pulse: set time_base as multiple of sample_rate

The stream time_base has been changed to be a multiple of the
device sample_rate. Now the frame_duration computation is simpler
and a message will warn about problematic frame_size values.

Signed-off-by: Federico Simoncelli <fsimonce@redhat.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avformat/nutdec: remove unused variable

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

v4l2: setting device parameters early

Setting the v4l2 device parameters may influence video properties
such as width, height and pixel format.
This patch moves v4l2_set_parameters to be executed before fetching
the video properties, avoiding errors such as (pal vs ntsc):

 The v4l2 frame is 829440 bytes, but 691200 bytes are expected

Signed-off-by: Federico Simoncelli <fsimonce@redhat.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  configure: Support preprocessor macros as header names

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

configure: Support preprocessor macros as header names

New versions of FreeType have moved the location of their API
header(s) and hide the location behind a macro.

Since the location changes between versions and no other way
to know the location exists, this workaround becomes necessary.

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

avformat/matroskaenc: warn when muxing video codec not supported by format

This occurs for video codecs that have codec_tag set, but are are not listed
in ff_codec_bmp_tags (e.g. AV_CODEC_ID_BINKAUDIO, AV_CODEC_ID_IFF_ILBM).

Fixes ticket #3269.

Signed-off-by: Peter Ross <pross@xvid.org>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  sdp: Check that fmt->oformat is non-null before accessing it

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

Merge commit '9d18a7d3ec09d6d933d648570643fde924aa391a'

* commit '9d18a7d3ec09d6d933d648570643fde924aa391a':
  fate: Update AAC ELD 5.1 ref for recent bugfixes.

also reenable fate-aac-er_ad6000np_44_ep0, as this commit updates the pcm
reference, so that the test works again

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

Merge commit '7f29644108c5fbd80f160930b31b78b8704c1a49'

* commit '7f29644108c5fbd80f160930b31b78b8704c1a49':
  aac: Fix low delay windowing.

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

Merge commit '42d1b41983971da63302ac3d12091cad1f3d6324'

* commit '42d1b41983971da63302ac3d12091cad1f3d6324':
  fate: Add a test vector for AAC ELD with TNS.

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

Merge commit 'b2212dec0f011893ec68eecaa990170fa24050d7'

* commit 'b2212dec0f011893ec68eecaa990170fa24050d7':
  aac: Fix TNS decoding for the 512 sample window family.

also temporarily disable fate-aac-er_ad6000np_44_ep0 as this commit
causes a mismatch with the reference pcm file
The test will be reenabled after all fixes and with a new pcm reference

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

Merge commit 'e2096e2eaa9e75663d6bf0c37d342752aa5a146d'

* commit 'e2096e2eaa9e75663d6bf0c37d342752aa5a146d':
  fate: Add a downsampled SBR testvector

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

avformat/mp3dec: fix start time in light of initial skip samples

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

lavf/file: fix help message first character casing for trunc option

doc/protocols/file: fix semantical reverse

Change "read" to "write" in introductory description.

Spotted-by: Alexander Strasser <eclipse7@gmx.net>

doc/protocols/file: document general file protocol URL syntax

Also drop confusing ff* tools reference about exceptions to the
file:FILENAME syntax, which is not ff* tool specific.

With various edits by Alexander Strasser <eclipse7@gmx.net>.

sdp: Check that fmt->oformat is non-null before accessing it

This avoids crashes when avserver tries to create an SDP, since
d77f4af.

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

fate: Update AAC ELD 5.1 ref for recent bugfixes.

aac: Fix low delay windowing.

AAC LD uses a low overlap sine window instead of a KBD window.

fate: Add a test vector for AAC ELD with TNS.

aac: Fix TNS decoding for the 512 sample window family.

fate: Add a downsampled SBR testvector

lavf/file: file_check: Handle file URLs that start with "file:"

Handle the URL analog to file_open, it may contain a "file:"
prefix. Skip it. Make access checks to file URLs starting
with "file:" work.

Fix part of ticket #3249.

Signed-off-by: Alexander Strasser <eclipse7@gmx.net>

avformat/oggparseopus: fix segmented timestamps

Fix timestamp calculation for code 3 Ogg Opus packets with less than
2 bytes in the last segment (e.g. packet length 255 or 256).
A sample that would seek incorrectly in ffplay can be created with:
  ffmpeg -i in.wav -b:a 34k -vbr off -frame_duration 60 out.opus
and libopus 1.1

Also do not read past the end of the buffer when a packet has length 0.

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

avformat/oggdec: dont read timestamps from EOS pages of ogm videos

Some muxers store invalid timestamps there, which breaks seeking
Fixes Ticket2739

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

avformat/utils: fix rounding error accumulation for generated dts in compute_pkt_fields()

This only updates the case where there is no B frame reordering delay.

Tested-by: Anssi Hannula <anssi.hannula@iki.fi>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avutil/rational: av_add_stable() test code

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

avutil/mathematics: add av_add_stable()

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

Changelog: add nuts side & metadata support

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

avformat/nut: add support for per frame side & meta data with version 4

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

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  arm: Use the matching endfunc macro instead of the assembler directive directly

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

Merge commit '2ad4ee345a4216aef3999f57dd14c56128d27a13'

* commit '2ad4ee345a4216aef3999f57dd14c56128d27a13':
  arm: Add a missing endfunc macro call

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

arm: Use the matching endfunc macro instead of the assembler directive directly

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

arm: Add a missing endfunc macro call

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

avfilter/lut3d: fix channel order in Iridas format parsing.

According to http://doc.iridas.com/index.php?title=LUT_Formats

    for 3D textures, the data is also RGB, and ordered in such a way
    that the red coordinate changes fastest, then the green coordinate,
    and finally, the blue coordinate changes slowest:

Fixes Ticket #3236.

avcodec/rv30: put the rpr check back in init with the max vs bits bug fixed

Its usefull to know immediatly if extradata is wrong and not just when later
the decoder tries to use it.

This check was removed by: a6a2282c25abe43e352010a7c3fbc92994c0bc1c
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/rv30: check rpr before use against maximum

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

avcodec/rv30: cleanup rpr handling

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

avcodec/rv10: cleanup rpr handling

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

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  mpegvideo: remove disabled bfin asm

Conflicts:
libavcodec/bfin/mpegvideo_bfin.c

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

Merge commit 'a4d0c6e0503562d4cc8f9f6d02d84d7b32583b15'

* commit 'a4d0c6e0503562d4cc8f9f6d02d84d7b32583b15':
  mpegvideo: move dct_unquantize functions up to avoid forward declarations

Conflicts:
libavcodec/mpegvideo.c

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

Merge commit 'aec25b1c4650944d32706bfd40eb02bbd5587303'

* commit 'aec25b1c4650944d32706bfd40eb02bbd5587303':
  mpegvideo: split the encoding-only parts of frame_start() into a separate function

Conflicts:
libavcodec/mpegvideo.c
libavcodec/mpegvideo_enc.c

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

Merge commit 'b318106fae65149356934fc72feafef3272fd4ea'

* commit 'b318106fae65149356934fc72feafef3272fd4ea':
  FATE: add a test for the lavr mixing case fixed in fc6a3ef40d34ce8443ae57c2452f3f273d7d4891

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

Merge commit 'a8cc88b1a23dc1515f27cfa98af16a273c539091'

* commit 'a8cc88b1a23dc1515f27cfa98af16a273c539091':
  tests/Makefile: allow FILTER* to be called with lists of filter names

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

Merge commit 'cc976a75dffa148d655b52604331679ff669e8a2'

* commit 'cc976a75dffa148d655b52604331679ff669e8a2':
  audio_mix: print (SKIP) instead of 0.0 for matrix columns removed along with output zeroing

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

Merge commit 'fc6a3ef40d34ce8443ae57c2452f3f273d7d4891'

* commit 'fc6a3ef40d34ce8443ae57c2452f3f273d7d4891':
  audio_mix: fix zeroing output channels in certain cases

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

Merge commit '3d95d27376e59de14f984e7a22a52e066d85df35'

* commit '3d95d27376e59de14f984e7a22a52e066d85df35':
  audio_mix: initialize the data pointers to NULL

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

Merge commit '58312b2472d3a44d7458865c459d59ef2e02bf1a'

* commit '58312b2472d3a44d7458865c459d59ef2e02bf1a':
  h264: reset data_partitioning if decoding the slice header for NAL_DPA fails

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

Merge commit 'e89aa4bf56e5b5c45f569eb12733519789e057da'

* commit 'e89aa4bf56e5b5c45f569eb12733519789e057da':
  lzw: switch to bytestream2

Conflicts:
libavcodec/lzw.c

See: 44c4170c52c10e3da3a7ea8e3435ef37c4edc2cc
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '1b5d065ca722eb8028c7a08e054b6da3419faf5d'

* commit '1b5d065ca722eb8028c7a08e054b6da3419faf5d':
  pmpdec: check that there is at least one audio packet.

Conflicts:
libavformat/pmpdec.c

See: 8b1cd25ca7e64e6128fa2902d78e48bfeeec9786
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '94a417acc05cc5151b473abc0bf51fad26f8c5a0'

* commit '94a417acc05cc5151b473abc0bf51fad26f8c5a0':
  mathematics: remove asserts from av_rescale_rnd()

Conflicts:
libavutil/mathematics.c

The asserts are left in place for now as no code checks the return
value, but we sure can change this if application developers
prefer

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

Merge commit '24057c83207d6ea8bfd824155ac37be8a33dfd0c'

* commit '24057c83207d6ea8bfd824155ac37be8a33dfd0c':
  eacmv: check the framerate before setting it.

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

Merge commit '5569146d48f06564e8fa393424782cceed510916'

* commit '5569146d48f06564e8fa393424782cceed510916':
  adx: check that the offset is not negative

See: 30bce34b6719ca99ad72c62e2fba3eade71f1eae
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'a6a2282c25abe43e352010a7c3fbc92994c0bc1c'

* commit 'a6a2282c25abe43e352010a7c3fbc92994c0bc1c':
  rv30: fix extradata size check.

Conflicts:
libavcodec/rv30.c

See: 09c5f990bc7629dfbee8c760fd485936c60a7b40
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit 'feded990e3ef9af4a0b827d5b6d8fe86f0b94942'

* commit 'feded990e3ef9af4a0b827d5b6d8fe86f0b94942':
  mpegvideo: set reference/pict_type on generated reference frames

Conflicts:
libavcodec/mpegvideo.c

The picture type is left at P type as it was explicitly set before

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

avformat/utils: fix order of buffers in timestamp update code

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

avcodec/mpegvideo: fix ac/dc_val and coded_block table sizes

With interlaced vc1 it was possible that accesses could happen outside these
tables before this.

Regression since 017e234c204f8ffb5f85a073231247881be1ac6f
Reproduced with a sample from Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Fixes (again) : 480i30__codec_WVC1__mode_2__framerate_29.970__type_2__preproc_17.SIGFPE.bfa.390.wmv
No releases are affected by this

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

avutil/mathematics: fix 2 typos in the doxy

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

doc/faq: remove "-profile option fails when encoding H.264 video with AAC audio" entry

The error was fixed in commit ce9bd303794fcc94aa232c116bece13c5994b55b
more than one year ago, and does not affect all the currently supported
versions.

In particular, fix trac issue #1535.

doc/ffserver: mention how to access streams through RTSP

mpegvideo: remove disabled bfin asm

It has been disabled as 'broken' over 3 years ago in b716a792

mpegvideo: move dct_unquantize functions up to avoid forward declarations

mpegvideo: split the encoding-only parts of frame_start() into a separate function

This introduces some code duplication. However, much of it should go
away once the decoders stop using MpegEncContext.

FATE: add a test for the lavr mixing case fixed in fc6a3ef40d34ce8443ae57c2452f3f273d7d4891

tests/Makefile: allow FILTER* to be called with lists of filter names

audio_mix: print (SKIP) instead of 0.0 for matrix columns removed along with output zeroing

audio_mix: fix zeroing output channels in certain cases

Specifically, when the corresponding input channel exists and its matrix
column is all-zero (which is necessary for zeroing the output), the
matrix column must be removed from the matrix.

This is not done currently, so the mixing code would end up using
uninitialized pointers from stack.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind

audio_mix: initialize the data pointers to NULL

This should make it easier to catch problems where some of those
pointers are used uninitialized, since reading from NULL should always
crash, while random numbers from stack can turn out to be valid
pointers, so random memory may be silently overwritten.

h264: reset data_partitioning if decoding the slice header for NAL_DPA fails

If it was set before then we can end up trying to decode a slice without
a valid slice header, which can lead to invalid memory access.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC:libav-stable@libav.org

lzw: switch to bytestream2

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC:libav-stable@libav.org

pmpdec: check that there is at least one audio packet.

The code cannot handle there being none, but that should not happen for
valid files.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC:libav-stable@libav.org

mathematics: remove asserts from av_rescale_rnd()

It is a public function, it must not assert on its parameters.

eacmv: check the framerate before setting it.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC:libav-stable@libav.org

adx: check that the offset is not negative

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC:libav-stable@libav.org

rv30: fix extradata size check.

It has been checking the number of bits in the offset instead of the
actual offset.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC:libav-stable@libav.org

mpegvideo: set reference/pict_type on generated reference frames

Otherwise the generic code will unref them, which can then result in
last_picture_ptr == current_picture_ptr, which causes deadlocks at least
in rv40.

Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC:libav-stable@libav.org

avcodec/avcodec: document that some video decoders do not support linesizes changing between frames.

Omission in the documentation found by: Sebastian Dröge
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avformat/bink: display audio track ids

Signed-off-by: Peter Ross <pross@xvid.org>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

MAINTAINERS: add myself for spdif* and hls.c

Signed-off-by: Anssi Hannula <anssi.hannula@iki.fi>

avformat/hls: decouple playlists from variants

Not all "sub-playlists" are variant playlists (containing the same
content with a different bitrate, etc) in the current version of the HLS
specification. They can now also be alternative renditions, containing
e.g. alternative audio tracks etc.

Decouple playlists from variants to prepare for handling the new
features.

Signed-off-by: Anssi Hannula <anssi.hannula@iki.fi>

avformat/http: always allow no-op seek

This also allows checking stream position as per ffurl_seek() doxy.

Signed-off-by: Anssi Hannula <anssi.hannula@iki.fi>

avformat/bink: seek to first frame

Some rare bink files include unused bytes between the frame table index
and first frame data.

Fixes ticket #3266.

This patch has also been tested with non-seekable protocols.

Signed-off-by: Peter Ross <pross@xvid.org>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

matroskadec: Export the MuxingApp element value as metadata

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

ffmpeg: do not fail when options are routed to libavformat and libavcodec and only one can be used

Fixes setting frame_size for pulse demuxer

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

h264: Clear ERContext.cur_pic when unref'ing current picture.

Signed-off-by: Dale Curtis <dalecurtis@chromium.org>

avcodec/mace: check buffer size

Fixes use of uninitialized memory
Fixes: msan_uninit-mem_7f9d2775e090_8782_surge-2-8-MAC3.mov
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

mp3on4: fail when the header check fails

It seems the code assumed that when one out of multiple blocks fail
some span of time is correct, its not, its some channels
that are correct

Fixes use of uninitialized memory
Fixes: msan_uninit-mem_7f44a03ba3ce_5692_id5_1.mp4
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/aacps: fix number of bands used with ipd/opd

Fixes use of uninitialized memory
Fixes: msan_uninit-mem_7f861d16355f_1664_File1_fixed.aac
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/iff: ensure that runs with insufficient input dont leave uninitialized bytes in the output

Fixes use of uninitialized memory
Fixes: msan_uninit-mem_7fa0dea15eae_8988_test.iff
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avformat/bink: recognize bink2 files

Support demuxing of audio streams in bink2 files.

Signed-off-by: Peter Ross <pross@xvid.org>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

doc/muxers/hls: change "can used" to "can be used"

Spotted-by: Alexander Strasser <eclipse7@gmx.net>

fate: Add OggVP8 demux test

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

atrac3plus: Add forgotten initialization of an internal struct

This patch also fixes trac ticket 3254.

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

doc/doxy: remove unused files

Reviewed-by: Stefano Sabatini <stefasab@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  Makefile: add html template files to the doxygen deps

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

doc/muxers/segment: clarify ambiguos reference

Replace "The segmenter muxer" with "This muxer". The "segmenter muxer"
was never defined before, and thus it is a slightly ambiguous term.

doc/muxers/segment: reference hls muxer

doc/muxers/hls: extend documentation

In particular, reference segment muxer and clarify difference between
segment filename number and playlist sequence number.

Should fix trac ticket #2601.

avformat/mux: fix "no tag found" check in validate_codec_tag()

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

avformat/vqf: check avio_read()s return value more completely

Fixes use of uninitialized memory
Fixes: msan_uninit-mem_7f180a523a71_5052_esvorbei_extd.vqf
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>