Merge commit 'c29da01ac95ea2c8c5c4b3a312a33aaaa8fb7068'

* commit 'c29da01ac95ea2c8c5c4b3a312a33aaaa8fb7068':
  svq3: Convert to the new bitstream reader

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

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

Merge commit 'acfa7a2178f08fd81b66279959cd55ec3ae237e2'

* commit 'acfa7a2178f08fd81b66279959cd55ec3ae237e2':
  configure: Drop weak dependencies on external libraries for webm muxer

This commit is a noop.

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

Merge commit '66988320794a107f2a460eaa71dbd9fab8056842'

* commit '66988320794a107f2a460eaa71dbd9fab8056842':
  configure: Add proper weak dependency of drawtext filter on libfontconfig

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

Merge commit '24d5680bbc01fc124709d522d348572ad4672563'

* commit '24d5680bbc01fc124709d522d348572ad4672563':
  configure: Simplify inline asm check with appropriate helper function

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

Merge commit 'b3825723dceffc64240da7b0e562bd1fd024da26'

* commit 'b3825723dceffc64240da7b0e562bd1fd024da26':
  configure: Merge compiler/libc/os hacks sections

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

Merge commit '577326d430593a25456393a75212b95d1cd94131'

* commit '577326d430593a25456393a75212b95d1cd94131':
  lavc: deprecate refcounted_frames field

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

Merge commit '3ad825793a43253154bed05827f27425fc0757df'

* commit '3ad825793a43253154bed05827f27425fc0757df':
  hwcontext_cuda: implement frames_get_constraints

This commit is a noop, c16fe1432d88f87a96be9e943e0f1229543ad61d

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

Merge commit 'fd9212f2edfe9b107c3c08ba2df5fd2cba5ab9e3'

* commit 'fd9212f2edfe9b107c3c08ba2df5fd2cba5ab9e3':
  Mark some arrays that never change as const.

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

Merge commit 'b420a27e74750b60d2e064236afb10be06a38ace'

* commit 'b420a27e74750b60d2e064236afb10be06a38ace':
  avconv: allow -b to be used with streamcopy

This commit is a noop. We already have this functionality.

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

Merge commit 'ab2539bd374fe7ddbc6e2f058b62645cd5076192'

* commit 'ab2539bd374fe7ddbc6e2f058b62645cd5076192':
  ffv1: Convert to the new bitstream reader
  h261dec: Convert to the new bitstream reader
  shorten: Convert to the new bitstream reader
  ralf: Convert to the new bitstream reader
  loco: Convert to the new bitstream reader
  fic: Convert to the new bitstream reader
  dirac: Convert to the new bitstream reader
  cavs: Convert to the new bitstream reader
  aic: Convert to the new bitstream reader
  golomb: Convert to the new bitstream reader

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

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

Merge commit 'ab87af41636b081dd3562423999351b5444fa09e'

* commit 'ab87af41636b081dd3562423999351b5444fa09e':
  configure: Add proper weak dependency of avformat on network

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

Merge commit '612cc0712836af2f025b0c68b11da29b9f259d5a'

* commit '612cc0712836af2f025b0c68b11da29b9f259d5a':
  pgssubdec: reset rle_data_len/rle_remaining_len on allocation error

This commit is a noop, see 842e98b4d83d8cf297e2bc2761f1f47eb89e49e4

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

Merge commit '708e84cda1bdbffb92847f3d6ccf6fbeb26d9948'

* commit '708e84cda1bdbffb92847f3d6ccf6fbeb26d9948':
  mov: Avoid memcmp of uninitialised data

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

Merge commit 'ca62236a89f47bd871eaf69d8d9e837c93c55a6c'

* commit 'ca62236a89f47bd871eaf69d8d9e837c93c55a6c':
  vaapi_encode: Add VP8 support
  vaapi_encode: Pass framerate parameters to driver
  vaapi_h264: Enable VBR mode
  vaapi_encode: Support VBR mode

This commit is a noop, see
ceb28c3cc4c7921935b48904db3c559eed1597fe
2201c02e6dc9f9652a8e27dec194915f05954ad0
be6546a4ff592785d039df6cbdd7659781d30b2c
d1acab8293054151157910eb081d5edcc7496e13

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

Merge commit 'c5c663541739cb813a2a5668ee8339b535b35d7d'

* commit 'c5c663541739cb813a2a5668ee8339b535b35d7d':
  doc: add dash muxer

Merged-by: Rodger Combs <rodger.combs@gmail.com>

Merge commit '7295b7373862ee54903b33d6ef3335531dfa93ad'

* commit '7295b7373862ee54903b33d6ef3335531dfa93ad':
  dashenc: add webm support

Merged-by: Rodger Combs <rodger.combs@gmail.com>

Merge commit '01f1f017d831cf14617aaaeafcec3ae3a81efce7'

* commit '01f1f017d831cf14617aaaeafcec3ae3a81efce7':
  dashenc: use avio_dynbuf instead of packet_write callback

Merged-by: Rodger Combs <rodger.combs@gmail.com>

Merge commit 'dce2929efa8e82b0832a828f7e8cb81ff8c20a4e'

* commit 'dce2929efa8e82b0832a828f7e8cb81ff8c20a4e':
  dashenc: copy language and role metadata from streams assigned to sets

Merged-by: Rodger Combs <rodger.combs@gmail.com>

Merge commit 'ca9bc9de690258d4761a19b0df6e9c9113b80115'

* commit 'ca9bc9de690258d4761a19b0df6e9c9113b80115':
  dashenc: default to one AdaptationSet per stream

Merged-by: Rodger Combs <rodger.combs@gmail.com>

Merge commit 'efd2fc41b3f0749f9715d50b581f22bbaa8c5b99'

* commit 'efd2fc41b3f0749f9715d50b581f22bbaa8c5b99':
  dashenc: allow assigning all streams of a media type to an AdaptationSet

Merged-by: Rodger Combs <rodger.combs@gmail.com>

Merge commit '3d23a5f96ad72961c14ba3a0c2add8f2ab374b61'

* commit '3d23a5f96ad72961c14ba3a0c2add8f2ab374b61':
  dashenc: add support for assigning streams to AdaptationSets

Merged-by: Rodger Combs <rodger.combs@gmail.com>

Merge commit '9df9309d233f59d9706444a1e24ac24139f2640d'

* commit '9df9309d233f59d9706444a1e24ac24139f2640d':
  dashenc: calculate stream bitrate from first segment if not available

Merged-by: Rodger Combs <rodger.combs@gmail.com>

Revert "lavf/dashenc: update bitrates on dash_write_trailer"

This reverts commit 89c0fda5f43d8a3d3a1c538ff8d72e6737bc7d8e.

A different solution will be committed instead.

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

lavf/tls_gnutls: fix warnings from version check

The GnuTLS version is checked through the macro GNUTLS_VERSION_NUMBER,
but this wasn't introduced before 2.7.2. Building with older versions
of GnuTLS (using icc) warns:

src/libavformat/tls_gnutls.c(38): warning #193: zero used for undefined preprocessing identifier "GNUTLS_VERSION_NUMBER"
  #if HAVE_THREADS && GNUTLS_VERSION_NUMBER <= 0x020b00

This adds a fallback to the older, deprecated LIBGNUTLS_VERSION_NUMBER
macro.

Signed-off-by: Moritz Barsnick <barsnick@gmx.net>

lavf/tls_gnutls: fix compilation with GnuTLS 2.x

Commit 598e41684066feba701d19ca7443d24b9e5efa77 added use of
GNUTLS_E_PREMATURE_TERMINATION, which wasn't introduced to GnuTLS
before 2.99.x / 3.x. This fixes compilation with older versions.

Signed-off-by: Moritz Barsnick <barsnick@gmx.net>

avcodec/dnxhdenc: fix DNxHR 444 encoding crashes

Fixes #6649.

avformat/img2enc: remove av_dup_packet() call

It's unnecessary after a call to av_packet_ref().

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

ffplay: replace use of av_copy_packet()

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

avformat: replace all uses of av_copy_packet()

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

avcodec/avpacket: deprecate av_copy_packet()

It does the same thing as av_packet_ref().

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

avfilter/interlace: rename two variables for consistency

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

avfilter/af_headphone: increase max ir length

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

fate/fits: add missing png & gif dependencies

Signed-off-by: Paras Chadha <paraschadha18@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/hapdec : add support HapAlphaOnly

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

libavformat : add mov dataformat tag for HapAlphaOnly and HapQAlpha

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

avcodec/takdec: Fix integer overflow in decode_lpc()

Fixes: runtime error: signed integer overflow: 16748560 + 2143729712 cannot be represented in type 'int'
Fixes: 3202/clusterfuzz-testcase-minimized-4988291642294272

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

avcodec/proresdec2: Check bits in DECODE_CODEWORD(), fixes invalid shift

Fixes: runtime error: shift exponent 42 is too large for 32-bit type 'unsigned int'
Fixes: 3410/clusterfuzz-testcase-minimized-5313377960198144

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

avcodec/takdec: Fix integer overflows in decode_subframe()

Fixes: runtime error: signed integer overflow: -1562477869 + -691460395 cannot be represented in type 'int'
Fixes: 3196/clusterfuzz-testcase-minimized-4528307146063872

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

avformat/libopenmpt: Query duration and metadata after selecting subsong

Duration depends on the selected subsong and thus must be queried after
selecting the subsong. There is no compelling reason to query other
metadata earlier either.

Tested with libopenmpt version: 0.2.8760-beta27
Libopenmpt configure options: --without-ogg --without-vorbis
--without-vorbisfile --without-portaudio --without-portaudiocpp
--without-mpg123 --without-pulseaudio --without-sndfile --without-flac

Signed-off-by: Jörn Heusipp <osmanx@problemloesungsmaschine.de>
Signed-off-by: Josh de Kock <josh@itanimul.li>

avcodec/mips: Improve avc chroma horiz mc msa functions

Replace generic with block size specific function.

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mips: Unrolled loops avc intra msa functions

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mips: Improve hevc uni-w copy mc msa functions

Load the specific destination bytes instead of MSA load and pack.
Pack the data to half word before clipping.
Use immediate unsigned saturation for clip to max saving one vector register.

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/flacenc: Replace "return -1" by named constant

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

avcodec/snowenc: Replace "return -1" by named constants

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

libavdevice/v4l2: fix invalid access to struct v4l2_buffer

In case we are short of queued buffers, at first v4l2_buffer was enqueued to kernel so it's not owned by
user-space anymore. After that it's timestamp field was read, but it might be overwritten by driver at
that moment. It resulted in invalid timestamp sometimes.

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

ffplay: zero initialize copy avpacket

Prevents potential use of uninitialized stack.

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

avformat/mpeg: zero initialize idx_pkt

Prevents use of uninitialized stack.

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

avcodec/v4l2_m2m: add missing header inclusions

Should fix checkheaders on systems where v4l2_m2m is enabled.

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

avcodec/Makefile: skip v4l2_m2m headers if needed

Fixes checkheaders on systems where v4l2_m2m is unavailable.

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

avcodec/frame_thread_encoder: use av_packet_alloc()

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

avfilter/interlace: add support for 10 and 12 bit

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Thomas Mundt <tmundt75@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avfilter/tinterlace: use drawutils for pad mode

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Thomas Mundt <tmundt75@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avformat/mxfdec: use the common packet pts setter function for opatom files

Fixes ticket #6631.

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

avformat/mxfdec: factorize packet pts setter function

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

avcodec/opusenc_psy: use av_clip_uintp2()

Fixes fate-source.

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

avcodec/opusenc_utils: add missing preprocessor guards

Fixes fate-source.

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

lavfi: make window_func an inline function

Eliminate lavc->lavfi dependency. The function isn't big and doesn't
deserve its own file.

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

opusenc: implement a psychoacoustic system

This commit implements a psychoacoustic system for the native Opus
encoder. Its unlike any other psychoacoustic system known since its
capable of using a lookahead to make better choices on how to treat the
current frame and how many bits to allocate for it (and future frames).

Also, whilst the main bulk of the analysis function has to run in a
single thread, the per-frame anaylsis functions does not modify the main
psychoacoustic context, so in the future it will be fairly trivial to
run those as slice threads.

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

libavcodec: v4l2: add support for v4l2 mem2mem codecs

    This patchset enhances Alexis Ballier's original patch and validates
    it using Qualcomm's Venus hardware (driver recently landed upstream
    [1]).

    This has been tested on Qualcomm's DragonBoard 410c and 820c
    Configure/make scripts have been validated on Ubuntu 10.04 and
    16.04.

    Tested decoders:
           - h264
           - h263
           - mpeg4
           - vp8
           - vp9
           - hevc

    Tested encoders:
           - h264
           - h263
           - mpeg4

    Tested transcoding (concurrent encoding/decoding)

    Some of the changes introduced:
        - v4l2: code cleanup and abstractions added
        - v4l2: follow the new encode/decode api.
        - v4l2: fix display size for NV12 output pool.
        - v4l2: handle EOS (EPIPE and draining)
        - v4l2: vp8 and mpeg4 decoding and encoding.
        - v4l2: hevc and vp9 support.
        - v4l2: generate EOF on dequeue errors.
        - v4l2: h264_mp4toannexb filtering.
        - v4l2: fixed make install and fate issues.
        - v4l2: codecs enabled/disabled depending on pixfmt defined
        - v4l2: pass timebase/framerate to the context
        - v4l2: runtime decoder reconfiguration.
        - v4l2: add more frame information
        - v4l2: free hardware resources on last reference being released
        - v4l2: encoding: disable b-frames for upstreaming (patch required)

    [1] https://lwn.net/Articles/697956/

    System Level view:
        v42l_m2m_enc/dec --> v4l2_m2m --> v4l2_context --> v4l2_buffers

Reviewed-by: Jorge Ramirez <jorge.ramirez-ortiz@linaro.org>
Reviewed-by: Alexis Ballier <aballier@gentoo.org>
Tested-by: Jorge Ramirez <jorge.ramirez-ortiz@linaro.org>
Signed-off-by: wm4 <nfxjfg@googlemail.com>

avformat/gif: use av_packet_alloc()

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

avformat/apngenc: use av_packet_alloc()

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

avcodec/dirac_dwt: Fix integer overflow in COMPOSE_FIDELITYi*()

Fixes: runtime error: signed integer overflow: 161 * 13872281 cannot be represented in type 'int'

Fixes: 3295/clusterfuzz-testcase-minimized-4738998142500864

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

avcodec/ffv1dec: Fix integer overflow in read_quant_table()

Fixes: runtime error: signed integer overflow: 2147483647 + 1 cannot be represented in type 'int'
Fixes: 3361/clusterfuzz-testcase-minimized-5065842955911168

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

avcodec/svq3: Fix overflow in svq3_add_idct_c()

Fixes: runtime error: signed integer overflow: 2147392585 + 524288 cannot be represented in type 'int'
Fixes: 3348/clusterfuzz-testcase-minimized-4809500517203968

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

avcodec/mips: Remove generic func use in hevc non-uni copy mc msa functions

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mips: preload data in hevc sao edge 90 degree filter msa functions

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/thumbnail_cuda: add cuda thumbnail filter

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

Fix several typos

"apix_fmts" found by Marc Péchaud.
"speedloss" found by Mikhail V.

Signed-off-by: Lou Logan <lou@lrcd.com>

Remove some unneeded casts of bit_rate.

avformat/hlsenc: support http method for hls fmp4 init file

Signed-off-by: Steven Liu <lq@onvideo.cn>

MAINTAINERS: modify the hlsenc description

change the hlsenc from hls encryption to hlsenc

Suggested-by: Aman Gupta <aman@tmm1.net>
Reviewed-by: Lou Logan <lou@lrcd.com>
Signed-off-by: Steven Liu <lq@onvideo.cn>

avformat/hlsenc: fix CID 1418106

fix the "Uninitialized scalar variable (UNINIT)" problem.

Signed-off-by: Steven Liu <lq@onvideo.cn>

pixdesc: Add API to map color property names to enum values

Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>

avcodec/mips: Reduced conditional cases in avc inter lpf msa functions

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mips: Unrolled loops and expanded functions in avc put mc 10 & 30 msa functions

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

GnuTLS: eat PREMATURE_TERMINATION error

GnuTLS is too strict on the SSL shutdown alert, and it's neither
mandatory in the spec or critical. As it's ignored in OpenSSL, we
should also suppress it in GnuTLS as well.

Ticket: #6667

Reviewed-by: wm4 <nfxjfg@googlemail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/cafdec: reject multichannel Opus streams

Multichannel Opus streams require channel mapping information in extradata.

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

lavf/caf: Support demuxing Opus.

Introduced in macOS High Sierra and iOS 11.

avcodec/mips: preload data in hevc sao edge 0 degree filter msa functions

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mips: Fixed rnd_val variable to 6 in hevc uni mc msa functions

Signed-off-by: Kaustubh Raste <kaustubh.raste@imgtec.com>
Reviewed-by: Manojkumar Bhosale <Manojkumar.Bhosale@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

MAINTAINERS: add myself for bwdif and (t)interlace

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

avcodec/pngdec: Clean up on av_frame_ref() failure

Fixes: memleak
Fixes: 3203/clusterfuzz-testcase-minimized-4514553595428864

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

lavf/mpegts: Consider stream_type 0x0f just a hint towards AAC.

It is also used in the wild to signal latm.

Fixes ticket #6657.

lavfi/stereo3d: Set SAR for every output frame.

Fixes ticket #6672.

lavf/utils: Do not force chapter end time before chapter start.

Fixes ticket #6671.

lavd/kmsgrab: Remove the mapping for AV_PIX_FMT_RGB8.

The definitions are different: 3:3:2 vs 2:3:3.

avutil/hwcontext_dxva2: return an error when buffer allocation fails

This also prevents the use of an uninitialized variable.

Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>

kmsgrab: Add more DRM plane formats

hwcontext_vaapi: Fix DRM format mapping

kmsgrab: Remove multiple-plane formats

The planes are unlikely to be contiguous, assuming they are results in
very broken output.  (Tested with NV12/NV16 on Rockchip.)

kmsgrab: Fix DRM format definitions

All DRM formats are defined in terms of little-endian words, so RGB formats
like XRGB actually have the elements in the opposite order order in memory
to the order they are in the name.

This does not affect YUYV and similar YUV 4:2:2 formats, which are in the
expected order.

vaapi: Disable deprecation warnings around use of old struct vaapi_context

vp9: move VP9SharedContext back to the top of VP9Context

VP9SharedContext needs to be the first member so its properties can be
safely accessed from hardware accelerators, without the need to share
the full VP9Context.

Fixes ticket #6674.

x86/exrdsp: optimize ff_reorder_pixels_avx2()

Tested with "checkasm --test=exrdsp -bench"

Before:
reorder_pixels_c: 5187.8
reorder_pixels_sse2: 377.0
reorder_pixels_avx2: 331.3

After:
reorder_pixels_c: 5181.5
reorder_pixels_sse2: 377.0
reorder_pixels_avx2: 313.8

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

avformat/mxfenc: cosmetic changes

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

fate: add mxf_dv25/dvcpro50 regression tests

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

avformat/mxfenc: fix aspect ratio when writing 16:9 DV frames

Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>

avfilter/tinterlace: Simplify checks for lowpass filtering flags

fate: add tinterlace lowpass filtering tests

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

lavc/frame_thread_encoder: Do not mix variable declaration and code.

Fixes a warning:
ISO C90 forbids mixed declarations and code