ffmpeg.git
8 years agoBetter threshold for the gcd based r_frame_rate calculation.
Michael Niedermayer [Tue, 15 Feb 2011 03:14:53 +0000 (04:14 +0100)]
Better threshold for the gcd based r_frame_rate calculation.
Fixes Gilmore Girls_Net 5_2011_01_08_16_10_53.wtv and RTL GP- WK Superbike - Qatar_RTL7_2009_03_14_15_49_23.wtv.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
8 years agoFix spelling error.
Tilman Brock [Mon, 14 Feb 2011 13:29:43 +0000 (14:29 +0100)]
Fix spelling error.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit b08f59dca493e7582ddfd08097c594bfb0ae6ff9)

8 years agommst: fix reading uninitialized data for ping packets.
Ronald S. Bultje [Mon, 14 Feb 2011 14:55:14 +0000 (09:55 -0500)]
mmst: fix reading uninitialized data for ping packets.

Fixes errors after a few minutes (first ping) when playing back
mmst://wm.bbc.co.uk/wms/bbc7coyopa/bbc7_-_friday_0430.wma
(cherry picked from commit 275189a2bd71cf49d66374c44d3de2262d323460)

8 years agoUpdate APIchanges for 910b5b8
Mans Rullgard [Mon, 14 Feb 2011 13:02:59 +0000 (13:02 +0000)]
Update APIchanges for 910b5b8

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 08a459f096106f13c03182eed4b7cb725dc4a0ae)

8 years agoRemove disabled legacy code in ffplay.
Stefano Sabatini [Sat, 12 Feb 2011 17:25:53 +0000 (18:25 +0100)]
Remove disabled legacy code in ffplay.

The code was setting aspect ratio from the DTG active format, and was
present and disabled since the creation of ffplay.

See thread:
Subject: [FFmpeg-devel] [PATCH] Cosmetics: adopt compact notation in
        disabled code.
Date: Sat, 12 Feb 2011 18:46:19 +0100

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit b7ec5cc4013242a9f30c429d2ce9c20ec01b03eb)

8 years agoaac: remove dead SBR input scaling
Alex Converse [Mon, 14 Feb 2011 00:42:38 +0000 (16:42 -0800)]
aac: remove dead SBR input scaling

It has been unused since 9d06d7bce3babb82ed650c13ed13a57f6f626a71

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 916dee9f45dab85cfc20b6bf63ce8722746e6508)

8 years agoasfdec: fix build for big endian targets
Mans Rullgard [Sun, 13 Feb 2011 22:10:45 +0000 (22:10 +0000)]
asfdec: fix build for big endian targets

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit aabc2682fcb70417abda0e78d27a425269a82fc8)

8 years agoac3enc: Add x86-optimized function to speed up log2_tab().
Justin Ruggles [Sun, 13 Feb 2011 19:49:50 +0000 (14:49 -0500)]
ac3enc: Add x86-optimized function to speed up log2_tab().

AC3DSPContext.ac3_max_msb_abs_int16() finds the maximum MSB of the absolute
value of each element in an array of int16_t.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit fbb6b49dabc3398440c6dfa838aa090a7a6ebc0d)

8 years agowtv: ignore empty data_guid chunks
Peter Ross [Sun, 13 Feb 2011 03:42:49 +0000 (14:42 +1100)]
wtv: ignore empty data_guid chunks

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 1a973feb45826a1998b4286ecfe1fa7a602b8780)

8 years agoasfdec: split reading stream properties out of asf_read_header()
Anton Khirnov [Wed, 9 Feb 2011 20:55:57 +0000 (21:55 +0100)]
asfdec: split reading stream properties out of asf_read_header()

This contains a rename from gsize->size

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 8bf6db1b29f3766732e2bb2bdd2681318fd8e215)

8 years agommst: print packet type with error status code message
Francesco Cosoleto [Sat, 12 Feb 2011 10:17:20 +0000 (11:17 +0100)]
mmst: print packet type with error status code message

Signed-off-by: Francesco Cosoleto <cosoleto@gmail.com>
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 70630e35a436d890f3e3fbbe73b70042a375f326)

8 years agoconfigure: fix non-standard regex used with expr
Mans Rullgard [Sun, 13 Feb 2011 21:00:03 +0000 (21:00 +0000)]
configure: fix non-standard regex used with expr

The colon operator of expr always anchors the pattern at the start
of the string.  An explicit ^ in the pattern has unspecified
behaviour, so remove it.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 3ad464bfc7df6d0f68c7df9c907089b84d92ab75)

8 years agoconfigure: report full gcc version string
Mans Rullgard [Sun, 13 Feb 2011 13:17:46 +0000 (13:17 +0000)]
configure: report full gcc version string

This makes the cc_ident value, which is used in FATE reports, include
all interesting parts of the gcc version string.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 5d57846bba7010dc6ee65c3503dd06600e5441a5)

8 years agoconfigure: add cleanws() function
Mans Rullgard [Sun, 13 Feb 2011 13:06:53 +0000 (13:06 +0000)]
configure: add cleanws() function

This function removes leading and trailing spaces and collapses
multiple spaces into one.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit bb7a22b5840e56c034227ad596b1165c03f2da27)

8 years agoFFT: factor a shuffle out of the inner loop and merge it into fft_permute.
Loren Merritt [Sat, 12 Feb 2011 11:48:16 +0000 (11:48 +0000)]
FFT: factor a shuffle out of the inner loop and merge it into fft_permute.

6% faster SSE FFT on Conroe, 2.5% on Penryn.

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit e6b1ed693ae4098e6b9eabf938fc31ec0b09b120)

8 years agolavf: print stream disposition in dump_stream_format
Anssi Hannula [Sat, 12 Feb 2011 18:11:11 +0000 (20:11 +0200)]
lavf: print stream disposition in dump_stream_format

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 3c33c0e26325f189dbf7f212f8e9042feb83bfb0)

8 years agospdifenc: fix byte order on big-endian systems
Anssi Hannula [Sat, 12 Feb 2011 13:57:01 +0000 (15:57 +0200)]
spdifenc: fix byte order on big-endian systems

There is a check for HAVE_BIGENDIAN when outputting the IEC 61937
stream. On big-endian systems the payload data is not byteswapped,
causing in effect the outputted payload data to be in a different byte
order on big-endian than on little-endian systems.

However, the IEC 61937 preamble (and the final odd byte if present) is
always outputted in the same byte order. This means that on big-endian
systems the headers have a different byte order than the payload,
preventing useful use of the output.

Fix that by outputting the data in a format suitable for sending to an
audio device in S16LE format by default. Output as big-endian (S16BE)
is added as an AVOption. This makes the muxer output the same on all
archs by default.

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 57f2c9aed9fecbb1e6798ec33613141afe74fd31)

8 years agospdifenc: IEC 61937 encapsulation of DTS-HD for HDMI
Anssi Hannula [Sat, 29 Jan 2011 01:36:30 +0000 (03:36 +0200)]
spdifenc: IEC 61937 encapsulation of DTS-HD for HDMI

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 482d98f69b2eb7a7b0b5054101a43db384e9432b)

8 years agobink: permute Bink version 'b' quant matrices by inverse scan order
Peter Ross [Sun, 13 Feb 2011 09:36:29 +0000 (09:36 +0000)]
bink: permute Bink version 'b' quant matrices by inverse scan order

This fixes visual glitches in Bink version 'b' files, as the quantization
tables were not being permuted.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 231539217431c32d4f22a16ee2834e24f54611f4)

8 years agobink: simplify how quantization matrices are passed to read_dct_coeffs()
Peter Ross [Sun, 13 Feb 2011 03:33:49 +0000 (03:33 +0000)]
bink: simplify how quantization matrices are passed to read_dct_coeffs()

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit cb62510cab8b745e65eb966c72ab7e9b122d564b)

8 years agoFix MMX rgb24 to yuv conversion with gcc 4.6
Mans Rullgard [Sun, 13 Feb 2011 00:19:06 +0000 (00:19 +0000)]
Fix MMX rgb24 to yuv conversion with gcc 4.6

When built with gcc 4.6, the MMX rgb24 to yuv conversion gives
wrong output.  The compiler produces this warning:

libswscale/swscale_template.c:1885:5: warning: use of memory input without lvalue in asm operand 4 is deprecated

Changing the memory operand to a register makes it work.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit f344903ca5ce28a833fdd656bc1ed5b16d97e7e9)

8 years agomovenc: Force first sample as SYNC
Takashi Mochizuki [Sun, 13 Feb 2011 00:18:45 +0000 (09:18 +0900)]
movenc: Force first sample as SYNC

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 44b11207248c6ca6edb7d3469c0fbdca45dbfa70)

8 years agoMake avfilter_config_links() propagate audio information of the source link if it...
Stefano Sabatini [Sun, 13 Feb 2011 21:59:55 +0000 (22:59 +0100)]
Make avfilter_config_links() propagate audio information of the source link if it is not explicitely set by the filter.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
8 years agoRestructure video filter implemenattion in ffmpeg.c.
Michael Niedermayer [Mon, 14 Feb 2011 22:02:10 +0000 (23:02 +0100)]
Restructure video filter implemenattion in ffmpeg.c.
This fixes several bugs like multiple outputs and -aspect mixed with -vf

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
8 years agoMake truncated mpeg4 GOP header check a bit more specific.
Michael Niedermayer [Mon, 14 Feb 2011 13:44:17 +0000 (14:44 +0100)]
Make truncated mpeg4 GOP header check a bit more specific.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
8 years agoApply minor cosmetics fixes to the av_audio_resample_init() doxy.
Stefano Sabatini [Mon, 14 Feb 2011 09:59:14 +0000 (10:59 +0100)]
Apply minor cosmetics fixes to the av_audio_resample_init() doxy.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
8 years agoDocument audio_resample_close().
Stefano Sabatini [Mon, 14 Feb 2011 09:59:25 +0000 (10:59 +0100)]
Document audio_resample_close().

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
8 years agoDetect null gop headers.
Michael Niedermayer [Sun, 13 Feb 2011 16:20:46 +0000 (17:20 +0100)]
Detect null gop headers.
Fix issue2592. / mpeg4-bad-gvop.raw

8 years agoMake tables generation insensitive to floating-point implementation
Vitor Sessak [Sat, 12 Feb 2011 09:15:58 +0000 (10:15 +0100)]
Make tables generation insensitive to floating-point implementation

Using doubles make the double -> int cast well defined for all the values
used, with the exception of when s[i]==1.0, which is special-cased.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 47d62c965b7bcf4715127fbf9bffab7b4c5c88b7)

8 years agoFix script command in a FAQ entry
Stefano Sabatini [Sat, 12 Feb 2011 09:12:41 +0000 (09:12 +0000)]
Fix script command in a FAQ entry

In the FAQ section "How do I encode single pictures into movies?", use
-s for generating symbolic links with the ln command.

The script was generating hard links, which is not likely what it was
supposed to do.

Fix issue 2488.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit e063f5886b27a7374f5060b7c2c21385685b2cb0)

8 years agoAllow "make clean" to clean files in tools
Stefano Sabatini [Sat, 12 Feb 2011 08:50:46 +0000 (08:50 +0000)]
Allow "make clean" to clean files in tools

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 40321376d8aea3347bec9c4d09925e8c5b8edb55)

8 years agoAdd lavfi-showfiltfmts and graph2dot to $(TOOLS)
Stefano Sabatini [Sat, 12 Feb 2011 08:50:35 +0000 (08:50 +0000)]
Add lavfi-showfiltfmts and graph2dot to $(TOOLS)

Allow make clean to remove the corresponding binaries.

Fix issue 2162.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit cd6a5a57b820c5fe008ea0f91a727861b7b2a331)

8 years agoFix qtrle regression test, actually test qtrle.
Baptiste Coudurier [Fri, 11 Feb 2011 23:20:10 +0000 (23:20 +0000)]
Fix qtrle regression test, actually test qtrle.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 646739a0a8368bd234ff077c093bbf01e9e7f44a)

8 years agortsp/rdt: Assign the RTSPStream index to AVStream->id
Martin Storsjö [Fri, 11 Feb 2011 21:36:47 +0000 (23:36 +0200)]
rtsp/rdt: Assign the RTSPStream index to AVStream->id

This is used for mapping AVStreams back to their corresponding
RTSPStream. Since d9c0510, the RTSPStream pointer isn't stored in
AVStream->priv_data any longer, breaking this mapping from AVStreams
to RTSPStreams.

Also, we don't need to clear the priv_data in rdt cleanup any longer,
since it isn't set to duplicate pointers.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit b2dd842d21a0b441bb9f7092357f479beb6b6f69)

8 years agoAdd check for Athlon64 and similar AMD processors with slow SSE2.
Justin Ruggles [Fri, 11 Feb 2011 20:17:32 +0000 (15:17 -0500)]
Add check for Athlon64 and similar AMD processors with slow SSE2.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 74b1f96859eb967222fcb3eb4c72d949b5165a89)

8 years agoconfigure: remove early check_deps $ARCH_EXT_LIST
Mans Rullgard [Fri, 11 Feb 2011 18:28:12 +0000 (18:28 +0000)]
configure: remove early check_deps $ARCH_EXT_LIST

The early disabling of irrelevant arch extensions is no longer
required, and removing it makes dependencies involving these
work as expected.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 4b884207eb7319577f8ceb94a56f42c5ac0ab294)

8 years agoVP8: ARM optimised decode_block_coeffs_internal
Mans Rullgard [Tue, 1 Feb 2011 22:38:15 +0000 (22:38 +0000)]
VP8: ARM optimised decode_block_coeffs_internal

Approximately 5% faster on Cortex-A8.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit a7878c9f73c12cfa685bd8af8f3afcca85f56a8b)

8 years agoARM optimised vp56_rac_get_prob()
Mans Rullgard [Wed, 22 Dec 2010 12:57:53 +0000 (12:57 +0000)]
ARM optimised vp56_rac_get_prob()

Approximately 3% faster on Cortex-A8.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 7da48fd0111adf504cfcfc5ebda7fd0681968041)

8 years agoUpdate overlay documentation after movie syntax update.
Stefano Sabatini [Fri, 11 Feb 2011 13:39:31 +0000 (13:39 +0000)]
Update overlay documentation after movie syntax update.

Overlay documentation is still using the old unsupported syntax.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit aa8ac53b51cd8b9446b50404660685041060fa4f)

8 years agobink: reindent after last commit
Peter Ross [Wed, 9 Feb 2011 12:05:19 +0000 (23:05 +1100)]
bink: reindent after last commit

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit adb1ad0d8005dfcbae863bccdda8fdd26bb63f30)

8 years agoBink version 'b' video decoder
Peter Ross [Fri, 11 Feb 2011 12:31:32 +0000 (23:31 +1100)]
Bink version 'b' video decoder

Based on original patch by Kostya Shishkov

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit e00f41d5742b3a0dc1877b030f4f6f58c19b7bbd)

8 years agolavf: deprecate AVFormatContext.index_built
Anton Khirnov [Thu, 10 Feb 2011 08:36:44 +0000 (09:36 +0100)]
lavf: deprecate AVFormatContext.index_built

it's not touched anywhere in ffmpeg, the code setting it was removed
over two years ago (e9b78eeba22b050810a507e69df1b652e56ab62b).

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit b0294c80d3a3981ec7ef9e4654962780e8566075)

8 years agoClarify that pts_correction* fields are internal to libavcodec
Michael Niedermayer [Fri, 11 Feb 2011 13:56:03 +0000 (14:56 +0100)]
Clarify that pts_correction* fields are internal to libavcodec
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
8 years agoSet maximum lowres value for the MJPEG decoder to 3.
Carl Eugen Hoyos [Sat, 12 Feb 2011 14:29:10 +0000 (15:29 +0100)]
Set maximum lowres value for the MJPEG decoder to 3.

While 4 works for some samples, 3 is the correct value since 8x8
DCT is used by (m)jpeg.

8 years agoffplay: stats: do not dereference NULL video
Nicolas George [Tue, 8 Feb 2011 10:57:08 +0000 (11:57 +0100)]
ffplay: stats: do not dereference NULL video

Also: cosmetic: split this overly long line

8 years agoVP8: optimized mv prediction and decoding
Jason Garrett-Glaser [Thu, 10 Feb 2011 06:43:25 +0000 (22:43 -0800)]
VP8: optimized mv prediction and decoding

Merge find_near_mvs and mv bitstream decoding: don't do prediction steps
until absolutely necessary.
(cherry picked from commit f3d09d44b70492022964e2ead5f439bb7c96eaa2)

8 years agoac3enc: Remove unneeded clipping of shift amount.
Justin Ruggles [Thu, 10 Feb 2011 22:12:16 +0000 (22:12 +0000)]
ac3enc: Remove unneeded clipping of shift amount.

s->windowed_samples will always have a range of [-32767,32767] due to the
window function, so the return value from log2_tab() will always be in the
range [0,14].

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 626264b11b5406c0c78c4056cabb63cb650e9bf2)

8 years agoac3enc: remove right shifting from lshift_tab() and make lshift unsigned.
Justin Ruggles [Thu, 10 Feb 2011 21:15:01 +0000 (21:15 +0000)]
ac3enc: remove right shifting from lshift_tab() and make lshift unsigned.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit d4582889eefeee4dd23face9e74b2829dbaaeae5)

8 years agoasfdec: deobfuscate reading video properties size
Anton Khirnov [Wed, 9 Feb 2011 20:55:56 +0000 (21:55 +0100)]
asfdec: deobfuscate reading video properties size

This code will be later split out into a function which takes a 'size'
argument, so I'm keeping the name 'sizeX' here.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 0b1d291a716dd79ca6862a95baf9ac574feba885)

8 years agoasfdec: split asf_read_header()
Anton Khirnov [Wed, 9 Feb 2011 20:55:55 +0000 (21:55 +0100)]
asfdec: split asf_read_header()

Only trivial splits are done here -- i.e. copy/paste + reindent +
missing variable declarations.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit c1fea2307011ac7a7bb98af98010f81abc789732)

8 years agowtv: mark streams intended for hearing or visual impaired persons
Peter Ross [Thu, 10 Feb 2011 06:26:10 +0000 (17:26 +1100)]
wtv: mark streams intended for hearing or visual impaired persons

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 68137ba386933243a84fc65f0deb3ad0ce5ff0ea)

8 years agomake av_find_best_stream() ignore streams marked with AV_DISPOSITION_*_IMPAIRED
Peter Ross [Thu, 10 Feb 2011 06:25:40 +0000 (17:25 +1100)]
make av_find_best_stream() ignore streams marked with AV_DISPOSITION_*_IMPAIRED

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 52091491575f015c09a32a745de4f7f7592fe6d4)

8 years agoadd AV_DISPOSITION_HEARING_IMPAIRED and AV_DISPOSITION_VISUAL_IMPAIRED
Peter Ross [Thu, 10 Feb 2011 06:25:13 +0000 (17:25 +1100)]
add AV_DISPOSITION_HEARING_IMPAIRED and AV_DISPOSITION_VISUAL_IMPAIRED

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 12c14cd4a8bf844f6f3d16ae18999bdb47999324)

8 years agoRemove final semicolon from some macros
Mans Rullgard [Thu, 10 Feb 2011 12:04:25 +0000 (12:04 +0000)]
Remove final semicolon from some macros

This avoids double semicolons after macro expansion.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 44adbebe1744c68d66d7f811c38270fdcc89665a)

8 years agodvbsubenc: Fix placement of the object version
Ronen Mizrahi [Thu, 10 Feb 2011 16:50:00 +0000 (11:50 -0500)]
dvbsubenc: Fix placement of the object version

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit df211c3ab73b36208ac4511a9cd7feb817f551fa)

8 years agodocument passing the fate samples location via make variable
Janne Grunau [Thu, 10 Feb 2011 08:57:18 +0000 (09:57 +0100)]
document passing the fate samples location via make variable
(cherry picked from commit 5c19f64c600ece050a1deb27f53963e4416beb31)

8 years agodvbsubdec: check against buffer overreads
Janne Grunau [Wed, 9 Feb 2011 22:23:22 +0000 (23:23 +0100)]
dvbsubdec: check against buffer overreads

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 493aa30adf88baf5bc734072592a22db586f0cfb)

8 years agoDo not fail DVB sub decoding because of a few padding bytes
Reimar Döffinger [Wed, 9 Feb 2011 18:32:07 +0000 (19:32 +0100)]
Do not fail DVB sub decoding because of a few padding bytes

Instead of returning an error when bytes are left over, just return
the number of actually used bytes as other decoders do.
Instead add a special case so an error will be returned when none
of the data looks valid to avoid making debugging a pain.

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 4a72765a1c94b05bd3053b1f34f8457a3b71d714)

8 years agoAdd x86-optimized versions of exponent_min().
Justin Ruggles [Thu, 10 Feb 2011 17:20:36 +0000 (12:20 -0500)]
Add x86-optimized versions of exponent_min().

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit dda3f0ef48aa5c3b03566b60b6bf63211e1fe579)

8 years agoRemove incorrect return statement from avcodec_thread_free()
Mans Rullgard [Thu, 10 Feb 2011 12:21:19 +0000 (12:21 +0000)]
Remove incorrect return statement from avcodec_thread_free()

The function return type is void, so a return statement with an
expression is forbidden (and pointless).

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit b4668274b944abae61759e796c5cc36ade510f24)

8 years agoivi_dsp: remove semicolons after function definitions
Mans Rullgard [Thu, 10 Feb 2011 12:04:52 +0000 (12:04 +0000)]
ivi_dsp: remove semicolons after function definitions

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit dbb09ec23f16cb5e4ec3d4e0911dae0beb95869b)

8 years agomov: remove stray semicolon
Mans Rullgard [Thu, 10 Feb 2011 11:42:21 +0000 (11:42 +0000)]
mov: remove stray semicolon

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 628b16f45f79aa5eb7514f11026d85e8aa34387f)

8 years agoasfdec: skip the stream bitrate list
Anton Khirnov [Wed, 9 Feb 2011 20:55:52 +0000 (21:55 +0100)]
asfdec: skip the stream bitrate list

Its contents aren't used for anything.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit d7a5106eb2dad33765b0e5f11fd8b1a87e5a9b4b)

8 years agoasfdec: use an ASFContext array for storing stream bitrates
Anton Khirnov [Wed, 9 Feb 2011 20:55:54 +0000 (21:55 +0100)]
asfdec: use an ASFContext array for storing stream bitrates

This will be useful for splitting asf_read_header()

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 7c7253802bbb4f7330ff9b32358bf05f9880cbdf)

8 years agoasfdec: move DAR list to ASFContext
Anton Khirnov [Wed, 9 Feb 2011 20:55:53 +0000 (21:55 +0100)]
asfdec: move DAR list to ASFContext

This will be useful for splitting asf_read_header()

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit d42b09723ed154d5269b95efcc20cb1874f60816)

8 years agoudp: Enable address reuse by default for multicast
David Fries [Mon, 7 Feb 2011 21:14:43 +0000 (23:14 +0200)]
udp: Enable address reuse by default for multicast

Keep the original corner case behaviour, where reuse is enabled
for the case where no argument is given to the reuse url option.

Signed-off-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 00952be424ba7403d71c338933354d2172df8e67)

8 years agoasfdec: remove some write-only values from the context
Anton Khirnov [Wed, 9 Feb 2011 20:55:51 +0000 (20:55 +0000)]
asfdec: remove some write-only values from the context

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 569ff02168dfe618599db5d5a327caf8b3051dd5)

8 years agoFix build with threading disabled
Mans Rullgard [Wed, 9 Feb 2011 20:15:25 +0000 (20:15 +0000)]
Fix build with threading disabled

The avcodec_thread_free() compatibility wrapper calls ff_thread_free(),
which is not defined when threading is disabled.  Make this call
conditional.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 9a77a92c2b6855781d2a4cfab14c67ae4025760c)

8 years agow32thread: add missing #include thread.h
Mans Rullgard [Wed, 9 Feb 2011 20:12:57 +0000 (20:12 +0000)]
w32thread: add missing #include thread.h

This should fix building with win32 threads.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit aef669cdfd984a737ad876b33ee1b160f87f5f9a)

8 years agoasf: split ASFContext into muxer and demuxer parts.
Anton Khirnov [Tue, 8 Feb 2011 20:46:47 +0000 (21:46 +0100)]
asf: split ASFContext into muxer and demuxer parts.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 4bc328a2bdf96d01afd3cf4532efb9db41f00985)

8 years agocheck sample_fmt in avcodec_open
Reimar Döffinger [Wed, 9 Feb 2011 15:52:33 +0000 (10:52 -0500)]
check sample_fmt in avcodec_open

check AVCodecContext->sample_fmt against AVCodec->sample_fmts[] to ensure
that the encoder supports the specified sample format. Error out if it doesn't.
Previously, it would continue and output garbage. Fixes issue 2587.
(cherry picked from commit 2cfa2d925808e6cc6fb7a7c133b7cb7622afd37e)

8 years agompegts: remove unused macro MAX_SCAN_PACKETS
Vladimir Pantelic [Wed, 9 Feb 2011 16:24:23 +0000 (16:24 +0000)]
mpegts: remove unused macro MAX_SCAN_PACKETS

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit f4c79d1e0b2e797012304db57903e4091b0c2d7c)

8 years agovp3: Frame-based multithreading support
Alexander Strange [Tue, 8 Feb 2011 02:15:45 +0000 (21:15 -0500)]
vp3: Frame-based multithreading support

Decode times for big_buck_bunny_720p_stereo:

1 thread:
real    1m14.227s
user    1m13.104s
sys     0m1.108s

2 threads: (33% faster)
real    0m49.329s
user    1m33.735s
sys     0m1.834s

3 threads: (44% faster)
real    0m41.593s
user    1m44.884s
sys     0m1.967s
(cherry picked from commit d23845f311f04e98777612cae1b55e28923ccdc6)

8 years agoAdd missing git rev hash.
Ronald S. Bultje [Wed, 9 Feb 2011 12:53:06 +0000 (07:53 -0500)]
Add missing git rev hash.
(cherry picked from commit f2146944fcd00a0a7918931c6f0f5c3f88cd6f1d)

8 years agoDeprecate avcodec_thread_init()
Alexander Strange [Tue, 8 Feb 2011 02:15:45 +0000 (21:15 -0500)]
Deprecate avcodec_thread_init()

As a side effect of the last commit, avcodec_open() now calls it automatically,
so there is no longer any need for clients to call it.
Instead they should set AVCodecContext.thread_count.

avcodec_thread_free() is deprecated, and will be removed from avcodec.h at the
next MAJOR libavcodec bump.

Rename the functions to ff_thread_init/free, since they are now internal.
Wrappers are provided to maintain API compatibility.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit c0b102ca03fe92250f1ce620aec3836f529fc1d6)

8 years agoAdd missing git rev hash.
Ronald S. Bultje [Wed, 9 Feb 2011 12:49:49 +0000 (07:49 -0500)]
Add missing git rev hash.
(cherry picked from commit 8e8cc52be3b515bc91cd9452daca7a65feaea5ad)

8 years agoFrame-based multithreading framework using pthreads
Alexander Strange [Tue, 8 Feb 2011 02:15:44 +0000 (21:15 -0500)]
Frame-based multithreading framework using pthreads

See doc/multithreading.txt for details on use in codecs.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
8 years agoAdd missing git revision hask.
Ronald S. Bultje [Wed, 9 Feb 2011 02:09:47 +0000 (21:09 -0500)]
Add missing git revision hask.
(cherry picked from commit c2bd7578af069206831a9c25fa68c9bbd5004619)

8 years agoasf: make ff_guidcmp inline and move it to asf.h
Anton Khirnov [Wed, 9 Feb 2011 12:27:25 +0000 (12:27 +0000)]
asf: make ff_guidcmp inline and move it to asf.h

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit e4e234fad77daa8628fc56155563248165d51ab4)

8 years agoUpdate qmin/qmax values for libx264 presets
Jason Garrett-Glaser [Tue, 8 Feb 2011 23:51:15 +0000 (15:51 -0800)]
Update qmin/qmax values for libx264 presets

Also allow qmin/qmax to go up to 69 (the current max value for libx264).  This
will have to increase when we add 9/10-bit support.
(cherry picked from commit c7ac200d15e3240972d269df768068f0e3bb9c2f)

8 years agoFix broken vbv_buffer_init handling in libx264.c
Jason Garrett-Glaser [Sun, 6 Feb 2011 13:10:15 +0000 (05:10 -0800)]
Fix broken vbv_buffer_init handling in libx264.c

Due to being pants-on-head retarded, libavcodec defaults this to zero, which
results in broken output.  This didn't affect ffmpeg.c, which sets it itself,
but caused problems for other calling apps using VBV.
(cherry picked from commit f7f8120fb9a06b5d7fbf5add48a0d8464319a332)

8 years agoVP8: idct_mb optimizations
Jason Garrett-Glaser [Sun, 6 Feb 2011 05:19:52 +0000 (21:19 -0800)]
VP8: idct_mb optimizations

Currently uses AV_RL32 instead of AV_RL32A, as the latter doesn't exist yet.
(cherry picked from commit 62457f9052ea15123688455aad866cb070634f13)

8 years agoFix ff_emu_edge_core_sse() on Win64.
Ronald S. Bultje [Tue, 8 Feb 2011 20:56:32 +0000 (15:56 -0500)]
Fix ff_emu_edge_core_sse() on Win64.

Fix emu_edge_v_extend_15 to be <128 bytes on Win64, by being more strict
on the size of registers and which registers are being used for operations
where multiple are available. This fixes segfaults in emulated_edge()
function calls on Win64.
(cherry picked from commit 17cf7c68ed26a4cb3c7adf7488a38c2e19118918)

8 years agolavf: rename ff_probe_input_buffer to make it public
Anssi Hannula [Mon, 7 Feb 2011 22:46:53 +0000 (00:46 +0200)]
lavf: rename ff_probe_input_buffer to make it public

It is useful for applications that hand input data directly to lavf via
a ByteIOContext.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 3940caad02fbc4f075e77bc605849b1d84ba1b8e)

8 years agolavf: simplify pb parameter of ff_probe_input_buffer
Anssi Hannula [Mon, 7 Feb 2011 22:46:52 +0000 (00:46 +0200)]
lavf: simplify pb parameter of ff_probe_input_buffer

There is no need to pass the ByteIOContext via a pointer to a pointer
anymore.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit aad216fd7e3209fe0a865d8751e680925f6654d3)

8 years agolavf: update ff_probe_input_buffer documentation
Anssi Hannula [Mon, 7 Feb 2011 22:46:51 +0000 (00:46 +0200)]
lavf: update ff_probe_input_buffer documentation

It never reopens the bytestream anymore.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 4d016dd4e5b288b3c0c7396935326cb07a289ac2)

8 years agooggdec: Fix incorrect assumption about header/data interleaving
Reimar Döffinger [Sat, 5 Feb 2011 16:32:35 +0000 (16:32 +0000)]
oggdec: Fix incorrect assumption about header/data interleaving

Currently (since the data_offset fix) the ogg demuxer assumes that
after the first non-header packets in any stream no more header packets
will follow.
This is not guaranteed, so change the code back again to wait until it
has finished the headers for all streams before returning from ogg_get_headers.

This fixes issue 2428.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit 6bd69e6adadf645fd07e909bebc150eab4b9c3c8)

8 years agoFix HOSTLIBS on Haiku
François Revol [Tue, 8 Feb 2011 21:09:44 +0000 (21:09 +0000)]
Fix HOSTLIBS on Haiku

Haiku does not have a separate libm, so do not try to link to it.

Signed-off-by: François Revol <revol@free.fr>
Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit f59c4bd625839dbe1e099c6a8b38988108883623)

8 years agoDocumentation updates for the git migration
Reinhard Tartler [Mon, 7 Feb 2011 16:17:30 +0000 (17:17 +0100)]
Documentation updates for the git migration

This cleanup patch updates the developer documentation with respect to
the migration to the git scm.
(cherry picked from commit 87800dc2bf8f2724a99e51bb079ad7fb4b9dfd3b)

8 years agologging: downgrade recoverable errors to warnings
Stefan Kost [Tue, 8 Feb 2011 12:16:49 +0000 (14:16 +0200)]
logging: downgrade recoverable errors to warnings

In all 3 cases, the decoding continues and thus a warning would be sufficient.
Helps application that catch them with own log handers to handle them
accordingly.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit ae2104791ffa8466d3f6972ed0793b2a165daabe)

8 years agoadd APIChanges entry for fe9a3fb
Janne Grunau [Tue, 1 Feb 2011 21:09:12 +0000 (22:09 +0100)]
add APIChanges entry for fe9a3fb
(cherry picked from commit 7ab8758baf743b3d525c07a770e13038b0885461)

8 years agoh264: define FF_PROFILE_H264_HIGH_444 to the correct value
Janne Grunau [Tue, 1 Feb 2011 20:56:13 +0000 (21:56 +0100)]
h264: define FF_PROFILE_H264_HIGH_444 to the correct value

It was removed in fe9a3fb since it had the wrong value. Add profile name
for it.
(cherry picked from commit 440b61691db2166abac3f58e01da1d4722f1f1e4)

8 years agolibavfilter: add video movie source
Stefano Sabatini [Sat, 29 Jan 2011 10:22:41 +0000 (11:22 +0100)]
libavfilter: add video movie source

See thread:
Subject: [PATCH] movie video source
Date: 2010-12-31 15:35:30 GMT

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
8 years agoApply misc fixes to the image2 muxer documentation.
Stefano Sabatini [Wed, 2 Feb 2011 23:49:07 +0000 (00:49 +0100)]
Apply misc fixes to the image2 muxer documentation.

The fixes were pointed out by Diego.

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 0cad24ce9b02037c9cdad84cae905f96b04d7d4c)

8 years agoAdd documentation for the framecrc muxer.
Stefano Sabatini [Tue, 1 Feb 2011 00:03:48 +0000 (01:03 +0100)]
Add documentation for the framecrc muxer.

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit 77d4ed7a127b88f9b8ddd741c65e678198feb285)

8 years agoAdd documentation for the crc muxer.
Stefano Sabatini [Mon, 31 Jan 2011 14:20:09 +0000 (15:20 +0100)]
Add documentation for the crc muxer.

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit a4effe432f4aa59d52465b88589f92addf3afd3e)

8 years agoDocument null muxer.
Stefano Sabatini [Mon, 31 Jan 2011 14:01:46 +0000 (15:01 +0100)]
Document null muxer.

Signed-off-by: Janne Grunau <janne-ffmpeg@jannau.net>
(cherry picked from commit f4acb837eb4af74996b678f3f55eaa66c47a66ee)

8 years agoVP8: ARM NEON optimisations for dsp functions
Mans Rullgard [Wed, 2 Feb 2011 16:26:20 +0000 (16:26 +0000)]
VP8: ARM NEON optimisations for dsp functions

This adds NEON optimised versions of all functions in VP8DSPContext.
Based on initial work by Rob Clark.

Signed-off-by: Mans Rullgard <mans@mansr.com>
(cherry picked from commit a1c1d3c003b0ec16fdb6574913781313fb2c7ab6)

8 years agoRemove Xan WC4 cruft from xan.c
Kostya [Thu, 3 Feb 2011 09:06:12 +0000 (10:06 +0100)]
Remove Xan WC4 cruft from xan.c

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 90e8a9c34f641d0cb9137190c1912a3da3095e01)

8 years agoXan4 decoder
Kostya Shishkov [Fri, 4 Feb 2011 22:41:07 +0000 (23:41 +0100)]
Xan4 decoder

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
(cherry picked from commit 44ddfd47d671d2587903d94c8b565f68f45bd4bc)