cuvid: support AVCodecContext.hw_device_ctx API

This is a newer API that is intended for decoders like the cuvid
wrapper. Until now, the wrapper required to set an awkward
"incomplete" hw_frames_ctx to set the device. Now the device
can be set directly, and the user can get AV_PIX_FMT_CUDA output
for a specific device simply by setting hw_device_ctx.

This still does a dummy ff_get_format() call at init time, and should
be fully backward compatible.

ffmpeg: check for unconnected outputs

Fixes e.g.:

ffmpeg -f lavfi -i testsrc -f lavfi -i testsrc -filter_complex "[0:v][1:v]psnr[out]" -f null none

compat/strtod: Add missing const qualifiers.

Fixes many warnings:
initialization discards 'const' qualifier from pointer target type

avcodec/mjpegdec: Fix runtime error: signed integer overflow: -24543 * 2031616 cannot be represented in type 'int'

Fixes: 943/clusterfuzz-testcase-5114865297391616
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/wavpack: Fix invalid shift and integer overflow

Fixes: 940/clusterfuzz-testcase-5200378381467648
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/flicvideo: Check for chunk overread

Fixes integer overflow
Fixes: 1292/clusterfuzz-testcase-minimized-5795512143839232
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/mpeg4videodec: Convert sprite_offset to 64bit

This avoids intermediates from overflowing (the final values are checked)
Fixes: runtime error: signed integer overflow: -167712 + -2147352576 cannot be represented in type 'int'
Fixes: 1298/clusterfuzz-testcase-minimized-5955580877340672
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/pngdec: Use ff_set_dimensions()

Fixes OOM
Fixes: 1314/clusterfuzz-testcase-minimized-4621997222920192
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/msvideo1: Check buffer size before re-getting the frame

Fixes timeout
Fixes: 1306/clusterfuzz-testcase-minimized-6152296217968640
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

configure: Do not add omit-frame-pointer for ossfuzz

ossfuzz works better without it

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

tools/target_dec_fuzzer: Fix memleak on open failure

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

configure: fix libopus detection

Prevents compilation failures on libopus < 1.0.3
Regression since 37941878.

avformat/concatdec: port to the new bitstream filter API

Reviewed-by: wm4 <nfxjfg@googlemail.com>
Reviewed-by: Aaron Levinson <alevinsn@aracnet.com>
Signed-off-by: James Almer <jamrial@gmail.com>

tools/target_dec_fuzzer: Use decoder and not codec_id as argument

This allows fuzzing decoders with the same codec_id
We also avoid register all to allow the linker to prune unused sections and symbols

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Merge commit '11a9320de54759340531177c9f2b1e31e6112cc2'

* commit '11a9320de54759340531177c9f2b1e31e6112cc2':
  build: Move build-system-related helper files to a separate subdirectory

"ffbuild" directory name is used instead of "avbuild".

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'f9edc734e0ca3f6ef06c1ad0bd2c19c0c66f1ffa'

* commit 'f9edc734e0ca3f6ef06c1ad0bd2c19c0c66f1ffa':
  ratecontrol: Drop xvid-rc-related struct members unused after a6901b9c6

This commit is a noop, see 99b35a51ccbe2fc98530e5cf7529e7d377e1165e

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '5b26d3b789bd19a32dbe1e9c7ccab9498de7ee9b'

* commit '5b26d3b789bd19a32dbe1e9c7ccab9498de7ee9b':
  nvenc: Update check for lookahead

This commit is a noop, see 67db4ff3b66e96a20ddf1c264d07e146334ae88e

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'a0c443a3980dc22eb02b067ac4cb9ffa2f9b04d2'

* commit 'a0c443a3980dc22eb02b067ac4cb9ffa2f9b04d2':
  aarch64: vp9itxfm: Use the offset parameter to movrel

This commit is a noop, see 8b11a89c06b94632d545f67ca508bd9c05c435ac

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'fc322d6a70189da24dbd445c710bb214eb031ce7'

* commit 'fc322d6a70189da24dbd445c710bb214eb031ce7':
  tta: Convert to the new bitstream reader
  mlp: Convert to the new bitstream reader
  unary: Convert to the new bitstream reader

This merge is a noop, see
http://ffmpeg.org/pipermail/ffmpeg-devel/2017-April/209609.html

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '45286a625c6ced1f5c4c842244cbb4509429abba'

* commit '45286a625c6ced1f5c4c842244cbb4509429abba':
  h264dec: make sure to only end a field if it has been started

This commit is a noop. Our h264 slicing handling is different and does
not seem to be affected by the issue.

Merged-by: Clément Bœsch <u@pkh.me>

avcodec/svq3: Reject dx/dy beyond 16bit

The code does use 16bit sized arrays later so larger deltas would not work

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/svq3: Increase offsets to prevent integer overflows

Fixes: 1280/clusterfuzz-testcase-minimized-6102353767825408
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavc/jpeg2000dec: Fix jp2 inner atom size used for overread checks.

avcodec/decode: also check for FF_CODEC_CAP_SETS_PKT_DTS in audio decoders

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>

avutil/softfloat: use ldexp(), fixes undefined shift

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

ffmpeg: count packets when queued

Because write_packet() fakely writes packets to muxer by queueing
them when muxer hasn't been initialized, it should also increment
frame_number fakely.
This is required because code in do_streamcopy() rely on
frame_number.

Should fix Ticket6227

Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Muhammad Faiz <mfcc64@gmail.com>

avcodec/decode: do not treat discarded frames as eof when draining

Fix fate failures:
    make fate-mov THREADS=32

Reviewed-by: wm4 <nfxjfg@googlemail.com>
Signed-off-by: Muhammad Faiz <mfcc64@gmail.com>

avfilter/interlace: add complex vertical low-pass filter

This complex (-1 2 6 2 -1) filter slightly less reduces interlace 'twitter' but better retain detail and subjective sharpness impression compared to the linear (1 2 1) filter.

Signed-off-by: Thomas Mundt <tmundt75@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

lavf/nutdec: Fix an impossible condition, regression since e0c53c34.

Fixes ticket #6362.

avcodec/internal: update FF_CODEC_CAP_SETS_PKT_DTS doxy

The code it refers to was moved to decode.c in 00fb745a10a151791ce2e49ba3e463bd17ea9251

tools/target_dec_fuzzer: Use avcodec_register_all() instead of register_all()

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavc/mips/iirfilter_mips: Include config.h.

Fixes the following warning:
libavcodec/mips/iirfilter_mips.c:57:5: warning: "HAVE_INLINE_ASM" is not defined

lavc/mips/hevc_idct_msa: Add missing const qualifier.

Fixes many warnings:
initialization discards 'const' qualifier from pointer target type

avcodec/indeo2: Check remaining bits in ir2_decode_plane()

Fixes: 1290/clusterfuzz-testcase-minimized-5815578902134784
Fixes: timeout
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/vp3: Check remaining bits in unpack_dct_coeffs()

Decreases the time spend decoding junk.

May fix: 1283/clusterfuzz-testcase-minimized-6221126759874560

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

doc/developer: Add terse documentation of assumed C implementation defined behavior

Suggested-by: "Ronald S. Bultje" <rsbultje@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/bmp: Use ff_set_dimensions()

Fixes out of memory

Fixes: 1282/clusterfuzz-testcase-minimized-5400131681648640
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavformat/http: Ignore expired cookies

Signed-off-by: Micah Galizia <micahgalizia@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

fate/exr : add test for negative float value

the tested sample contain negative value in the red channel
need to be clip to zero, and not set to MAX_RED

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

fate/exr : add tests for piz uncompress

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/exr : fix piz uncompress on big endian

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/options: do a more thorough clean up in avcodec_copy_context()

Free coded_frame, coded_side_data and unref hw_device_ctx to prevent
potential leaks.

Reviewed-by: Aaron Levinson <alevinsn@aracnet.com>
Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/options: factorize avcodec_copy_context() cleanup code

Reviewed-by: Aaron Levinson <alevinsn@aracnet.com>
Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>

avformat/matroskaenc: add support for writing Content Light Level elements

Signed-off-by: James Almer <jamrial@gmail.com>

avformat/matroskadec: export Content Light Level metadata

Based on a patch by Hendrik Leppkes

Signed-off-by: James Almer <jamrial@gmail.com>

Merge commit 'c2fa6bb0e8703a7a6aa10e11f9ab36094416d83f'

* commit 'c2fa6bb0e8703a7a6aa10e11f9ab36094416d83f':
  mpeg12dec: move setting first_field to mpeg_field_start()

This commit is a noop, see 2f6f2f4f73c068979829e785d08cb1dd345c7fc8

Merged-by: James Almer <jamrial@gmail.com>

tools: Eliminate codec_type complexity from fuzzer

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libswscale/tests/swscale: Fix uninitialized variables

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/af_crystalizer: add support for more sample formats

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

avcodec/pthread_frame, decode: allow errors to happen on draining

So, all frames and errors are correctly reported in order.
Also limit the numbers of error during draining to prevent infinite loop.

This fix fate failure with THREADS>=4:
  make fate-h264-attachment-631 THREADS=4
This also reverts a755b725ec1d657609c8bd726ce37e7cf193d03f.

Suggested-by: wm4, Ronald S. Bultje, Marton Balint
Reviewed-by: w4 <nfxjfg@googlemail.com>
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Muhammad Faiz <mfcc64@gmail.com>

avfilter: add video oscilloscope filter

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

avfilter: add pixscope filter

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

Merge commit 'e807491fc6a336e4becc0cbc981274a8fde18aba'

* commit 'e807491fc6a336e4becc0cbc981274a8fde18aba':
  mpeg12dec: avoid signed overflow in bitrate calculation
  mpegvideo_parser: avoid signed overflow in bitrate calculation

This merge is a noop.

2017-04-29 12:54:15     @ubitux michaelni: is 740959fdbfbf804ccd8a6e426b1b1ba321fe5cfb enough to fix the overflow fixed in 58405de0951a843765625159402870c1eea3c3b1?
2017-04-29 12:55:53     @ubitux same question with e807491fc6a336e4becc0cbc981274a8fde18aba
2017-04-29 13:21:45     michaelni       ubitux, the libav code refered to is wrong for us and i doubt the problem it fixes applies to us.
2017-04-29 13:24:29     @ubitux michaelni: ok, for both commits?
2017-04-29 13:33:55     michaelni       yes, they do more or less the same thing

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '46191a2da16f751e53d93646ae1388d421d12bee'

* commit '46191a2da16f751e53d93646ae1388d421d12bee':
  mov: fix a possible invalid read in mov_read_mac_string()

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'cfa4eb4fba782f3f37a33be997b27a91a07053c9'

* commit 'cfa4eb4fba782f3f37a33be997b27a91a07053c9':
  vaapi_decode: use the correct logging context

This commit is a noop, see 79307ae56374b35cf12563a7c8e3e759658f847e

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'ea8b730d8e67152107d7fcdd5590bbb51ec236b1'

* commit 'ea8b730d8e67152107d7fcdd5590bbb51ec236b1':
  hevcdec: add a VAAPI hwaccel

This commit is a noop, see adb54e59c18db347f39e55832104fc3e40a3c42b

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '1783d7ec03d730c5bd96c07bc5fa7aa566f85c66'

* commit '1783d7ec03d730c5bd96c07bc5fa7aa566f85c66':
  Changelog: add some missing entries

Merged-by: Clément Bœsch <u@pkh.me>

doc/filters: add one lowpass filter example

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

avformat/movenc: Explicitly address potential division by zero.

find_fps attempts to infer framerate from AVCodec's timebase. When this
results in a frame rate that isn't explicitly marked as supported in
av_timecode_check_frame_rate, find_fps returns the AVStream's
avg_frame_rate, which, per avformat.h, _may_ be set (or not).

mov_get_mpeg2_xdcam_codec_tag, mov_get_h264_codec_tag and
find_compressor attempt to call av_q2d on the return value of find_fps,
which in the above case, may result in division by zero and therefore,
an undefined frame rate when NaN is converted to int.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libvpxenc: allow aq-mode 4 (equator360)

this was added in 1.6.0

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Zern <jzern@google.com>

avfilter/lavfutils: use image2pipe demuxer on ff_load_image

allow protocols other than file to be used
for example, use data protocol to embed a file in script

Signed-off-by: Muhammad Faiz <mfcc64@gmail.com>

avformat/hlsenc: fix CID 1405135

Fixes Coverity CID: 1405135

Reviewed-by: Rodger Combs <rodger.combs@gmail.com>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

avformat/hlsenc: hold old key info when append list

fix ticket id: #6353

Signed-off-by: Steven Liu <lq@chinaffmpeg.org>

opus_pvq: use function pointers for recursion

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

avcodec/opus_pvq: fix recursive inlining compilation failures

They were introduced by f16180f44859587e0211170bd82b1bfd26f24b5a

avcodec/mdec: Fix runtime error: left shift of negative value -127

Fixes undefined behavior
Fixes: 1275/clusterfuzz-testcase-minimized-6718162017976320
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

tools/target_dec_fuzzer: Fix return code on open failure

Fixes: 1271/clusterfuzz-testcase-minimized-6095220498235392
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/x86/vc1dsp_init: Fix build failure with --disable-optimizations and clang

compilers doing DCE at -O0 do not necessarily understand "complex" boolean expressions
Build succeeds with this change, this was the only failure

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/exr : fix float to uint16 conversion for negative float value

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

opus_pvq: merge band encoding and decoding into one function

Most code between the 2 functions was duplicated which made keeping
both in sync difficult.

This also fixes some discovered issues with encoding (incorrect
TF switching buffers) and reduces stack usage (reuse the already
allocated CeltFrame->scratch buffer for the quantized coefficients).

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

opus_pvq: remove outdated/incorrect comments and redundant variables

Removes the last style issues with opus_pvq.c

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

opus_pvq: minor cleanups

Removes unneeded variables, renames confusing and innacurate variables
and rewrites and slightly optimizes hadamard interleave/deinterleave
functions.

Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>

avcodec/nvenc: surface allocation reduction

This patch aims to reduce the number of input/output surfaces
NVENC allocates per session. Previous default sets allocated surfaces to 32
(unless there is user specified param or lookahead involved). Having large
number of surfaces consumes extra video memory (esp for higher resolution
encoding). The patch changes the surfaces calculation for default, B-frames,
lookahead scenario respectively.

The other change involves surface selection. Previously, if a session
allocates x surfaces, only x-1 surfaces are used (due to combination
of output delay and lock toggle logic). To prevent unused surfaces,
changing surface rotation to using predefined fifo.

Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>

lavc/hevcdec: fix invalid use of ff_get_format()

Regression since 76cc100afba25308eaa909acd8804cc0b42057f6.

Spotted-by: James Almer <jamrial@gmail.com>

Merge commit 'd4a91e65343be5d79a4afa61c791191e1b57499a'

* commit 'd4a91e65343be5d79a4afa61c791191e1b57499a':
  pthread_frame: do not run hwaccel decoding asynchronously unless it's safe

This commit is a noop, see e0cd598bc4684654d63942e9ff4872c0b48a7dc2

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '8dfba25ce89b62c80ba83e2116d549176c376144'

* commit '8dfba25ce89b62c80ba83e2116d549176c376144':
  pthread_frame: ensure the threads don't run simultaneously with hwaccel

This commit is a noop, see 14bb15bfd56d6e907fabe4620206c1ee152b7a20

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '373fd76b4dbd9aa03ed28e502f33f2ca8c1ce19a'

* commit '373fd76b4dbd9aa03ed28e502f33f2ca8c1ce19a':
  hevcdec: do not set decoder-global SPS prematurely

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '0fea8555ae25124c21f4c4f55a5fa76e9169aa03'

* commit '0fea8555ae25124c21f4c4f55a5fa76e9169aa03':
  v4l2: use codec descriptors for mapping a codec name to id

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'ee480790c7eeb03c9cebd8971c46e0cb7db65277'

* commit 'ee480790c7eeb03c9cebd8971c46e0cb7db65277':
  build: Add name parameter to check_lib() helper function

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '1faffe7e8fab21186a233011bc8a62f47962e2cd'

* commit '1faffe7e8fab21186a233011bc8a62f47962e2cd':
  configure: Disentangle vfw32 and user32 lib handling

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'f7174d7ed045445d00a6d557236737d09ad32343'

* commit 'f7174d7ed045445d00a6d557236737d09ad32343':
  configure: fix linking with MSVC when using --disable-optimizations

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'ef9a711be718ed3802a263d1d9ed340a4aaef224'

* commit 'ef9a711be718ed3802a263d1d9ed340a4aaef224':
  configure: put d3d11 check in alphabetical order

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'a4fec9a7eab842ea5eea1b1ee98624356cb31422'

* commit 'a4fec9a7eab842ea5eea1b1ee98624356cb31422':
  rtmppkt: Check for packet size mismatches

See 7d57ca4d9a75562fa32e40766211de150f8b3ee7

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'cdcfa97dc49d83b5eefd0a651db6bb0a6f98e8f2'

* commit 'cdcfa97dc49d83b5eefd0a651db6bb0a6f98e8f2':
  libavformat: Fix a faulty api deprecation guard in prepare_input_packet

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '2425d7329fdccfa9954faba748f3865151354f0c'

* commit '2425d7329fdccfa9954faba748f3865151354f0c':
  arm64: replace 'bic' with immediate with 'and' with inverted immediate

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '05a603a94e4b3eeefa5e18ae653a848001461e89'

* commit '05a603a94e4b3eeefa5e18ae653a848001461e89':
  ppc: Merge types_altivec.h into util_altivec.h

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'ea7ee4b4e381e0fa731458de0cbf740430eeb013'

* commit 'ea7ee4b4e381e0fa731458de0cbf740430eeb013':
  ppc: Centralize compiler-specific altivec.h #include handling in one place

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '39929e55eb13eeb8dfbe1bc99301fecf6b8942dd'

* commit '39929e55eb13eeb8dfbe1bc99301fecf6b8942dd':
  ppc: hevcdsp: Use shorthands for vector types

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '554e55bbf0e4a3640a784cb512b816e776c56333'

* commit '554e55bbf0e4a3640a784cb512b816e776c56333':
  decode.h: Add missing headers to fix standalone compilation

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '343e2833994655c252d5236a3394bf6db7a4d8b1'

* commit '343e2833994655c252d5236a3394bf6db7a4d8b1':
  pthread_frame: use better memory orders for frame progress

This commit is a noop, see c358c62550e60a150c49f192d72631142a6eedd6

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'e94b9313b21c3d91a36ef064f7fe3e867616f47f'

* commit 'e94b9313b21c3d91a36ef064f7fe3e867616f47f':
  fate: Add h264 test for frame num gaps

This commit is a noop, see b7e4ea0c80f4b3e060625fd97ffdd3b9689bfcd1

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '5c7f2cf81df06614f255f061850132355a01d75e'

* commit '5c7f2cf81df06614f255f061850132355a01d75e':
  h264_slice: Wait for refs to be available before we use them in error concealment

This commit is a noop, see 4413e950b26396bfb797168e4f81164a022cc6f3

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '86157e6db2c7a9222f77fa7e7f50fb9aebc3aa81'

* commit '86157e6db2c7a9222f77fa7e7f50fb9aebc3aa81':
  hevc: decouple calling get_format() from exporting the SPS parameters

See 786032cad8ecabe577d9cff0356da6e9e9488a2d (which has been reverted
and replaced with Anton's version to reduce diffs between the two
projects).

Merged-by: Clément Bœsch <u@pkh.me>

avformat/concatdec: fix the h264 annexb extradata check

The start code can be either in the first three or four bytes.

Merge commit '730c02326094bcfb1fa67f10a7e7b22f03f5a88f'

* commit '730c02326094bcfb1fa67f10a7e7b22f03f5a88f':
  binkaudio: switch to the new send/receive API

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'fa1749dd34c55fb997c97dfc4da9383c9976ab91'

* commit 'fa1749dd34c55fb997c97dfc4da9383c9976ab91':
  vp9: split superframes in the filtering stage before actual decoding

This commit is a noop.

2017-04-24 20:45:04     @ubitux BBB: btw, do you think you can get the bsf thing this week or we should skip it to give you more time and go on with the merges?
2017-04-24 20:45:20     @BBB    I’m not sure I’ll finish it that soon
2017-04-24 20:45:26     @BBB    I’d skip it and leave it for later
2017-04-24 20:45:35     @BBB    I’ll do it, I promise, but I Can’t guarantee it’ll be done by $date

Merged-by: Clément Bœsch <u@pkh.me>

avfilter/af_biquads: allow filtering only selected channels

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

fate/exr : fix pix_fmt

rgb_scanline_pxr24_half_uint32_13x9.exr doesn't have alpha

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

tools/target_dec_fuzzer: Remove FuzzerInterface.h dependancy

The header is not always available in the docker build environment

Suggested-by: Kostya Serebryany
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

fate: Add test for pkt_size of ffprobe

Suggested-by: James Almer
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>