lavf/matroskadec: A_QUICKTIME and fourcc 0x00000000

In many older QuickTime files, the audio format, or "fourcc", is
0x00000000. The QuickTime File Format Specification states the following
regarding this situation:

"This format descriptor should not be used, but may be found in some
files. Samples are assumed to be stored in either 'raw ' or 'twos'
format, depending on the sample size field in the sound description."

MPlayer handles this logic by itself, but FFmpeg/FFplay currently does
not.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavf/mov: Audio and fourcc 0x00000000

In many older QuickTime files, the audio format, or "fourcc", is
0x00000000. The QuickTime File Format Specification states the following
regarding this situation:

"This format descriptor should not be used, but may be found in some
files. Samples are assumed to be stored in either 'raw ' or 'twos'
format, depending on the sample size field in the sound description."

MPlayer handles this logic by itself, but FFmpeg/FFplay currently does
not.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

x86/vf_w3fdif: 32-bit compatibility for w3fdif_simple_high

avfilter/avf_showspectrum: add option to draw legend

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

ffmpeg: check fclose return values

In the spirit of commit a956840cbc. Simple method to reproduce:
pass -vstats_file /dev/full to ffmpeg.

All raw fclose usages in ffmpeg.c taken care of here.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

lavfi/avf_showspectrum: replace pow(x, 0.25) by sqrt(sqrt(x))

This is faster; precision assured as result is a float.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

configure: remove unused bulitin check

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

lavu: rename and move ff_parity to av_parity

av_popcount is not defined in intmath.h.

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

lavc/g729dec: use ff_parity()

lavc/ccaption_dec: use ff_parity()

lavu: add ff_parity()

lavc/ccaption_dec: fix always true condition

No idea why this wasn't ever detected by a static analyzer.

lavu/common: add an explanation to FF_CEIL_RSHIFT()

lavf/uncodedframecrcenc: use FF_CEIL_RSHIFT()

lavfi/deshake: use FF_CEIL_RSHIFT()

lavu/common: fix FF_CEIL_RSHIFT() range comment

x86/intmath: add sse optimized av_clipf and av_clipd

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avfilter/formats: fix leak of formats on error

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

ffmpeg: replace log2 by av_log2

Slight change in output deemed irrelevant here.

Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

avcodec/utils: Check bits_per_raw_sample on video encoder open

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/mpegvideo_enc: Clip bits_per_raw_sample within valid range

Fixes out of array read
Fixes: test_case-mdc.264 (b47be15a120979f5a1a945c938cbef33)
Found-by: Tyson Smith <twsmith@mozilla.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/ccaption_dec: rewrite packet handler as case statement; remove COR3 macro

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/ccaption_dec: clean up and standardize white space

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

configure: use warn() for mips unknown cpu warning

Found-by: ubitux
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

mips: display a warning message when using an unknown CPU

Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec: Remove libstagefright

It serves absolutely no purpose other than to confuse potentional
Android developers about how to use hardware acceleration properly
on the the platform. The stagefright "API" is not public, and the
MediaCodec API is the proper way to do this.

Furthermore, stagefright support in avcodec needs a series of
magic incantations and version-specific stuff, such that
using it actually provides downsides compared just using the actual
Android frameworks properly, in that it is a lot more work and confusion
to get it even running. It also leads to a lot of misinformation, like
these sorts of comments (in [1]) that are absolutely incorrect.

[1] http://stackoverflow.com/a/29362353/3115956

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

avfilter/vf_pad: fix direct padding

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

lavf/matroskadec: correct codec_tag for "SMI" SVQ3 files

This corrects the codec_tag for some SVQ3 files

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavfi/af_compensationdelay: replace pow(x,0.5) by sqrt(x)

sqrt is faster, and is sometimes more accurate depending on the libm.

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

ffmpeg: check return value of avio_closep for progress report

avio_closep is not guaranteed to succeed, and its return value can
contain information regarding failure of preceding writes and silent
loss of data (man 2 close, man fclose). Users should know when the
progress was not successfully logged, and so a diagnostic is printed
here.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

avpacket: fix size check in packet_alloc

The previous check only caught sizes from -AV_INPUT_BUFFER_PADDING_SIZE
to -1.

This fixes ubsan runtime error: signed integer overflow: 2147483647 + 32
cannot be represented in type 'int'

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

parser: add av_assert1 to make sure the codec matches

Otherwise this can have some surprising effects (crashes), so let's
better not allow it.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

lavf/matroskaenc: Assume 48kHz sample rate for Opus initial padding.

Analyzed by Timothy B. Terriberry in Mozilla bug 1227153.
Fixes ticket #5121.

avcodec/motion_est: Fix mv_penalty table size

Fixes out of array read

Found-by: Tyson Smith <twsmith@mozilla.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/ccaption_dec: remove unnecessary include

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/avf_avectorscope: add line drawing support

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

avfilter/window_func: add tukey window function

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

avfilter/avf_showspectrum: use av_clip/lrintf

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

avfilter/avf_showspectrum: add yet another color map

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

avcodec/dxv: Check idx in dxv_decompress_dxt5()

Fixes potential out of array read

No testcase available

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/dxv: Check idx in CHECKPOINT()

Fixes out of array read

Fixes Ticket5098
Fixes Ticket5099

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavc/pcm_tablegen: slight speedup of table generation

This gets rid of some branches to speed up table generation slightly
(impact higher on mulaw than alaw). Tables are identical to before,
tested with FATE.

Sample benchmark (Haswell, GNU/Linux+gcc):
old:
 313494 decicycles in build_alaw_table,    4094 runs,      2 skips
 315959 decicycles in build_alaw_table,    8190 runs,      2 skips

 323599 decicycles in build_ulaw_table,    4095 runs,      1 skips
 318849 decicycles in build_ulaw_table,    8188 runs,      4 skips

new:
 261902 decicycles in build_alaw_table,    4096 runs,      0 skips
 266519 decicycles in build_alaw_table,    8192 runs,      0 skips

 209657 decicycles in build_ulaw_table,    4096 runs,      0 skips
 232656 decicycles in build_ulaw_table,    8192 runs,      0 skips

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>

avcodec/h264_slice: Fix integer overflow in implicit weight computation

Fixes mozilla bug 1230423

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

swscale/utils: Use normal bilinear scaler if fast cannot be used due to tiny dimensions

Fixes Ticket4960

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/h264_refs: Fix and add back trace code removed in e1c5170c638e39a8017827339ac687c5f6ff9b35

swscale/arm/yuv2rgb: add ff_yuv422p_to_{argb,rgba,abgr,bgra}_neon_{16,32}

swscale/arm/yuv2rgb: fix typo

avfilter/avf_showspectrum: add option to set gain for calculating color intensity

Inspired by Sonic Visualizer.

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

brstm: fix missing closing brace

Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

brstm: also allocate b->table in read_packet

This fixes NULL pointer dereferencing if the codec is forced to
adpcm_thp even though a different one was detected.

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

brstm: make sure an ADPC chunk was read for adpcm_thp

This fixes NULL pointer dereferencing.

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

vorbisdec: reject rangebits 0 with non-0 partitions

This causes non-unique elements in floor_setup->data.t1.list, which
makes the stream undecodable according to the specification.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

Merge commit '2080bea4a74b6b1598caa806a11ea807ba546f94'

* commit '2080bea4a74b6b1598caa806a11ea807ba546f94':
  h264_refs: Remove broken trace debug code

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '58170b027144f4840018c7a60fb8d07924cc6242'

* commit '58170b027144f4840018c7a60fb8d07924cc6242':
  lavc: Add missing #includes for ff_dlog()

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '3b6473b43eb69fc3faaf69f7fd0b83b51db7607f'

* commit '3b6473b43eb69fc3faaf69f7fd0b83b51db7607f':
  qsvdec: properly handle the warning from MFXVideoCORE_SyncOperation

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit 'a43905f4ae261bdde87c300901d867b31961f57b'

* commit 'a43905f4ae261bdde87c300901d867b31961f57b':
  qsvenc_hevc: improve the default settings

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '9d74012761bc3ee676fe43321d5699e4877fde5b'

* commit '9d74012761bc3ee676fe43321d5699e4877fde5b':
  h264: improve behaviour with invalid reference lists

Not merged, as we have a different solution.

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

lavf/qtpalette: Ignore greyscale bit in certain cases

The QuickTime File Format Specification states the following:

"Depth: A 16-bit integer that indicates the pixel depth of the
compressed image. Values of 1, 2, 4, 8 ,16, 24, and 32 indicate the
depth of color images. The value 32 should be used only if the image
contains an alpha channel. Values of 34, 36, and 40 indicate 2-, 4-, and
8-bit grayscale, respectively, for grayscale images."

There is no mention of value 33, i.e. 1-bit video (0x01) with the
greyscale bit (0x20) set. I therefore suggest that we ignore the
greyscale bit when processing 1-bit video. Another reason to do this is
that the first 1-bit sample file below will be displayed properly with
blue colors in QuickTime in Windows or Mac *in spite of* the greyscale
bit being set.

Also, QuickTime in Windows or Mac ignores the greyscale bit if the
video sample description contains a palette, regardless of bit depth.
This is undocumented behaviour, but I think we should do the same, and
it seems pretty logical after all, since one wouldn't really bother
putting a customized palette into a grayscale file anyway. See the
second 8-bit sample file below, which has the greyscale bit set, and
which contains a palette in the video sample description. In Windows or
Mac, it will be displayed with the palette in the sample description, in
spite of the greyscale bit being set.

Sample file 1 (1-bit QuickTime Animation):
https://drive.google.com/open?id=0B3_pEBoLs0faTThSek1EeXQ0ZHM
Earth Spin 1-bit qtrle orig.mov

Sample file 2 (8-bit QuickTime Animation):
https://drive.google.com/open?id=0B3_pEBoLs0fad2s0V1YzUWo5aDA
quiz-palette+gs.mov

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavc/qtrle: Use AV_PIX_FMT_PAL8 for 1-bit video

This commit fixes the lack of palettized display of 1-bit video
in the qtrle decoder. It is related to my commit of
lavf/qtpalette, which added 1-bit video to the "palettized video"
category. As far as I can see, everything works fine, but comments are
of course welcome.

Below are links to sample files, which should now be displayed properly
with bluish colors, but which were previously displayed in black &
white.

Matroska:
https://drive.google.com/open?id=0B3_pEBoLs0faNjI0cHBMWDhYY2c
Earth Spin 1-bit qtrle.mkv

QuickTime (mov):
https://drive.google.com/open?id=0B3_pEBoLs0faUlItWm9KaGJSTEE
Earth Spin 1-bit qtrle.mov

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/h264: Fix regression caused by removial of default_ref_list

This fixes a regression of the sample from Ticket 2371

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

vorbisdec: reject channel mapping with less than two channels

It causes the angle channel number to equal the magnitude channel
number, which makes the stream undecodable according to the
specification.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

avfilter/avf_showspectrum: finally fix log scaler

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

avfilter/avf_showspectrum: switch to FFT

Rationale: supports >16 bit audio, more than 96dB SNR.

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

lavfi/drawtext: Fix microsecond display.

Fixes ticket #4792.

h264_refs: Remove broken trace debug code

lavc: Add missing #includes for ff_dlog()

ffmdec: reset packet_end in case of failure

This fixes segmentation faults caused by passing a packet_ptr of NULL to
memcpy.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

lavc: use get_bitsz to simplify the code

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

get_bits: add get_bitsz for reading 0-25 bits

This can be used to simplify code in a couple of places.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

avformat/ffmdec: Add {} to nested if/else

This preempts potential bugs if this is changed and the indention
ends up different from C interpretation

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavfi/vf_decimate: do not compare the first frame to itself.

This is a waste of computing power and will result to 0,
making it always dropped.
Use maximum difference values instead.

x86/vf_stereo3d: remove a few unnecessary movas

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

configure: bump copyright year to 2016

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

ffmdec: change type of state and id to unsigned

This fixes ubsan runtime error: left shift by 8 places cannot be
represented in type 'int'

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>

avcodec/put_bits: Always check buffer end before writing

This causes a overall slowdown of 0.1 % (tested with mpeg4 single thread encoding of matrixbench at QP=3)

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

doc/encoders: Fix application name

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avutil/cpu: add missing entry for vfp_vm to av_parse_cpu_caps

avcodec/dca: remove unused float quant table

dca: adjust decoding of the XBR extension for integer core decoding

avfilter/avf_showspectrum: add 4th and 5th root scaler

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

avfilter/avf_showspectrum: make some helper functions

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

avfilter/avf_showspectrum: fix null pointer dereference if allocation fails

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

avfilter/avf_showspectrum: add fiery color map

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

avfilter: add showspectrumpic filter

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

avfilter/avf_showspectrum: store win_size in private context and calculate it only once

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

avfilter/vf_zoompan: do not free frame we pushed to lavfi

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

Merge commit '40d949677335a564f769823f4afdb7e7a3da8d6b'

* commit '40d949677335a564f769823f4afdb7e7a3da8d6b':
  dca: use defines for subband related constants

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '8563f9887194b07c972c3475d6b51592d77f73f7'

* commit '8563f9887194b07c972c3475d6b51592d77f73f7':
  x86: use emms after ff_int32_to_float_fmul_scalar_sse

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit 'f0f54117c8f206e8045d301c2eb975b26e9f263d'

* commit 'f0f54117c8f206e8045d301c2eb975b26e9f263d':
  checkasm: x86: post commit review fixes

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit 'f4f27e4cf1013c55b2c7df359ce8d58ee922662c'

* commit 'f4f27e4cf1013c55b2c7df359ce8d58ee922662c':
  x86: zero extend the 32-bit length in int32_to_float_fmul_scalar implicitly

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '69a68593ce5684409c3c4dd9a901bfd8b16925b1'

* commit '69a68593ce5684409c3c4dd9a901bfd8b16925b1':
  Remove stray line breaks from avpriv_{report_missing_feature|request_samples}

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '5049f6b772891cdf4030a9d572362efc8f7ae97f'

* commit '5049f6b772891cdf4030a9d572362efc8f7ae97f':
  rtpdec_jpeg: Coalesce redundant error checks

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '2008f76054906e9ff6bf744800af0e5a5bfe61be'

* commit '2008f76054906e9ff6bf744800af0e5a5bfe61be':
  dca: remove unused decode_hf function and quant_d tables

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit 'aebf07075f4244caf591a3af71e5872fe314e87b'

* commit 'aebf07075f4244caf591a3af71e5872fe314e87b':
  dca: change the core to work with integer coefficients.

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '85990140e7302d1e7fcc9fc0eea316178c19fe03'

* commit '85990140e7302d1e7fcc9fc0eea316178c19fe03':
  dca: Add math helpers.

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '489e6add4478b0f5717dbf644234c6f3a3baf02c'

* commit '489e6add4478b0f5717dbf644234c6f3a3baf02c':
  checkasm: add fmtconvert tests

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '568a4323fbde03665b2b23a98068d02b39121812'

* commit '568a4323fbde03665b2b23a98068d02b39121812':
  checkasm: add synth_filter test

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit 'e71b747e9dc56cb84f8a06ec8214d5f3bd98bb6d'

* commit 'e71b747e9dc56cb84f8a06ec8214d5f3bd98bb6d':
  checkasm: add tests for dcadsp

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

Merge commit '9d218d573f8088c606d873e80df572582e6773ef'

* commit '9d218d573f8088c606d873e80df572582e6773ef':
  checkasm: add float comparison util functions

Merged-by: Hendrik Leppkes <h.leppkes@gmail.com>

doc: make apidoc output independent of SRC_PATH

Previously it included the SRC_PATH in every title.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>