jpeg2000: remove unneeded variable initialization

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

jpeg2000: tag_tree_size: replace check by assert

This function is never called with values large enough for the
error condition to occur

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

j2k/jpeg2000: split data pointer in int & float.

This fixes a TODO item and unifies both decoders structures
It also fixes undefined behavior due to aliasing violations

I choose 2 fields instead of a union because mistakely using the
wrong type with a union will lead to hard to debug "wrong output"
while with 2 fields mistakely using the wrong type will crash
with a null pointer derefernce which is much easier to debug

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

Use -lX11 -lXext when linking Xv output device.

Reviewed-by: Stefano Sabatini

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  arm: Include hpeldsp_neon.o if h264qpel is enabled

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

Merge commit 'efb7968cfe8b285ab4f27b363719b7c92d19ec74'

* commit 'efb7968cfe8b285ab4f27b363719b7c92d19ec74':
  arm: Don't unconditionally build dsputil files

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

Merge commit '36a7df8cf1115aa37a1b0d42324ecde5ab6c2304'

* commit '36a7df8cf1115aa37a1b0d42324ecde5ab6c2304':
  arm: Only build the FFT init files if FFT is enabled

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

Merge commit '186599ffe0a94d587434e5e46e190e038357ed99'

* commit '186599ffe0a94d587434e5e46e190e038357ed99':
  build: cosmetics: Place unconditional before conditional OBJS lines

Conflicts:
libavcodec/x86/Makefile

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

doc/encoders: remove duplicated "the"

lavd: add xv output device

Based on the work of Jeff Moguillansky <Jeff.Moguillansky@am.sony.com>.

See thread:
Subject: [FFmpeg-devel] x11 output device for libavdevice
Date: Wed, 10 Apr 2013 23:10:47 +0000

Merge commit '9b9b2e9f3036abfd42916bcf734af14b4cb686aa'

* commit '9b9b2e9f3036abfd42916bcf734af14b4cb686aa':
  build: arm: cosmetics: Place all OBJS declarations in alphabetical order

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

Merge commit '701966730ce10290fd49c5ccedd73f505680f764'

* commit '701966730ce10290fd49c5ccedd73f505680f764':
  vmd: drop incomplete chunks and spurious samples

Conflicts:
libavcodec/vmdav.c

2 of the changes are replaced by assert0s, as they should
be impossible.
The actual bug is likely caused by a invalid block_align
which is checked for and thus impossible in ffmpeg.

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

Merge commit '676da248cad49debc40720baa13214f0b94dcc71'

* commit '676da248cad49debc40720baa13214f0b94dcc71':
  vmd: refactor the inner decode loop

Conflicts:
libavcodec/vmdav.c

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

Merge commit 'c8f3cb9119c2183680d44a509a1b5a9817a3bee9'

* commit 'c8f3cb9119c2183680d44a509a1b5a9817a3bee9':
  vmd: return meaningful errors

Conflicts:
libavcodec/vmdav.c

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

Merge commit '91a6944e56236234f0a7ba162404665753cbcb51'

* commit '91a6944e56236234f0a7ba162404665753cbcb51':
  vmd: use the PALETTE_COUNT constant uniformly

Conflicts:
libavcodec/vmdav.c

No change as the changed line does not exist in ffmpegs vmdav.c

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

fate: fix smvjpeg test

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

libavcodec/smvjpegdec: pass flags / idct algo into jpeg avctx

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

fate: add smv test

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

arm: Include hpeldsp_neon.o if h264qpel is enabled

A few of the h264qpel neon functions are shared with other
hpeldsp functions in this file.

This fixes standalone compilation of the h264 decoder on arm.

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

arm: Don't unconditionally build dsputil files

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

arm: Only build the FFT init files if FFT is enabled

This fixes build errors in cases where FFT is disabled.

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

build: cosmetics: Place unconditional before conditional OBJS lines

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

build: arm: cosmetics: Place all OBJS declarations in alphabetical order

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

avformat/mov: Try to detect and support files with invalid sample sizes

Fixes Ticket2605

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

Add AV_HASH_MAX_SIZE.

Makes it easier to use static allocation for the result buffer.

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

wamenc: handle failure to encode.

This is necessary since the switch to floating point
input means there is no longer sufficient input data
validation.
There is a good chance that other encoders are affected
by similar issues.
This problem can be triggered by trying to encode
extremely large values and probably also with Inf and
possibly also NaNs.

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

vmd: drop incomplete chunks and spurious samples

Odd chunk size makes no sense for stereo and incomplete chunks are
not supported.

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

wtvenc: use ffio_fill()

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

j2kdec: remove unused variables

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

mpegvideo: fix forgotten lowres op_index limits

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

mpegvideo: implement ff_put_h264_chroma_mc1 & ff_avg_h264_chroma_mc2

These are needed for lowres 3

This may fix Ticket2538

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

lavfi/noise: fix out of array access

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

lavfi/noise: support slice threading

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

wvenc: remove flush call, not needed since 4f112a8e3

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

remove APEv1 tag writer

Not used anymore.

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

apetag: do not create invalid APE tags

APEv2 specifications forbids non-ascii keys.

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

tiff: support inverted fill_order for packbits compression

Fixes #1995.

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

tiff: add helper function for fill_order case

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

alac: use init_get_bits8()

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

lavc/tta: use init_get_bits8()

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

lavf/flacenc: use ffio_fill()

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

vmd: refactor the inner decode loop

Simplify a little, assume empty frames are acceptable and
do not pointlessly reinit the bytestream2 contexts using
possibly wrong size values.

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

vmd: return meaningful errors

CC: libav-stable@libav.org

vmd: use the PALETTE_COUNT constant uniformly

While at it drop useless parentheses.

Merge remote-tracking branch 'qatar/master'

* qatar/master:
  doc: Mention the target_samples and ld variables for fate configs
  fate.sh: Allow specifying --as via a specific variable

Conflicts:
doc/fate.texi

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

Merge commit '9f30fb5a773d59298d8d45c741b3fd971d84c97b'

* commit '9f30fb5a773d59298d8d45c741b3fd971d84c97b':
  configure: Don't pass -mthumb or -march= to MSVC
  mpeg12: skip frames consistently

Conflicts:
libavcodec/mpeg12dec.c

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

Merge commit '2d2d6a4883479403798f4ed46941d5b365823570'

* commit '2d2d6a4883479403798f4ed46941d5b365823570':
  lavf: add a raw WavPack muxer.
  apetag: add support for writing APE tags
  matroskaenc: support muxing WavPack

Conflicts:
libavformat/Makefile
libavformat/allformats.c
libavformat/apetag.h
libavformat/version.h
libavformat/wvenc.c

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

Merge commit '794ca87d2bff2513118de8b97595b3e23070e67d'

* commit '794ca87d2bff2513118de8b97595b3e23070e67d':
  wvdec: split block header parsing into a separate file

Conflicts:
libavformat/Makefile
libavformat/wvdec.c

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

Merge commit '0a1a94450a28eef854162f859e79ecfb9f97915b'

* commit '0a1a94450a28eef854162f859e79ecfb9f97915b':
  lavf: rename wv.c to wvdec.c

Conflicts:
libavformat/Makefile

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

doc: Mention the target_samples and ld variables for fate configs

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

fate.sh: Allow specifying --as via a specific variable

This simplifies specifying a value containing spaces for this
parameter.

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

Merge commit '89806691b1c39181c63d95e0fddc30f11e2a7b04'

* commit '89806691b1c39181c63d95e0fddc30f11e2a7b04':
  wavpack: check that all the channels were coded.
  wavpack: check that there aren't too many blocks per packet

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

Merge commit '7d039e70a5ff23a7deaa866684d2e8872acc5169'

* commit '7d039e70a5ff23a7deaa866684d2e8872acc5169':
  wavpack: extract channel information from the bitstream

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

Merge commit '78f75b6fa421dd39a715588e9487579f1ce5bada'

* commit '78f75b6fa421dd39a715588e9487579f1ce5bada':
  wavpack: extract sample rate from the bitstream
  configure: support gcc-4.8 instrumentation

Conflicts:
libavcodec/wavpack.c

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

Merge commit 'ba13606ca6adbc74b4db4a72b0769397d6408791'

* commit 'ba13606ca6adbc74b4db4a72b0769397d6408791':
  fate: Add a --target-samples path parameter

Conflicts:
configure
tests/fate/audio.mak
tests/fate/cover-art.mak
tests/fate/demux.mak
tests/fate/ea.mak
tests/fate/filter-video.mak
tests/fate/h264.mak
tests/fate/image.mak
tests/fate/lossless-audio.mak
tests/fate/lossless-video.mak
tests/fate/microsoft.mak
tests/fate/pcm.mak
tests/fate/prores.mak
tests/fate/qt.mak
tests/fate/real.mak
tests/fate/screen.mak
tests/fate/video.mak
tests/fate/voice.mak
tests/fate/vpx.mak
tests/fate/vqf.mak

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

Merge commit '4a27a52a1f74016095b7aee1b4a422cf62217ade'

* commit '4a27a52a1f74016095b7aee1b4a422cf62217ade':
  fate: Don't use files from SRC_PATH in the actual tests
  indeo4: reuse context block VLC for band instead of defaulting

Conflicts:
tests/Makefile

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

Merge commit '6647aa0426e73839b9b1d1c9d86188f469167531'

* commit '6647aa0426e73839b9b1d1c9d86188f469167531':
  indeo4: add missing Haar and slanted transforms
  dxtory v2 support

Conflicts:
libavcodec/indeo4.c
libavcodec/ivi_dsp.c
libavcodec/version.h

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

Merge commit '728214992e3698305550c1762f973d2ac567f016'

* commit '728214992e3698305550c1762f973d2ac567f016':
  vc1dec: Remove interlaced warning
  vc1dec: Fix mixed field/frame intensity compensation
  vc1dec: Match addressing between compensation and MC in vc1_mc_4mv_chroma4
  vc1dec: Handle top and bottom blocks in vc1_mc_4mv_chroma4() differently if needed
  vc1dec: Fix doxy for vc1_mc_4mv_chroma4()
  vc1dec: Drop old use_ic code from vc1_b_mc
  vc1: Use shuffled use_ic instead of equally shuffled mv_mode
  vc1dec: Implement intensity compensation for vc1_interp_mc()

Conflicts:
Changelog
libavcodec/vc1dec.c
libavcodec/version.h

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

Merge commit '28243b0d35b47bbf9abbd454fc444a6e0a9e7b71'

* commit '28243b0d35b47bbf9abbd454fc444a6e0a9e7b71':
  vc1dec: Redesign the intensity compensation

Conflicts:
libavcodec/vc1.c
libavcodec/vc1dec.c

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

Merge commit '93b1281264b87961f53c3e9c134cc2727ecd91ed'

* commit '93b1281264b87961f53c3e9c134cc2727ecd91ed':
  vc1dec: Shuffle field MVs after decoding, not before

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

j2k/jpeg2000: merge float DWT and related code

the float conditions are largely changed to check the DWT
instead of the bitexact flag, which previously was wrong
(DWT53 is always int)

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

configure: Don't pass -mthumb or -march= to MSVC

The modern MSVC for ARM always builds for thumb, and it can't be
disabled.

Also just use the default arch instead of trying to map the -march
parameter to MSVC's -arch parameter (which only takes the values
ARMv7VE and VFPv4).

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

j2kdec:merge mct_decode from jpeg2000

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

jpeg2000dec: mct_decode: remove unused return

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

ffmpeg: use actual packet pts when discarding for the copypriorss option

The ist->pts value at this point does not necessarily match the actual packet
pts, which is what should be used to decide whether to discard the packet.
Without this change, some video packets that have pts on or after the start
time are getting discarded when they should not be.

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

j2k/jpeg2000: split stepsize in float & int variables

This is more clear and less prone to mistakes.

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

jpeg2000dec: remove unneeded returns

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

j2k/jpeg2000: fix typo in lut_gain index

no testcase

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

lavfi/drawutils: attempt to fix subsampling.

mpeg12: skip frames consistently

The decoder did not start predicted frames with missing references but
called the end of frame handling on them anyway. This caused an
assertion in the VA API HW accelorator since it tried to render a
picture with uninitialized buffers.

j2kdec/jpeg2000dec: partially merge quantization code

The quantization code needs more work, not so much work
merging but more work investigating what is correct.

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

j2kdec: Try to fix 8bps output case

Ive no test samples for which this makes a difference but it
matches the 16bit implementation.

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

lavf: add a raw WavPack muxer.

apetag: add support for writing APE tags

This will be useful in the WavPack muxer.

matroskaenc: support muxing WavPack

wvdec: split block header parsing into a separate file

It will be reused by other muxers and demuxers.

lavf: rename wv.c to wvdec.c

wv.c will be used for shared wavpack functions.

wavpack: check that all the channels were coded.

wavpack: check that there aren't too many blocks per packet

wavpack: extract channel information from the bitstream

This way we don't require the caller to parse the block header. It also
allows the channel configuration to change mid-stream.

wavpack: extract sample rate from the bitstream

This way we don't require the caller to parse the block header. It also
allows the sample rate to change mid-stream.

configure: support gcc-4.8 instrumentation

Since version 4.8 gcc sports asan and tsan integration that can
be leveraged when hunting bugs.

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

fate: Add a --target-samples path parameter

This allows having the samples accessible via different paths
on the target and on the host.

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

fate: Don't use files from SRC_PATH in the actual tests

If building out of tree, make sure the filter scripts are copied
into the build tree before running tests. This makes sure that
SRC_PATH doesn't need to exist on the remote system (or doesn't
need to exist at the same path).

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

indeo4: reuse context block VLC for band instead of defaulting

Currently if no custom block VLC is specified for band, a default block VLC
will be used, while the global block VLC stored in the context should be used
instead.

This fixes decoding of one sample I have.

indeo4: add missing Haar and slanted transforms

That involves fixing INV_HAAR4() macro and changing it to work with different
input and output like INV_HAAR8() instead of in-place transform.

tta: make probing more robust

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

dxtory v2 support

vc1dec: Remove interlaced warning

Also add a note about the feature in the changelog.

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

vc1dec: Fix mixed field/frame intensity compensation

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

vc1dec: Match addressing between compensation and MC in vc1_mc_4mv_chroma4

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

vc1dec: Handle top and bottom blocks in vc1_mc_4mv_chroma4() differently if needed

Now it can use different references for those blocks and even use
averaging.

This fixes several chroma artifacts in several videos.

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

vc1dec: Fix doxy for vc1_mc_4mv_chroma4()

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

vc1dec: Drop old use_ic code from vc1_b_mc

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

vc1: Use shuffled use_ic instead of equally shuffled mv_mode

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

vc1dec: Implement intensity compensation for vc1_interp_mc()

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

vc1dec: Redesign the intensity compensation

Use the intensity-compensated reference frame for subsequent
fields/B-frames.

Since we currently don't change the reference frame we have to
maintain lookup tables for intensity compensation in the following
dependent frames.

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

vc1dec: Shuffle field MVs after decoding, not before

This simplifies the code since copying MVs to the reference is not needed
anymore (and maybe something about fixing artifacts).

Also remove the unused mv_f_last.

Fixes a small number of artifacts in black_screen_VC-1.mkv
and several more artifacts in other videos.

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

avcodec/vp8: Cosmetics, maintain alphabetical order in threading headers

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

os2threads: move from lavc to compat/

For useage in other places besides lavc. Needed after commit
90f9a5830b5d332de7ebb1ab45589f1870cbd65d

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