avcodec/x86: use function pointers for {put,add}_pixels_clamped

Same behavior as in simple_idct.
This way the best optimized versions available will be used instead.

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

libavutil/opt: fix av_opt_set_channel_layout() to access correct memory address

Signed-off-by: Philip DeCamp <decamp@mit.edu>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

x86/videodsp: add ff_emu_edge_{hfix,hvar}_avx2

~15% faster than sse2.

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

libavformat/ffmetadec: Make unchanged function arguments const

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

libavformat/dv: Make unchanged function arguments const

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

libavformat/avienc: Make unchanged function arguments const

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

libavformat/avidec: Make unchanged function arguments const

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

Merge commit 'c463dfc7e49929a9891884312b23b27d14729c51'

* commit 'c463dfc7e49929a9891884312b23b27d14729c51':
  rtpdec_hevc: Drop a duplicated, nonstandard entry

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

doc/APIchanges: fix typo

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

postproc: add basic deblock filter visualization support

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

ffplay: use av_codec_get_pkt_timebase()

Reviewed-by: Benoit Fouet <benoit.fouet@free.fr>
Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

rtpdec_hevc: Drop a duplicated, nonstandard entry

The RFC spec draft only specifies the "H265" name - there is no
specification saying how to interpret "HEVC" (if such a packet
format is specified it could be an entirely different format).

Since this is a very new standard (still a draft), there is little
need for compatibility with existing, broken implementations. Therefore
remove the extra alias, to avoid the risk of encouraging incorrect
usage.

Intentionally keeping the ff_hevc_dynamic_handler name for the
handler, to use "hevc" consistently as name for the codec instead
of "h265" within the library internals as long as there only is one
single variant in actual use.

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

configure: Check if "-D_POSIX_C_SOURCE=200112 -D_XOPEN_SOURCE=600" is needed for localtime_r()

Fixes build with musl

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

configure: fail if SDL is unavailable and --enable-sdl is specified

Reviewed-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

configure: reindent the --disable-sdl related code

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

configure: support --disable-sdl

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

hevc: move intermediate bidir buffer

Other buffers are already there.

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

hevc: remove unneeded buffer.

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

x86/videodsp: fix warning about discarded 'const' qualifier

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

avcodec/libvpxdec: don't check for formats other than i420 when vp9 decoding is disabled

Should fix ticket #3958.

Reviewed-by: James Zern <jzern@google.com>
Signed-off-by: James Almer <jamrial@gmail.com>

tests/fate-run.sh: Cat .err file in case of error with V>0

This may make fate failures where only the console output is available
easier to analyze

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

Skip decklink_common_c.h when running make checkheaders.

Merge remote-tracking branch 'cus/stable'

* cus/stable:
  ffplay: move pts calculating code to generic decoder
  ffplay: factorize decoder functions to generic decoder
  ffplay: handle fullscreen status on resize event

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

Fix compile error on arm4/arm5 platform

Since these commits
http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=adf8227cf4e7b4fccb2ad88e1e09b6dc00dd00ed
http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=db7f1c7c5a1d37e7f4da64a79a97bea1c4b6e9f8

compilation on arm4/arm5 fails:

libavcodec/libavcodec.so: undefined reference to
`ff_startcode_find_candidate_armv6'

Because libavcodec/arm/Makefile contains
ARMV6-OBJS-$(CONFIG_STARTCODE)         += arm/startcode_armv6.o
function ff_startcode_find_candidate_armv6 is not included for older ARM
archs. The bug was found during automatic buildroot builds:

http://autobuild.buildroot.net/results/ec7/ec71e4f16ee9106747dff5f15999cbd17903e76f//build-end.log
Quote from configure summary:
ARCH                      arm (armv4t)
big-endian                no
runtime cpu detection     yes
ARMv5TE enabled           no
ARMv6 enabled             no
ARMv6T2 enabled           no

http://autobuild.buildroot.net/results/be7/be72eb182eaccf0064a32c9dfc2ac1c0d6555506/build-end.log
ARCH                      arm (armv5te)
big-endian                no
runtime cpu detection     yes
ARMv5TE enabled           yes
ARMv6 enabled             no
ARMv6T2 enabled           no

This patch provides the necessary #if clauses as discussed with Michael:
https://ffmpeg.org/pipermail/ffmpeg-devel/2014-September/163329.html

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffplay: move pts calculating code to generic decoder

Signed-off-by: Marton Balint <cus@passwd.hu>

ffplay: factorize decoder functions to generic decoder

Signed-off-by: Marton Balint <cus@passwd.hu>

ffplay: handle fullscreen status on resize event

In some cases resize events can happen even during fullscreen, so let's pass
the proper flag to SDL_SetVideoMode.

Fixes ticket #3964.

Reported-by: Jeff Dwork <jeff@reggie.com>
Signed-off-by: Marton Balint <cus@passwd.hu>

avformat/riffenc: Filter out "BottomUp" in ff_put_bmp_header()

Fixes Ticket1304

Commit message and extradata size bugfix by commiter
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/webp: fix default palette color 0xff000000 -> 0x00000000

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

avcodec/asvenc: fix AAN scaling

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

avcodec/asvenc: Fix integer overflow in level

Warn if the qscale is too low for the input data and clip levels to
minimize artifacts

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

x86/vp9: add AVX and AVX2 MC

Roughly 25% faster MC than ssse3 for blocksizes 32 and 64.

Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

ffmpeg: replace impossible dts/pts combinations by a reasonable guess instead of hard failing

Fixes stream copy from tserror270.ts

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

doc/decoders: adding documentation for lavc/dvdsubdec.c option "forced_subs_only"

Signed-off-by: Nicholas Robbins <nickrobbins@yahoo.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libavcodec/dvdsubdec: Add option forced_subs_only to only decode forced subtitle frames.

Signed-off-by: Nicholas Robbins <nickrobbins@yahoo.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avformat/asfenc: Make asf_write_indexs index argument const

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

avformat/adtsenc: buf isnt changed in adts_decode_extradata(), make it const

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

avcodec/4xm: Make src of decode_p_block() const

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

avcodec/snow: Make block argumrnt of ff_snow_pred_block() const

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

postproc/postprocess_template: mark unchanged function arguments const

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

postprocess: prefetch* dont change anything, thus their arguments should be const

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

Revert "configure: Refactor setting of feature test macro _XOPEN_SOURCE"

This fixes build on netbsd and probably other bsd based Platforms

This reverts commit 56b8d106769c5c5f9174dcc426f1ee2f991428b1.

Add missing entry for maintainer of rtpenc_hevc.*

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

avformat/avidec: ensure that palette does not contain the BottomUp info.

Considering the palette is located at the end of extradata may be flawed
when the extradata contains the palette followed by the BottomUp field.
When the BottomUp field is present, exclude it from the palette.
Fixes part of ticket #1304

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

avformat/img2: remove useless 'pix' duplicated entry.

The second entry will never be selected, so let's save some bytes in the
library.

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

rtpenc_h263_rfc2190: avoid misleading error output

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

avdevice/version: bump minor for the decklink input

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

avdevice: add decklink input support

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

avdevice/decklink: move general code of decklink encoder to common file

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

configure: Refactor setting of feature test macro _XOPEN_SOURCE

Some C libraries, eg glibc, uclibc, and musl, uses feature test macros
to expose definitions conforming to the standards ISO C, POSIX and
extensions. According to which feature test macros are defined by the
user or the compiler, a header file, eg <features.h>, used by these
libraries internally defines various other macros.

glibc and uclibc also defines release test macros, eg __GLIBC__ and
__UCLIBC__ in <features.h>. musl does not have (and does not want) a
macro __MUSL__. Therefore it is not possible to check for the musl
library.

However, building FFmpeg with musl needs the feature test macro
_XOPEN_SOURCE=600 to be defined.

Signed-off-by: Jörg Krause <jkrause@posteo.de>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

lavu/bprint: add const to av_bprint_is_complete() argument.

avformat/assenc: mux all extradata at once

Before this commit, the code was muxing up to the 2nd line after
"[Events]" (assuming it to be the "Format:" line). The remaining are
generally "Comment:" directives which can stay in that place. mkvextract
behaves that way so it seems there is no reason for that extra
complexity.

avformat/assenc: return correct error code

avcodec/webvttdec: add some memory checks

avcodec/textdec: add some memory checks

avcodec/subviewerdec: add some memory checks

avcodec/srtdec: add some memory checks

avcodec/samidec: add some memory checks

avcodec/realtextdec: add some memory checks

avcodec/mpl2dec: add some memory checks

avcodec/movtextdec: add some memory checks

avcodec/microdvddec: add some memory checks

avcodec/jacosubdec: add some memory checks

avcodec/ass: add ff_ass_add_rect_bprint() helper

rtpdec_hevc: do not print an error message if the received packet has a valid header but lacks additional bytes as payload

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

avformat/mp3dec: avoid early EOF with concatenated gapless mp3s

Consider a file created with something like:

    cat file1.mp3 file2.mp3 > result.mp3

Then if file2.mp3 has gapless information, result.mp3 would stop playing
something in the middle. This happens because the gapless info directs
the decoder to discard all samples after a certain position. To make
matters worse, the gapless info of file2.mp3 will be used when playing
the file1.mp3 part, because the gapless info is located at the end of
the file.

While handling concatenated gapless files correctly would be insane and
a lot of effort (especially without scanning the whole file on opening),
it's easy to prevent at least early EOF. Playback will happen to work,
even if it's slightly broken.

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

mlpdec: support TrueHD streams with an Atmos substream

The fourth substream is being discarded, since its not raw audio data,
but an encoded Atmos stream which needs a specialized decoder.

Fixes decoding of the true hd stream from Transformers\ -\ Age\ of\ Extinction\ 2014\ 1080P-003.mkv

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

mlpdec: support major sync headers with optional extension blocks

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

rtpenc: HEVC/H.265 support

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

avfilter/vf_scale: Allow chroma samples to be above and to the left of luma samples

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

swscale: Allow chroma samples to be above and to the left of luma samples

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

Merge commit '103391ca90b2f7c56ae756d76c76f7c3dfa28dd4'

* commit '103391ca90b2f7c56ae756d76c76f7c3dfa28dd4':
  dca: Remove some commented-out cruft

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

avcodec/alacenc: Remove unused variable

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

alacenc: remove unneeded masking

The extra bits have already been masked, so this was not doing anything.

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

alacenc: fix incorrect buffer use

The issue lies in actually dead code ("for now it's not used").

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

avformat/mp3dec: fix gapless audio support

The code already had skipping of initial padding, but discarding
trailing frame padding was missing.

This is somewhat questionable, because it will make the decoder discard
any data after the declared file size in the LAME header. But note that
skipping full frames at the end of the stream is required. Encoders
actually create such files.

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

dca: Remove some commented-out cruft

avfilter/idet: typo fix: PROGRSSIVE -> PROGRESSIVE

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

x86/me_cmp: port mmxext vsad functions to yasm

Also add mmxext versions of vsad8 and vsad_intra8, and sse2 versions of
vsad16 and vsad_intra16.
Since vsad8 and vsad16 are not bitexact, they are accordingly marked as
approximate.

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

avcodec/dvdsubdec: fix VD/SD identifier name

Found-by: Nicholas Robbins <nickrobbins-at-yahoo.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avformat/os_support: Add _DEFAULT_SOURCE to hide warning about _SVID_SOURCE depreciation

As suggested, this just adds _DEFAULT_SOURCE while preserving
_SVID_SOURCE for whatever old systems still need it.

Signed-off-by: Daniel Bomar <dbdaniel42@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

tests/utils: make arguments const that arent changed

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

avcodec/dv_profile: deprecate internal function that shouldn't be public

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

postproc/postprocess: Avoid some ifs in do_a_deblock_C()

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

libavcodec/webp: treat out-of-bound palette index as translucent black

See https://code.google.com/p/webp/issues/detail?id=206
for a description of the problem/fix.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This patch makes the decoder follow the recommendation of the spec.
There is some disagreement (see "[FFmpeg-devel] [PATCH]: libavcodec/webp")
about what would be best to be written in the spec, so in case the spec
is changed again, this potentially would need to be amended or reverted

avcodec/libilbc: support for latest git of libilbc

in the latest git commits of libilbc developers removed WebRtc_xxx typedefs

This commit uses int types instead,
it's safe to apply also for previous versions since
WebRtc_Word16 was always a typedef of int16_t and
WebRtc_UWord16 a typedef of uint16_t

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

postproc: Replace CLIP by av_clip_uint8

drop "#define CLIP av_clip_uint8"

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

postproc/postprocess: remove redundant casts

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

postproc/postprocess: Use FF_ARRAY_ELEMS() in pp_free_context() instead of hard-coding their size

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

postproc/postprocess: Remove unused argument from reallocAlign()

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

Merge branch 'postprocwork'

* postprocwork:
  postproc/postprocess: use av_strtok()
  postprocess: make some variables in pp_get_mode_by_name_and_quality() const
  postproc: simplify forwarding return codes
  libpostproc/postprocess: avoid some if()
  fate: add fate-filter-pp1

This is merged instead of just fast forward pushed due to a bug in the git hook
which does not allow commits to change filter-video.mak except merge commits.
filter-video.mak contains a few tabs, which are needed due to Makefile syntax

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

postproc/postprocess: use av_strtok()

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

postprocess: make some variables in pp_get_mode_by_name_and_quality() const

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

postproc: simplify forwarding return codes

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

libpostproc/postprocess: avoid some if()

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

fate: add fate-filter-pp1

This tests a few more filters with forced quantizers

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

x86/me_cmp: combine sad functions into a single macro

No point in having the sad8 functions separate now that the loop is no
longer unrolled.

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

Merge commit '2bb2c2bd75e5f4b28a945511cda77e0a1a44c758'

* commit '2bb2c2bd75e5f4b28a945511cda77e0a1a44c758':
  rtpenc_chain: Pass the initial time_base hint on to the chained muxer

See: [FFmpeg-devel] [PATCH] avformat/rtpenc_chain: Set timebase
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '3f2c70355ab722bc9f741bd3ed8224c7cfb62379'

* commit '3f2c70355ab722bc9f741bd3ed8224c7cfb62379':
  configure: Use the right variables in check_host_cpp

See: 453944cc77b9ea27c74011eb7c0e20c887b321cf
Merged-by: Michael Niedermayer <michaelni@gmx.at>

vf_deshake: rename Transform.vector to Transform.vec to avoid compiler confusion

The token 'vector' is a keyword in the Vector/SIMD Multimedia Extension data types and thus should not be used as a variable name.

This fixes building on powerpc/ppc64el.

Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>