platform/upstream/libav.git
13 years agortsp: Accept options via private avoptions instead of URL options
Martin Storsjö [Thu, 23 Jun 2011 22:49:09 +0000 (01:49 +0300)]
rtsp: Accept options via private avoptions instead of URL options

Eventually, the old way of passing options by adding
stuff to the URL can be dropped.

This avoids having to tamper with the user-specified URL to
pass options on the transport mode. This also works better
with redirects, since the options don't need to be parsed out
from the URL.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agortsp: Simplify AVOption definitions
Martin Storsjö [Thu, 13 Oct 2011 11:00:06 +0000 (14:00 +0300)]
rtsp: Simplify AVOption definitions

Use defines for shortening common parts, omit the .dbl named
initializer (since it's the first element in the union).

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agortsp: Merge the AVOption lists
Martin Storsjö [Thu, 23 Jun 2011 23:04:54 +0000 (02:04 +0300)]
rtsp: Merge the AVOption lists

This eases adding options that are common for both. The
AV_OPT_FLAG_EN/DECODING_PARAM still indicates whether they belong
to the muxer or demuxer.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agolavfi: port libmpcodecs delogo filter
Stefano Sabatini [Wed, 10 Aug 2011 16:58:49 +0000 (18:58 +0200)]
lavfi: port libmpcodecs delogo filter

The ported filter supports named option parsing and more YUV formats.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agolavfi: port boxblur filter from libmpcodecs
Stefano Sabatini [Sat, 9 Jul 2011 16:13:10 +0000 (18:13 +0200)]
lavfi: port boxblur filter from libmpcodecs

With the following additions:
* support to gray format
* support to yuva420p format
* parametric luma/chroma/alpha radius
* consistency check on the radius values, avoid crashes with invalid values

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agolavfi: add negate filter
Stefano Sabatini [Sat, 28 May 2011 20:00:26 +0000 (22:00 +0200)]
lavfi: add negate filter

This filter is a simple wrapper around the LUT filter.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agolavfi: add LUT (LookUp Table) generic filters
Stefano Sabatini [Sat, 28 May 2011 19:51:03 +0000 (21:51 +0200)]
lavfi: add LUT (LookUp Table) generic filters

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agoAVOptions: don't segfault on NULL parameter in av_set_options_string()
Anton Khirnov [Sat, 1 Oct 2011 12:42:53 +0000 (14:42 +0200)]
AVOptions: don't segfault on NULL parameter in av_set_options_string()

13 years agoavio: Check for invalid buffer length.
Alex Converse [Sat, 15 Oct 2011 01:27:59 +0000 (18:27 -0700)]
avio: Check for invalid buffer length.

13 years agompegenc/mpegtsenc: add muxrate private options.
Anton Khirnov [Sun, 9 Oct 2011 12:40:11 +0000 (14:40 +0200)]
mpegenc/mpegtsenc: add muxrate private options.

Deprecate AVFormatContext.mux_rate.

13 years agolavf: deprecate AVFormatContext.file_size
Anton Khirnov [Sun, 9 Oct 2011 12:12:14 +0000 (14:12 +0200)]
lavf: deprecate AVFormatContext.file_size

It's too unreliable to be useful. avio_size() should be called instead.

13 years agomov: add support for TV metadata atoms tves, tvsn and stik
Raivo Hool [Fri, 14 Oct 2011 17:14:18 +0000 (20:14 +0300)]
mov: add support for TV metadata atoms tves, tvsn and stik

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agow32threads: support for frame multithreading
Steven Walters [Sun, 9 Oct 2011 19:38:35 +0000 (21:38 +0200)]
w32threads: support for frame multithreading

Replace our incomplete w32threads implementation with x264's pthreads
w32threads wrapper.
Relicensed to LGPL with kind permission by Pegasys Inc.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agoavcodec: remove stray @deprecated comment
Mans Rullgard [Sun, 16 Oct 2011 15:16:55 +0000 (16:16 +0100)]
avcodec: remove stray @deprecated comment

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoprores: get correct size for coded V plane if alpha is present
Mans Rullgard [Sat, 15 Oct 2011 13:35:49 +0000 (14:35 +0100)]
prores: get correct size for coded V plane if alpha is present

The size check must be updated to take into account both manners
in which v_data_size might be set.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoprores: do not set pixel format on codec init
Mans Rullgard [Sat, 15 Oct 2011 13:35:10 +0000 (14:35 +0100)]
prores: do not set pixel format on codec init

The pixel format is not known until the frame header is parsed.
Guessing it here only causes trouble for the caller if the guess
turns out to be wrong (and actually causes very wrong output by
avconv/avplay).

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agopthread: prevent updating AVCodecContext from itself in frame_thread_free
Ronald S. Bultje [Fri, 14 Oct 2011 21:47:45 +0000 (23:47 +0200)]
pthread: prevent updating AVCodecContext from itself in frame_thread_free

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agopthread: copy coded frame dimensions in update_context_from_thread
Ronald S. Bultje [Fri, 14 Oct 2011 21:46:06 +0000 (23:46 +0200)]
pthread: copy coded frame dimensions in update_context_from_thread

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agovp8: prevent read from uninitialized memory in decode_mvs
Ronald S. Bultje [Fri, 14 Oct 2011 21:43:29 +0000 (23:43 +0200)]
vp8: prevent read from uninitialized memory in decode_mvs

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agovp8: force reallocation in update_thread_context after frame size change
Ronald S. Bultje [Fri, 14 Oct 2011 21:27:52 +0000 (23:27 +0200)]
vp8: force reallocation in update_thread_context after frame size change

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agovp8: fix return value if update_dimensions fails
Ronald S. Bultje [Fri, 14 Oct 2011 21:21:46 +0000 (23:21 +0200)]
vp8: fix return value if update_dimensions fails

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agomatroskadec: fix out of bounds write
Ronald S. Bultje [Fri, 14 Oct 2011 22:03:55 +0000 (00:03 +0200)]
matroskadec: fix out of bounds write

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agoadpcmdec: calculate actual number of output samples for each decoder.
Justin Ruggles [Sun, 2 Oct 2011 14:18:17 +0000 (10:18 -0400)]
adpcmdec: calculate actual number of output samples for each decoder.

This also allows for removing some of the buf_size checks and using the
sample count for some of the decoding loops.

13 years agoadpcmdec: check remaining buffer size before decoding next block in the
Justin Ruggles [Sun, 2 Oct 2011 20:07:55 +0000 (16:07 -0400)]
adpcmdec: check remaining buffer size before decoding next block in the
ADPCM IMA WAV decoder.

13 years agoadpcmdec: do not terminate early in ADPCM IMA Duck DK3 decoder.
Justin Ruggles [Sun, 2 Oct 2011 19:58:54 +0000 (15:58 -0400)]
adpcmdec: do not terminate early in ADPCM IMA Duck DK3 decoder.

There are still 2 nibbles to decode once the last byte in the packet has been
read. Updated FATE reference.

13 years agoadpcmdec: remove unneeded buf_size==0 check.
Justin Ruggles [Sun, 2 Oct 2011 18:43:28 +0000 (14:43 -0400)]
adpcmdec: remove unneeded buf_size==0 check.

This is already done by avcodec_decode_audio3()

13 years agoadpcmdec: remove unneeded zeroing of *data_size
Justin Ruggles [Sun, 2 Oct 2011 18:41:34 +0000 (14:41 -0400)]
adpcmdec: remove unneeded zeroing of *data_size

13 years agodnxhdenc: fixed signed multiplication overflow
Mans Rullgard [Sat, 8 Oct 2011 00:37:20 +0000 (01:37 +0100)]
dnxhdenc: fixed signed multiplication overflow

The low 32 bits of a multiplication are the same for signed
and unsigned operands.  Casting to unsigned before multiplying
is thus equivalent while avoiding signed overflow, which is
undefined by the C99 standard.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoproresdsp: fix function prototypes.
Ronald S. Bultje [Fri, 14 Oct 2011 16:28:46 +0000 (09:28 -0700)]
proresdsp: fix function prototypes.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agoprores-idct: fix overflow in c code.
Ronald S. Bultje [Fri, 14 Oct 2011 16:27:59 +0000 (09:27 -0700)]
prores-idct: fix overflow in c code.

Fix the fate ref for prores-422_proxy by reverting the changes to it
in commit f492df0927c42da174edb674857670fc50abc5dc

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agofate: update prores-alpha ref after changing pix_fmt to yuv444p10le
Janne Grunau [Fri, 14 Oct 2011 19:20:46 +0000 (21:20 +0200)]
fate: update prores-alpha ref after changing pix_fmt to yuv444p10le

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agoprores: add missing feature warning for alpha
Janne Grunau [Fri, 14 Oct 2011 10:18:23 +0000 (12:18 +0200)]
prores: add missing feature warning for alpha

13 years agomov: 10l: Terminate string with 0 not '0'
Alex Converse [Fri, 14 Oct 2011 17:38:42 +0000 (10:38 -0700)]
mov: 10l: Terminate string with 0 not '0'

13 years agomov: Prevent illegal writes when chapter titles are very short.
Alex Converse [Thu, 13 Oct 2011 21:47:06 +0000 (14:47 -0700)]
mov: Prevent illegal writes when chapter titles are very short.

13 years agoprores: add appropriate -fix_fmt parameter to FATE command
Diego Biurrun [Fri, 14 Oct 2011 14:11:32 +0000 (16:11 +0200)]
prores: add appropriate -fix_fmt parameter to FATE command

Also correct CRCs for Sequence_1-Apple_ProRes_422_Proxy.mov, previous
CRCs were generated with the non-bitexact x86_64 assembler IDCT.

13 years agoriff: always generate a proper WAVEFORMATEX structure in ff_put_wav_header
John Brooks [Thu, 13 Oct 2011 08:34:20 +0000 (08:34 +0000)]
riff: always generate a proper WAVEFORMATEX structure in ff_put_wav_header

The cbSize field should be included in all cases, even with PCM where
its value is ignored.

Fixes encoding PCM audio in Matroska for some players which insist on
a full WAVEFORMATEX structure for A_MS/ACM audio.

Since fate uses wav files for the audio test a larger number of tests
has changed checksums or shifted positions due to the 2 byte longer
wave header.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agolavc: add a flag-based error_recognition field to AVCodecContext and deprecate non...
Dustin Brody [Thu, 13 Oct 2011 07:26:53 +0000 (03:26 -0400)]
lavc: add a flag-based error_recognition field to AVCodecContext and deprecate non-flag-based ER field

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agolavc: rename deprecation symbol FF_API_VERY_AGGRESSIVE to FF_API_ER
Dustin Brody [Thu, 13 Oct 2011 07:26:27 +0000 (03:26 -0400)]
lavc: rename deprecation symbol FF_API_VERY_AGGRESSIVE to FF_API_ER

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agoprores: add FATE tests
Diego Biurrun [Fri, 23 Sep 2011 00:06:23 +0000 (02:06 +0200)]
prores: add FATE tests

13 years agoid3v2: reduce the scope of some non-globally-used symbols/structures
Diego Biurrun [Mon, 10 Oct 2011 13:51:02 +0000 (15:51 +0200)]
id3v2: reduce the scope of some non-globally-used symbols/structures

13 years agoid3v2: cosmetics: move some declarations before the places they are used
Diego Biurrun [Mon, 10 Oct 2011 13:48:59 +0000 (15:48 +0200)]
id3v2: cosmetics: move some declarations before the places they are used

13 years agoshorten: remove the flush function.
Justin Ruggles [Sat, 17 Sep 2011 02:21:47 +0000 (22:21 -0400)]
shorten: remove the flush function.

The shorten decoder does not support seeking.

13 years agoshn: do not allow seeking in the raw shn demuxer.
Justin Ruggles [Sun, 2 Oct 2011 16:03:22 +0000 (12:03 -0400)]
shn: do not allow seeking in the raw shn demuxer.

The demuxer does not read the seektable, a parser is not possible without a
full decode, and no shorten decoder can handle random seeking because it needs
side info from the seektable.

13 years agoavformat: add AVInputFormat flag AVFMT_NO_BYTE_SEEK.
Justin Ruggles [Sun, 2 Oct 2011 15:57:53 +0000 (11:57 -0400)]
avformat: add AVInputFormat flag AVFMT_NO_BYTE_SEEK.

This will prevent seeking by bytes for formats that cannot handle this.

13 years agoavformat: update AVInputFormat allowed flags
Justin Ruggles [Sun, 2 Oct 2011 15:29:20 +0000 (11:29 -0400)]
avformat: update AVInputFormat allowed flags

13 years agoavformat: don't unconditionally call ff_read_frame_flush() when trying to seek.
Justin Ruggles [Fri, 16 Sep 2011 23:02:23 +0000 (19:02 -0400)]
avformat: don't unconditionally call ff_read_frame_flush() when trying to seek.

This prevents flushing the packet buffer when the input format does not
support seeking.

13 years agotruespeech: use sizeof() instead of hardcoded sizes
Justin Ruggles [Thu, 13 Oct 2011 18:03:17 +0000 (14:03 -0400)]
truespeech: use sizeof() instead of hardcoded sizes

13 years agotruespeech: remove unneeded variable, 'consumed'
Justin Ruggles [Thu, 13 Oct 2011 04:21:17 +0000 (00:21 -0400)]
truespeech: remove unneeded variable, 'consumed'

increment the 'buf' pointer instead, and consume the whole packet.

13 years agotruespeech: simplify truespeech_read_frame() by using get_bits()
Justin Ruggles [Thu, 13 Oct 2011 04:17:05 +0000 (00:17 -0400)]
truespeech: simplify truespeech_read_frame() by using get_bits()

13 years agotruespeech: decode directly to output buffer instead of a temp buffer
Justin Ruggles [Thu, 13 Oct 2011 03:27:32 +0000 (23:27 -0400)]
truespeech: decode directly to output buffer instead of a temp buffer

13 years agotruespeech: check to make sure channels == 1
Justin Ruggles [Thu, 13 Oct 2011 03:23:18 +0000 (23:23 -0400)]
truespeech: check to make sure channels == 1

13 years agotruespeech: check for large enough output buffer rather than truncating output
Justin Ruggles [Thu, 13 Oct 2011 03:15:00 +0000 (23:15 -0400)]
truespeech: check for large enough output buffer rather than truncating output

13 years agotruespeech: remove unneeded zero-size packet check.
Justin Ruggles [Thu, 13 Oct 2011 03:10:41 +0000 (23:10 -0400)]
truespeech: remove unneeded zero-size packet check.

This is already checked in avcodec_decode_audio3()

13 years agomlpdec: return meaningful error codes instead of -1
Justin Ruggles [Thu, 13 Oct 2011 02:47:50 +0000 (22:47 -0400)]
mlpdec: return meaningful error codes instead of -1

13 years agomlpdec: remove unnecessary wrapper function
Justin Ruggles [Thu, 13 Oct 2011 02:37:12 +0000 (22:37 -0400)]
mlpdec: remove unnecessary wrapper function

13 years agomlpdec: only calculate output size once
Justin Ruggles [Wed, 21 Sep 2011 16:14:28 +0000 (12:14 -0400)]
mlpdec: only calculate output size once

13 years agomlpdec: validate that the reported channel count matches the actual output
Justin Ruggles [Thu, 13 Oct 2011 02:30:44 +0000 (22:30 -0400)]
mlpdec: validate that the reported channel count matches the actual output
channel count

13 years agopcm: reduce pointer type casting
Mans Rullgard [Sat, 8 Oct 2011 11:47:53 +0000 (12:47 +0100)]
pcm: reduce pointer type casting

Making 'samples' a pointer to uint8_t simplifies the DECODE
macro and reduces the amount of type casting overall.

This also fixes some signed overflows on left shift.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agomathops: remove undefined behaviour from sign_extend()
Mans Rullgard [Sun, 9 Oct 2011 11:57:08 +0000 (12:57 +0100)]
mathops: remove undefined behaviour from sign_extend()

This function intentionally overflows the signed range on
the left shift.  Using this type-punning avoids errors from
the overflow checker without disabling this test globally.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agovc1: explicitly zero interlaced mode coding variables for progressive mode
Kostya Shishkov [Thu, 13 Oct 2011 08:12:42 +0000 (10:12 +0200)]
vc1: explicitly zero interlaced mode coding variables for progressive mode

Both v->fcm and v->field_mode are used in common code, now they won't be
reset for progressive frame after interlaced one causing writing past the
frame end for example.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
13 years agotiffenc: initialize forgotten avctx.
Jean First [Fri, 23 Sep 2011 21:39:51 +0000 (23:39 +0200)]
tiffenc: initialize forgotten avctx.

13 years agoavplay: free the active audio packet at exit.
Justin Ruggles [Wed, 12 Oct 2011 18:44:02 +0000 (14:44 -0400)]
avplay: free the active audio packet at exit.

fixes a memleak

13 years agoavplay: free rdft data used for spectrogram analysis.
Justin Ruggles [Wed, 12 Oct 2011 18:38:02 +0000 (14:38 -0400)]
avplay: free rdft data used for spectrogram analysis.

fixes a memleak

13 years agolog.h: make AVClass a named struct
Jindrich Makovicka [Wed, 12 Oct 2011 20:30:56 +0000 (22:30 +0200)]
log.h: make AVClass a named struct

'struct AVClass' is used in the code since
641c7afe3c17334b81e3e2eef88f1751eb68f89f, but AVClass is typedeffed as
an anonymous struct.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agofix ac3 encoder documentation
Justin Ruggles [Tue, 11 Oct 2011 19:04:33 +0000 (15:04 -0400)]
fix ac3 encoder documentation

some of the mode values were reversed

13 years agovc1: more prettyprinting cosmetics
Kostya Shishkov [Tue, 11 Oct 2011 09:59:57 +0000 (11:59 +0200)]
vc1: more prettyprinting cosmetics

Signed-off-by: Diego Biurrun <diego@biurrun.de>
13 years agovc1: prettyprint some tables
Kostya Shishkov [Tue, 11 Oct 2011 10:01:04 +0000 (12:01 +0200)]
vc1: prettyprint some tables

Signed-off-by: Diego Biurrun <diego@biurrun.de>
13 years agovc1: K&R formatting cosmetics
Kostya Shishkov [Tue, 11 Oct 2011 10:00:18 +0000 (12:00 +0200)]
vc1: K&R formatting cosmetics

Signed-off-by: Diego Biurrun <diego@biurrun.de>
13 years agoAVOptions: bump minor and add APIchanges entry.
Anton Khirnov [Tue, 4 Oct 2011 06:13:40 +0000 (08:13 +0200)]
AVOptions: bump minor and add APIchanges entry.

13 years agocmdutils/avtools: simplify show_help() by using av_opt_child_class_next()
Anton Khirnov [Tue, 4 Oct 2011 12:50:00 +0000 (14:50 +0200)]
cmdutils/avtools: simplify show_help() by using av_opt_child_class_next()

13 years agoAVOptions: rename FF_OPT_TYPE_* => AV_OPT_TYPE_*
Anton Khirnov [Tue, 4 Oct 2011 05:38:01 +0000 (07:38 +0200)]
AVOptions: rename FF_OPT_TYPE_* => AV_OPT_TYPE_*

13 years agoRemove all uses of deprecated AVOptions API.
Anton Khirnov [Mon, 3 Oct 2011 19:10:06 +0000 (21:10 +0200)]
Remove all uses of deprecated AVOptions API.

13 years agoAVOptions: add av_opt_next, deprecate av_next_option.
Anton Khirnov [Mon, 3 Oct 2011 18:04:07 +0000 (20:04 +0200)]
AVOptions: add av_opt_next, deprecate av_next_option.

Just for naming consistency, no functional changes.

13 years agoAVOptions: add functions for evaluating option strings.
Anton Khirnov [Tue, 23 Aug 2011 13:02:25 +0000 (15:02 +0200)]
AVOptions: add functions for evaluating option strings.

13 years agoAVOptions: split get_number().
Anton Khirnov [Tue, 23 Aug 2011 12:25:56 +0000 (14:25 +0200)]
AVOptions: split get_number().

Split actual writing to read_number() in the same way as write_number().
Allows set_string_number() to write to a caller-provided destination.

13 years agoAVOptions: add av_opt_get*, deprecate av_get*.
Anton Khirnov [Mon, 5 Sep 2011 19:37:02 +0000 (21:37 +0200)]
AVOptions: add av_opt_get*, deprecate av_get*.

New functions can get values from child objects, properly report error
codes and have consistent naming and signatures.

13 years agoAVOptions: add av_opt_set*().
Anton Khirnov [Mon, 5 Sep 2011 06:15:32 +0000 (08:15 +0200)]
AVOptions: add av_opt_set*().

Deprecate av_set_*

New functions support setting values on children, return error codes
instead of options and have consistent naming and signatures.

13 years agoAVOptions: add new API for enumerating children.
Anton Khirnov [Mon, 3 Oct 2011 17:49:12 +0000 (19:49 +0200)]
AVOptions: add new API for enumerating children.

This will allow the caller to enumerate child contexts in a generic way
and since the API is recursive, it also allows for deeper nesting (e.g.
AVFormatContext->AVIOContext->URLContext)

This will also allow the new setting/reading API to transparently apply
to children contexts.

13 years agorv34: move inverse transform functions to DSP context
Janne Grunau [Sat, 24 Sep 2011 10:54:28 +0000 (12:54 +0200)]
rv34: move inverse transform functions to DSP context

13 years agoflvenc: Write the right metadata entry count
Martin Storsjö [Wed, 21 Sep 2011 13:58:07 +0000 (16:58 +0300)]
flvenc: Write the right metadata entry count

No application rely on this count being correct as far as
I know, but if we write a nonzero count value, it might just
as well be the right one.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agortsp: Request that dynamic rate is disabled
Martin Storsjö [Mon, 27 Dec 2010 07:51:09 +0000 (09:51 +0200)]
rtsp: Request that dynamic rate is disabled

DSS enables this automatically if streaming VOD over TCP. If
enabled, the server feeds packets faster than realtime, screwing
up RTCP NTP based timestamps.

Also, DSS doesn't indicate that this was indicated, if it was
enabled automatically (although if it was requested to be enabled,
a header saying that it was enabled is added, but this isn't
added if it is enabled automatically), making it even harder
to detect and work around properly without explicitly asking
for it to be disabled(/enabled, if we were able to support it).

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agortsp: Parse the x-Accept-Dynamic-Rate header
Martin Storsjö [Sun, 2 Jan 2011 12:32:18 +0000 (14:32 +0200)]
rtsp: Parse the x-Accept-Dynamic-Rate header

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agortpdec: Add ff_ prefix to all nonstatic symbols
Martin Storsjö [Wed, 12 Oct 2011 09:37:42 +0000 (12:37 +0300)]
rtpdec: Add ff_ prefix to all nonstatic symbols

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agoh264: change unsupported bit depth message to error level
Mans Rullgard [Wed, 12 Oct 2011 11:25:10 +0000 (12:25 +0100)]
h264: change unsupported bit depth message to error level

Unsupported bit depth is certainly an error the user will
want to know about.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agortpdec: Read the packet length for all RTCP packet types
John Brooks [Wed, 12 Oct 2011 06:53:12 +0000 (00:53 -0600)]
rtpdec: Read the packet length for all RTCP packet types

This allows skipping past unsupported RTCP packet types, as
RFC 3550 section 6.1 mandates.

Currently this only has any practical effect if a sender puts
an unrecognized type before RTCP_BYE in a compounded packet, or
(incorrectly) does not put RTCP_SR first.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agortpdec: Fix the minimum packet length for RTCP SR packets
John Brooks [Wed, 12 Oct 2011 08:06:26 +0000 (11:06 +0300)]
rtpdec: Fix the minimum packet length for RTCP SR packets

We actually read 20 bytes of these packets.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agoqdm2: fix signed multiplication overflow
Mans Rullgard [Tue, 11 Oct 2011 16:45:55 +0000 (17:45 +0100)]
qdm2: fix signed multiplication overflow

This code relies on the result wrapping as for unsigned
values, and the sign is not used.  Thus an unsigned type
is proper here.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoflvdec: Don't export metadata entries that aren't proper metadata
Martin Storsjö [Tue, 11 Oct 2011 21:03:20 +0000 (00:03 +0300)]
flvdec: Don't export metadata entries that aren't proper metadata

This avoids writing these entries doubly if transcoding from
flv to flv, since the muxer blindly writes any and all metadata
keys set, in addition to the fixed fields that the muxer
always writes.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agoflvdec: Split out setting of numeric fields from storing metadata
Martin Storsjö [Tue, 11 Oct 2011 20:56:39 +0000 (23:56 +0300)]
flvdec: Split out setting of numeric fields from storing metadata

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agolibx264: support yuv422/444 output.
Anton Khirnov [Sun, 9 Oct 2011 17:34:20 +0000 (19:34 +0200)]
libx264: support yuv422/444 output.

13 years agosws: support yuv444p9/10 output.
Anton Khirnov [Sun, 9 Oct 2011 17:33:30 +0000 (19:33 +0200)]
sws: support yuv444p9/10 output.

13 years agofix AC3ENC_OPT_MODE_ON/OFF
John Stebbins [Tue, 11 Oct 2011 18:57:58 +0000 (14:57 -0400)]
fix AC3ENC_OPT_MODE_ON/OFF

The values were reversed.

Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
13 years agoh264: fix HRD parameters parsing
Mans Rullgard [Tue, 11 Oct 2011 15:00:21 +0000 (16:00 +0100)]
h264: fix HRD parameters parsing

The bit_rate_value_minus1 and cpb_size_value_minus1 elements
allow a wider range than get_ue_golomb() supports.  This
adds a get_ue_golomb_long() function supporting up to 31
leading zeros, which is the maximum for these syntax
elements, and uses it in decode_hrd_parameters().

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoprores: implement multithreading.
Ronald S. Bultje [Thu, 6 Oct 2011 14:17:44 +0000 (07:17 -0700)]
prores: implement multithreading.

60% speedup (overall decoding time) at 2 threads, and another 60%
speedup at 4 threads.

13 years agoprores: idct sse2/sse4 optimizations.
Ronald S. Bultje [Fri, 30 Sep 2011 12:37:11 +0000 (14:37 +0200)]
prores: idct sse2/sse4 optimizations.

~3.0-3.5x as fast as original C version, 1.6x as fast overall.

13 years agoswscale: use aligned move for storage into temporary buffer.
Ronald S. Bultje [Sun, 9 Oct 2011 04:40:01 +0000 (21:40 -0700)]
swscale: use aligned move for storage into temporary buffer.

The intermediate buffer is always aligned.

13 years agoprores: extract idct into its own dspcontext and merge with put_pixels.
Ronald S. Bultje [Thu, 6 Oct 2011 15:03:38 +0000 (08:03 -0700)]
prores: extract idct into its own dspcontext and merge with put_pixels.

13 years agoh264: fix invalid shifts in init_cavlc_level_tab()
Mans Rullgard [Tue, 11 Oct 2011 11:58:31 +0000 (12:58 +0100)]
h264: fix invalid shifts in init_cavlc_level_tab()

The level_code expression includes a shift which is invalid in
those cases where the value is not used.  Moving the calculation
to the branch where the result is used avoids these.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agointfloat_readwrite: fix signed addition overflows
Mans Rullgard [Sat, 8 Oct 2011 01:16:29 +0000 (02:16 +0100)]
intfloat_readwrite: fix signed addition overflows

These additions might overflow the signed range for large
input values.  Converting to unsigned before the addition
rather than after avoids such undefined behaviour.  The
result under normal two's complement wraparound remains
unchanged.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agomov: do not misreport empty stts
Luca Barbato [Tue, 11 Oct 2011 08:14:06 +0000 (10:14 +0200)]
mov: do not misreport empty stts

Return -1 instead of ENOMEM if entries is 0.
Fixes a av_malloc(0) crash in macosx.