avformat/mxfenc: Support storing signal standard for D10 muxing

Reviewed-by: tim nicholson <nichot20@yahoo.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/mips: Restructure as per avutil/mips/generic_macros_msa.h

This patch modifies H264 loopfilter, weighted & bi-weighted prediction MIPS-SIMD optimized code according to improved version of generic macros.
Also there are minor code alignment changes.

Overall, this patch is just upgrading the code with styling changes and will bring it in sync with MIPS-SIMD optimized latest codebase at our end.

Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avutil/mips: Restructure of generic macros

This patch includes restructuring of existing macros and addition of more generic macros.

This change was necessary to avoid repeated review comments in remaining patches which we were about to submit.
Also this patch reduces number of code lines due to maximum use of generic macros, allows better code alignment & readability etc.

These modifications in commonly used .libavutil/mips/generic_macros_msa.h. impacts the already accepted code, hence re-submitting it in 2/4,3/4 & 4/4.
Overall, this patch set is just upgrading the code with styling changes and will bring it in sync with MIPS-SIMD optimized latest codebase at our end.

Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

network: Move variable declaration under an #if

Avoids an unused variable warning.

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

avcodec/ffv1dec: Fix skip_alpha

Fixes Ticket4322

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

x86inc: Clear __SECT__

This commit silences warning(s) like:

    libavcodec/x86/fft.asm:93: warning: section flags ignored on section
    redeclaration

The cause of this warning is that because `struc` and `endstruc` attempts to
revert to the previous section state [1]. The section state is stored in the
macro __SECT__, defined by x86inc.asm to be `.note.GNU-stack ...`, through the
`SECTION` directive [2].  Thus, the `.note.GNU-stack` section is defined twice
(once in x86inc.asm, once during `endstruc`), causing the warning.

That is the first part of the commit: using the primitive `[section]` format
for .note.GNU-stack etc., which does not update `__SECT__` [2].

That fixes only half of the problem. Even without any `SECTION` directives,
`__SECT__` is predefined as `.text`, which conflicting with the later
`SECTION_TEXT` (which expands to `.text align=16`).

[1]: http://www.nasm.us/doc/nasmdoc6.html#section-6.4
[2]: http://www.nasm.us/doc/nasmdoc6.html#section-6.3

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

Revert "Move struc FFTContext below SECTION_RODATA"

This reverts commit 599888a48040f4086c9086aae958b7231b37afcc.

The commit does not silence the warning on ELF-based systems, and will be
fixed in the subsequent commit.

Conflicts:
libavcodec/x86/fft_mmx.asm

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

Merge commit '57cde2b180fcec0eaf60aad65f436ab6420546f5'

* commit '57cde2b180fcec0eaf60aad65f436ab6420546f5':
  lavf: move TLS-related ifdeffery to library specific files

Conflicts:
libavformat/network.c
libavformat/tls.h
libavformat/tls_openssl.c

See: a9f1d584e53fb39d983201585cb136986a85cac8
Merged-by: Michael Niedermayer <michaelni@gmx.at>

lavf: move TLS-related ifdeffery to library specific files

There is no need to have this mess in network.c.

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

Merge commit 'd8ffb2055f0e0fcb5d025bab72eb19c2a886c125'

* commit 'd8ffb2055f0e0fcb5d025bab72eb19c2a886c125':
  lavf: split tls.c

Conflicts:
libavformat/tls.c
libavformat/tls_gnutls.c
libavformat/tls_openssl.c

See: 4a006b9eb7e7d736fd00e6045b5612978ef6404b
Merged-by: Michael Niedermayer <michaelni@gmx.at>

lavf: split tls.c

Move the OpenSSL and GnuTLS implementations to their own files. Other
than the connection code (including options) and some boilerplate, no
code is actually shared.

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

avformat/mpsubdec: Use double instead of float for timestamp calculations

This provides higher precission

Reviewed-by: Clément Bœsch <u@pkh.me>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avformat/mp3dec: Ensure the buffer is large enough to avoid seeks from the first frame search

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

avformat/mp3: skip junk at the beginning of mp3 files

Apparently it can happen that a mp3 file has junk data between id3 tag
and actual mp3 data. Skip this to avoid outputting nonsense timestamps.
(Two packets had the same timestamps, because the mp3 parser failed to
compute a frame duration.)

In this case, the junk consisted of 1044 bytes of zero, which
incidentally is the same size as normal mp3 frames in this stream. I
suspect the mp3 was edited with some tool which wiped the Xing/LAME
headers. Data near the end of the file suggests it was encoded with
"LAME3.97", but the normal Xing/LAME headers are missing. So this could
be "normal". mpg123 also attempts to skip at least 64KB of junk data by
scanning for headers.

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

avformat/mpsubdec: change multipler to int, it only stores 1 and 100

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

avcodec: loongson3 optimized h264dsp weighted mc with mmi

Signed-off-by: ZhouXiaoyong <zhouxiaoyong@loongson.cn>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

doc/build_system: Document GEN, text taken from the webpage

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

fate: Update exif tests after 09ec31ce3aab41407965e7a95b7d145702434c26

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

ffprobe: print stream_index with frame data

This makes it easier to relate frame data to its associated stream.

x86/cpu: add AV_CPU_FLAG_AVXSLOW flag

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

avfilter/x86/vf_hqdn3d: Fix register types

Fixes Ticket4301

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

avfilter/vf_hqdn3d: Initialize the whole LUT

With bps > 8 more than 255..255 are used
The initialized table content is left unchanged,
But it could also be adjusted for the slight difference of
the maximum

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

avcodec/nvenc: Fix typo: 1204 -> 1024

Fixes Ticket4508

Signed-off-by: Philip Langdale <philipl@overt.org>

avformat/movenc: Allow muxing mp3 with samplerate < 16khz if the user sets strict to -1

Fixes Ticket4267

Approved-by: Baptiste Coudurier <baptiste.coudurier@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

ffmpeg: do not print misleading recommanditions on 1pass vpx encoding

Fixes part of Ticket4295

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

Merge commit 'cf1f3d837e1266034a487de5b575bd76426c6b10'

* commit 'cf1f3d837e1266034a487de5b575bd76426c6b10':
  doc: Fix spelling of 'Transmission'

See: 9898bd9a82ad16f7ef2dcc26542827ff92255821
Merged-by: Michael Niedermayer <michaelni@gmx.at>

lavf: move TLS-related ifdeffery to library specific files

There is no need to have this mess in network.c.

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

lavf: split tls.c

Move the OpenSSL and GnuTLS implementations to their own files. Other
than the connection code (including options) and some boilerplate, no
code is actually shared.

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

avformat/matroskaenc: Avoid floats in default duration calculation

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

avformat/ircamenc: Avoid floats

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

avutil: Add av_q2intfloat()

This function allows writing AVRationals as IEEE floats without the need
of platform dependant float operations

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

avformat/movenc: Avoid floats & float rounding in tmcd nb_frames calculation

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

avformat/asfdec: Avoid float usage in duration calculation

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

doc: Fix spelling of 'Transmission'

Signed-off-by: Sebastian Ramacher <sramacher@debian.org>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>

avformat/mpegts: Avoid float in bitrate calculation

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

avcodec/h264: Fix HWACCEL_MAX for D3D11

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

avcodec/hevc: Fix HWACCEL_MAX for D3D11

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

avcodec/dxva2_h264: Fix "may be used uninitialized" warnings

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

avcodec/vdpau: Re-factor pre-hwaccel helper functions into separate header

h264.h and hevc.h are mutually exclusive due to defining some of the same
names. As such, we need to avoid forcing h264.h to be included if we want
hevc decode acceleration to be possible.

However, some of the pre-hwaccel helper functions need h264.h. To avoid
messy collisions, let's move the declaration of all those helpers to
a separate header which we will exclude for the hevc support (which will
be hwaccel-only).

Signed-off-by: Philip Langdale <philipl@overt.org>

avcodec/dxva2: Fix "may be used uninitialized" warnings

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

configure: Check if ID3D11VideoDecoder exists, not just the header to enable d3d11va

Fixes build with "mingw-w64-x86-64-dev 3.2.0-2"

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

avfilter/x86/vf_fspp: Fix invalid combination of opcode and operands

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

lavf/dnxhd: Autodetect more files that can be decoded.

mov: reject zero bytes_per_frame with non-zero samples_per_frame

In this case the mov demuxer can return a large number of empty packets.

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

configure: d3d11va is auto-detected like the others

Fix the help text accordingly

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

lavf/riff: Add 0x729A as TwoCC for G.729.

Reported by forum user muchuan.

avcodec/dxva2: Fix build without D3D11

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

Merge commit 'd8039ef8d221ea273aa4f1e62e5df21bf618c772'

* commit 'd8039ef8d221ea273aa4f1e62e5df21bf618c772':
  D3D11va: add a Direct3D11 video decoder similar to DXVA2

Conflicts:
Changelog
configure
libavcodec/Makefile
libavcodec/allcodecs.c
libavcodec/dxva2_vc1.c
libavcodec/version.h
libavutil/pixdesc.c
libavutil/pixfmt.h
libavutil/version.h

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

Improve upon dynamic arrays- movtext subtitles

Signed-off-by: Niklesh <niklesh.lalwani@iitb.ac.in>

libshine: fix support for shine 3.0

shine_encode_buffer expects written to be an int pointer, while the
previous shine_encode_frame expected it to be a long pointer.

Thus encoding with libshine currently always fails with
"internal buffer too small", because a negative return value of
shine_encode_buffer is interpreted as a very large long value.

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

avcodec/mjpegenc_common: Use ff_mpv_reallocate_putbitbuffer()

Fixes assertion failure
Fixes Ticket4396

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

avcodec/mpegvideo: Factor ff_mpv_reallocate_putbitbuffer() out

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

avformat/rmenc: Avoid floats in duration calculation

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

avformat/mpegenc: Replace *0.7 by *7/10

This avoids another float computation, avoiding a potential source of
rounding issues

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

avformat/mpegenc: Do not use floats for vcd_padding_bitrate

This reduces the risk for rounding differences.

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

ffmpeg_opt: Set the video VBV parameters only for the video stream from -target

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

D3D11va: add a Direct3D11 video decoder similar to DXVA2

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

avcodec/put_bits: Assert that size in set_put_bits_buffer_size() does not cause integer overflows

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

avcodec/bitstream: Assert that there is enough space left in avpriv_copy_bits()

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

avcodec/put_bits: Assert that there is enough space left in skip_put_bytes()

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

avcodec/mpegvideo_enc: Update the buffer size as more slices are merged

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

avcodec/put_bits: Update size_in_bits in set_put_bits_buffer_size()

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

avformat/mov: Mark avio context of decompressed atoms as seekable

Fixes Ticket4329

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

avcodec/libutvideoenc: Fix memleak

Fixes: CID1257657
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

libdcadec: search for frames that start late in a packet

Based on commit 4ae15605f6298d8369a1f1374596b52a3839c703

Reviewed-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/ituh263enc: Pass PutBitContext into h263p_encode_umotion() instead of MpegEncContext

This avoids the need to dereference MpegEncContext->pb if it is
already available outside h263p_encode_umotion()

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

avcodec: Pass PutBitContext into ff_h263_encode_motion() instead of MpegEncContext

This avoids the need to dereference MpegEncContext->pb if it is
already available outside ff_h263_encode_motion()

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

avcodec/h263: Remove unused argument of h263_get_motion_length()

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

avcodec/mpeg4video: Reorder operations to reduce accesses to err_recognition

About 9 cpu cycle faster mpeg4_decode_mb()

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

mov: always check avio_read return value

If avio_read fails, the buffer can contain uninitialized data.

This fixes 'Conditional jump or move depends on uninitialised value(s)'
valgrind warnings.

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

mov: fix two memleaks

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

mov: abort on EOF in ff_mov_read_chan

Otherwise the loop can take a lot of time if num_descr is very large.

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

avformat/movenc: Remove float use from rgb_to_yuv()

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

avformat/swfenc: Remove another unneeded float computation

fewer chances for rounding differences between platforms is better

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

avformat/swfenc: Remove unneeded floats

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

avformat/avienc: Avoid float usage

reduces risk for rounding differences between platforms

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

avformat/4xm: Change fps to AVRational

This fixes the video timebase as well

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

avcodec/y41pdec: Avoid using float for size test

Floats are not bitexact

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

libwebp: simplify AVCodec.close functions

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

configure: simplify libwebp checks

The libwebpmux pkg-config file already has an explicit dependecy on libwebp >= 0.2.0.
Also remove the warning and silently disable the anim encoder when libwebpmux is not new enough.
This is more in line with other library components, like libvpx-vp9

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

nutdec: fix various memleaks on failure

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

avcodec/dcadec: Search and decode frame in case it starts later in a packet

This fixes decoding the first frame of some dts files

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

avformat/wavdec: Increase dts packet threshold to fix more misdetections

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

avformat/wavdec: Increase probe_packets limit

Fixes DTS detection of b2429e5ba9.dts

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

nutdec: always check the get_str return value

If it fails, the buffers can be (partially) uninitialized.

This fixes 'Conditional jump or move depends on uninitialised value(s)'
valgrind warnings.

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

libwebp: use a separate AVClass for each encoder

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

libwebpenc_common: add header guards

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

libwebp: remove unneeded defines

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

WebP encoder: use WebPAnimEncoder API when available.

WebPAnimEncoder API is a combination of encoder (WebPEncoder) and muxer
(WebPMux). It performs several optimizations to make it more efficient
than the combination of WebPEncode() and native ffmpeg muxer.

When WebPAnimEncoder API is used:
- In the encoder layer: we use WebPAnimEncoderAdd() instead of
  WebPEncode().
- The muxer layer: works like a raw muxer.

On the other hand, when WebPAnimEncoder API isn't available, the old code is
used as it is:
- In the codec layer: WebPEncode is used to encode each frame
- In the muxer layer:  ffmpeg muxer is used

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

avformat/nutdec: Check X in 2nd branch of index reading

Prevents read of uninitialized variable

Based on patch by: Andreas Cadhalpun <andreas.cadhalpun@googlemail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '4e17946f10d39eec6cc03fb249ae8147373141b6'

* commit '4e17946f10d39eec6cc03fb249ae8147373141b6':
  mpegvideo: Rework various functions not to use MpegEncContext directly

Conflicts:
libavcodec/mpegvideo.c

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

Merge commit 'a3f4c930ac3f49f47b6e6ffda925d0dcf80320e2'

* commit 'a3f4c930ac3f49f47b6e6ffda925d0dcf80320e2':
  mpegvideo: Have ff_mpeg_ref_picture use AVCodecContext directly

Conflicts:
libavcodec/mpegvideo.c

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

Merge commit 'd528045558825f01472e9bee873f60c98d661e53'

* commit 'd528045558825f01472e9bee873f60c98d661e53':
  mpegvideo: Have ff_mpeg_unref_picture use AVCodecContext directly

Conflicts:
libavcodec/mpegvideo.c
libavcodec/mpegvideo_enc.c

The memset is left in place

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

Merge commit '6f54dc43cee6b2f5d183acf98b32a3cf8be4a4fc'

* commit '6f54dc43cee6b2f5d183acf98b32a3cf8be4a4fc':
  mpegvideo: Drop stream_codec_tag

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

Merge commit '9c1db92ad372d4cd69e0490e691c56e4097cb193'

* commit '9c1db92ad372d4cd69e0490e691c56e4097cb193':
  mpegvideo: Drop err_recognition

Conflicts:
libavcodec/h263dec.c
libavcodec/ituh263dec.c
libavcodec/mpeg4video.h
libavcodec/mpeg4videodec.c
libavcodec/msmpeg4dec.c

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

Merge commit '848e86f74d3e6e87fa592ee8ba8c184cc5fd9a42'

* commit '848e86f74d3e6e87fa592ee8ba8c184cc5fd9a42':
  mpegvideo: Drop flags and flags2

Conflicts:
libavcodec/mpeg12dec.c
libavcodec/mpeg12enc.c
libavcodec/mpegvideo.c
libavcodec/mpegvideo_enc.c
libavcodec/mpegvideo_motion.c
libavcodec/ratecontrol.c
libavcodec/vc1_block.c
libavcodec/vc1_loopfilter.c
libavcodec/vc1_mc.c
libavcodec/vc1dec.c

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

avcodec/movtextdec: Add support for large boxes(>32 bit)

Signed-off-by: Niklesh <niklesh.lalwani@iitb.ac.in>

mpegvideo: Rework various functions not to use MpegEncContext directly

mpegvideo: Have ff_mpeg_ref_picture use AVCodecContext directly

mpegvideo: Have ff_mpeg_unref_picture use AVCodecContext directly

This skips setting the memory to 0 but allows for reuse on different
contextes. Oracle did not report any unsual activity because of it.