if (x1 != lx0 && y1 != ly0)
{
constrain_line (lx0, &x1, ly0-1, &y1, width, height);
vp8_blit_line (lx0, x1, ly0-1, y1, y_buffer, y_stride);
constrain_line (lx0, &x1, ly0+1, &y1, width, height);
vp8_blit_line (lx0, x1, ly0+1, y1, y_buffer, y_stride);
}
else
vp8_blit_line (lx0, x1, ly0, y1, y_buffer, y_stride);
}
mi++;
}
mi++;
}
}
/* Color in block modes */
if (flags & VP8D_DEBUG_CLR_BLK_MODES)
{
int y, x;
YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
int width = post->y_width;
int height = post->y_height;
unsigned char *y_ptr = oci->post_proc_buffer.y_buffer;
unsigned char *u_ptr = oci->post_proc_buffer.u_buffer;
unsigned char *v_ptr = oci->post_proc_buffer.v_buffer;
int y_stride = oci->post_proc_buffer.y_stride;
MODE_INFO *mi = oci->mi;
for (y = 0; y < height; y += 16)
{
for (x = 0; x < width; x += 16)
{
int Y = 0, U = 0, V = 0;
if (mi->mbmi.mode == B_PRED)
{
int by, bx;
unsigned char *yl, *ul, *vl;
B_MODE_INFO *bmi = mi->bmi;
yl = y_ptr + x;
ul = u_ptr + (x>>1);
vl = v_ptr + (x>>1);
for (by = 0; by < 16; by += 4)
{
for (bx = 0; bx < 16; bx += 4)
{
Y = B_PREDICTION_MODE_colors[bmi->mode][0];
U = B_PREDICTION_MODE_colors[bmi->mode][1];
V = B_PREDICTION_MODE_colors[bmi->mode][2];
POSTPROC_INVOKE(RTCD_VTABLE(oci), blend_b)
(yl+bx, ul+(bx>>1), vl+(bx>>1), Y, U, V, 0xc000, y_stride);
bmi++;
}
yl += y_stride*4;
ul += y_stride*1;
vl += y_stride*1;
}
}
else
{
Y = MB_PREDICTION_MODE_colors[mi->mbmi.mode][0];
U = MB_PREDICTION_MODE_colors[mi->mbmi.mode][1];
V = MB_PREDICTION_MODE_colors[mi->mbmi.mode][2];
POSTPROC_INVOKE(RTCD_VTABLE(oci), blend_mb_inner)
(y_ptr+x, u_ptr+(x>>1), v_ptr+(x>>1), Y, U, V, 0xc000, y_stride);
}
mi++;
}
y_ptr += y_stride*16;
u_ptr += y_stride*4;
v_ptr += y_stride*4;
mi++;
}
}
/* Color in frame reference blocks */
if (flags & VP8D_DEBUG_CLR_FRM_REF_BLKS)
{
int y, x;
YV12_BUFFER_CONFIG *post = &oci->post_proc_buffer;
int width = post->y_width;
int height = post->y_height;
unsigned char *y_ptr = oci->post_proc_buffer.y_buffer;
unsigned char *u_ptr = oci->post_proc_buffer.u_buffer;
unsigned char *v_ptr = oci->post_proc_buffer.v_buffer;
int y_stride = oci->post_proc_buffer.y_stride;
MODE_INFO *mi = oci->mi;
for (y = 0; y < height; y += 16)
{
for (x = 0; x < width; x +=16)
{
int Y = 0, U = 0, V = 0;
Y = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][0];
U = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][1];
V = MV_REFERENCE_FRAME_colors[mi->mbmi.ref_frame][2];
POSTPROC_INVOKE(RTCD_VTABLE(oci), blend_mb_outer)
(y_ptr+x, u_ptr+(x>>1), v_ptr+(x>>1), Y, U, V, 0xc000, y_stride);
mi++;
}
y_ptr += y_stride*16;
u_ptr += y_stride*4;
v_ptr += y_stride*4;
mi++;
}
}
*dest = oci->post_proc_buffer;
// handle problem with extending borders
/* handle problem with extending borders */
dest->y_width = oci->Width;
dest->y_height = oci->Height;
dest->uv_height = dest->y_height / 2;

40
vp8/common/postproc.h
View File

@@ -24,6 +24,18 @@
char whiteclamp[16], char bothclamp[16],\
unsigned int w, unsigned int h, int pitch)
#define prototype_postproc_blend_mb_inner(sym)\
void sym (unsigned char *y, unsigned char *u, unsigned char *v,\
int y1, int u1, int v1, int alpha, int stride)
#define prototype_postproc_blend_mb_outer(sym)\
void sym (unsigned char *y, unsigned char *u, unsigned char *v,\
int y1, int u1, int v1, int alpha, int stride)
#define prototype_postproc_blend_b(sym)\
void sym (unsigned char *y, unsigned char *u, unsigned char *v,\
int y1, int u1, int v1, int alpha, int stride)
#if ARCH_X86 || ARCH_X86_64
#include "x86/postproc_x86.h"
#endif
@@ -48,16 +60,36 @@ extern prototype_postproc(vp8_postproc_downacross);
#endif
extern prototype_postproc_addnoise(vp8_postproc_addnoise);
#ifndef vp8_postproc_blend_mb_inner
#define vp8_postproc_blend_mb_inner vp8_blend_mb_inner_c
#endif
extern prototype_postproc_blend_mb_inner(vp8_postproc_blend_mb_inner);
#ifndef vp8_postproc_blend_mb_outer
#define vp8_postproc_blend_mb_outer vp8_blend_mb_outer_c
#endif
extern prototype_postproc_blend_mb_outer(vp8_postproc_blend_mb_outer);
#ifndef vp8_postproc_blend_b
#define vp8_postproc_blend_b vp8_blend_b_c
#endif
extern prototype_postproc_blend_b(vp8_postproc_blend_b);
typedef prototype_postproc((*vp8_postproc_fn_t));
typedef prototype_postproc_inplace((*vp8_postproc_inplace_fn_t));
typedef prototype_postproc_addnoise((*vp8_postproc_addnoise_fn_t));
typedef prototype_postproc_blend_mb_inner((*vp8_postproc_blend_mb_inner_fn_t));
typedef prototype_postproc_blend_mb_outer((*vp8_postproc_blend_mb_outer_fn_t));
typedef prototype_postproc_blend_b((*vp8_postproc_blend_b_fn_t));
typedef struct
{
vp8_postproc_inplace_fn_t down;
vp8_postproc_inplace_fn_t across;
vp8_postproc_fn_t downacross;
vp8_postproc_addnoise_fn_t addnoise;
vp8_postproc_inplace_fn_t down;
vp8_postproc_inplace_fn_t across;
vp8_postproc_fn_t downacross;
vp8_postproc_addnoise_fn_t addnoise;
vp8_postproc_blend_mb_inner_fn_t blend_mb_inner;
vp8_postproc_blend_mb_outer_fn_t blend_mb_outer;
vp8_postproc_blend_b_fn_t blend_b;
} vp8_postproc_rtcd_vtable_t;
#if CONFIG_RUNTIME_CPU_DETECT

19
vp8/common/ppflags.h
View File

@@ -13,14 +13,17 @@
#define __INC_PPFLAGS_H
enum
{
VP8D_NOFILTERING = 0,
VP8D_DEBLOCK = 1,
VP8D_DEMACROBLOCK = 2,
VP8D_ADDNOISE = 4,
VP8D_DEBUG_LEVEL1 = 8,
VP8D_DEBUG_LEVEL2 = 16,
VP8D_DEBUG_LEVEL3 = 32,
VP8D_DEBUG_LEVEL4 = 64,
VP8D_NOFILTERING = 0,
VP8D_DEBLOCK = 1<<0,
VP8D_DEMACROBLOCK = 1<<1,
VP8D_ADDNOISE = 1<<2,
VP8D_DEBUG_LEVEL1 = 1<<3,
VP8D_DEBUG_LEVEL2 = 1<<4,
VP8D_DEBUG_LEVEL3 = 1<<5,
VP8D_DEBUG_LEVEL4 = 1<<6,
VP8D_DEBUG_DRAW_MV = 1<<7,
VP8D_DEBUG_CLR_BLK_MODES = 1<<8,
VP8D_DEBUG_CLR_FRM_REF_BLKS = 1<<9
};
#endif

47
vp8/common/recon.c
View File

@@ -106,8 +106,24 @@ void vp8_recon2b_c
}
}
void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
void vp8_recon_mby_c(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
#if ARCH_ARM
BLOCKD *b = &x->block[0];
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
/*b = &x->block[4];*/
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
/*b = &x->block[8];*/
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
/*b = &x->block[12];*/
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
#else
int i;
for (i = 0; i < 16; i += 4)
@@ -116,10 +132,36 @@ void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
#endif
}
void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
void vp8_recon_mb_c(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
#if ARCH_ARM
BLOCKD *b = &x->block[0];
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
/*b = &x->block[16];*/
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
#else
int i;
for (i = 0; i < 16; i += 4)
@@ -135,4 +177,5 @@ void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
#endif
}

27
vp8/common/recon.h
View File

@@ -12,11 +12,18 @@
#ifndef __INC_RECON_H
#define __INC_RECON_H
#include "blockd.h"
#define prototype_copy_block(sym) \
void sym(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch)
#define prototype_recon_block(sym) \
void sym(unsigned char *pred, short *diff, unsigned char *dst, int pitch);
void sym(unsigned char *pred, short *diff, unsigned char *dst, int pitch)
#define prototype_recon_macroblock(sym) \
void sym(const struct vp8_recon_rtcd_vtable *rtcd, MACROBLOCKD *x)
struct vp8_recon_rtcd_vtable;
#if ARCH_X86 || ARCH_X86_64
#include "x86/recon_x86.h"
@@ -56,9 +63,20 @@ extern prototype_recon_block(vp8_recon_recon2);
#endif
extern prototype_recon_block(vp8_recon_recon4);
#ifndef vp8_recon_recon_mb
#define vp8_recon_recon_mb vp8_recon_mb_c
#endif
extern prototype_recon_macroblock(vp8_recon_recon_mb);
#ifndef vp8_recon_recon_mby
#define vp8_recon_recon_mby vp8_recon_mby_c
#endif
extern prototype_recon_macroblock(vp8_recon_recon_mby);
typedef prototype_copy_block((*vp8_copy_block_fn_t));
typedef prototype_recon_block((*vp8_recon_fn_t));
typedef struct
typedef prototype_recon_macroblock((*vp8_recon_mb_fn_t));
typedef struct vp8_recon_rtcd_vtable
{
vp8_copy_block_fn_t copy16x16;
vp8_copy_block_fn_t copy8x8;
@@ -66,6 +84,8 @@ typedef struct
vp8_recon_fn_t recon;
vp8_recon_fn_t recon2;
vp8_recon_fn_t recon4;
vp8_recon_mb_fn_t recon_mb;
vp8_recon_mb_fn_t recon_mby;
} vp8_recon_rtcd_vtable_t;
#if CONFIG_RUNTIME_CPU_DETECT
@@ -74,9 +94,6 @@ typedef struct
#define RECON_INVOKE(ctx,fn) vp8_recon_##fn
#endif
#include "blockd.h"
void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
void vp8_recon_intra_mbuv(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x);
#endif

57
vp8/common/reconinter.c
View File

@@ -18,9 +18,10 @@
#include "onyxc_int.h"
#endif
// use this define on systems where unaligned int reads and writes are
// not allowed, i.e. ARM architectures
//#define MUST_BE_ALIGNED
/* use this define on systems where unaligned int reads and writes are
* not allowed, i.e. ARM architectures
*/
/*#define MUST_BE_ALIGNED*/
static const int bbb[4] = {0, 2, 8, 10};
@@ -255,7 +256,7 @@ void vp8_build_inter_predictors_mbuv(MACROBLOCKD *x)
}
}
//encoder only
/*encoder only*/
void vp8_build_inter_predictors_mby(MACROBLOCKD *x)
{
@@ -491,15 +492,16 @@ void vp8_build_uvmvs(MACROBLOCKD *x, int fullpixel)
}
// The following functions are wriiten for skip_recon_mb() to call. Since there is no recon in this
// situation, we can write the result directly to dst buffer instead of writing it to predictor
// buffer and then copying it to dst buffer.
/* The following functions are wriiten for skip_recon_mb() to call. Since there is no recon in this
* situation, we can write the result directly to dst buffer instead of writing it to predictor
* buffer and then copying it to dst buffer.
*/
static void vp8_build_inter_predictors_b_s(BLOCKD *d, unsigned char *dst_ptr, vp8_subpix_fn_t sppf)
{
int r;
unsigned char *ptr_base;
unsigned char *ptr;
//unsigned char *pred_ptr = d->predictor;
/*unsigned char *pred_ptr = d->predictor;*/
int dst_stride = d->dst_stride;
int pre_stride = d->pre_stride;
@@ -535,8 +537,8 @@ static void vp8_build_inter_predictors_b_s(BLOCKD *d, unsigned char *dst_ptr, vp
void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
{
//unsigned char *pred_ptr = x->block[0].predictor;
//unsigned char *dst_ptr = *(x->block[0].base_dst) + x->block[0].dst;
/*unsigned char *pred_ptr = x->block[0].predictor;
unsigned char *dst_ptr = *(x->block[0].base_dst) + x->block[0].dst;*/
unsigned char *pred_ptr = x->predictor;
unsigned char *dst_ptr = x->dst.y_buffer;
@@ -546,26 +548,26 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
unsigned char *ptr_base;
unsigned char *ptr;
unsigned char *uptr, *vptr;
//unsigned char *pred_ptr = x->predictor;
//unsigned char *upred_ptr = &x->predictor[256];
//unsigned char *vpred_ptr = &x->predictor[320];
/*unsigned char *pred_ptr = x->predictor;
unsigned char *upred_ptr = &x->predictor[256];
unsigned char *vpred_ptr = &x->predictor[320];*/
unsigned char *udst_ptr = x->dst.u_buffer;
unsigned char *vdst_ptr = x->dst.v_buffer;
int mv_row = x->mode_info_context->mbmi.mv.as_mv.row;
int mv_col = x->mode_info_context->mbmi.mv.as_mv.col;
int pre_stride = x->dst.y_stride; //x->block[0].pre_stride;
int pre_stride = x->dst.y_stride; /*x->block[0].pre_stride;*/
ptr_base = x->pre.y_buffer;
ptr = ptr_base + (mv_row >> 3) * pre_stride + (mv_col >> 3);
if ((mv_row | mv_col) & 7)
{
x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7, dst_ptr, x->dst.y_stride); //x->block[0].dst_stride);
x->subpixel_predict16x16(ptr, pre_stride, mv_col & 7, mv_row & 7, dst_ptr, x->dst.y_stride); /*x->block[0].dst_stride);*/
}
else
{
RECON_INVOKE(&x->rtcd->recon, copy16x16)(ptr, pre_stride, dst_ptr, x->dst.y_stride); //x->block[0].dst_stride);
RECON_INVOKE(&x->rtcd->recon, copy16x16)(ptr, pre_stride, dst_ptr, x->dst.y_stride); /*x->block[0].dst_stride);*/
}
mv_row = x->block[16].bmi.mv.as_mv.row;
@@ -588,8 +590,9 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
}
else
{
//note: this whole ELSE part is not executed at all. So, no way to test the correctness of my modification. Later,
//if sth is wrong, go back to what it is in build_inter_predictors_mb.
/* note: this whole ELSE part is not executed at all. So, no way to test the correctness of my modification. Later,
* if sth is wrong, go back to what it is in build_inter_predictors_mb.
*/
int i;
if (x->mode_info_context->mbmi.partitioning < 3)
@@ -597,7 +600,7 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
for (i = 0; i < 4; i++)
{
BLOCKD *d = &x->block[bbb[i]];
//vp8_build_inter_predictors4b(x, d, 16);
/*vp8_build_inter_predictors4b(x, d, 16);*/
{
unsigned char *ptr_base;
@@ -609,11 +612,11 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
if (d->bmi.mv.as_mv.row & 7 || d->bmi.mv.as_mv.col & 7)
{
x->subpixel_predict8x8(ptr, d->pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst_ptr, x->dst.y_stride); //x->block[0].dst_stride);
x->subpixel_predict8x8(ptr, d->pre_stride, d->bmi.mv.as_mv.col & 7, d->bmi.mv.as_mv.row & 7, dst_ptr, x->dst.y_stride); /*x->block[0].dst_stride);*/
}
else
{
RECON_INVOKE(&x->rtcd->recon, copy8x8)(ptr, d->pre_stride, dst_ptr, x->dst.y_stride); //x->block[0].dst_stride);
RECON_INVOKE(&x->rtcd->recon, copy8x8)(ptr, d->pre_stride, dst_ptr, x->dst.y_stride); /*x->block[0].dst_stride);*/
}
}
}
@@ -627,7 +630,7 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
{
//vp8_build_inter_predictors2b(x, d0, 16);
/*vp8_build_inter_predictors2b(x, d0, 16);*/
unsigned char *ptr_base;
unsigned char *ptr;
unsigned char *pred_ptr = d0->predictor;
@@ -659,7 +662,7 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
if (d0->bmi.mv.as_int == d1->bmi.mv.as_int)
{
//vp8_build_inter_predictors2b(x, d0, 8);
/*vp8_build_inter_predictors2b(x, d0, 8);*/
unsigned char *ptr_base;
unsigned char *ptr;
unsigned char *pred_ptr = d0->predictor;
@@ -669,11 +672,15 @@ void vp8_build_inter_predictors_mb_s(MACROBLOCKD *x)
if (d0->bmi.mv.as_mv.row & 7 || d0->bmi.mv.as_mv.col & 7)
{
x->subpixel_predict8x4(ptr, d0->pre_stride, d0->bmi.mv.as_mv.col & 7, d0->bmi.mv.as_mv.row & 7, dst_ptr, x->dst.y_stride);
x->subpixel_predict8x4(ptr, d0->pre_stride,
d0->bmi.mv.as_mv.col & 7,
d0->bmi.mv.as_mv.row & 7,
dst_ptr, x->dst.uv_stride);
}
else
{
RECON_INVOKE(&x->rtcd->recon, copy8x4)(ptr, d0->pre_stride, dst_ptr, x->dst.y_stride);
RECON_INVOKE(&x->rtcd->recon, copy8x4)(ptr,
d0->pre_stride, dst_ptr, x->dst.uv_stride);
}
}
else

46
vp8/common/reconintra.c
View File

@@ -14,9 +14,9 @@
#include "reconintra.h"
#include "vpx_mem/vpx_mem.h"
// For skip_recon_mb(), add vp8_build_intra_predictors_mby_s(MACROBLOCKD *x) and
// vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x).
/* For skip_recon_mb(), add vp8_build_intra_predictors_mby_s(MACROBLOCKD *x) and
* vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x).
*/
void vp8_recon_intra_mbuv(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
int i;
@@ -42,7 +42,7 @@ void vp8_build_intra_predictors_mby(MACROBLOCKD *x)
yleft_col[i] = x->dst.y_buffer [i* x->dst.y_stride -1];
}
// for Y
/* for Y */
switch (x->mode_info_context->mbmi.mode)
{
case DC_PRED:
@@ -156,14 +156,14 @@ void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x)
int r, c, i;
int y_stride = x->dst.y_stride;
ypred_ptr = x->dst.y_buffer; //x->predictor;
ypred_ptr = x->dst.y_buffer; /*x->predictor;*/
for (i = 0; i < 16; i++)
{
yleft_col[i] = x->dst.y_buffer [i* x->dst.y_stride -1];
}
// for Y
/* for Y */
switch (x->mode_info_context->mbmi.mode)
{
case DC_PRED:
@@ -204,11 +204,11 @@ void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x)
expected_dc = 128;
}
//vpx_memset(ypred_ptr, expected_dc, 256);
/*vpx_memset(ypred_ptr, expected_dc, 256);*/
for (r = 0; r < 16; r++)
{
vpx_memset(ypred_ptr, expected_dc, 16);
ypred_ptr += y_stride; //16;
ypred_ptr += y_stride; /*16;*/
}
}
break;
@@ -222,7 +222,7 @@ void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x)
((int *)ypred_ptr)[1] = ((int *)yabove_row)[1];
((int *)ypred_ptr)[2] = ((int *)yabove_row)[2];
((int *)ypred_ptr)[3] = ((int *)yabove_row)[3];
ypred_ptr += y_stride; //16;
ypred_ptr += y_stride; /*16;*/
}
}
break;
@@ -233,7 +233,7 @@ void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x)
{
vpx_memset(ypred_ptr, yleft_col[r], 16);
ypred_ptr += y_stride; //16;
ypred_ptr += y_stride; /*16;*/
}
}
@@ -256,7 +256,7 @@ void vp8_build_intra_predictors_mby_s(MACROBLOCKD *x)
ypred_ptr[c] = pred;
}
ypred_ptr += y_stride; //16;
ypred_ptr += y_stride; /*16;*/
}
}
@@ -418,8 +418,8 @@ void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x)
unsigned char *vabove_row = x->dst.v_buffer - x->dst.uv_stride;
unsigned char vleft_col[20];
unsigned char vtop_left = vabove_row[-1];
unsigned char *upred_ptr = x->dst.u_buffer; //&x->predictor[256];
unsigned char *vpred_ptr = x->dst.v_buffer; //&x->predictor[320];
unsigned char *upred_ptr = x->dst.u_buffer; /*&x->predictor[256];*/
unsigned char *vpred_ptr = x->dst.v_buffer; /*&x->predictor[320];*/
int uv_stride = x->dst.uv_stride;
int i, j;
@@ -472,14 +472,14 @@ void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x)
}
//vpx_memset(upred_ptr,expected_udc,64);
//vpx_memset(vpred_ptr,expected_vdc,64);
/*vpx_memset(upred_ptr,expected_udc,64);*/
/*vpx_memset(vpred_ptr,expected_vdc,64);*/
for (i = 0; i < 8; i++)
{
vpx_memset(upred_ptr, expected_udc, 8);
vpx_memset(vpred_ptr, expected_vdc, 8);
upred_ptr += uv_stride; //8;
vpred_ptr += uv_stride; //8;
upred_ptr += uv_stride; /*8;*/
vpred_ptr += uv_stride; /*8;*/
}
}
break;
@@ -491,8 +491,8 @@ void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x)
{
vpx_memcpy(upred_ptr, uabove_row, 8);
vpx_memcpy(vpred_ptr, vabove_row, 8);
upred_ptr += uv_stride; //8;
vpred_ptr += uv_stride; //8;
upred_ptr += uv_stride; /*8;*/
vpred_ptr += uv_stride; /*8;*/
}
}
@@ -505,8 +505,8 @@ void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x)
{
vpx_memset(upred_ptr, uleft_col[i], 8);
vpx_memset(vpred_ptr, vleft_col[i], 8);
upred_ptr += uv_stride; //8;
vpred_ptr += uv_stride; //8;
upred_ptr += uv_stride; /*8;*/
vpred_ptr += uv_stride; /*8;*/
}
}
@@ -538,8 +538,8 @@ void vp8_build_intra_predictors_mbuv_s(MACROBLOCKD *x)
vpred_ptr[j] = predv;
}
upred_ptr += uv_stride; //8;
vpred_ptr += uv_stride; //8;
upred_ptr += uv_stride; /*8;*/
vpred_ptr += uv_stride; /*8;*/
}
}

Some files were not shown because too many files have changed in this diff Show More