Merge commit '4fd34e639d15b44e02686c9b4ef58c9c3c9b0a69'

* commit '4fd34e639d15b44e02686c9b4ef58c9c3c9b0a69':
  h264: remove pointless setting of some variables in loop_filter

Merged-by: Clément Bœsch <clement@stupeflix.com>

Merge commit '9c858ce33fa9b94ebc320dd9d9fa423e708e90cc'

* commit '9c858ce33fa9b94ebc320dd9d9fa423e708e90cc':
  h264: remove a pointless comment

Merged-by: Clément Bœsch <clement@stupeflix.com>

Merge commit 'b77fffa127663028169c5ed543956af4b9496c29'

* commit 'b77fffa127663028169c5ed543956af4b9496c29':
  h264: make slice threading work with deblocking_filter=1

again label, and SLICE_SINGLETHREAD error handling are preserved as that
SLICE_SINGLETHREAD can be raised on a remaining case.

slice_context_count is also kept since it's still in use.

Merged-by: Clément Bœsch <clement@stupeflix.com>

Merge commit '370ddc7b38d6b27b54fc2f5ee5f3dd9506f8c7c8'

* commit '370ddc7b38d6b27b54fc2f5ee5f3dd9506f8c7c8':
  h264: remove H264Context.pict_type

Merged-by: Clément Bœsch <clement@stupeflix.com>

Merge commit '56087ec0a29314d1860f6f0e6f40fbb9b40feccd'

* commit '56087ec0a29314d1860f6f0e6f40fbb9b40feccd':
  h264: drop a pointless indirection

Merged-by: Clément Bœsch <clement@stupeflix.com>

lavc/audiotoolboxdec: Forward extradata for QDMC and QDM2.

Fixes audiotoolbox decoding of QDMC and QDM2.

Merge commit '0e7772c5e4f1b31e2a3dda714ba4f89b1cca644a'

* commit '0e7772c5e4f1b31e2a3dda714ba4f89b1cca644a':
  h264: remove unused H264SliceContext.rbsp_buffer

Merged-by: Clément Bœsch <clement@stupeflix.com>

Merge commit '7f045c4429e91688f1f2335dd347203431901c06'

* commit '7f045c4429e91688f1f2335dd347203431901c06':
  h264: merge ff_h264_free_context() into h264_decode_end()

Merged-by: Clément Bœsch <clement@stupeflix.com>

Merge commit '99c554efc8b09c3f1bb2fb41c3da5431085f7470'

* commit '99c554efc8b09c3f1bb2fb41c3da5431085f7470':
  h264: eliminate low_delay

ff_print_debug_info2() is adjusted to allow a NULL pointer as low_delay.
It's only useful for MPEG codecs with the exception of H264.

Merged-by: Clément Bœsch <clement@stupeflix.com>

lavc/h264_parser: replace AVCodecContext with logging ctx in scan_mmco_reset()

lavc/h264: add a logging ctx to ff_h264_pred_weight_table()

lavc/mediacodecdec{,_h264}: set FF_CODEC_CAP_SETS_PKT_DTS capability

And sets frames pkt_dts to AV_NOPTS_VALUE as we do not want lavc/utils
to overwrite the field with incorrect values as the decoder is
asynchronous.

lavc/videotoolbox: Fix videotoolbox compile error on OS X 10.8.

Fixes error:
libavcodec/videotoolbox.c:511:18: error: implicit declaration of function
'CMVideoFormatDescriptionCreateFromH264ParameterSets' is invalid in C99

This was added in 10.9:
https://developer.apple.com/reference/coremedia/1489818-cmvideoformatdescriptioncreatefr?language=objc

avformat/mux: do not call write_packet with a flush packet if header is not written

Signed-off-by: Marton Balint <cus@passwd.hu>

avformat/mux: do not call write_header multiple times if it fails the first time

Signed-off-by: Marton Balint <cus@passwd.hu>

doc/filters.texi: Move bwdif to correct alphabetical position

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

tests/api/api-codec-param-test: Do not directly access caps_internal

The caps_internal field has moved without major bump and direct
access causes crashes, found when testing 3.1

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

avformat/version: Change the version bumping comment

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

ffplay: ensure that we buffer at least 1 second of content

In order to do that, we keep track of the total duration of packets in a packet
queue.

Signed-off-by: Marton Balint <cus@passwd.hu>

ffplay: factorize checking if a stream needs additional packets

Signed-off-by: Marton Balint <cus@passwd.hu>

avformat/udp: replace packet_gap with bitrate option

We haven't had a stable release since the packet_gap addition, so probably it
is worth reworking the option to something that makes more sense to the end
user. Also add burst_bits option to specify maximum length of bit bursts.

Signed-off-by: Marton Balint <cus@passwd.hu>

avformat/udp: do not accumulate packet_gap delay errors

Signed-off-by: Marton Balint <cus@passwd.hu>

avfilter/vf_bwdif: Change default to deinterlace all frames

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

lavc/magicyuv: fix undefined behaviour introduced in 8a135a55b

Order of evaluation of parameters in C is not defined.

lavc/cafenc: Allow QDMC muxing.

avcodec/qsvenc_h264: declare the QSVEncContext variable before using it

Should fix compilation.

Regression introduced by commit 63adb3602d3b35c5d1df14cf1e477bc458f96b7b

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

Merge commit '5b35b290dd76db64f079b3940f6148c8f273fc17'

* commit '5b35b290dd76db64f079b3940f6148c8f273fc17':
  h264: remove a stale comment

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '89ae244e78a8d636fbb8427a2da052b04cf710bd'

* commit '89ae244e78a8d636fbb8427a2da052b04cf710bd':
  h264_refs: remove an unused parameter from ff_h264_fill_mbaff_ref_list()

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '755f79f84cbeb5d749fb120e55e0098a2d7663a0'

* commit '755f79f84cbeb5d749fb120e55e0098a2d7663a0':
  h264_refs: make the H264Context const where possible

ff_h264_decode_ref_pic_list_reordering() and h264_initialise_ref_list()
do not have a const H264Context * as they modify the default_ref inside
that context.

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit 'a2fd547839d3ef8f342b6f2c096d0887675d3b1a'

* commit 'a2fd547839d3ef8f342b6f2c096d0887675d3b1a':
  h264_refs: reorder functions to avoid forward declarations

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '0ba471d7d864c712f45d7ac6aca4829aba025adc'

* commit '0ba471d7d864c712f45d7ac6aca4829aba025adc':
  h264: eliminate copy_fields

Added a copy of default_ref which isn't present in Libav.

Merged-by: Clément Bœsch <u@pkh.me>

Merge commit '72da8d9bb24d1b1bf74c2f1108650c0da0054d2e'

* commit '72da8d9bb24d1b1bf74c2f1108650c0da0054d2e':
  h264_parser: remove the remaining dependencies on the h264 decoder

Merged-by: Clément Bœsch <u@pkh.me>

lavc/h264: move history parsing variable to H264ParseContext

This makes h264_find_frame_end() not depend on H264Context anymore.

lavc/h264_parser: pass logctx to h264_find_frame_end()

This helps removing the H264Context from the H264ParseContext.

Merge commit '98c97994c5b90bdae02accb155eeceeb5224b8ef'

* commit '98c97994c5b90bdae02accb155eeceeb5224b8ef':
  h264: decouple extradata parsing from the decoder

Main changes:

- move get_avc_nalsize() inside h264_parser.c and make it use
  H264ParseContext instead of H264Context. This helps fixing
  fate-flv-demux.

- Also use is_avc/nal_length_size from the H264ParseContext in various
  places instead of the H264Context one as that's the fields now filled
  by ff_h264_decode_extradata()

- h264_parse: dont fail decode_extradata_ps() due to nal split failure.
  Change by Michael to fix decoding of h264/ref_10.avi.

Merged-by: Clément Bœsch <u@pkh.me>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/magicyuv: check dimensions

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

libavcodec: factor out SEI generation for A53 captions

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

Revert "avfilter/trim: add compatibility layer to not break ABI used by ffmpeg"

This reverts commit 9219ec93b145725ac74fbfbde7f67ac5135b85cf.

Fixes Ticket 5411

lavc/h264_parse: remove tabs introduced in a2922b5d

Merge commit '728d90a0c1973661a9e73da697bf4f90c9d19577'

* commit '728d90a0c1973661a9e73da697bf4f90c9d19577':
  h264: decouple h264_sei from the h264 decoder

Main changes:

- SEI decoding doesn't have access to the debug flag in the codec context so a
  few logging are dropped.

- naming of quincunx_sampling_flag and frame_packing_arrangement_type are kept
  as they are in FFmpeg instead of respectively quincunx_subsampling and
  arrangement_type used in Libav because the former match the specifications.

- don't reset the x264 build info once read in order to fix
  fate-h264-lossless (change by Hendrik)

- H264Context.has_recovery_point and deprecated
  AVCodecContext.dtg_active_format are set after ff_h264_sei_decode()
  based on the SEI state since ff_h264_sei_decode() doesn't have access
  to H264Context anymore.

- frame_packing_arrangement_type is not checked against <= 0 in
  decode_postinit() since it is always read as a positive value with
  get_bits(). This fixes a -Wtype-limits warning by GCC spotted by
  Michael.

Side Notes:

- tested that ffprobe on the file from ticket #3652 still returns 4
  keyframes
- tested that playback from ticket #3063 still works

Merged-by: Clément Bœsch <clement@stupeflix.com>
Signed-off-by: Hendrik Leppkes <h.leppkes@gmail.com>

avcodec/ffv1enc: silence warning about deprecated coded_frame

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

avcodec/snow: Document BlockNode

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

avcodec/snow: Remove unused function

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

lavc/videotoolbox: remove config_info maximum capacity

In case of extradata, config_info will contain two entries instead of
one.

lavc/videotoolbox: CFRelease() session

"When you are done with a decompression session you created, call
VTDecompressionSessionInvalidate to tear it down and then CFRelease to
release your object reference."

fate/h264: add test for ticket #631

Add experimental support for vp9 in iso-bmff

Implemented according to the draft specification
"VP Codec ISO Media File Format Binding":
http://www.webmproject.org/vp9/#draft-vp-codec-iso-media-file-format-binding

'-strict -2' is required to use this feature.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

swresample/resample: do not increase phase_count on exact_rational

high phase_count is only useful when dst_incr_mod is non zero
in other word, it is only useful on soft compensation

on init, it will build filter with low phase_count
but when soft compensation is enabled, rebuild filter
with high phase_count

this approach saves lots of memory

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

swresample/resample: add support for odd phase_count

because exact_rational does not guarantee
that phase_count is even

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

d3d11va: don't keep the context lock while waiting for a frame

also fixes a deadlock found by Денис Кулаков <kudesnik33ra@gmail.com>

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

lavc/videotoolboxenc: remove unnecessary logging

Signed-off-by: Rick Kern <kernrj@gmail.com>

MAINTAINERS: Add myself for videotoolbox.c, remove vda* maintainer

Person in MAINTAINERS hasn't responded to a patch on the ML or private
email, and doesn't maintain the files according to git.

Signed-off-by: Rick Kern <kernrj@gmail.com>

lavc/videotoolbox: fix H.264 hwaccel init issue

Fixes VTDecompressionSessionCreate() error.

Signed-off-by: Rick Kern <kernrj@gmail.com>

lavc/audiotoolboxenc: fix dropped frames on iOS

AudioConverterFillComplexBuffer() doesn't always call its callback. A frame
queue is used to prevent skipped audio samples.

Signed-off-by: Rick Kern <kernrj@gmail.com>

avformat/http: change seek_ret to int64

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

avutil/threadmessage.h: Fix swapped comments

Fix swapped descriptions of av_thread_message_queue_set_err_send
and av_thread_message_queue_set_err_recv.

Signed-off-by: Jan Sebechlebsky <sebechlebskyjan@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>

avodec/h264dec: fix compilation with -DTRACE

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

fate: add test for agate

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

avcodec/sheervideo: check build_vlc for failure

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

lavd/v4l2: allow devices not implementing VIDIOC_G_PARM

Not all v4l2 devices implement the VIDIOC_G_PARM ioctl. This patch allow
ffmpeg to open such device and treat it the same as devices that do
implement the ioctl but returns that it do not implement the
V4L2_CAP_TIMEPERFRAME capability.

Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Benoit Fouet <benoit.fouet@free.fr>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

doc/APIchanges: Fill in some missing things

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

fate: add swr-resample_exact_async tests

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

lavc/dpx: Support decoding 12 bit colourspace with transparency information.

lavc/mediacodec: re-indent after previous commit

lavc/mediacodec: refactor ff_AMediaCodecList_getCodecByType

Allows to select a codec (encoder or decoder) only if it supports a
specific profile.

Adds ff_AMediaCodecProfile_getProfileFromAVCodecContext to convert an
AVCodecContext profile to a MediaCodec profile. It only supports H264
for now.

The codepath using MediaCodecList.findDecoderForFormat() (Android >= 5.0)
has been dropped as this method does not allow to select a decoder
compatible with a specific profile.

avfilter/af_hdcd: Use int32_t instead of int for gaintable in hdcd filter.

Signed-off-by: Benjamin Steffes <benjaminst123@gmail.com>
(comment by ronald)
prevent the theoretical case where the container type (int)
would be 64 bit on some platforms, which would waste some space

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

vf_codecview: added new options

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

hls muxer doc: clarify segment splitting option

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

os_support: use the appropriate stat functions matching the stat type

The stat struct is defined to stati64, which requires using the appropriate wstati/stati functions as well.
Fixes a whole bunch of compiler warnings as well as build breakage with the decklink avdevice.

Fixes trac #5640

avformat/utils: in debug mode, print number of streams found before avformat_find_stream_info()

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

avformat/mpegts: enhance logging in trace mode

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

avfilter/vf_telecine: Make frame writable before writing into it

Fixes Ticket 5627

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

x86/aacpsdsp: optimize add_squares loop

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

lavc/dpx: Support decoding 10 bit colourspace with transparency information.

Fixes one sample from ticket #5639.

doc/muxers/mpegts: apply misc consistency/formatting fixes

lavf/utils: fix timestamp trace log in estimate_timings() function

Use the correct timebase, also replace "stream" with "format" in the
context of the format.

lavc/h264_sei: fix broken style around green metadata code

lavc/h264: rename GreenMetaData to H264SEIGreenMetaData

Reduces diff for the next merge with Libav.

lavc/h264_sei: reduce scope of parameters for green meta decode

This is again will help the merge as ff_h264_decode_sei will not have
access to H264Context anymore.

lavc/h264: move green meta logging out of the sei decoding

This will simplify the next Libav merge where SEI decoding doesn't have
access to the debug level anymore.

rtpdec_vp9: Support parsing the scalability structure

We still only support one single layer though, but this allows
receiving streams that have this structure present even for
single layer streams.

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

rtpdec_h264: Use avpriv_report_missing_feature instead of a manual av_log

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

rtpdec: Use AVERROR_PATCHWELCOME instead of AVERROR(ENOSYS) for
unimplemented features

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

rtpdec_vp9: Update header parsing to spec draft 02

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

rtpdec_vp9: Make sure to free the temp buffer on close

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

fate: fate-filter-acrossfade needs samples

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

avformat/mpegts: adjust probe score for low check_count

Fixes mis-detection of tiff as mpegts
Fixes Ticket5565

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

avformat/mux: factorize header writing code

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Marton Balint <cus@passwd.hu>

avformat/mux: call deinit if write_header fails

Docs clearly states that av_write_trailer should only be called if
avformat_write_header was successful, therefore we have to deinit if we return
failure.

Signed-off-by: Marton Balint <cus@passwd.hu>

avformat/mxfdec: check if source_package is NULL

Fixes ticket #5554.

Reviewed-by: Tomas Härdin <tomas.hardin@codemill.se>
Signed-off-by: Marton Balint <cus@passwd.hu>

lavc/h264_sei: fix green metadata typo

avcodec/ffv1dec: fix some unsupported pix_fmt

When checking pix_fmt mapping, some bitstreams are mapped to an
incorrect pix_fmt instead of being rejected (ENOSYS).
Actually, such bitstreams are not supported (FFmpeg encoder does not
produce such bitstream, such bitstream may come only from another
encoder for the moment).

- JPEG 2000 RCT 11/13/15/16 bit depths are mapped to a 8-bit FFmpeg
pix_fmt (e.g. bgr0), which is not expected.
- JPEG 2000 RCT 9/10/12/14 bit depths with alpha are mapped to a
FFmpeg pix_fmt without alpha (e.g. AV_PIX_FMT_GBRP9 for 9-bit with
alpha), which is not expected.

The order for choosing the pix_fmt is changed to the one used by YCbCr
selection (<=8 bit first).
" && !f->transparency" is added to the other lines.

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

avformat/au: Read MetaData from AU Sun audio file header

recognize title= album= artist= genre= track=

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

avformat/dump: Show coded dimensions again

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

rtpdec_vc2hq: avoid magic number

rtpenc: packetizer for VP9 RTP payload format (draft v2)

avformat/utils: Export coded dimensions unconditionally

This fixes a API regression
Probably fixes Ticket5451

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

avfilter/graphparser: remove '\n' from parse_filter

this allow a filter to be written like this:
aformat =
    sample_fmts  = fltp|flt:
    sample_rates = 44100|44800

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

doc/resampler: add exact_rational option

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

fate: add libswresample exact_rational test

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