Merge commit 'e52567c2954f627d420b30f75f71af2f2e4afe80'

* commit 'e52567c2954f627d420b30f75f71af2f2e4afe80':
  build: Strip spurious labels

Conflicts:
configure

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

Merge commit '25cb0c1a1e66edacc1667acf6818f524c0997f10'

* commit '25cb0c1a1e66edacc1667acf6818f524c0997f10':
  x86inc: activate REP_RET automatically

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

avcodec/x86/vp9dsp: Fix compilation with nasm.

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

ffmpeg: add max_error_rate to allow tuning the point decoding errors are considered a failure of the whole transcoding.

Suggested-by: Daemon404
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avfilter/vf_deshake: fix block_contrast() lower brightness value

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

avfilter/deshake

The old row index looks weird and wrong to me

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

avcodec/options_table: add field_order

This allows setting and overriding the field order

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

avcodec/dpx: support for 8 and 16 bit luma only files

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

avcodec/dpx: refactor pixel format selection

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

avfilter/lswsutils: dont override the default scaler

This should make no difference as this code is only used by removelogo which
seems to refuse to use masks of different size.

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

build: Strip spurious labels

The implementation of 25cb0c1a involves lots of spurious labels.

The effect of keeping those labels around is making debugging harder.
Those labels are meaningless, and complicate the disassembly. Also,
gdb can't tell the difference between them and function entry points.

This new strip command is irrelevant to any usage of Libav that would
have used the old fully stripped version, because the old one was for
non-debug use.

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

x86inc: activate REP_RET automatically

Now RET checks whether it immediately follows a branch, so the
programmer dosen't have to keep track of that condition. REP_RET
is still needed manually when it's a branch target, but that's
much rarer.

The implementation involves lots of spurious labels, but that's OK
because we strip them.

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

doc/indevs: make pulse dev formatting consistent with other devices

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

lavu/parseutils: add more resolutions

See http://en.wikipedia.org/wiki/Graphics_display_resolution

Signed-off-by: Niv Sardi <xaiki@evilgiggle.com>
Signed-off-by: Stefano Sabatini <stefasab@gmail.com>

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  x86: fdct: Employ more specific ifdefs

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

Merge commit '9adbc3f3a1770fec9d24b8f5be3438a6c8e9e6a6'

* commit '9adbc3f3a1770fec9d24b8f5be3438a6c8e9e6a6':
  bmv: Remove unused variable

See: 63139f62f35944b9abd42505bf13a740676978e9
Merged-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/dpx: return different error code for unsupported depths

Also give better message in such cases.

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

avfilter/vf_drawtext: remove dead initialization

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

avfilter/vf_phase: remove dead initialization

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

ffmpeg: dont detect slight (0.1sec) backward moving dts as discontinuity

avformat/oggparseflac: fix handling of old flac in ogg

Improves handling of the file in Ticket1617

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

avcodec/flac_parser: export sample_rate also when PARSER_FLAG_COMPLETE_FRAMES is set

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

avcodec/dpx: make .long_name more descriptive

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

avcodec: add dpx parser

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

x86: fdct: Employ more specific ifdefs

This avoids building mmxext and sse2 code when disabled by configure.

bmv: Remove unused variable

configure: filters should select the parts they need not use deps

otherwise enabling the filter would not work if the part hasnt been
enabled by other means already

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

fate: fix dpx on big endian

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

avfilter/vf_scale: dont duplicate default.

use the defaults set by swscale

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

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  oggparsevorbis: return meaningful errors

Conflicts:
libavformat/oggparsevorbis.c

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

Merge commit '2ddb35b91131115c094d90e04031451023441b4d'

* commit '2ddb35b91131115c094d90e04031451023441b4d':
  x86: dsputil: Separate ff_add_hfyu_median_prediction_cmov from dsputil_mmx

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

Merge commit '6b3ff6f91a535d6383f41ca7bdf760165dcb6015'

* commit '6b3ff6f91a535d6383f41ca7bdf760165dcb6015':
  swscale: provide a default scaler if none is set

Conflicts:
libswscale/utils.c

The default is left at bicubic until someone has compared the scalers
properly speed and quality wise.

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

Merge commit '258414d0771845d20f646ffe4d4e60f22fba217c'

* commit '258414d0771845d20f646ffe4d4e60f22fba217c':
  x86: fdct: Initialize optimized fdct implementations in the standard way

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

Merge commit '0b8b2ae5e93d616c2ece59f7175f483154cff918'

* commit '0b8b2ae5e93d616c2ece59f7175f483154cff918':
  x86: xviddct: Employ more specific ifdefs

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

Merge commit '4272bb6ef1533846a788c259cc498562d0704444'

* commit '4272bb6ef1533846a788c259cc498562d0704444':
  doxy: Document avlog

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

avfilter/vf_scale: change the default scaler to bicubic

See: [FFmpeg-devel] [PATCH] swscale/options: switch default to bicubic
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

swscale/options: switch default to bicubic

Suggested-by: Ronald S. Bultje
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avformat/matroskadec: only set r_frame_rate if the value is within reasonable limits

Fixes Ticket2451

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

oggparsevorbis: return meaningful errors

Signed-off-by: Diego Biurrun <diego@biurrun.de>

avfilter/f_interleave: use the name 's' for the pointer to the private context

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

x86: dsputil: Separate ff_add_hfyu_median_prediction_cmov from dsputil_mmx

The function does not depend on MMX and compilation without MMX enabled
fails if the function is compiled conditional on MMX availability.

configure: add --tempprefix option

With this and ccache configure execution time goes down from 3.9 to 2.4 seconds

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

swscale: provide a default scaler if none is set

Lanczos for general case, sinc for upscaling, Gaussian for
downscaling. According to current literature these scalers
should be the best quality-wise algorithms for each case.

Inspired from a patch by wm4 <nfxjfg@googlemail.com>

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

x86: fdct: Initialize optimized fdct implementations in the standard way

x86: xviddct: Employ more specific ifdefs

This avoids building mmxext and sse2 code when disabled by configure.

doxy: Document avlog

Provide some information for every function and add a group.

avcodec/dpx: cosmetics: fix identation

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

avcodec/dpx: reformat avpriv_report_missing_feature messages

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

avcodec/dpx: return proper error code for unsupported files

Also use avpriv_report_missing_feature().

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

avfilter/af_silencedetect: support more sample formats

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

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  ismindex: Change the duration field to int64_t

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

Merge commit '583a287f6f48f263859df142e6f90993e4217c72'

* commit '583a287f6f48f263859df142e6f90993e4217c72':
  ismindex: Calculate the file duration among the included tracks

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

Merge commit 'd347a7b248d4ffdc278373fecf033b0ade030343'

* commit 'd347a7b248d4ffdc278373fecf033b0ade030343':
  ismindex: Use the individual stream duration instead of the global one

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

Merge commit 'b698542ad83284fbb8c22404e3cafeb2dd739d38'

* commit 'b698542ad83284fbb8c22404e3cafeb2dd739d38':
  mov: Don't allocate arrays with av_malloc that will be realloced

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

avcodec/dpx: check packing

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

avcodec/dpx: check encoding

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

avcodec/dpx: fix reading of the packing field

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

avcodec/dpx: fix reading 12bit dpx images, which have non zero padding bits

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

avformat/gifdec: make GIF_APP_EXT_LABEL parsing more robust

Fixes Ticket3021

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

lavd/pulse: add MAINTAINERS entry

Signed-off-by: Lukasz Marek <lukasz.m.luki@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ismindex: Change the duration field to int64_t

This reduces the risk for overflow in pathlogical cases.

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

ismindex: Calculate the file duration among the included tracks

If the input file contains other tracks (non video/audio) that
aren't included in ismindex, the global file duration as returned
by libavformat might not be equal to the maximum of the duration
of the actual included tracks.

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

ismindex: Use the individual stream duration instead of the global one

The stream duration is used for calculating the duration of the
last fragment easily without manually parsing anything else than
the mfra/tfra atoms. When the global file duration was used
previously, the duration of the last fragment could end up wrong
if the streams weren't equally long.

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

mov: Don't allocate arrays with av_malloc that will be realloced

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

avfilter/vf_histogram: use const for variables that are not supposed to change

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

avformat/mov: force parsing of headers if stts is absent

Fixes Ticket2991

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

lavu/opt: fix doxy for av_opt_get* functions about return value

Success code must be >= 0 and not == 0, consistently with the
implementation.

lavu/common: add warning to GET_UTF8 doxy

Should prevent wrong uses, or at least decrease their chance.

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  x86: fdct: Only build fdct code if encoders have been enabled

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

Merge commit 'cd818b3a5709b9b08bd5901cb8863a8b61be265e'

* commit 'cd818b3a5709b9b08bd5901cb8863a8b61be265e':
  rtmpproto: Validate the embedded flv packet size before copying

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

Merge commit '8921e32f730c191543b84e61338bc9d549aa05a3'

* commit '8921e32f730c191543b84e61338bc9d549aa05a3':
  rtmpproto: Readjust the end of the flv buffer if handle_metadata exited early

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

Merge commit '24fee95321c1463360ba7042d026dae021854360'

* commit '24fee95321c1463360ba7042d026dae021854360':
  rtmpproto: Move the flv header/trailer addition to append_flv_data

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

Merge commit '72540e514ceaaed8ecb97ed63637b2a2a4447ab8'

* commit '72540e514ceaaed8ecb97ed63637b2a2a4447ab8':
  rtmpproto: Clear the flv allocation size on reallocp failures

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

Merge commit '4d6d70292e91a7ef027824d731b6b6570ceabf2f'

* commit '4d6d70292e91a7ef027824d731b6b6570ceabf2f':
  rtmpproto: Pass the 'live' parameter in the right unit

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

Merge commit 'a6b361325f2bfc8d9d4e5f761d6c1a07b209c4fb'

* commit 'a6b361325f2bfc8d9d4e5f761d6c1a07b209c4fb':
  rtmpproto: Print the error code string if there's no description

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

cosmetics: group remaining .name and .long_name.

See b2bed9325.

Merge commit 'b2bed9325dbd6be0da1d91ffed3f513c40274fd2'

* commit 'b2bed9325dbd6be0da1d91ffed3f513c40274fd2':
  cosmetics: Group .name and .long_name together in codec/format declarations

Conflicts:
libavcodec/8svx.c
libavcodec/alac.c
libavcodec/cljr.c
libavcodec/dnxhddec.c
libavcodec/dnxhdenc.c
libavcodec/dpxenc.c
libavcodec/dvdec.c
libavcodec/dvdsubdec.c
libavcodec/dvdsubenc.c
libavcodec/ffv1dec.c
libavcodec/flacdec.c
libavcodec/flvdec.c
libavcodec/fraps.c
libavcodec/frwu.c
libavcodec/g726.c
libavcodec/gif.c
libavcodec/gifdec.c
libavcodec/h261dec.c
libavcodec/h263dec.c
libavcodec/iff.c
libavcodec/imc.c
libavcodec/libopencore-amr.c
libavcodec/libopenjpegdec.c
libavcodec/libopenjpegenc.c
libavcodec/libspeexenc.c
libavcodec/libvo-amrwbenc.c
libavcodec/libvorbisenc.c
libavcodec/libvpxenc.c
libavcodec/libx264.c
libavcodec/libxavs.c
libavcodec/libxvid.c
libavcodec/ljpegenc.c
libavcodec/mjpegbdec.c
libavcodec/mjpegdec.c
libavcodec/mpeg12dec.c
libavcodec/mpeg4videodec.c
libavcodec/msmpeg4dec.c
libavcodec/pgssubdec.c
libavcodec/pngdec.c
libavcodec/pngenc.c
libavcodec/proresdec_lgpl.c
libavcodec/proresenc_kostya.c
libavcodec/ra144enc.c
libavcodec/rawdec.c
libavcodec/rv10.c
libavcodec/sp5xdec.c
libavcodec/takdec.c
libavcodec/tta.c
libavcodec/v210dec.c
libavcodec/vp6.c
libavcodec/wavpack.c
libavcodec/xbmenc.c
libavcodec/yop.c

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

Merge commit '49fe9c05f97bc2cfafd8fdcfea2f313d7c8e2c48'

* commit '49fe9c05f97bc2cfafd8fdcfea2f313d7c8e2c48':
  oggparsevorbis: K&R formatting cosmetics

Conflicts:
libavformat/oggparsevorbis.c

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

Merge commit '47b6cfc2a0333fb24f074d27830bf35ae5007050'

* commit '47b6cfc2a0333fb24f074d27830bf35ae5007050':
  avformat/output-example: Declare link dependency on libswscale in the Makefile

Conflicts:
libavformat/Makefile

Changes to libavformat/Makefile are not merged as our build system for the examples
is different.

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

Merge commit '80fefbed623491b92fe59ead99225f99c0d0ca08'

* commit '80fefbed623491b92fe59ead99225f99c0d0ca08':
  x86: cpu: Restore some explanatory comments removed in 7160bb7

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

Merge commit '6a5bfd4d49e7433f6de869ccfe682940dbca8740'

* commit '6a5bfd4d49e7433f6de869ccfe682940dbca8740':
  atrac: Add missing av_cold

See: 852241561d040e0b147b5c866fd4f288324fbe48
Merged-by: Michael Niedermayer <michaelni@gmx.at>

avfilter/vf_frei0r: fix out of array read

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

avformat/vobsub: raise packet even if apparently incomplete.

This restore the latest packet which was dropped from the FATE test in
dbfe61100.

x86: fdct: Only build fdct code if encoders have been enabled

fdct is only initialized if encoders are enabled.

rtmpproto: Validate the embedded flv packet size before copying

This wasn't an issue prior to 58404738, when the whole RTMP packet
was copied at once and the length of the individual embedded flv
packets only were validated by the flv demuxer.

Prior to this patch, this could lead to reads and writes out of bound.

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

rtmpproto: Readjust the end of the flv buffer if handle_metadata exited early

If the embedded flv packets were incomplete and we aborted the
copying loop early, make sure the flv buffer is trimmed to
only contain full packets.

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

rtmpproto: Move the flv header/trailer addition to append_flv_data

update_offset is also called from handle_metadata, where the
packet header sizes is already included in the size.

Previously this lead to flv_data/flv_size including 15 uninitialized
bytes at the end after each call to handle_metadata, making the
flv demuxer lose sync with the stream.

Also remove leftover copying in handle_metadata. This is a leftover
from the refactoring in 5840473. (Previously this final mempcy was
the one that copied all the packets at once, while this is done
within the loop right now.) After making sure flv_size is set to
the right size, this write was out of bounds.

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

rtmpproto: Clear the flv allocation size on reallocp failures

This was overlooked in d872fb0f7 since I assumed all the realloc
issues in the rtmp code was fixed already.

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

rtmpproto: Pass the 'live' parameter in the right unit

The current magic numbers passed are values in seconds, while the
parameter itself should be passed over the wire in milliseconds.

This makes (some/all?) live streams from Red5 work correctly, that
previously returned StreamNotFound even with "-rtmp_live live". After
this commit, the default 'any' also works on these streams.

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

rtmpproto: Print the error code string if there's no description

On (certain streams/setups at least on) Red5, the description string
actually is present, but empty. Therefore, first try loading the
description, but if not found or empty, load the code string instead.
The code string is quite understandable in most cases anyway (like
"NetStream.Play.StreamNotFound").

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

avformat/vobsub: fix several issues.

Here is an extract of fate-samples/sub/vobsub.idx, with an additional
text at the end of each line to better identify each bitmap:

    timestamp: 00:04:55:445, filepos: 00001b000 Ace!
    timestamp: 00:05:00:049, filepos: 00001b800 Wake up, honey!
    timestamp: 00:05:02:018, filepos: 00001c800 I gotta go to work.
    timestamp: 00:05:02:035, filepos: 00001d000 <???>
    timestamp: 00:05:04:203, filepos: 00001d800 Look after Clayton, okay?
    timestamp: 00:05:05:947, filepos: 00001e800 I'll be back tonight.
    timestamp: 00:05:07:957, filepos: 00001f800 Bye! Love you.
    timestamp: 00:05:21:295, filepos: 000020800 Hey, Ace! What's up?
    timestamp: 00:05:23:356, filepos: 000021800 Hey, how's it going?
    timestamp: 00:05:24:640, filepos: 000022800 Remember what today is? The 3rd!
    timestamp: 00:05:27:193, filepos: 000023800 Look over there!
    timestamp: 00:05:28:369, filepos: 000024800 Where are they going?
    timestamp: 00:05:28:361, filepos: 000025000 <???>
    timestamp: 00:05:29:946, filepos: 000025800 Let's go see.
    timestamp: 00:05:31:230, filepos: 000026000 I can't, man. I got Clayton.

Note the two "<???>": they are basically split subtitles (with the
previous one), which the dvdsub decoder is now supposed to reconstruct
with a previous commit. But also note that while the first chunk has
increasing timestamps,

    timestamp: 00:05:02:018, filepos: 00001c800
    timestamp: 00:05:02:035, filepos: 00001d000

...it's not the case of the second one (and this is not an exception in the
original file):

    timestamp: 00:05:28:369, filepos: 000024800
    timestamp: 00:05:28:361, filepos: 000025000

For the dvdsub decoder, they need to be "filepos'ed" ordered, but the
FFDemuxSubtitlesQueue is timestamps ordered, which is the reason of the
introduction of a sub sort method in the context, to allow giving
priority to the position, and then the timestamps. With that change, the
dvdsub decoder get fed with ordered packets.

Now the packet size estimation was also broken: the filepos differences
in the vobsub index defines the full data read between two subtitles
chunks, and it is necessary to take into account what is read by the
mpegps_read_pes_header() function since the length returned by that
function doesn't count the size of the data it reads. This is fixed with
the introduction of total_read, and {old,new}_pos. By doing this change,
we can drop the unreliable len16 heuristic and simplify the whole loop.
Note that mpegps_read_pes_header() often read more than one PES packet
(typically in one call it can read 0x1ba and 0x1be chunk along with the
relevant 0x1bd packet), which triggers the "total_read + pkt_size >
psize" check. This is an expected behaviour, which could be avoided by
having a more chunked version of mpegps_read_pes_header().

The latest change is the extraction of each stream into its own
subtitles queue. If we don't do this, the maximum size for a subtitle
chunk is broken, and the previous changes can not work. Having each
stream in a different queue requires some little adjustments in the
seek code of the demuxer.

This commit is only meaningful as a whole change and can not be easily
split. The FATE test changes because it uses the vobsub demuxer.

avcodec/dvdsubdec: reconstruct incomplete SPU packets.

avcodec/dvdsubdec: extract every subtitle in a different file (debug).

Even thought a generic solution would be more relevant, this is
currently helpful while debugging.

avcodec/dvdsubdec: factor out rectangles reset.

cosmetics: Group .name and .long_name together in codec/format declarations

oggparsevorbis: K&R formatting cosmetics

Signed-off-by: Diego Biurrun <diego@biurrun.de>

avformat/output-example: Declare link dependency on libswscale in the Makefile

x86: cpu: Restore some explanatory comments removed in 7160bb7

atrac: Add missing av_cold

Signed-off-by: Diego Biurrun <diego@biurrun.de>

ffplay: avoid direct access to lowres use av_codec_g/set_lowres()

Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>