Merge commit 'eb800f120d2d42590fde2302fe828c847e41e773'

* commit 'eb800f120d2d42590fde2302fe828c847e41e773':
  libavcodec: set AVFrame colorspace fields on decoding

Conflicts:
libavcodec/utils.c

See: a80e622924c89df69fb1c225ba432fe12fe6648e
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '8c02adc62d71dfbb079a04753d8c16152c49de88'

* commit '8c02adc62d71dfbb079a04753d8c16152c49de88':
  lavu: add all color-related enums to AVFrame

Conflicts:
libavcodec/avcodec.h
libavutil/frame.c
libavutil/frame.h
libavutil/version.h

The version check is changed so they are available with the current ABI
FFmpeg libs should have no problems with added fields, nor should any
application using the libs, and we regularly added fields in the past.
We also moved 2 of these fields to AVFrame already previously without issues.

See: a80e622924c89df69fb1c225ba432fe12fe6648e
Merged-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/cljr: require (v)strict -1 to use width %4 != 0

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

Merge commit 'd2ef708c95ace2518deffe830a9c439aeb9edd5d'

* commit 'd2ef708c95ace2518deffe830a9c439aeb9edd5d':
  matroskaenc: Allow VP9 and Opus in webm

Conflicts:
libavformat/matroskaenc.c

See: 820ffaed0f64b7ec27275d093c6f81e181c86e94
Merged-by: Michael Niedermayer <michaelni@gmx.at>

libavcodec: set AVFrame colorspace fields on decoding

Signed-off-by: Anton Khirnov <anton@khirnov.net>

lavu: add all color-related enums to AVFrame

Signed-off-by: Anton Khirnov <anton@khirnov.net>

matroskaenc: Allow VP9 and Opus in webm

Bug-Id: 695

Signed-off-by: Anton Khirnov <anton@khirnov.net>

avcodec/mjpegenc: disallow encoding amv with height%16!=0

I do not know on which side to place the padding to encode with 16x16 MBs
If someone knows or has a known to be correct sample, please contact me

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

avcodec/mjpegdec: request a AMV sample with non mod 16 height

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

x86/motion_est: enable sad16_sse2 on k10 CPUs

The check is meant for k8 CPUs. sad16_sse2 is ~20% faster than sad16_mmxext on k10.

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/cljr: support width%4 encoding

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

avcodec/cljr: remove coded_frame code

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

lavc/dsputil: add missing vsad8 and vsse8 functions

the mpeg encoder would try to use them if vsad or vsse were selected for frame_skip_cmp
and frame_skip_threshold or frame_skip_factor were set to values != 0

example: "ffmpeg -i INPUT -c:v mpeg2video -skipcmp vsad -skip_threshold 1 -f null -"

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge remote-tracking branch 'cehoyos/master'

* cehoyos/master:
  Read mov files where the moov atom is hidden within a free atom.

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

avcodec/asvenc: fix encoding dimensions %16 != 0

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

avutil/frame: frame_copy_video: support copying from a smaller to a larger frame

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

avcodec/snow: fix 1 pixel block sizes

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

Read mov files where the moov atom is hidden within a free atom.

Fixes ticket #1551 / issue 2226.

Fix compilation with --disable-everything --enable-demuxer=matroska.

Fixes ticket #3693.

lavc/jpeglsenc: Enable frame-threading.

lavc/lclenc: Enable frame-threading.

lavc/ljpegenc: Enable frame-threading.

lavc/tiffenc: Enable frame-threading.

lavc/utvideoenc: Enable frame-threading.

Use rc_max_rate in av_find_best_stream() if bitrate == 0.

Fixes ticket #2042.

libavformat: output cues for each subtitle block in MKV muxer

Signed-off-by: John Peebles <johnpeeb@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Revert "avformat/movenchint: Fix movhinting with ff_rtp_chain_mux_open() from abb810db036628e11a5171134ebe320b187ee6d6"

This has become unneeded due to f797b134cad4d248b1c8955659997980d0668bc3

This reverts commit 1fe40e1b0538a8eb8425123c47775b5d141ae084.

Merge commit 'f797b134cad4d248b1c8955659997980d0668bc3'

* commit 'f797b134cad4d248b1c8955659997980d0668bc3':
  rtpenc_chain: Don't copy the time base to the source stream by default

See: 1fe40e1b0538a8eb8425123c47775b5d141ae084
Merged-by: Michael Niedermayer <michaelni@gmx.at>

rtpenc_chain: Don't copy the time base to the source stream by default

Only copy it manually in the muxers where it makes sense (rtspenc,
sapenc). Don't touch the original AVStream in movenchint, where
the original AVStream should be kept untouched.

This fixes the normal tracks in RTP hinted files after
abb810db - the hint tracks were ok while the normal media tracks
were broken, noticed by Michael Niedermayer.

This reverts abb810db but achieves the same effect for the other
muxers.

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

huffyuv: increase VLC_BITS to 12

May require a different tuning for arm chips?

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

avformat/flvenc: fix assertion failure after 4h muxing

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

build: fix compilation of svq1enc_mmx.c with --disable-mmx

It's needed for ff_svq1enc_init_x86() even if simd functions are disabled.

Alternatively, svq1enc_init.c could be made and the relevant code moved there.

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

x86/huffyuvdsp: fix some prototypes

Remove duplicate prototypes and fix int -> intptr_t in another

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'd6d767d93e532246cacf3567e6bcad76a821f838'

* commit 'd6d767d93e532246cacf3567e6bcad76a821f838':
  ppc: Fix compilation for ppc64le (ELFv2)

Conflicts:
libavcodec/ppc/asm.S

See: 41b928c5fad7d4ed8477b68cb0c2d54a70b0afac
Merged-by: Michael Niedermayer <michaelni@gmx.at>

avformat/movenchint: Fix movhinting with ff_rtp_chain_mux_open() from abb810db036628e11a5171134ebe320b187ee6d6

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

Merge commit 'abb810db036628e11a5171134ebe320b187ee6d6'

* commit 'abb810db036628e11a5171134ebe320b187ee6d6':
  Revert "rtpenc_chain: Don't copy the time_base back to the caller"

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

ppc: Fix compilation for ppc64le (ELFv2)

CC: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>

Revert "rtpenc_chain: Don't copy the time_base back to the caller"

While it strictly isn't necessary to copy the time base (since
any use of it is scaled in ff_write_chained), it still is better
to signal the actual time base to the caller, avoiding one
unnecessary rescaling. This also lets the caller know what the
actual internal time base is, in case that is useful info
for some caller.

This reverts commit 397ffde115f4e0482a007b672f40457596cebfc4.

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

x86: huffyuvdsp: fewer functions for x86_64

When there are 2 functions that are <= SSE2, only one is needed for x86_64.

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

avcodec/options: leave priv_data in avcodec_copy_context() intact

previously it was freed but the pointer not cleared

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

avcodec/proresenc_anatoliy: simplify get()

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

avcodec/proresenc_anatoliy: load 4 samples at a time in get()

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

avcodec/proresenc_anatoliy: Use fdct from DSPContext instead of direct call.

Based-on: a55546f48d55e3d1155840541b2be5f4f8cf18ab by Diego Biurrun
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/proresenc_anatoliy: Set bits_per_raw_sample

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

avformat/nutenc: Suggest genpts if AV_NOPTS_VALUE

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

Fix VDPAU decoders, regression since f2f99f07.

x86: dsputilenc: convert ff_sse{8, 16}_mmx() to yasm

Signed-off-by: Timothy Gu <timothygu99@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

x86: dsputilenc: move all the function prototypes together

Signed-off-by: Timothy Gu <timothygu99@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

tests: Add MPEG4 nsse test

Signed-off-by: Timothy Gu <timothygu99@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

x86: huffyuvdsp: add_hfyu_left_pred_bgr32

          C   MMX   SSE2
Cycles: 3092  1053  578

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

avcodec/huffyuvdsp: Change w to intptr in add_hfyu_median_pred() and add_hfyu_left_pred()

This avoids potential issues with the high 32bits being random in x86-64 asm

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

x86: huffyuvdsp: add SSE2 median prediction

From 5010c to 4566 on lagarith YUY2.

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

avformat/matroskadec: fix off by 1 error in matroska_read_seek()

Fixes out of array read

Fixes: vp9-opus-crash.webm
Found-by: Dale Curtis <dalecurtis@google.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/indeo4: Decode both parts of IP frames

Fixes part of Ticket845

av_frame_move_ref() idea by Anton Khirnov

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

avcodec/indeo4: Move frametypes into common header.

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

Do not overwrite the sample format with AV_SAMPLE_FMT_NONE in libspeex_decode_init().

Fixes a regression since ef48ac65 when decoding speex in f4v, reported
by irc user massdos.

Improve frame size calculation in caf muxer to fix adpcm_ms remuxing.

Fixes ticket #3645.

avcodec/qpeldsp: Restore author attribution

See: 368f50359eb328b0b9d67451f56fda20b3255f9a
See:
similarity index 51%
copy from libavcodec/dsputil.c
copy to libavcodec/qpeldsp.c
index 6d41dfc..0beeab8 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/qpeldsp.c
@@ -1,9 +1,5 @@
 /*
- * DSP utils
- * Copyright (c) 2000, 2001 Fabrice Bellard
- * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
- *
- * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>
+ * quarterpel DSP functions
  *
  * This file is part of FFmpeg.

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

avcodec/x86/qpeldsp_init: Restore author attribution

See: 368f50359eb328b0b9d67451f56fda20b3255f9a
See: 44eb49512888143905860af2de2932ab002cdbf7, and many others
See:
similarity index 83%
copy from libavcodec/x86/dsputil_init.c
copy to libavcodec/x86/qpeldsp_init.c
index ebbf97f..8f296a1 100644
--- a/libavcodec/x86/dsputil_init.c
+++ b/libavcodec/x86/qpeldsp_init.c
@@ -1,6 +1,5 @@
 /*
- * Copyright (c) 2000, 2001 Fabrice Bellard
- * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
+ * quarterpel DSP functions
  *
  * This file is part of FFmpeg.
  *

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

qpel_template: Restore author attribution

See: 368f50359eb328b0b9d67451f56fda20b3255f9a
See: 59fe111e8122acc614ace5618e1a4ba7416e9875
See:
similarity index 98%
rename from libavcodec/dsputil_template.c
rename to libavcodec/qpel_template.c
index 95d966e..500c75b 100644
--- a/libavcodec/dsputil_template.c
+++ b/libavcodec/qpel_template.c
@@ -1,9 +1,5 @@
 /*
- * DSP utils
- * Copyright (c) 2000, 2001 Fabrice Bellard
- * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
- *
- * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>
+ * quarterpel DSP function templates
  *
  * This file is part of FFmpeg.
  *

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

Merge commit '368f50359eb328b0b9d67451f56fda20b3255f9a'

* commit '368f50359eb328b0b9d67451f56fda20b3255f9a':
  dsputil: Split off quarterpel bits into their own context

Conflicts:
configure
libavcodec/dsputil.c
libavcodec/h263dec.c
libavcodec/mpegvideo.c
libavcodec/mpegvideo_enc.c
libavcodec/vc1dec.c
libavcodec/vc1dsp.c
libavcodec/x86/dsputil_init.c
libavcodec/x86/qpeldsp.asm

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

Merge commit 'cb52a17cb62297802b0a77e5af9850d87c3f925b'

* commit 'cb52a17cb62297802b0a77e5af9850d87c3f925b':
  dsputil: Move Bink-specific add_pixels8 to binkdsp

Conflicts:
libavcodec/dsputil.c

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

Merge commit '054013a0fc6f2b52c60cee3e051be8cc7f82cef3'

* commit '054013a0fc6f2b52c60cee3e051be8cc7f82cef3':
  dsputil: Move APE-specific bits into apedsp

Conflicts:
libavcodec/arm/int_neon.S
libavcodec/x86/dsputil.asm

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

Merge commit '256da0770e495176d1b2699ec6e9c7993c2a6d7b'

* commit '256da0770e495176d1b2699ec6e9c7993c2a6d7b':
  dsputil: Move mspel_pixels_tab to the only place it is used

Conflicts:
libavcodec/dsputil.c

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

Merge commit '3832a6590156e6abc575bed51cbffa78617ee5fd'

* commit '3832a6590156e6abc575bed51cbffa78617ee5fd':
  dsputil: Move ff_alternate_*_scan tables to mpegvideo

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

avcodec/x86/svq1enc_mmx: Add author attribution

See: 5900637219ccccdd39ddafa4e7181da20b8e1f1b
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '65d5d5865845f057cc6530a8d0f34db952d9009c'

* commit '65d5d5865845f057cc6530a8d0f34db952d9009c':
  dsputil: Move SVQ1 encoding specific bits into svq1enc

Conflicts:
libavcodec/x86/Makefile

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

Merge commit '2ea2612df508abdd1f97c6a6ef56275a52c5c41e'

* commit '2ea2612df508abdd1f97c6a6ef56275a52c5c41e':
  svq1enc: Rename SVQ1Context to SVQ1EncContext

Conflicts:
libavcodec/svq1enc.c

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

ffmpeg: fix check for muxing overhead being unknown

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

x86/dsputilenc: add missing guards to ff_pix_sum16_xop

XOP support was added in Yasm 1.0.0 and Nasm 2.06, and we still
support older versions.

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

x86: huffyuvdsp: port add_bytes to yasm

          C   MMX  SSE2
Cycles: 2972  587  302

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

x86: hpeldsp: better factorization

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

tests: allow passing dimensions to videogen

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

rename add_hfyu_left_prediction_int16 to add_hfyu_left_pred_int16

This makes the naming more consistent with the 8bit variant

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

rename add_hfyu_median_prediction_int16 to add_hfyu_median_pred_int16

This makes the naming more consistent with the 8bit variant

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

rename sub_hfyu_median_prediction_int16 to sub_hfyu_median_pred_int16

This makes the naming more consistent with the 8bit variant

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

x86/dsputilenc: implement XOP version of pix_sum16

SSE2: 137 cycles
XOP:   87 cycles
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '08cd92144e73195eecc28ed0348e66e255516b82'

* commit '08cd92144e73195eecc28ed0348e66e255516b82':
  aarch64: Use the correct syntax for relocations

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

avcodec/dpx_parser: Allow frame size to be too small

Fixes Ticket3402

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

avcodec/dpx_parser: fix flushing end out

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

avcodec/dpx_parser: reset index when finding a startcode, not after

This is simpler

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

dsputil: Split off quarterpel bits into their own context

dsputil: Move Bink-specific add_pixels8 to binkdsp

dsputil: Move APE-specific bits into apedsp

dsputil: Move mspel_pixels_tab to the only place it is used

dsputil: Move ff_alternate_*_scan tables to mpegvideo

dsputil: Move SVQ1 encoding specific bits into svq1enc

svq1enc: Rename SVQ1Context to SVQ1EncContext

This allows making it visible without name clashes.

huffyuv: change left prediction access in BGRA

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

huffyuv: avoid duplicated defines

Move the defines to the dsp header.

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

aarch64: Use the correct syntax for relocations

This fixes building in PIC mode with gas. The examples in the gas
manual showed using a # here even though gas itself actually didn't
support that syntax (and the gas test suite only tests it without
the extra hash sign).

CC: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>

Merge commit '962d63157322466a9a82f9f9d84c1b6f1b582f65'

* commit '962d63157322466a9a82f9f9d84c1b6f1b582f65':
  matroskaenc: set the stream timebase earlier

Conflicts:
tests/ref/lavf/mkv
tests/ref/seek/lavf-mkv

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

Merge commit '43e7f0797f9f821a3866a20f05e512e13c82076a'

* commit '43e7f0797f9f821a3866a20f05e512e13c82076a':
  flvenc: only write the framerate tag based on avg_frame_rate

Conflicts:
tests/ref/lavf/flv_fmt
tests/ref/seek/lavf-flv_fmt
tests/ref/seek/vsynth2-flv
tests/ref/vsynth/vsynth1-flashsv
tests/ref/vsynth/vsynth1-flv
tests/ref/vsynth/vsynth2-flashsv
tests/ref/vsynth/vsynth2-flv

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

Merge commit 'cf6977712c9e5abe6dc55289f6322ccbf10321a9'

* commit 'cf6977712c9e5abe6dc55289f6322ccbf10321a9':
  movenc: write avg_frame_rate as the framerate, not the codec timebase

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

Merge commit '81eec081afea9fc017a175581ceea7c420a0dfc3'

* commit '81eec081afea9fc017a175581ceea7c420a0dfc3':
  matroskaenc: base DefaultDuration on the framerate, not the codec timebase

Conflicts:
libavformat/matroskaenc.c
tests/ref/lavf/mkv
tests/ref/seek/lavf-mkv

See: ea83b032af825d10f7d6db8a71933edce5b842a3
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'a53551cba86bb67efcb6105fdc337a36c43132bd'

* commit 'a53551cba86bb67efcb6105fdc337a36c43132bd':
  frame: fix the error path in av_frame_copy_props()

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

matroskaenc: set the stream timebase earlier

Fixes calculating the ts offset for audio codecs with delay.

flvenc: only write the framerate tag based on avg_frame_rate

Do not fall back on the codec timebase, since that can be anything for
VFR video.

movenc: write avg_frame_rate as the framerate, not the codec timebase

matroskaenc: base DefaultDuration on the framerate, not the codec timebase

This results in DefaultDuration not being written when the framerate is
not known, but as this field is purely informative, this should not
break any sane demuxers.