Merge remote-tracking branch 'cigaes/master'

* cigaes/master:
  ffmpeg: use thread message API.
  lavu: add thread message API.
  compat/w32pthreads: add return value to pthread_cond_init().

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

configure: uClibc native pthread requires -ldl

In the configure script, add a pthread detection with
-ldl added to cflags, because uClibc requires -ldl to
link with native libpthread.

Tested with a custom ARM toolchain with uClibc 0.9.33.2 and gcc 4.6.3.

Signed-off-by: Neil Armstrong <narmstrong@neotion.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

x86: hpeldsp: propagate changes across codecs

Some codecs still use mmx versions, so have them use the versions
with newer instruction sets.

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

ffmpeg: use thread message API.

lavu: add thread message API.

compat/w32pthreads: add return value to pthread_cond_init().

avformat/mov: allow seeking back to the begin even if nothing is marked as keyframe

Fixes Ticket 3663

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

doc/APIchanges: fix typo in date

update APIchanges & version for c37d179

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

avformat: add format_probesize to allow tuning the maximum amount of bytes to identify the filetype

Currently probesize is cliped at 1mb before being used for format detection.
Alternatively this cliping could be removed but this would then tie various
things like stream analysis to the file detection.

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

avcodec/sgirledec: fix () in RBG323_TO_BGR8() macro

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

avcodec/rv34: Fix () in GET_PTS_DIFF() macro

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

avcodec/psymodel: Fix () in AAC_CUTOFF() macro

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

avcodec/proresenc_anatoliy: Fix () in macros

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

Revert "x86: dsputilenc: convert ff_sse{8, 16}_mmx() to yasm"

This reverts commit ad733089b024e4a3ff8f024d247a032f79a50ac8.

breaks with --disable-yasm
revert requested by: Christophe Gisquet <christophe.gisquet@gmail.com>

avcodec/msvideo1enc: Fix () in MKRGB555() macro

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

avcodec/opus: Fix () in ROUND_MULL() macro

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

avcodec/opus_celt: Fix () in CELT_PVQ_V macro

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

avcodec/pngdec: fix () in OP_* macros

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

x86: dsputilenc: convert ff_sse{8, 16}_mmx() to yasm

Signed-off-by: Timothy Gu <timothygu99@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

configure: try to fix the disappearing "r" bug in fate

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

Merge commit 'e0d01dc7d7fc3ce4c23f286a10870e9599c8b8b9'

* commit 'e0d01dc7d7fc3ce4c23f286a10870e9599c8b8b9':
  smoothstream: check malloc calls

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

avcodec/webvttenc: do not use EOVERFLOW.

According to doc/errno.txt EOVERFLOW is not available everywhere. The
use of -1 is consistent with avcodec/srtenc.c.

avcodec/mss4: Fix () in MKVAL() macro

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

avcodec/mss34dsp: fix () in SOP* macros

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

avcodec/mlpdec: fix () in MSB_MASK() macro

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

avcodec/ivi_dsp: add some missing () to macros

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

avcodec/hevc: fix () in macros

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

avcodec/h264_slice: fix macro ()

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

avcodec/h264: fix () in macros

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

avcodec/golomb-test: fix () in EXTEND() macro

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

avcodec/aac: fix () in IS_CODEBOOK_UNSIGNED macro

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

avcodec/webvttenc: add webvtt encoder

Based off the srt encoder. The following features are unimplemented:

- fonts, colors, sizes
- alignment and positioning

The rest works well. For example, use ffmpeg to convert subtitles into the .vtt format:

  ffmpeg -i input.srt output.vtt

Signed-off-by: Aman Gupta <ffmpeg@tmm1.net>
Signed-off-by: Clément Bœsch <u@pkh.me>

x86/dsputilenc: make the SUM_ABS_DCTELEM macro more readable

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

avcodec/opusdec: check alignment, misalignment could lead to crashes with avx

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

x86/dsputilenc: port sum_abs_dctelem functions to yasm

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

avcodec/asvenc: dont use a negative global_quality

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

avcodec/aacenc: dont use global quality if its negative

Some applications used a negative value as default for "not set"

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

x86: hpeldsp: implement SSSE3 version of _xy2

Loading pb_1 rather than pw_8192 was benchmarked to be more efficient.
Loading of the 2 yields no advantage. Loading of one saves ~11 cycles.

decicycles count:
put8:  3223(mmx)    -> 2387
avg8:  2863(mmxext) -> 2125
put16: 4356(sse2)   -> 3553
avg16: 4481(sse2)   -> 3513

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

Merge commit '8692e6284f5169257a537c8fc25addf32fc67c87'

* commit '8692e6284f5169257a537c8fc25addf32fc67c87':
  rdt: check malloc calls

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

avformat/vobsub: do not create empty streams.

Fixes NULL dereference in vobsub_read_packet() on vobsub->q[i]->subs.

Fixes ticket #3669.

smoothstream: check malloc calls

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

rdt: check malloc calls

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

Add metadata injection to blackdetect

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

x86: hpeldsp: implement SSE2 put_pixels16_xy2

This is obviously equivalent to the avg version, without the avg.

3223(mmx) -> 2006(sse2)

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

x86: hpeldsp: implement SSE2 versions

Those are mostly used in codecs older than H.264, eg MPEG-2.

put16 versions:
      mmx  mmx2  sse2
x2:  1888  1185   552
y2:  1778  1092   510

avg16 xy2: 3509(mmx2) -> 2169(sse2)

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

MAINTAINERS: remove myself as website maintainer

I never implemented the plans I had for the site, and I prefer to work
in other areas of the project.

Signed-off-by: Lou Logan <lou@lrcd.com>

x86/hevc_deblock: improve chroma functions register allocation

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

ffplay: support rotated video.

with -f lavfi -i testsrc=s=hd1080 as source:
  rotate=90*PI/180  vs transpose=clock:   42fps -> 64fps
  rotate=180*PI/180 vs vflip,hflip:       75fps -> 77fps
  rotate=270*PI/180 vs transpose=cclock:  43fps -> 63fps

avformat/mov: use av_display_rotation_get() for rotate metadata.

avutil/display: use hypot().

x86/dsputil: fix argument declaration in vector_clipf

Should fix fate failures in msvc x86_64

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

x86/dsputil: fix VECTOR_CLIP_INT32 macro

The inline loop was incrementing and using the value of %%i
the wrong way.

Disassembly of ff_vector_clip_int32_sse2 before and after
this patch:

    movdqa (%rdx),%xmm0      |  movdqa (%rdx),%xmm0
    movdqa 0x10(%rdx),%xmm1  |  movdqa 0x10(%rdx),%xmm1
    movdqa 0x20(%rdx),%xmm2  |  movdqa 0x20(%rdx),%xmm2
    movdqa 0x30(%rdx),%xmm3  |  movdqa 0x30(%rdx),%xmm3
[...]                        |
    movdqa %xmm0,(%rcx)      |  movdqa %xmm0,(%rcx)
    movdqa %xmm1,0x10(%rcx)  |  movdqa %xmm1,0x10(%rcx)
    movdqa %xmm2,0x20(%rcx)  |  movdqa %xmm2,0x20(%rcx)
    movdqa %xmm3,0x30(%rcx)  |  movdqa %xmm3,0x30(%rcx)
    movdqa (%rdx),%xmm0      |  movdqa 0x40(%rdx),%xmm0
    movdqa 0x20(%rdx),%xmm1  |  movdqa 0x50(%rdx),%xmm1
    movdqa 0x40(%rdx),%xmm2  |  movdqa 0x60(%rdx),%xmm2
    movdqa 0x60(%rdx),%xmm3  |  movdqa 0x70(%rdx),%xmm3
[...]                        |
    movdqa %xmm0,(%rcx)      |  movdqa %xmm0,0x40(%rcx)
    movdqa %xmm1,0x20(%rcx)  |  movdqa %xmm1,0x50(%rcx)
    movdqa %xmm2,0x40(%rcx)  |  movdqa %xmm2,0x60(%rcx)
    movdqa %xmm3,0x60(%rcx)  |  movdqa %xmm3,0x70(%rcx)
    add    $0x80,%rdx        |  add    $0x80,%rdx
    add    $0x80,%rcx        |  add    $0x80,%rcx

Other versions were unaffected.

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

avcodec/dcadec: dont use hex float, msvc doesnt support it

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

avcodec/dcadec: use a constant instead of assuming every compiler can optimize pow(2,-15)

Idea-by: nevcairiel and wm4
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/dcadec: make dca_dmix_code() 650% faster

This effectively replaces the ldexpf() function call by a multiplication
with a constant.

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

Merge remote-tracking branch 'cehoyos/master'

* cehoyos/master:
  Refuse to mux H.264 with fourcc H264 into avi without startcode.

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

Merge commit '21f68c2489cba2a1a4a41d0c5c828266e6162800'

* commit '21f68c2489cba2a1a4a41d0c5c828266e6162800':
  avcodec: bump version after rotation api

Conflicts:
libavcodec/version.h

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

dcadec: fix xxch_dmix_coeff and xxch_dmix_sf after merge

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

Merge commit '6b7b8585d73548f08b98959df39d80b97677662f'

* commit '6b7b8585d73548f08b98959df39d80b97677662f':
  dca: Convert dca_dmixtable to integers

Conflicts:
libavcodec/dcadec.c

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

avcodec: bump version after rotation api

Signed-off-by: Kostya Shishkov <kostya.shishkov@gmail.com>

Refuse to mux H.264 with fourcc H264 into avi without startcode.

Fixes ticket #3638.

dca: Convert dca_dmixtable to integers

Also include zero in the table, eliminating a special case in the
decoder.

Signed-off-by: Niels Möller <nisse@southpole.se>
Signed-off-by: Anton Khirnov <anton@khirnov.net>

x86/diracdsp: mark all functions as yasm

No inline asm dirac code remains in the tree, so replace every relevant check.
This also moves all the dirac functions from dsputil_mmx.c to diracdsp_mmx.c

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

avformat/mov: Dont blindly trust the stream duration in seting chapter times

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

avformat/utils: Check end time in avpriv_new_chapter()

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

avformat/movenc: Check that packet duration is valid in ff_mov_write_packet()

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

x86/dsputil: port ff_vector_clipf_sse to yasm

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

Fix the mime types for MPEG and MPEG-TS formats

As per the RFCs:
http://tools.ietf.org/html/rfc3555#page-38
http://tools.ietf.org/html/rfc3003

x86: hpeldsp: avg_pixels_xy2 for mmx2&3dnow

This is a port of the inline assembly of the mmx version to use the
pavg(us|)b instruction.

        8    16
mmx   1498  4355
mmx2  1242  3509

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

x86: hpeldsp: mark _xy2 versions as approximate

Currently, only the mmx version is bitexact, the others (mmxext and
3dnow) are not, in spite of their naming.

Therefore, make their name more obvious. Also restore a comment that
was removed in 71155d7b.

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

x86: hpeldsp: kill hpeldsp_mmx.c

before:
1987 decicycles in 8_x2, 262121 runs, 23 skips

after:
1902 decicycles in 8_x2, 262112 runs, 32 skips

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

maintainers: adding myself as a maintainer to webm dash

adding myself as a maintainer to webm dash (part of matroskaenc.c)

Signed-off-by: Vignesh Venkatasubramanian <vigneshv@google.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libavformat/matroskaenc: Add WebM DASH support

WebM DASH specification [1] requires the Clusters and Cues to be output in a
specific way. Adding a flag to matroskaenc that will enable support for
creating WebM/Mkv files conforming to the WebM DASH specification.

[1] http://wiki.webmproject.org/adaptive-streaming/webm-dash-specification

Signed-off-by: Vignesh Venkatasubramanian <vigneshv@google.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge remote-tracking branch 'cehoyos/master'

* cehoyos/master:
  Autodetect Asterisk raw pcm with extension "sln".

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

Autodetect Asterisk raw pcm with extension "sln".

avdevice/v4l2: Use av_malloc_array()

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

avcodec/proresdec2: Use av_malloc_array()

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

avcodec/pgssubdec: Use av_malloc(z)_array()

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

avcodec/vdpau_internal: move "struct vdpau_picture_context" up to avoid "#else"

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

Fix compilation error around struct visibility when VDPAU disabled.

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

x86/dsputil: port ff_put_signed_pixels_clamped_mmx to yasm

Also add an SSE2 version

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

x86/dsputil: port clear_block functions to yasm

Signed-off-by: James Almer <jamrial@gmail.com>
Reviewed-by: Christophe Gisquet <christophe.gisquet@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Don't overwrite codec_id with 0 in ff_mp4_read_dec_config_descr().

Fixes ticket #3660.

sws: document color range >8bit oddity

ffmpeg: use av_stream_get_end_pts()

Simplifies code and should correct timing values when -*sync drop is used

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

avformat: add av_stream_get_end_pts()

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

avcodec/utils: add error message for the recode_subtitle() without iconv case

Based on the assert command which was removed in ef0c466a0fef4657d3f46cc9adea97e76a8e4faa

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

Replace assert with AVERROR when recode_subtitle called w/o ICONV.

recode_subtitle() is called implicitly by avformat_find_stream_info().
As such, clients which disable ICONV always crash if a file contains
subtitles; even if they don't care about them.

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

avcodec/atrac3plusdec: Use av_mallocz_array()

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

avcodec/atrac3: Use av_mallocz_array()

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

ffprobe: Use av_malloc_array()

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

avcodec/x86/hpeldsp_init: remove redundant if()

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

hpeldsp: fix build without inline asm

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

x86: kill fpel_mmx.c

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

avcodec/libvorbisenc: dont add the duration to AV_NOPTS_VALUE

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

Fix compilation errors when deprecated features are disabled.

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

avcodec/h264_picture: Fix build without error resilience

Found-by: Dale Curtis <dalecurtis@chromium.org>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fix build with --enable-ladspa --disable-frei0r.

https://bugs.gentoo.org/show_bug.cgi?id=510516

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

configure: arm: do not enable fast-unaligned if it is explicitly disabled via commandline

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