avcodec/atrac3plusdec: use av_freep(), do not leave stale pointers in memory

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

avcodec/bfi: use av_freep(), do not leave stale pointers in memory

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

ffserver: use av_freep(), do not leave stale pointers in memory

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

avcodec/vorbisdec: use av_freep(), do not leave stale pointers in memory

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

Merge commit '2d91abade29e43bb45c881d45909b8ee77e904e2'

* commit '2d91abade29e43bb45c881d45909b8ee77e904e2':
  x86: h264_intrapred: Don't treat 32-bit integers as 64-bit

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

Merge commit '82f8eb5f773d87e3d9c0066ef19b501a461ddfae'

* commit '82f8eb5f773d87e3d9c0066ef19b501a461ddfae':
  jpeg2000: split off inverse MCT as Jpeg2000DSP

Conflicts:
libavcodec/Makefile
libavcodec/jpeg2000dec.c
libavcodec/jpeg2000dsp.c
libavcodec/jpeg2000dsp.h

See: c9f2ec8a3464718641742a105179f828ccbfb392
Merged-by: Michael Niedermayer <michaelni@gmx.at>

x86: h264_intrapred: Don't treat 32-bit integers as 64-bit

The upper halves are not guaranteed to be zero in x86-64.

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

jpeg2000: split off inverse MCT as Jpeg2000DSP

This makes the addition of arch optimized functions easier.

Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>

avcodec/h264: Undefined behavior (left shift of 12852653 by 8 places cannot be represented in type 'int')

Fixes: asan_heap-oob_84f75d_8_asan_heap-oob_a2a00a_341_mbc.ts
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

swresample/audioconvert: Fix undefined behavior (left shift of negative value)

Fixes: asan_heap-oob_4da4f3_8_asan_heap-oob_4da4f3_419_scene1a.mm
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/mmvideo: Fix undefined behavior (left shift of negative value)

Fixes: asan_heap-oob_4da4f3_8_asan_heap-oob_4da4f3_419_scene1a.mm
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffprobe: add pixel format chroma sub-sampling output

Adds output of chroma sub-sampling information to
ffprobe -show_pixel_formats option.

Note: This patch just prints the "raw" log2_charoma_w/h values instead
of the previously suggested mapping into a chroma sub-sampling pattern
string like "4:2:2".

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

ffprobe: add pixel format component bit depth output

Adds output of bit depth per pixel format component to
ffprobe -show_pixel_formats option.

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

ffprobe: add pixel format flags output

Adds output of pixel format flags to ffprobe -show_pixel_formats option.

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

ffprobe: add -show_pixel_formats option

Adds option -show_pixel_formats to ffprobe which lists all
available pixel formats with some details.

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

avfilter: add w/h parameters in timeline

Fixes Ticket #4008.

avformat/tiertexseq: use av_freep(), do not leave stale pointers in memory

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

avfilter/af_adelay: use av_freep(), do not leave stale pointers in memory

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

Merge commit '9a03c2323593173a201cb75edd1b49887cf811ed'

* commit '9a03c2323593173a201cb75edd1b49887cf811ed':
  h263dec: Force padding bug workaround for H.263.

This is not merged as it breaks a good part of the error concealment/resilience for H.263
Also, messenger.h263 plays fine in ffmpeg.
If anyone has any other h263 files that do not work, please open an issue on trak or
mail me!

See: d225b0f7aaa65eafccc87165130e1c4bab71708b
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'd208d1eba3799c58fd6d3602d31de3e686f14aec'

* commit 'd208d1eba3799c58fd6d3602d31de3e686f14aec':
  hevc_mvs: make sure to always initialize the temporal MV fully

Conflicts:
libavcodec/hevc_mvs.c

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

Merge commit '79a60c8e779242b5ba4c531b2c706c871e8e6420'

* commit '79a60c8e779242b5ba4c531b2c706c871e8e6420':
  hevc_mvs: initialize the temporal MV in case of missing ref

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

h263dec: Force padding bug workaround for H.263.

Fixes decoding of http://samples.mplayerhq.hu/V-codecs/h263/h263-raw/messenger.h263

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

hevc_mvs: make sure to always initialize the temporal MV fully

The spec requires this.

Fixes uninitialized reads on some samples.

Remove now unnecessary initialization of the whole merge candidate list.

hevc_mvs: initialize the temporal MV in case of missing ref

The caller expects the MV to always be initialized.

avformat/hls: use av_freep(), do not leave stale pointers in memory

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

avcodec/escape124: use av_freep(), do not leave stale pointers in memory

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

avcodec/proresenc_kostya: use av_freep(), do not leave stale pointers in memory

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

avcodec/avpacket: use av_freep(), do not leave stale pointers in memory

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

avcodec/avpacket: simplify freeing pkt->data

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

avformat/mxfdec: read reel_name and source timecode from physical source package

Reviewed-by: Tomas Härdin <tomas.hardin@codemill.se>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '28816050e47b6dba430a52e429d21a864cffda8e'

* commit '28816050e47b6dba430a52e429d21a864cffda8e':
  lavf: Set the stream time base hint properly for chained muxers

Conflicts:
libavformat/segment.c

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

Merge commit 'b76249443864c88ffb2d41ab8d1de7432e985dc7'

* commit 'b76249443864c88ffb2d41ab8d1de7432e985dc7':
  sdp: Simplify parsing/conversion of H264 extradata

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

Merge commit '153fadc390d05aa47e5e2c56290401898fe41a23'

* commit '153fadc390d05aa47e5e2c56290401898fe41a23':
  ff_get_format: fix infinite loop

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

lavf: Set the stream time base hint properly for chained muxers

This avoids warnings about using the codec time base as time
base hint.

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

sdp: Simplify parsing/conversion of H264 extradata

By using ff_avc_write_annexb_extradata instead of the h264_mp4toannexb
BSF, the code for doing the conversion itself is kept much shorter,
there's less state to restore at the end, we don't risk leaving the
AVCodecContext in an inconsistent state if returning early due to
errors, etc.

Also add a missing free if the base64 encoding fails.

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

Merge commit '577899a6458ccad9026eb268f10dc0b39c224c8d'

* commit '577899a6458ccad9026eb268f10dc0b39c224c8d':
  lavc: specify the behavior of av_lockmgr_register on failure.

Conflicts:
libavcodec/avcodec.h
libavcodec/utils.c
libavcodec/version.h

See: a950edb472e8823e34832c7313ba447b2db76f27
Merged-by: Michael Niedermayer <michaelni@gmx.at>

ff_get_format: fix infinite loop

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

lavc: specify the behavior of av_lockmgr_register on failure.

The register function now specifies that the user callback should
leave things in the same state that it found them on failure but
that failure to destroy is ignored by the library.  The register
function is now explicit about its behavior on failure
(it unregisters the previous callback and destroys all mutex).

Signed-off-by: Manfred Georg <mgeorg@google.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>

avcodec/vdpau: do not dereference hwctx before checking it for NULL

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

Merge commit '8de1d67967a9f9e22c66cb0c1e518ae4f30d07dd'

* commit '8de1d67967a9f9e22c66cb0c1e518ae4f30d07dd':
  avconv_vdpau: update to new VDPAU interface

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

avcodec/vdpau: fix render2() check

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

ffmpeg: add vdpau_old to allow continued testing of the older (but not oldest) API

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

Merge commit 'bef067f88c74190cdf7e76d12f02e12e069974aa'

* commit 'bef067f88c74190cdf7e76d12f02e12e069974aa':
  vdpau: check video surface and decoder capabilities

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

Merge commit 'e3e158e81f0666b8fe66be9ce1cad63a535920e0'

* commit 'e3e158e81f0666b8fe66be9ce1cad63a535920e0':
  vdpau: add av_vdpau_bind_context()

Conflicts:
doc/APIchanges
libavcodec/vdpau.h
libavcodec/version.h

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

Merge commit '502cde409ca5ee97ef70c2cdede88b9101746ff6'

* commit '502cde409ca5ee97ef70c2cdede88b9101746ff6':
  vdpau: force reinitialization when output resolution changes

Conflicts:
libavcodec/vdpau.c

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

avcodec/vdpau: warn if the user application has not setup avctx->hwaccel_context instead of potentially crashing

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

Merge commit '89ac99ba5f2dc9f69ad3bc294753930eb0b3e4a4'

* commit '89ac99ba5f2dc9f69ad3bc294753930eb0b3e4a4':
  vdpau: pass codec-specific parameters from hwaccel

Conflicts:
libavcodec/vdpau.c

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

avcodec/vdpau_internal: add comment to #endif

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

Merge commit 'ce083282f0a8b7d63c4047c30b7bac498f9806dd'

* commit 'ce083282f0a8b7d63c4047c30b7bac498f9806dd':
  vdpau: common support for managing the VdpDecoder in avcodec

Conflicts:
libavcodec/vdpau.c

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

Merge commit 'fcc1022611f79c2f3aa2f392a5ce14c74be9c1d7'

* commit 'fcc1022611f79c2f3aa2f392a5ce14c74be9c1d7':
  vdpau: factor out common end-of-frame handling

Conflicts:
libavcodec/vdpau.c
libavcodec/vdpau_h264.c

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

Merge commit '319424d25c53f82b87187ce03ed984d438f1bee6'

* commit '319424d25c53f82b87187ce03ed984d438f1bee6':
  vdpau: add helper for VDPAU to libav error codes conversion

Conflicts:
libavcodec/vdpau.c

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

avconv_vdpau: update to new VDPAU interface

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

vdpau: check video surface and decoder capabilities

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

vdpau: add av_vdpau_bind_context()

This function provides an explicit VDPAU device and VDPAU driver to
libavcodec, so that the application is relieved from codec specifics
and VdpDevice life cycle management.

A stub flags parameter is added for future extension. For instance, it
could be used to ignore codec level capabilities (if someone feels
dangerous).

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

vdpau: force reinitialization when output resolution changes

This is necessary to recreate the decoder with the correct parameters,
as not all codecs invoke get_format() in this case.

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

swresample/swresample: replace always true if() by av_assert0()

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

swresample/swresample: fix sample drop loop end condition

Fixes Ticket3985

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

avformat/mov: do not overwrite extradata in mov_read_glbl()

Fixes regression from Ticket3962

This basically favors the first global header while since 34751f8313a41bf94e004ffa9491f35d96763558
it was the last. If this heuristic turns out worse, do not hesitate to revert this and
reopen 3962 as a feature request for multiple STSD

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

swscale: remove obsolete FF_API_SWS_FORMAT_NAME cruft

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

avformat: remove obsolete FF_API_READ_PACKET cruft

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

avformat: remove obsolete FF_API_CLOSE_INPUT_FILE cruft

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

avformat: remove obsolete FF_API_SET_PTS_INFO cruft

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

avformat: remove obsolete FF_API_NEW_STREAM cruft

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

avformat: remove obsolete FF_API_FORMAT_PARAMETERS cruft

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

avformat: remove obsolete FF_API_ALLOC_OUTPUT_CONTEXT cruft

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

avutil: remove obsolete FF_API_INTFLOAT cruft

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

avutil: remove obsolete FF_API_SAMPLES_UTILS_RETURN_ZERO cruft

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

avutil: remove obsolete FF_API_FIND_OPT cruft

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

avutil: remove obsolete FF_API_GET_BITS_PER_SAMPLE_FMT cruft

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

avcodec: remove obsolete FF_API_FAST_MALLOC cruft

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

avformat/options_table: add FF_COMPLIANCE_UNOFFICIAL

Fixes ticket 3959

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

configure: add -lrt to libavutil when needed for nanosleep

This is similar to 7d97511e13ef30cd99bae0e36400de055b72918d

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

avcodec/h264_mp4toannexb_bsf: reset the new IDR flag when SPS/PPS is seen.

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

avcodec/webp: add assert to ensure palette is not larger than 256

it should not be possible to be larger as its stored as 8bit value
but we would be overwriting a stack buffer if it is ...

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

avcodec/webp: add optimization: use local palette with extra padding

for big enough pictures.

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

vdpau: pass codec-specific parameters from hwaccel

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

vdpau: common support for managing the VdpDecoder in avcodec

Using the not so new init and uninit callbacks, avcodec can now take
care of creating and destroying the VDPAU decoder instance.

The application is still responsible for creating the VDPAU device
and allocating video surfaces - this is necessary to keep video
surfaces on the GPU all the way to the output. But the application
will no longer needs to care about any codec-specific aspects.

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

vdpau: factor out common end-of-frame handling

Also add error handling.

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

vdpau: add helper for VDPAU to libav error codes conversion

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

configure: Set large address aware.

Enables >2GB memory usage for 32 bit compiles on Windows.

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

avformat/img2dec: Attempt to detect non-escaped glob patterns too (-pattern_type glob)

Fixes ticket #3948

Based-on-patch-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Alexander Strasser <eclipse7@gmx.net>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '2a5ac99e6e06078713f684fee2466c91f677b303'

* commit '2a5ac99e6e06078713f684fee2466c91f677b303':
  x11grab: Check the XQueryPointer return value

Conflicts:
libavdevice/x11grab.c

See: 69c34a6ac986e31b5286a1d566617ec25b93e6a7
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '5b9c817dc7577b6d44acc94d73b9c77c52cda489'

* commit '5b9c817dc7577b6d44acc94d73b9c77c52cda489':
  x11grab: Check XFixesGetCursorImage return value

Conflicts:
libavdevice/x11grab.c

See: a65c0a3fe822386be30fd3371af9f0d008b02874

The warning with adjusted text is kept from a65c0a3fe822386be30fd3371af9f0d008b02874
but drawing the cursor is not disabled in case XFixesGetCursorImage() fails

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

Merge commit '89fa2b5616274194d90bde0aeebf6adbd1245193'

* commit '89fa2b5616274194d90bde0aeebf6adbd1245193':
  x11grab: Check for XFixes availability at runtime

See: a65c0a3fe822386be30fd3371af9f0d008b02874
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '68211c70afcfd4d675029cc85fba7dee8de53bdf'

* commit '68211c70afcfd4d675029cc85fba7dee8de53bdf':
  vc1: Do not assume seek happens after decoding

See: 6801eb0a0981f113f5f09ed4799d9ae805af62a3
Merged-by: Michael Niedermayer <michaelni@gmx.at>

avformat/img2dec: Support -loop with pipes

Fixes Ticket3976

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

x11grab: Check the XQueryPointer return value

The paint_mouse_pointer() code uses XFixes to retrieve the cursor
coordinates, but XFixes gives no information about which screen the
pointer is on; this results in always drawing the cursor on the
captured screen even if the mouse pointer was on another screen.

For example, when capturing from screen 1 (i.e. -f x11grab -i ":0.1")
the cursor was being drawn in the captured image even when the mouse
pointer was actually on screen 0, which is wrong and visually confusing.

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

x11grab: Check XFixesGetCursorImage return value

It could return NULL if the cursor is outside the screen, the connection
timed out or the system is out of memory.

CC: libav-stable@libav.org

x11grab: Check for XFixes availability at runtime

vc1: Do not assume seek happens after decoding

If a seek is requested before the decoding start there is no
current picture.

CC: libav-stable@libav.org

avcodec/on2avc: Check number of channels

Fixes out of array access
Fixes: asan_heap-oob_4da4f3_7_asan_heap-oob_4da4f3_173_Xmen_avc_500.vp6
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '36f3aec3630f27df64f4ff2b52a1c9ced760eb52'

* commit '36f3aec3630f27df64f4ff2b52a1c9ced760eb52':
  fate-vc1_ilaced_twomv: use -flags +bitexact

Conflicts:
tests/ref/fate/vc1_ilaced_twomv

See: 233d1acb7095b3dc89975cd4598c294cc4c08887
Merged-by: Michael Niedermayer <michaelni@gmx.at>

x86/hevc: get rid off packusdw for ssse3 compatibility

cherry picked from commit df8ebe304df453f26c28ff8f11d607f49b90a4c2

Fixes out of array access
Fixes: asan_stack-oob_1046454_9_asan_stack-oob_15a9e7c_170_WP_MAIN10_B_Toshiba_3.bit
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/hevc: fix chroma transform_add size

Fixes accessing misaligned address
Fixes: signal_sigsegv_1feb99c_10_signal_sigsegv_2d1d35c_79_cov_691940146_EXT_A_ericsson_3.bit
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

lavc/libx264: Fix compilation with libx264 version < 142.

avcodec/h264: Check mode before considering mixed mode intra prediction

Fixes out of array read
Fixes: asan_heap-oob_e476fc_2_asan_heap-oob_1333ec6_61_CAMACI3_Sony_C.jsv
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avformat/assdec: make sure pos is initialized

Fixes use of uninitialized memory
Fixes: signal_sigsegv_504fb0_10_signal_sigsegv_504fb0_343_mewmew_ssa.avi
Found-by: Mateusz j00ru Jurczyk and Gynvael Coldwind

Merge commit '7d97511e13ef30cd99bae0e36400de055b72918d'

* commit '7d97511e13ef30cd99bae0e36400de055b72918d':
  configure: Add -lrt dependency to avutil's pc file.

Conflicts:
configure

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

fate-vc1_ilaced_twomv: use -flags +bitexact

Also updates the reference since it was generated by the non-bitexact
x86 specific code.

configure: Add -lrt dependency to avutil's pc file.

This is needed for clock_gettime.

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