Merge commit '8267f9363532b23b3b8b4e27491ee2d412a8aec3'

* commit '8267f9363532b23b3b8b4e27491ee2d412a8aec3':
  configure: Set default HOSTCFLAGS/HOSTCPPFLAGS after compiler detection

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

avcodec/cinepakenc: fixes and improvements

Fixed/improved multistrip adaptivity.
Improved R/D by optimized codebook generation.
Support for the correct color space.
Vintage compatibility.

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

Adding Cinepak encoder

With permission of Tomas Härdin applied by Rl aetey.se

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

configure: Set default HOSTCFLAGS/HOSTCPPFLAGS after compiler detection

This prevents the default HOSTCFLAGS from getting clobbered by flags
passed to configure on the command line.

Merge remote-tracking branch 'lukaszmluki/master'

* lukaszmluki/master:
  lavf/ftp: remove double space
  lavf/libssh: fix seek with whence==SEEK_CUR
  lavf/libssh: rename context variable from s into libssh
  lavf/libssh: add av_cold attributes
  lavf/libssh: factorize create_sftp_session function
  lavf/libssh: factorize create_ssh_session function
  lavf/libssh: factorize file_stat function
  lavf/libssh: factorize open_file function
  lavf/libssh: add private_key option

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

lavf/ftp: remove double space

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

lavf/libssh: fix seek with whence==SEEK_CUR

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

lavf/libssh: rename context variable from s into libssh

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

lavf/libssh: add av_cold attributes

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

lavf/libssh: factorize create_sftp_session function

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

lavf/libssh: factorize create_ssh_session function

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

lavf/libssh: factorize file_stat function

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

lavf/libssh: factorize open_file function

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

lavf/libssh: add private_key option

Allows to specify private key to use during authorization.

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>

avformat/mov: redesign the fps calculation

This uses STTS and TRUN information to calculate the fps
It seems to work with more files than the previous code

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

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  rgb2rgb_template: add MMX/SSE2/AVX-optimized deinterleaveBytes

Conflicts:
libswscale/x86/rgb2rgb_template.c

See: 3033cd7555921242cbdc06d0d8aee9f8925d6a3c
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '7597e6efe492cb2449bb771054d64cc7fdf62ff5'

* commit '7597e6efe492cb2449bb771054d64cc7fdf62ff5':
  swscale/x86/rgb2rgb: add support for AVX

Conflicts:
libswscale/x86/rgb2rgb_template.c

See: 4729b529e60fb99356a1d9e13793835af50b3e87
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '9047491f8bcd87673eed55fb310647a03b0981e9'

* commit '9047491f8bcd87673eed55fb310647a03b0981e9':
  swscale: add nv12/nv21->yuv420 converter

Conflicts:
libswscale/rgb2rgb.c
libswscale/rgb2rgb_template.c

See: ef627bf9ecddfa888d4c6e4fd633852798c3b4a9
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'eb01a25fe1452740a7f3ae2cbaff356a5c6e7806'

* commit 'eb01a25fe1452740a7f3ae2cbaff356a5c6e7806':
  swscale: fix stride used in planarToNv12Wrapper()

Conflicts:
libswscale/swscale_unscaled.c

See: 94d7ca2b58af13c379d90aad595e82448743abc7
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'e0ab5078a7d865f8f6fd6a6d3cbe0f380ead4a3d'

* commit 'e0ab5078a7d865f8f6fd6a6d3cbe0f380ead4a3d':
  lavc: do not force the emu edge flag

Conflicts:
libavcodec/utils.c

There should be no such bugs in ffmpeg, but merging it anyway
as its safer.

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

Merge commit '025fd76e1a2623c858d8c686a73cc30980a314b0'

* commit '025fd76e1a2623c858d8c686a73cc30980a314b0':
  dxtory: change error code for unexpected slice configuration

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

Merge commit 'e91a3f1bdba9b4945e42c191d2e35e9844625fb4'

* commit 'e91a3f1bdba9b4945e42c191d2e35e9844625fb4':
  dxtory: correctly handle YUV slices with average odd height

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

Merge commit 'ade4ecb42d2dacd18d04eb8df2afa8131e5ad653'

* commit 'ade4ecb42d2dacd18d04eb8df2afa8131e5ad653':
  dxva2: Use correct printf format strings

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

Merge commit '766df7ca89a2398e71182f5f2b46053e3aa9bd69'

* commit '766df7ca89a2398e71182f5f2b46053e3aa9bd69':
  dxva2: Add missing #includes

See: 9056d0c94af5c09478acf0e1bc129a9c539070f0
Merged-by: Michael Niedermayer <michaelni@gmx.at>

lavf/segment: drop pointless variable oc from seg_write_packet()

lavf/segment: remove duplicated and inconsistent cleanup code in seg_write_packet()

In particular, avoid to leave around the seg->avf pointer to freed
structure, and fix crash with:
ffmpeg -f lavfi -i testsrc -c:v h264 -map 0 -f segment foo-%d.ts

lavf/mpegtsenc: fix weird indent

examples/filtering_audio,video: drop call to avcodec_get_frame_defaults()

The avcodec_get_frame_defaults() function is deprecated and its use
doesn't seem required.

avformat/mov: Ignore the last frame for duration and fps calculation if it looks suspect

The used heuristic will potentially need to be finetuned
Fixes daemon404s fps.mov

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

lossless_videodsp.asm: fix compilation.

Fixes these errors with nasm:
libavcodec/x86/lossless_videodsp.asm:86: error: invalid combination of opcode and operands
libavcodec/x86/lossless_videodsp.asm:88: error: invalid combination of opcode and operands
I don't know whether movd or movq was meant, but either way
maskq vs. maskd must match the mov size.

Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>

rgb2rgb_template: add MMX/SSE2/AVX-optimized deinterleaveBytes

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

swscale/x86/rgb2rgb: add support for AVX

This does not yet include any actual AVX code

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

swscale: add nv12/nv21->yuv420 converter

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

swscale: fix stride used in planarToNv12Wrapper()

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

lavc: do not force the emu edge flag

The default get_buffer2() implementation (and possibly some
user ones) does not allocate edges when this flag is set, which may
expose bugs in some decoders. Until the 10 release is out, it is safer
to remove this part.

dxtory: change error code for unexpected slice configuration

dxtory: correctly handle YUV slices with average odd height

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  dxva2: Log errors verbosely

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

Merge commit '5b2b23f2d69e05c5fcd1c933e383fe60e185574d'

* commit '5b2b23f2d69e05c5fcd1c933e383fe60e185574d':
  dxva2: Retry IDirectXVideoDecoder_BeginFrame()

This also adds a missig #include to prevent breaking building

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

Merge commit 'e22ebd04bcab7f86548794556c28ecca46d9c2ac'

* commit 'e22ebd04bcab7f86548794556c28ecca46d9c2ac':
  hevc: Bound check cu_qp_delta

Conflicts:
libavcodec/hevc.c

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

Merge commit '0d999333f96a34903448579bf13a3209deaee9da'

* commit '0d999333f96a34903448579bf13a3209deaee9da':
  hevc: Bound check slice_qp

Conflicts:
libavcodec/hevc.c

See: aead772b5814142b0e530804486ff7970ecd9eef
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '838740e6420538ad45982da6b1d3aa3ae91307f5'

* commit '838740e6420538ad45982da6b1d3aa3ae91307f5':
  hevc: Prevent some integer overflows

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

Merge commit 'faf03ecba03155bb1f5416713bd01da043863b43'

* commit 'faf03ecba03155bb1f5416713bd01da043863b43':
  hevc: Remove useless clip

See: b00a8b4d194f1bf23343f3f42138affa1fe26641
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'a246d06fe0dc6c2ea65e95327624b4537ff9bd0d'

* commit 'a246d06fe0dc6c2ea65e95327624b4537ff9bd0d':
  hevc: clip pixels when transquant bypass is used

See: c9fe0caf7a1abde7ca0b1a359f551103064867b1
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'b37e796082b2d787aff3cd5631bb89c4fd374708'

* commit 'b37e796082b2d787aff3cd5631bb89c4fd374708':
  hevc: Use uint64 to check for tile dimensions

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

Merge commit 'd5c15ebeaf1914ea5e3e0599d4316d7c4cf74434'

* commit 'd5c15ebeaf1914ea5e3e0599d4316d7c4cf74434':
  hevc: Fix modulo operations

Conflicts:
libavcodec/hevc.c
libavcodec/hevc_filter.c

See: 4ced5d7780fea2ea49444d6686d26f26b3a2160f
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'a7a07cc98ac548297b5b0628cb81280e11952e3f'

* commit 'a7a07cc98ac548297b5b0628cb81280e11952e3f':
  hevc: check that VPS referenced from SPS exists

Conflicts:
libavcodec/hevc_ps.c

See: d66bab0a69ac1860e78dd951ad8db1a507e75642
Merged-by: Michael Niedermayer <michaelni@gmx.at>

dxva2: Use correct printf format strings

dxva2: Add missing #includes

dxva2: Log errors verbosely

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

dxva2: Retry IDirectXVideoDecoder_BeginFrame()

If the function returns E_PENDING retry for a fixed number of times.

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

hevc: Bound check cu_qp_delta

The T-REC-H.265-2013044 page 91 states it has to be in the range
[-(26 + s->sps->qp_bd_offset / 2), (25 + s->sps->qp_bd_offset / 2)].

hevc: Bound check slice_qp

The T-REC-H.265-2013044 page 79 states it has to be in the range
[-s->sps->qp_bd_offset, 51].

Sample-Id: 00001386-google
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org

hevc: Prevent some integer overflows

get_ue_golomb_long() returns an unsigned.

Sample-Id: 00001541-google
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org

hevc: Remove useless clip

The src buffer should only contain values in the interval
[0, (1 << BIT_DEPTH) - 1].

Since shift = (BIT_DEPTH - 5), src[x] >> shift must be in
the interval [0, 31], so no clip is needed.

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

hevc: clip pixels when transquant bypass is used

The reconstructed picture should always be clipped (see section 8.6.5),
previously we did not clip coding units where
cu_transquant_bypass_flag == 1.

Sample-Id: 00001325-google
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

hevc: Use uint64 to check for tile dimensions

And use unsigned datatypes.

Otherwise it would overflow.

Sample-Id: 00001315-google
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org

hevc: Fix modulo operations

Keep qp fields within the range.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

hevc: check that VPS referenced from SPS exists

This matches how its done for SPS/PPS.
Fixes null pointer dereference.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

examples/muxing: reduce duration, remove wrong and misleading comment

Set duration to 10 seconds, after it was increased from 5 to 200 seconds
in 8d80f3cb877b890889b1673029387229648901ed.

200 seconds will generate too much data which is annoying especially when
testing.

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  vc1: Always reset numref when parsing a new frame header.

Conflicts:
libavcodec/vc1.c

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

Merge commit '89564be444d24f75ea5add8b6987e414cf7aa7d5'

* commit '89564be444d24f75ea5add8b6987e414cf7aa7d5':
  rtmpproto: Send a full, absolute timestamp if it isn't monotonically growing

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

Merge commit '24eb3c791606fe98a1591c13a8b2ba6c342bb3b5'

* commit '24eb3c791606fe98a1591c13a8b2ba6c342bb3b5':
  rtmpproto: Avoid using uninitialized memory

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

Merge commit 'd01e684186bc1631bc176f06b89d33c27ec0d24d'

* commit 'd01e684186bc1631bc176f06b89d33c27ec0d24d':
  mov: do not set avg_frame_rate in the demuxer

Conflicts:
libavformat/mov.c

Not merged
The default fallback code that sets avg_frame_rate guesses the value
from a very small number of frames at the videos start and is wrong in
some cases for the file as a whole.
Now that makes it guess vs guess, but the code in mov is much faster

Also ideally the mov code should be fixed and give a exact avg_frame_rate
based on the actual cts from the index (CTTS/STTS)

If someone has sample(s) which show wrong fps values, please open a ticket
on trac or mail me if its private.

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

fate: add a few tests for >8bpc ffvhuff

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

avcodec/lossless_videodsp: fix diff_int16_c on MIPS

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

avcodec/x86/lossless_videodsp: Port lorens add_hfyu_left_prediction_ssse3/sse4 to 16bit

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

avcodec/x86/lossless_videodsp: use SPLATW in add_int16

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

vc1: Always reset numref when parsing a new frame header.

Fixes an issue where the B-frame coding mode switches from interlaced
fields to interlaced frames, causing incorrect decisions in the motion
compensation code and resulting in visual artifacts.

CC: libav-stable@libav.org
Signed-off-by: Tim Walker <tdskywalker@gmail.com>

Move sub_hfyu_median_prediction_int16() to losslessviddsp

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

Move add_hfyu_median_prediction_int16() to losslessviddsp

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

Move add_hfyu_left_prediction_int16 to losslessviddsp

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

avcdoec/huffyuvenc: optimize sub_left_prediction()

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

Move add/diff_int16 to lossless_videodsp

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

rtmpproto: Send a full, absolute timestamp if it isn't monotonically growing

The normal differential timestamps can't handle negative
differences, thus send a full packet header with an absolute
timestamp in these cases.

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

rtmpproto: Avoid using uninitialized memory

If the url ends with .flv, we stripped it but didn't initialize
rt->playpath, doing av_strlcat on an uninitialized buffer.

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

mov: do not set avg_frame_rate in the demuxer

The track duration is often not reliable or is not the duration
represented by the number of frames. In those cases, avg_frame_rate
was reported incorrectly. Removing this code falls back to the
default calculation in avformat_find_stream_info().

This is a partial revert of commit c3aeaa540.

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

Merge remote-tracking branch 'cehoyos/master'

* cehoyos/master:
  Avoid a possible overflow when reading Nikon avi files.
  Add h263dsp dependency to mpeg1video and mpeg2video encoders.
  Fix compilation with --disable-hwaccel=mpeg1_xvmc,mpeg2_xvmc

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

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  lavc: deprecate CODEC_FLAG_EMU_EDGE and avcodec_get_edge_width().

Conflicts:
doc/APIchanges
libavcodec/utils.c
libavcodec/version.h

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

avcodec/jpeg2000dec: fix error detection in pix_fmt_match()

Fixes out of array accesses with CODEC_FLAG_EMU_EDGE

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

avcodec/jpeg2000dec: calculate planar and pixelsize from pixel format descriptor

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

avcodec/vmnc: Check  that rectangles are within the picture

Prevents out of array accesses with CODEC_FLAG_EMU_EDGE

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

avcodec/diracdec: allocate edges internally instead of depending on CODEC_FLAG_EMU_EDGE being not set

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

Merge commit '024db24912a39316b0ef0b7d793307d62da038f4'

* commit '024db24912a39316b0ef0b7d793307d62da038f4':
  mpegvideo: allocate edges when encoding.

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

avcodec/mpegvideo_enc: disable direct mode in load_input_picture() for dimensions%16 != 0

We currently read a whole 16x16 block from the input at a few places

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

Merge commit 'f7e85ee996b3886c2b13e928b83277382311af96'

* commit 'f7e85ee996b3886c2b13e928b83277382311af96':
  lavc: allow the caller to override dimensions in ff_get_buffer()

Conflicts:
libavcodec/utils.c

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

examples/muxing: remove redundant {}

examples/muxing: change error checks, from "ret != 0" to "ret < 0"

More consistent and more future-proof.

examples/muxing: factorize write_interleave code

Also log output packet information.

examples: add remuxing example

Show how to perform streamcopy from one container to another.

lavc: deprecate CODEC_FLAG_EMU_EDGE and avcodec_get_edge_width().

mpegvideo: allocate edges when encoding.

Allocating edges will be removed from generic code, so the encoder must
handle this explicitly.

lavc: allow the caller to override dimensions in ff_get_buffer()

This will be useful for allocating edges in the encoders.

Avoid a possible overflow when reading Nikon avi files.

Suggested-by: Reimar

Add h263dsp dependency to mpeg1video and mpeg2video encoders.

Fixes compilation with:
configure --disable-everything --enable-encoder=mpeg2video --disable-asm

Fix compilation with --disable-hwaccel=mpeg1_xvmc,mpeg2_xvmc

Merge remote-tracking branch 'cehoyos/master'

* cehoyos/master:
  Allow decoding of slightly broken Nikon avi files.

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

avcodec/x86/dsp: add_int16_mmx / add_int16_sse2

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

avcodec/dsputil/huffyuv: move diff int16 and add int16 to dsputil

This also fixes masking the bits

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

Allow decoding of slightly broken Nikon avi files.

Fixes ticket #3330.