h264: log debug output for slightly truncated streams

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

h264: try to better handle h264 streams that are slightly truncated

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

bmv: fix apparent sign error in the frame_off check

Fixes part of Ticket1373

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

bmv: fix integer overflows in vlc decoder.

Fixes part of Ticket1373

Found-by: Piotr Bandurski <ami_stuff@o2.pl>
Based-on-patch-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffv1: fix integer overflow in quant table initialization

Fixes part of Ticket1372

Found-by: Piotr Bandurski <ami_stuff@o2.pl>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffv1: fix crash caused by version becoming inconsistent

Fixes part of Ticket1372

Found-by: Piotr Bandurski <ami_stuff@o2.pl>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fraps: fix version 0/1 input data size check.

Fixes array overread.
Fixes Ticket1371

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

wmv1: check that the input buffer is large enough

Fixes null ptr deref
Fixes Ticket1367

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffprobe: add flat output format.

ffprobe/xml: use writer multiple_sections instead of local context variable.

ffprobe: move packets_and_frames work to writer context.

Let ffmpeg select the null and anull filters.

They are needed since 16e25ef.

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  http: Add the url_shutdown function for https, too
  http: Simplify code by removing a local variable
  http: Clear the old URLContext pointer when closed
  tcp: Try enabling SO_REUSEADDR when listening
  tcp: Check the return values from bind and accept
  avisynth: Make sure the filename passed to avisynth is in the right code page

Conflicts:
libavformat/http.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

dct-test: give emms a memory clober.

Without this, the compiler in theory could attempt to keep
things in float registers across the emms which would cause
bad things to happen.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

dct-test: perform emms before gettime().

This fixes the hypothetical case of a gettime implementation using floats

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: enable fate-mapchan-6ch-extract-2-downmix-mono

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffmpeg: move pan filter insertion up.

This fixes one (disabled) fate test.

Thanks-to: ubitux to pointing at the odd filter order and stereo/mono messup
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

http: Add the url_shutdown function for https, too

Signed-off-by: Martin Storsjö <martin@martin.st>

http: Simplify code by removing a local variable

Signed-off-by: Martin Storsjö <martin@martin.st>

http: Clear the old URLContext pointer when closed

This fixes issues with opening http urls that have authentication
or redirects, introduced in commit e999b641.

Signed-off-by: Martin Storsjö <martin@martin.st>

rv20: fix lowres out of array read

Fixes Ticket1239

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

yopdec: check frame oddness to be within supported limits

Fixes Ticket1365

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

yopdec: check that palette fits in the packet

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

http: use ffurl_closep()

Fixes Ticket1383

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

url: add ffurl_closep() which also sets the context pointer to NULL

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

lavf: use av_freep() in ffurl_close()

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

tcp: Check the return values from bind and accept

Signed-off-by: Martin Storsjö <martin@martin.st>

log: add category to AVClass

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffmpeg: fix memleak

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

dnxhd: replace asserts by av_asserts

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: only check stddev for acodec-ra144

ra144 uses floats so bitexactness cannot be guranteed
This should fix a long standing issue with icc

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Let ffmpeg select the setpts filter.

Do not disable ffmpeg with --disable-everything.

swr: fix a few typo.

fate: try to fix 3rd md5 implementation

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

8svx: fix crash

Fixes Ticket1377

Found-by: Piotr Bandurski <ami_stuff@o2.pl>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

sgienc: add a limit for maximum supported resolution

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

tcp: Try enabling SO_REUSEADDR when listening

Signed-off-by: Martin Storsjö <martin@martin.st>

tcp: Check the return values from bind and accept

Signed-off-by: Martin Storsjö <martin@martin.st>

qdm2: fix incorrect error spam

Fixes Ticket1375

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libmp3lame: add missing layout terminator

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

h264: Fail on DPC its not fully implemented

Fixes part of Ticket1369

Found-by: ami_stuff
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

dv-demux: dont mess with codec values

Fixes part of Ticket1369

Found-by: ami_stuff
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

jacosub_probe: slightly increase the score to limit misdetection.

jacosub_probe: speedup by making only one call to sscanf

avisynth: Make sure the filename passed to avisynth is in the right code page

avisynth is a non-unicode application and cannot accept UTF-8
characters. Therefore, the input filename should be converted to
the correct code page that it expects.

Signed-off-by: Martin Storsjö <martin@martin.st>

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  fate: Fix fate-ac3-fixed-encode for pre-ssse3 x86 machines
  http: Pass the proper return code of net IO operations
  http: Add 'post_data', a new option which sets custom HTTP post data
  lavfi: amix: check active input count before calling request_samples
  vp8: move block coeff arithcoder on stack.
  mp3/ac3 probe: search for PES headers to prevent probing MPEG-PS as MP3.

Conflicts:
libavformat/ac3dec.c
libavformat/mp3dec.c
tests/fate/ac3.mak

Merged-by: Michael Niedermayer <michaelni@gmx.at>

h264: move q0 scan tables into context

This fixes out of global array reads.
The alternative solutions of checking the index or modifying the VLC tables
to prevent the index going outside are each about 1-2 cpu cyclces slower
per coded 4x4 block.
The alternative of padding the global tables directly is more ugly and
moving them to the context should benefit cache locality.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

oggparsevorbis: fix null ptr dereference

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

mpeg4videodec: Check that cplx_estimation_* fits in the available space

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

vf_overlay: rewrite request/push logic.

The old implementation, upon receiving a frame on the main
input, would request an overlay frame if necessary. This could
generate an infinite recursion, off-by-one frame mismatch and
other harmful effects, especially it the split filter is present
upgraph.

The new implementation uses the linear approach: it uses two
buffer queues for frames received out of turn and forwards
request_frame calls to the input where a frame is necessary.

lavfi: add a generic API for buffer queues.

sink_buffer: guard against NULL frames.

This only happens if some filter somewhere is bogus,
but it helps debugging.

gifenc: support resolutions up to 65535x65535

Maybe someone can add a check in the second gif encoder (rgb24), because I'm not sure where it should be added.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

mjpegenc: add a limit for maximum supported resolution

jpeg does not allow more than 65000 pixels for width and height

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

mxfenc: support 50 and 60 frame rates

Reviewed-by: Tomas Härdin <tomas.hardin@codemill.se>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ipmovie_probe: speedup by avoiding memcmp() call

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ac3_probe: speedup by checking for header earlier

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

probetest: allow specifying parameters on the command line

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

yop_probe: check more fields to avoid detecting random data as yop

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffmpeg: allow overriding input top field first

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

binkaudio: check number of channels

Fixes #1380.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

indeo5: check quant_mat

prevents out of array read

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: dont freopen() stdout

it doesnt work and breaks most BSDs

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

range coder test code: decoding errors are av_log_error not debug.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffv1: fix reading global header with CRC

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avidec: fix odd extradata size case.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libavformat: ff_get_bmp_header: return esize too

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: Fix fate-ac3-fixed-encode for pre-ssse3 x86 machines

The default mmxext and sse implementations of apply_window_int16
aren't bitexact.

Signed-off-by: Martin Storsjö <martin@martin.st>

http: Pass the proper return code of net IO operations

Signed-off-by: Martin Storsjö <martin@martin.st>

http: Add 'post_data', a new option which sets custom HTTP post data

This allows doing http posts with a content-length header sent
in advance, avoiding chunked encoding.

Signed-off-by: Martin Storsjö <martin@martin.st>

lavfi: amix: check active input count before calling request_samples

fixes use of the amix filter with only 1 input

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  pcm-mpeg: improve log message wording
  fate: add missing $(TARGET_PATH) to ac3-fixed-encode
  fate: fix md5sum replacement on some systems
  avprobe: correctly set the default formatter
  lavr: add x86-optimized function for mixing 2 to 1 s16p with q8 coeffs
  lavr: add x86-optimized functions for mixing 2 to 1 s16p with float coeffs
  lavr: add C functions for mixing 2 to 1 channels with s16p format
  avprobe: move formatter functions in the context

Conflicts:
ffprobe.c
libavcodec/pcm-mpeg.c
tests/fate/ac3.mak

Merged-by: Michael Niedermayer <michaelni@gmx.at>

vp8: move block coeff arithcoder on stack.

This prevents gcc from assuming that contents of it may have changed
between calls to vp56_range_get_prob(), thus preventing countless (and
unnecessary) movs. Decoding of sintel trailer goes from (avg+SG) 9.796
+/- 0.003 to 9.635 +/- 0.010.

mp3/ac3 probe: search for PES headers to prevent probing MPEG-PS as MP3.

pcm-mpeg: improve log message wording

We support every defined value for channel layout, bitrate and sample depth.
All other values are not unsupported, but reserved.
Update comments to say "are used" instead of "are known or exist".

Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>

h264_cavlc: check prefix before using it.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

h264: increase scantable sizes to avoid overread

We could also check the index but this would slow speed critical code
down.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

truemotion1: Check index, fix out of array read

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffmpeg: disable autoinsertion for -async on the output side of the filtergraph.

This fixes the issue that since the last merge -async caused 2 aresample
filters to be inserted
users that generate funny timestamps in a filtergraph should probably
manually insert a aresample filter where they need it.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

cmdutils: replace 1 line #ifs by if()

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Allow building of command line utils without swresample lib

This allows building ffprobe without swresample

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

exr: make message about missing feature more useful

Signed-off-by: Paul B Mahol <onemda@gmail.com>

fate: fix long standing bug in the g723_1 test

The test compared files of mismatching sample rate

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: add a fate-acodec-dca test that works on all platforms  second try

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: add bitexact flag to ac3-fixed

cant hurt, and should be there for all tests

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

iff: check if there is extradata

Fixes #1368.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

exr: alpha support

Signed-off-by: Paul B Mahol <onemda@gmail.com>

fate: add missing $(TARGET_PATH) to ac3-fixed-encode

Signed-off-by: Mans Rullgard <mans@mansr.com>

fate: fix md5sum replacement on some systems

On systems where the 'md5' command is used, there is a conflict
with the md5() shell function in fate-run.sh.  Using the 'command'
keyword bypasses the shell function for correct behaviour.

Signed-off-by: Mans Rullgard <mans@mansr.com>

fate: disable fate-acodec-dca

the new test doesnt work on all platforms, thus disable it until a better solution is found.
the dca2 test remains and provides partial testing until this is resolved

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: remove unused functions

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: remove tests/codec-regression.sh

This was somehow forgotten in todays merge

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: resurrect acodec-dca

This was lost in todays merge
Ideas to do this less ugly are welcome!

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

lavc: clarify experimental codec message

Should be easier for new users to get a working output.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: try to fix fate-nellymoser-aref-encode make dependancies

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

fate: make idroq test run again.

Didnt investigate what broke it

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  avconv: extend -r to work on any input stream.
  doc/avconv: expand documentation for the -s option.
  avconv: don't print filters inserted by avconv in stream mappings.
  avconv: merge configuration code for complex and simple filters
  avconv: split configuring input filters out of configure_complex_filter

Conflicts:
configure
doc/ffmpeg.texi
ffmpeg.c
tests/ref/fate/idroq-video-encode

Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'f919cc7df6ab844bc12f89fe7bef4fb915a47725'

* commit 'f919cc7df6ab844bc12f89fe7bef4fb915a47725':
  fate: fix acodec/vsynth tests for make 3.81
  pcm_mpeg: fix number of consumed bytes to include the header.
  avfilter: include required header file avfilter.h in video.h
  x86: Avoid movs on BUTTERFLYPS when in AVX mode
  x86: use new schema for ASM macros
  fate: convert codec-regression.sh to makefile rules
  fate: allow tests to specify unit size for psnr comparison
  fate: teach videogen/rotozoom to output a single raw video stream
  http: Add support for reusing the http socket for subsequent requests
  http: Add support for using persistent connections

Merged-by: Michael Niedermayer <michaelni@gmx.at>

ffprobe: fix C escaping.

avprobe: correctly set the default formatter