Ben Avison [Mon, 21 Jul 2014 13:53:09 +0000 (14:53 +0100)]
vc-1: Optimise parser (with special attention to ARM)
The previous implementation of the parser made four passes over each input
buffer (reduced to two if the container format already guaranteed the input
buffer corresponded to frames, such as with MKV). But these buffers are
often 200K in size, certainly enough to flush the data out of L1 cache, and
for many CPUs, all the way out to main memory. The passes were:
1) locate frame boundaries (not needed for MKV etc)
2) copy the data into a contiguous block (not needed for MKV etc)
3) locate the start codes within each frame
4) unescape the data between start codes
After this, the unescaped data was parsed to extract certain header fields,
but because the unescape operation was so large, this was usually also
effectively operating on uncached memory. Most of the unescaped data was
simply thrown away and never processed further. Only step 2 - because it
used memcpy - was using prefetch, making things even worse.
This patch reorganises these steps so that, aside from the copying, the
operations are performed in parallel, maximising cache utilisation. No more
than the worst-case number of bytes needed for header parsing is unescaped.
Most of the data is, in practice, only read in order to search for a start
code, for which optimised implementations already existed in the H264 codec
(notably the ARM version uses prefetch, so we end up doing both remaining
passes at maximum speed). For MKV files, we know when we've found the last
start code of interest in a given frame, so we are able to avoid doing even
that one remaining pass for most of the buffer.
In some use-cases (such as the Raspberry Pi) video decode is handled by the
GPU, but the entire elementary stream is still fed through the parser to
pick out certain elements of the header which are necessary to manage the
decode process. As you might expect, in these cases, the performance of the
parser is significant.
To measure parser performance, I used the same VC-1 elementary stream in
either an MPEG-2 transport stream or a MKV file, and fed it through avconv
with -c:v copy -c:a copy -f null. These are the gperftools counts for
those streams, both filtered to only include vc1_parse() and its callees,
and unfiltered (to include the whole binary). Lower numbers are better:
Before After
File Filtered Mean StdDev Mean StdDev Confidence Change
M2TS No 861.7 8.2 650.5 8.1 100.0% +32.5%
MKV No 868.9 7.4 731.7 9.0 100.0% +18.8%
M2TS Yes 250.0 11.2 27.2 3.4 100.0% +817.9%
MKV Yes 149.0 12.8 1.7 0.8 100.0% +8526.3%
Yes, that last case shows vc1_parse() running 86 times faster! The M2TS
case does show a larger absolute improvement though, since it was worse
to begin with.
This patch has been tested with the FATE suite (albeit on x86 for speed).
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Ben Avison [Mon, 21 Jul 2014 13:53:08 +0000 (14:53 +0100)]
vc-1: Add platform-specific start code search routine to VC1DSPContext.
Initialise VC1DSPContext for parser as well as for decoder.
Note, the VC-1 code doesn't actually use the function pointer yet.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Ben Avison [Mon, 21 Jul 2014 15:25:48 +0000 (16:25 +0100)]
h264: Move start code search functions into separate source files.
This permits re-use with parsers for codecs which use similar start codes.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Diego Biurrun [Sun, 3 Aug 2014 13:21:15 +0000 (06:21 -0700)]
fate: Only generate tests/pixfmts.mak if some pixfmts fate test is run
Diego Biurrun [Mon, 4 Aug 2014 10:39:34 +0000 (03:39 -0700)]
avcodec: Suppress deprecation warnings from DTG code scheduled for removal
Vittorio Giovara [Wed, 23 Jul 2014 20:04:50 +0000 (21:04 +0100)]
fate: png-suite
Every supported format is converted to RGB.
Carl Eugen Hoyos [Sun, 20 Jul 2014 06:34:19 +0000 (02:34 -0400)]
tiff: support reading gray+alpha at 8 bits
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Vittorio Giovara [Sun, 20 Jul 2014 21:54:35 +0000 (22:54 +0100)]
tiff: support reading gray+alpha at 16 bits
Vittorio Giovara [Sun, 20 Jul 2014 21:54:27 +0000 (22:54 +0100)]
png: support reading gray+alpha at 16 bits
Vittorio Giovara [Fri, 1 Aug 2014 18:13:55 +0000 (19:13 +0100)]
png: disable broken MMX/SIMD code for bpp <= 2
The decoder was producing different results when ASM was disabled.
Based on a long debug session with Kostya.
Vittorio Giovara [Sun, 20 Jul 2014 05:05:35 +0000 (01:05 -0400)]
swscale: support AV_PIX_FMT_YA16 as input
Based on a long debug session with Kostya.
Vittorio Giovara [Tue, 29 Jul 2014 13:06:22 +0000 (14:06 +0100)]
avutil: add AV_PIX_FMT_YA16 pixel format
Vittorio Giovara [Tue, 29 Jul 2014 13:02:09 +0000 (14:02 +0100)]
avutil: rename AV_PIX_FMT_Y400A to AV_PIX_FMT_YA8
The rationale is that you have a packed format in form
<greyscale sample> <alpha sample> <greyscale sample> <alpha sample>
and shortening greyscale to 'G' might make one thing about Greenscale instead.
An alias pixel format and color space name are provided for compatibility.
Vittorio Giovara [Wed, 23 Jul 2014 20:05:31 +0000 (21:05 +0100)]
avutil: add alias names for gray 8/16 colour spaces
Luca Barbato [Thu, 24 Jul 2014 21:17:17 +0000 (23:17 +0200)]
pixdesc: Support pixelformat aliases
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Kostya Shishkov [Sun, 3 Aug 2014 09:28:49 +0000 (10:28 +0100)]
swscale: correctly pad destination buffer in rgb conversion
Bug-Id: 772
CC: libav-stable@libav.org
Found-By: Justin Ruggles <justin.ruggles@gmail.com>
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Janne Grunau [Mon, 4 Aug 2014 08:04:08 +0000 (10:04 +0200)]
fate: explicitly set the default THREADS value
This makes the default of '1' more explicit than defaulting to '1' in
fate-run.sh and regression-funcs.sh if THREADS is not set.
Fixes the reported thread count in fate-cpu if THREADS is not set.
Marvin Scholz [Mon, 28 Jul 2014 09:55:38 +0000 (11:55 +0200)]
Add Icecast protocol
Icecast is basically a convenience wrapper around the HTTP protocol.
Signed-off-by: Martin Storsjö <martin@martin.st>
Diego Biurrun [Sun, 3 Aug 2014 18:06:06 +0000 (11:06 -0700)]
imgutils: Do not declare avpriv_set_systematic_pal2 in the public header
Kieran Kunhya [Sun, 3 Aug 2014 18:24:56 +0000 (19:24 +0100)]
avcodec: Deprecate dtg_active_format field in favor of avframe side-data
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Diego Biurrun [Sun, 3 Aug 2014 19:19:10 +0000 (12:19 -0700)]
huffyuv: Check and propagate function return values
Bug-Id: CVE-2013-0868
inspired by a patch from Michael Niedermayer <michaelni@gmx.at>
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Diego Biurrun <diego@biurrun.de>
CC: libav-stable@libav.org
Diego Biurrun [Sun, 3 Aug 2014 18:22:20 +0000 (11:22 -0700)]
huffyuv: Return proper error codes
Diego Biurrun [Sun, 3 Aug 2014 18:02:22 +0000 (11:02 -0700)]
huffyuv: Use avpriv_report_missing_feature() where appropriate
Diego Biurrun [Sun, 3 Aug 2014 17:57:01 +0000 (10:57 -0700)]
huffyuv: Eliminate some pointless casts
Diego Biurrun [Sun, 3 Aug 2014 17:49:40 +0000 (10:49 -0700)]
huffyuv: K&R formatting cosmetics
Anton Khirnov [Sun, 3 Aug 2014 11:59:03 +0000 (04:59 -0700)]
mpeg4video: Initialize xvididct for all threads
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Janne Grunau [Wed, 23 Jul 2014 08:06:15 +0000 (10:06 +0200)]
aarch64: use MACH-O const data asm directive in const macro
Janne Grunau [Thu, 24 Jul 2014 12:50:46 +0000 (14:50 +0200)]
aarch64: add ',' between assembler macro arguments where missing
llvm's integrated assembler does not accept spaces as macro argument
delimiter when targeting darwin. Using a explicit delimiter is a good
idea in principle since it makes case like 'macro 4 -2' vs 'macro 4 - 2'
clear.
Janne Grunau [Sun, 3 Aug 2014 11:03:13 +0000 (13:03 +0200)]
cpu-test: test av_cpu_count
Add CPU count and number threads as informative values for fate.
Janne Grunau [Thu, 30 Jan 2014 12:48:35 +0000 (13:48 +0100)]
fate: add informative cpu test
libavutil/cpu-test prints raw and effective cpu flags to STDERR. Detected
cpu flags can be useful for debugging fate errors.
No comparison of the result against a expected result since that would
require fate config specific references.
Luca Barbato [Sat, 2 Aug 2014 11:29:02 +0000 (13:29 +0200)]
http: Refactor http_open_cnx
Split return value handling from the actual opening.
Incidentally fixes the https -> http redirect issue reported by
Compn on behalf of rcombs.
CC: libav-stable@libav.org
Justin Ruggles [Fri, 18 Jul 2014 21:39:01 +0000 (17:39 -0400)]
lavr: Do not change the sample format for mono audio
This treats mono as planar internally within libavresample rather
than changing the sample format.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Diego Biurrun [Fri, 25 Jul 2014 23:00:33 +0000 (16:00 -0700)]
avcodec: Deprecate unused defines and options
Diego Biurrun [Sat, 2 Aug 2014 19:37:49 +0000 (12:37 -0700)]
avcodec: options: Add missing deprecation ifdefs around emu_edge
Diego Biurrun [Sat, 2 Aug 2014 19:51:48 +0000 (12:51 -0700)]
filtfmts: Replace deprecated uses of AVFilterPad
Also add missing mem.h header for av_freep().
Diego Biurrun [Sat, 2 Aug 2014 19:39:22 +0000 (12:39 -0700)]
examples: filter_audio: Add missing mem.h header for av_freep()
Diego Biurrun [Sat, 2 Aug 2014 19:30:11 +0000 (12:30 -0700)]
vf_fps: Replace use of deprecated AVFilterBufferRef by AVFrame
Diego Biurrun [Sat, 2 Aug 2014 13:14:55 +0000 (06:14 -0700)]
lcl: Disentangle pointers to input data and decompression buffer
This is cleaner and avoids a cast plus a related const qualifier warning.
Diego Biurrun [Sat, 2 Aug 2014 15:47:46 +0000 (08:47 -0700)]
tiff: Replace deprecated PIX_FMT names by modern ones
Diego Biurrun [Sat, 2 Aug 2014 15:45:01 +0000 (08:45 -0700)]
dv: Update DV-profile-related functions to current public API
Diego Biurrun [Thu, 31 Jul 2014 11:46:50 +0000 (04:46 -0700)]
ppc: fft: Build AltiVec optimizations in the standard way
Luca Barbato [Fri, 1 Aug 2014 21:32:40 +0000 (23:32 +0200)]
af_channelmap: Set the frame channel layout
Otherwise the frame would show the first layout matching the
channel count.
Luca Barbato [Fri, 1 Aug 2014 20:27:13 +0000 (22:27 +0200)]
af_join: Set the output frame format
Luca Barbato [Fri, 1 Aug 2014 01:03:20 +0000 (03:03 +0200)]
flv: Index the audio stream
And leverage the video index if the video is just disabled as wm4
did in an initial patch.
Vittorio Giovara [Wed, 30 Jul 2014 18:33:36 +0000 (19:33 +0100)]
h264: prevent theoretical infinite loop in SEI parsing
Properly address CVE-2011-3946 and parse bitstream as described in the spec.
CC: libav-stable@libav.org
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Luca Barbato [Mon, 28 Jul 2014 20:28:13 +0000 (22:28 +0200)]
pixdesc: K&R formatting cosmetics
Also change some comments to multiline.
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Vittorio Giovara [Thu, 31 Jul 2014 13:25:30 +0000 (14:25 +0100)]
pngdec: correctly indent macros
Vittorio Giovara [Wed, 30 Jul 2014 20:12:14 +0000 (21:12 +0100)]
Revert "vf_interlace: deprecate lowpass option"
This reverts commit
35b05c5184fb3aa1191e2d1e7f1cae1e11f344a4.
A warning is introduced in case this option is used.
Diego Biurrun [Tue, 29 Jul 2014 16:57:48 +0000 (09:57 -0700)]
tscc: Eliminate pointless variable indirections in decode_frame()
Diego Biurrun [Tue, 29 Jul 2014 15:43:40 +0000 (08:43 -0700)]
pngenc: Drop pointless pointer cast in png_write_row()
Diego Biurrun [Mon, 21 Jul 2014 20:13:21 +0000 (13:13 -0700)]
idct: Split off Xvid IDCT
The Xvid IDCT is only required to decode some Xvid-encoded MPEG-4 files,
so there is no point in having it as an unconditional part of idctdsp.
Diego Biurrun [Thu, 31 Jul 2014 11:48:07 +0000 (04:48 -0700)]
ppc: idctdsp: Immediately return if no AltiVec is available
This is how all the other init functions operate.
Diego Biurrun [Tue, 29 Jul 2014 09:41:28 +0000 (02:41 -0700)]
fate: Split fate-pixdesc tests and dispatch them through Make
This allows running all the tests individually and/or in parallel.
Michael Niedermayer [Thu, 31 Jul 2014 01:31:19 +0000 (21:31 -0400)]
pgssubdec: Check RLE size before copying
Make sure the buffer size does not exceed the expected
RLE size.
Prevent an out of array bound write.
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Bug-Id: CVE-2013-0852
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Diego Biurrun [Wed, 30 Jul 2014 12:10:18 +0000 (05:10 -0700)]
configure: Globally add ZLIB_CONST to CPPFLAGS if zlib is enabled
Diego Biurrun [Tue, 29 Jul 2014 12:43:04 +0000 (05:43 -0700)]
vf_select: Drop a debug av_log with an unchecked double to enum conversion
CC: libav-stable@libav.org
Nidhi Makhijani [Tue, 29 Jul 2014 15:54:29 +0000 (21:24 +0530)]
mpegvideo: move vol_control_parameters to the only place it is used
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Luca Barbato [Thu, 13 Mar 2014 21:14:43 +0000 (22:14 +0100)]
matroska: Register mime types
Luca Barbato [Thu, 13 Mar 2014 21:12:18 +0000 (22:12 +0100)]
aac: Register the mime type
Speed up probing ADTS live streams that are not frame-aligned such
as http://mp3.streampower.be/radio1.aac .
Luca Barbato [Thu, 13 Mar 2014 21:11:12 +0000 (22:11 +0100)]
avformat: Use the mime type information in input probe
It should provide a quicker guess for elementary streams provided
by http.
Luca Barbato [Tue, 29 Jul 2014 13:29:13 +0000 (14:29 +0100)]
avstring: Expose the simple name match function
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Luca Barbato [Thu, 13 Mar 2014 18:54:00 +0000 (19:54 +0100)]
avformat: Move av_probe_input* to format.c
Diego Biurrun [Mon, 28 Jul 2014 16:21:57 +0000 (09:21 -0700)]
caf: Use correct printf conversion specifiers for POSIX int types
Diego Biurrun [Sat, 26 Jul 2014 12:12:48 +0000 (05:12 -0700)]
sanm: Use correct printf conversion specifiers for POSIX int types
Diego Biurrun [Sat, 26 Jul 2014 12:11:18 +0000 (05:11 -0700)]
dump: Use correct printf conversion specifiers for POSIX int types
Diego Biurrun [Sat, 26 Jul 2014 22:04:44 +0000 (15:04 -0700)]
x86: build: Restore ordering of OBJS lines
Luca Barbato [Thu, 24 Jul 2014 21:50:38 +0000 (23:50 +0200)]
configure: Use require_pkg_config for Speex
Distributors and integrators nowadays have less problems by leveraging
pkg-config files than having to set custom CFLAGS and LDFLAGS.
Marc-Antoine Arnaud [Tue, 22 Jul 2014 16:08:04 +0000 (18:08 +0200)]
mxf: Extract origin information from material and source track
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Marc-Antoine Arnaud [Tue, 22 Jul 2014 14:08:52 +0000 (16:08 +0200)]
mxf: Detect Vanc/Vbi SMPTE-436M mxf track
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Bernhard Übelacker [Sun, 27 Jul 2014 15:38:59 +0000 (08:38 -0700)]
video4linux2: Avoid a floating point exception
This avoids a segfault in avconv_opt.c:opt_target when trying to
determine the norm.
Anton Khirnov [Sun, 27 Jul 2014 10:59:40 +0000 (10:59 +0000)]
hevc: eliminate the last element from TransformTree
Replace it by passing an additional parameter to transform_unit()
Anton Khirnov [Sun, 27 Jul 2014 10:49:27 +0000 (10:49 +0000)]
hevc: eliminate unnecessary cbf_c{b,r} arrays
They are replaced by passing additional parameters to the transform
functions.
Anton Khirnov [Sun, 27 Jul 2014 09:49:37 +0000 (09:49 +0000)]
hevc: do not store the transform inter_split flag in the context
It does not need to be preserved.
Anton Khirnov [Sun, 27 Jul 2014 09:19:16 +0000 (09:19 +0000)]
hevc: simplify splitting the transform tree blocks
Anton Khirnov [Sun, 27 Jul 2014 08:17:29 +0000 (08:17 +0000)]
hevc: eliminate an unnecessary array
We do not need to store the value of the split flag.
Anton Khirnov [Sun, 20 Jul 2014 11:34:30 +0000 (11:34 +0000)]
codec_desc: fix some typos in long codec names
The rv20 typo spotted by Hendrik Leppkes <h.leppkes@gmail.com>
Anton Khirnov [Sun, 20 Jul 2014 11:26:13 +0000 (11:26 +0000)]
lavc: add a property for marking codecs that support frame reordering
Janne Grunau [Sat, 26 Jul 2014 21:29:46 +0000 (23:29 +0200)]
fate: support testing of release branches
Adding 'branch=release/10' to the fate config file will check the
release/10 branch instead of master. If no branch is specified it will
use 'master' so that existing config are still valid.
The server side changes are already deployed, see
https://fate.libav.org/v10/ for an example. The server supports only the
release/* branches.
The server enforces that a single slot tests always the same branch.
Please append "-v$RELEASE" to the slot of release branch configs or make
the slot otherwise unique.
A different fate samples dir is needed for each release branch. make
fate-rsync has the correct URL in each branch.
Anton Khirnov [Sun, 20 Jul 2014 12:06:47 +0000 (12:06 +0000)]
eamad: use the bytestream2 API instead of AV_RL
This is safer and possibly fixes invalid reads on truncated data.
CC:libav-stable@libav.org
Diego Biurrun [Sat, 26 Jul 2014 09:50:59 +0000 (02:50 -0700)]
avformat: Mark AVOutputFormat argument in avformat_query_codec as const
Diego Biurrun [Sat, 26 Jul 2014 09:47:41 +0000 (02:47 -0700)]
avformat: Mark argument in av_{i|o}format_next/ffurl_protocol_next as const
Diego Biurrun [Sat, 26 Jul 2014 09:46:40 +0000 (02:46 -0700)]
avcodec: Mark argument in av_{parser|hwaccel|bitstream_filter}_next as const
Pierre Edouard Lepere [Wed, 18 Jun 2014 03:57:16 +0000 (05:57 +0200)]
hevc: SSE2 and SSSE3 loop filters
Additional contributions by James Almer <jamrial@gmail.com>,
Carl Eugen Hoyos <cehoyos@ag.or.at>, Fiona Glaser <fiona@x264.com> and
Anton Khirnov <anton@khirnov.net>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Anton Khirnov [Sat, 12 Jul 2014 08:36:03 +0000 (08:36 +0000)]
hevcdsp: remove an unneeded variable in the loop filter
beta0 and beta1 will always be the same
Anton Khirnov [Thu, 24 Jul 2014 17:47:26 +0000 (17:47 +0000)]
output example: convert audio to the format supported by the encoder
Anton Khirnov [Wed, 23 Jul 2014 11:09:16 +0000 (11:09 +0000)]
output example: set the stream timebase
This is required by the new API.
Diego Biurrun [Wed, 23 Jul 2014 20:53:54 +0000 (13:53 -0700)]
mpegvideo: Move QMAT_SHIFT* defines to the only place they are used
Diego Biurrun [Wed, 23 Jul 2014 20:50:35 +0000 (13:50 -0700)]
mpegvideo: Move ME_MAP_* defines to the only place they are used
Diego Biurrun [Wed, 23 Jul 2014 20:49:16 +0000 (13:49 -0700)]
mpegvideo: Drop unused MPEG_BUF_SIZE and CHROMA_444 defines
Diego Biurrun [Wed, 23 Jul 2014 20:38:35 +0000 (13:38 -0700)]
fft-test: Pass the right struct members instead of casting
Diego Biurrun [Thu, 24 Jul 2014 23:22:44 +0000 (16:22 -0700)]
vc1dsp: Add wrappers for {avg|put}_vc1_mspel_mc00_c
This avoids invoking the wrapped functions with too many arguments.
Diego Biurrun [Thu, 24 Jul 2014 23:19:45 +0000 (16:19 -0700)]
qpeldsp: Mark source pointer in qpel_mc_func function pointer const
Diego Biurrun [Wed, 23 Jul 2014 14:36:19 +0000 (07:36 -0700)]
dct/rdft: Remove duplicate typedefs for context structs
The typedefs also exist in the avfft.h header and since typedefs cannot be
legally redefined in C, the code fails to compile with some compilers.
This reverts commits
11c7155cce and
57f1b1dcc7.
Nidhi Makhijani [Thu, 24 Jul 2014 14:52:43 +0000 (20:22 +0530)]
mpegenccontext: Remove unused opaque pointer
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Diego Biurrun [Tue, 22 Jul 2014 18:42:03 +0000 (11:42 -0700)]
http: K&R formatting cosmetics
Also comment some #endifs and reshuffle headers into canonical order.
Diego Biurrun [Wed, 23 Jul 2014 14:33:28 +0000 (07:33 -0700)]
fft-test: Pass correct struct members to (i)dct functions
This fixes a number of incompatible pointer type warnings.
Diego Biurrun [Fri, 18 Jul 2014 12:37:11 +0000 (05:37 -0700)]
fft-test: K&R formatting cosmetics
Also rearrange #includes into canonical order.
Diego Biurrun [Mon, 21 Jul 2014 09:12:32 +0000 (02:12 -0700)]
fft-test: Check memory allocations
Diego Biurrun [Thu, 26 Jun 2014 23:14:30 +0000 (16:14 -0700)]
fft-test: Drop unnecessary pointer indirection for context structs
Diego Biurrun [Thu, 26 Jun 2014 00:35:03 +0000 (17:35 -0700)]
fft-test: Reduce variable scope where sensibly possible
Also merge variable declaration and initialization.