platform/upstream/libav.git
13 years agodpcm: output AV_SAMPLE_FMT_U8 for Sol DPCM subcodecs 1 and 2.
Justin Ruggles [Sun, 11 Sep 2011 16:04:46 +0000 (12:04 -0400)]
dpcm: output AV_SAMPLE_FMT_U8 for Sol DPCM subcodecs 1 and 2.

Uses the native sample format for the codec instead of left-shifting all
samples by 8.

13 years agodpcm: calculate and check actual output data size prior to decoding.
Justin Ruggles [Sun, 11 Sep 2011 15:51:08 +0000 (11:51 -0400)]
dpcm: calculate and check actual output data size prior to decoding.

13 years agodpcm: factor out the stereo flag calculation
Justin Ruggles [Sun, 11 Sep 2011 15:19:00 +0000 (11:19 -0400)]
dpcm: factor out the stereo flag calculation

13 years agodpcm: cosmetics: rename channel_number to ch
Justin Ruggles [Sun, 11 Sep 2011 15:13:04 +0000 (11:13 -0400)]
dpcm: cosmetics: rename channel_number to ch

Make the code easier to read.

13 years agoavserver: Fix a bug where the socket is IPv4, but IPv6 is autoselected for the loopba...
Nathan Caldwell [Wed, 28 Sep 2011 00:48:43 +0000 (18:48 -0600)]
avserver: Fix a bug where the socket is IPv4, but IPv6 is autoselected for the loopback address.

This fixes bind(8080): Address family not supported by protocol.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agolavf: Avoid using av_malloc(0) in av_dump_format
Martin Storsjö [Fri, 30 Sep 2011 17:30:35 +0000 (20:30 +0300)]
lavf: Avoid using av_malloc(0) in av_dump_format

On OS X, av_malloc(0) returns pointers that cause crashes when
freed.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agodxva2_h264: pass the correct 8x8 scaling lists
Carl Eugen Hoyos [Wed, 21 Sep 2011 20:42:48 +0000 (20:42 +0000)]
dxva2_h264: pass the correct 8x8 scaling lists

Copy the Inter 8x8 scaling list as second 8x8 matrix into DXVA2's
quantization matrix data structure instead of a potentially unset
Intra chroma scaling matrix.

Fix dxva2 decoding for some H264 samples.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agodca: NEON optimised high freq VQ decoding
Mans Rullgard [Wed, 28 Sep 2011 20:53:03 +0000 (21:53 +0100)]
dca: NEON optimised high freq VQ decoding

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoavcodec: reject audio packets with NULL data and non-zero size
Justin Ruggles [Sat, 24 Sep 2011 01:54:44 +0000 (21:54 -0400)]
avcodec: reject audio packets with NULL data and non-zero size

There is no valid reason the user should ever send such packets in the
first place, but the documentation for CODEC_CAP_DELAY states that the
codec is guaranteed not to get a NULL packet unless that capability is
set. That isn't true without preventing this case.

13 years agodxva: Add ability to enable workaround for older ATI cards
Joakim Plate [Mon, 4 Jul 2011 22:18:32 +0000 (00:18 +0200)]
dxva: Add ability to enable workaround for older ATI cards

The workaround needs to be enabled per PCI ID which cannot be detected inside
libavcodec. So add a flag to manually enable the alternate behavior.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agolatmenc: Set latmBufferFullness to largest value to indicate it is not used
Kieran Kunhya [Fri, 23 Sep 2011 15:51:07 +0000 (10:51 -0500)]
latmenc: Set latmBufferFullness to largest value to indicate it is not used

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agoadpcm: move codec-specific variable declarations to the sections for the correspondin...
Justin Ruggles [Sat, 10 Sep 2011 18:37:09 +0000 (14:37 -0400)]
adpcm: move codec-specific variable declarations to the sections for the corresponding codecs.

13 years agoadpcm: check buffer size in Funcom ISS decoder before reading header.
Justin Ruggles [Sat, 10 Sep 2011 18:24:00 +0000 (14:24 -0400)]
adpcm: check buffer size in Funcom ISS decoder before reading header.

Also use the post-header data size to control termination of the main
decoding loop.

13 years agoadpcm: simplify reading of Funcom ISS frame header.
Justin Ruggles [Sat, 10 Sep 2011 18:21:39 +0000 (14:21 -0400)]
adpcm: simplify reading of Funcom ISS frame header.

13 years agoadpcm: check buffer size in IMA DK4 decoder before reading header.
Justin Ruggles [Sat, 10 Sep 2011 17:54:02 +0000 (13:54 -0400)]
adpcm: check buffer size in IMA DK4 decoder before reading header.

Also use the post-header data size to control termination of the main
decoding loop.

13 years agoadpcm: simplify reading of IMA DK4 frame header.
Justin Ruggles [Sat, 10 Sep 2011 17:52:33 +0000 (13:52 -0400)]
adpcm: simplify reading of IMA DK4 frame header.

13 years agoadpcm_ms: clean up reading of predictor coefficients
Justin Ruggles [Sat, 10 Sep 2011 03:18:03 +0000 (23:18 -0400)]
adpcm_ms: clean up reading of predictor coefficients

13 years agoadpcm_4xm: process planar packets sequentially rather than simultaneously.
Justin Ruggles [Fri, 9 Sep 2011 22:43:19 +0000 (18:43 -0400)]
adpcm_4xm: process planar packets sequentially rather than simultaneously.

Also properly clip the right channel step_index.

13 years agoadpcm_ima_wav: process channel-interleaved blocks sequentially rather than simultaneo...
Justin Ruggles [Fri, 9 Sep 2011 20:26:11 +0000 (16:26 -0400)]
adpcm_ima_wav: process channel-interleaved blocks sequentially rather than simultaneously.

Speeds up the ADPCM IMA WAV decoder by 15-20% overall.

13 years agoadpcm: update reference links
Justin Ruggles [Fri, 9 Sep 2011 18:57:36 +0000 (14:57 -0400)]
adpcm: update reference links

Add Multimedia Wiki link.
Mark dead links with [dead]. Some can still be accessed through archive.org.
Update URLs for pages which have moved.
Replace duplicated links in adpcmenc.c with a note to see the ADPCM decoder
reference documents.

13 years agoadpcm: simplify packet size bounds checking in the ADPCM IMA QT decoder.
Justin Ruggles [Thu, 8 Sep 2011 22:57:56 +0000 (18:57 -0400)]
adpcm: simplify packet size bounds checking in the ADPCM IMA QT decoder.

This is easier to understand. It also avoids returning existing samples mixed
with new samples when the packet is too small.

13 years agoadpcm: simplify and speed up several ADPCM decoders.
Justin Ruggles [Thu, 8 Sep 2011 20:03:44 +0000 (16:03 -0400)]
adpcm: simplify and speed up several ADPCM decoders.

13 years agoadpcm: pretty-print tables
Justin Ruggles [Thu, 8 Sep 2011 03:08:57 +0000 (23:08 -0400)]
adpcm: pretty-print tables

13 years agoprores: Handle 0 or fewer bits left
Alex Converse [Thu, 29 Sep 2011 00:50:51 +0000 (17:50 -0700)]
prores: Handle 0 or fewer bits left

show_bits() is undefined when the number of bits is less than or equal to
zero.

13 years agompeg probe: check the 2/4-bit synchronization value found after a pack_start_code.
Alex Converse [Wed, 28 Sep 2011 22:43:24 +0000 (15:43 -0700)]
mpeg probe: check the 2/4-bit synchronization value found after a pack_start_code.

13 years agodca: clear inactive subbands only once in qmf_32_subbands()
Mans Rullgard [Wed, 28 Sep 2011 20:00:35 +0000 (21:00 +0100)]
dca: clear inactive subbands only once in qmf_32_subbands()

Writing zeros to the high entries in the array need only be
done once as the cutoff position is constant throughout the
loop.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agovf_unsharp: set default chroma size value to 5x5
Stefano Sabatini [Sat, 13 Aug 2011 14:30:44 +0000 (16:30 +0200)]
vf_unsharp: set default chroma size value to 5x5

The previous default value 0x0 was not good, since it is not even
valid.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agovf_unsharp: fix out-of-buffer read
Stefano Sabatini [Fri, 12 Aug 2011 06:42:35 +0000 (08:42 +0200)]
vf_unsharp: fix out-of-buffer read

In apply_unsharp(), when y is >= height, prevent out-of-buffer reading
from src, read from the last buffer line in src2 instead.

The check was implemented in the original unsharp libmpcodecs code and
lost in the port.

This also fixes output discrepancy between the two filters.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agoavconv: Reformat s16 volume adjustment.
Alex Converse [Tue, 27 Sep 2011 17:24:28 +0000 (10:24 -0700)]
avconv: Reformat s16 volume adjustment.

13 years agoARM: NEON optimised vector_fmac_scalar()
Mans Rullgard [Wed, 28 Sep 2011 13:34:54 +0000 (14:34 +0100)]
ARM: NEON optimised vector_fmac_scalar()

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agodca: use vector_fmac_scalar from dsputil
Mans Rullgard [Wed, 28 Sep 2011 13:34:04 +0000 (14:34 +0100)]
dca: use vector_fmac_scalar from dsputil

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agodsputil: add vector_fmac_scalar()
Mans Rullgard [Wed, 28 Sep 2011 13:33:21 +0000 (14:33 +0100)]
dsputil: add vector_fmac_scalar()

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agolatmenc: Fix private options
Kieran Kunhya [Wed, 28 Sep 2011 12:46:10 +0000 (07:46 -0500)]
latmenc: Fix private options

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agovf_unsharp: store hsub/vsub in the filter context
Stefano Sabatini [Fri, 12 Aug 2011 07:30:17 +0000 (09:30 +0200)]
vf_unsharp: store hsub/vsub in the filter context

Also drop obfuscating CHROMA_WIDTH/HEIGHT macros.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agovf_unsharp: adopt a more natural order of params in apply_unsharp()
Stefano Sabatini [Fri, 12 Aug 2011 07:22:31 +0000 (09:22 +0200)]
vf_unsharp: adopt a more natural order of params in apply_unsharp()

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agovf_unsharp: rename method "unsharpen" to "apply_unsharp"
Stefano Sabatini [Fri, 12 Aug 2011 06:47:09 +0000 (08:47 +0200)]
vf_unsharp: rename method "unsharpen" to "apply_unsharp"

More consistent with the original libmpcodecs code, and the name
"unsharpen" was confusing.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agovf_scale: apply the same transform to the aspect during init that is applied per...
Michael Niedermayer [Fri, 26 Aug 2011 23:49:55 +0000 (01:49 +0200)]
vf_scale: apply the same transform to the aspect during init that is applied per frame

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agovf_pad: fix "vsub" variable value computation
Stefano Sabatini [Mon, 4 Jul 2011 09:15:14 +0000 (11:15 +0200)]
vf_pad: fix "vsub" variable value computation

It was shifting 2 rather than 1, +10l.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agovf_scale: add a "sar" variable
Stefano Sabatini [Sat, 2 Jul 2011 15:27:31 +0000 (17:27 +0200)]
vf_scale: add a "sar" variable

Also create a "dar" alias for the "a" variable, for avoiding possible
confusion between dar/sar.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agolavfi: fix realloc size computation in avfilter_add_format()
Stefano Sabatini [Sun, 19 Jun 2011 20:07:18 +0000 (22:07 +0200)]
lavfi: fix realloc size computation in avfilter_add_format()

Replace sizeof((*avff)->formats)
with    sizeof(*(*avff)->formats)

as the size of the array element is given by the pointed element
rather than by its pointer.

In particular fix computation with the pending patch when
sizeof(int64_t) != sizeof(int64_t *).

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agovsrc_color: use internal timebase
Stefano Sabatini [Fri, 17 Jun 2011 23:52:40 +0000 (01:52 +0200)]
vsrc_color: use internal timebase

Avoid timescale conversion, simplify.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agolavfi: fix signature for avfilter_graph_parse() and avfilter_graph_config()
Stefano Sabatini [Sat, 11 Jun 2011 09:41:49 +0000 (11:41 +0200)]
lavfi: fix signature for avfilter_graph_parse() and avfilter_graph_config()

Require "void *" rather than "AVClass *" for the log context type.

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agographparser: prefer void * over AVClass * for log contexts
Stefano Sabatini [Sat, 11 Jun 2011 12:33:09 +0000 (14:33 +0200)]
graphparser: prefer void * over AVClass * for log contexts

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agoavfiltergraph: use meaningful error codes
Stefano Sabatini [Sat, 11 Jun 2011 09:15:40 +0000 (11:15 +0200)]
avfiltergraph: use meaningful error codes

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agoavconv: Initialize return value for codec copy path.
Alex Converse [Tue, 27 Sep 2011 22:27:18 +0000 (15:27 -0700)]
avconv: Initialize return value for codec copy path.

13 years agofate: use 'run' helper for seek-test
Mans Rullgard [Wed, 28 Sep 2011 01:13:31 +0000 (02:13 +0100)]
fate: use 'run' helper for seek-test

This is simpler, and the actual seek-test command is printed with V=1.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agofate: remove seek-mpeg2reuse test
Mans Rullgard [Wed, 28 Sep 2011 01:11:14 +0000 (02:11 +0100)]
fate: remove seek-mpeg2reuse test

The input file for this test is no longer generated.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoFix memory (re)allocation in matroskadec.c, related to MSVR-11-0080.
Michael Niedermayer [Thu, 28 Jul 2011 12:59:54 +0000 (14:59 +0200)]
Fix memory (re)allocation in matroskadec.c, related to MSVR-11-0080.

Whitespace of the patch cleaned up by Aurel
Some of the issues have been reported by Steve Manzuik / Microsoft Vulnerability Research (MSVR)
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
(cherry picked from commit 956c901c68eff78288f40e3c8f41ee2fa081d4a8)

Further suggestions from Kostya <kostya.shishkov@gmail.com> have been
implemented by Reinhard Tartler <siretart@tauware.de>

Signed-off-by: Reinhard Tartler <siretart@tauware.de>
13 years agoavconv: Fix volume adjustment for non-s16 sample formats
Alex Converse [Tue, 27 Sep 2011 17:15:07 +0000 (10:15 -0700)]
avconv: Fix volume adjustment for non-s16 sample formats

13 years agoavconv: Make samples void*.
Alex Converse [Tue, 27 Sep 2011 17:21:39 +0000 (10:21 -0700)]
avconv: Make samples void*.

Different sample formats are different sizes.

13 years agoavconv: Use the size of the decode sample format when allocating the audio samples...
Alex Converse [Tue, 27 Sep 2011 17:20:23 +0000 (10:20 -0700)]
avconv: Use the size of the decode sample format when allocating the audio samples buffer.

13 years agoavconv: use different variables for decoded and filtered frame.
Anton Khirnov [Fri, 23 Sep 2011 14:17:42 +0000 (16:17 +0200)]
avconv: use different variables for decoded and filtered frame.

Makes the code less obfuscated and fixes encoding one video stream to
several outputs.

Also use avcodec_alloc_frame() instead of allocating AVFrame on stack.

Breaks me_threshold in avconv, as motion vectors aren't passed through
lavfi. They could be copied manually, but I don't think this misfeature
is useful enough to justify ugly hacks.

13 years agoavconv: add support for copying attachments.
Anton Khirnov [Tue, 27 Sep 2011 08:37:22 +0000 (10:37 +0200)]
avconv: add support for copying attachments.

13 years agomatroskaenc: write attachments.
Anton Khirnov [Sat, 30 Jul 2011 11:37:36 +0000 (13:37 +0200)]
matroskaenc: write attachments.

13 years agomatroskadec: export mimetype of attachments as metadata.
Anton Khirnov [Tue, 27 Sep 2011 08:36:53 +0000 (10:36 +0200)]
matroskadec: export mimetype of attachments as metadata.

13 years agoavconv: factorize common code from new_*_stream()
Anton Khirnov [Tue, 27 Sep 2011 05:42:29 +0000 (07:42 +0200)]
avconv: factorize common code from new_*_stream()

13 years agodoc/avconv: expand documentation for some options.
Anton Khirnov [Mon, 12 Sep 2011 19:03:52 +0000 (21:03 +0200)]
doc/avconv: expand documentation for some options.

13 years agodoc/avconv: document -timelimit.
Anton Khirnov [Mon, 12 Sep 2011 18:40:29 +0000 (20:40 +0200)]
doc/avconv: document -timelimit.

13 years agofate: allow testing with libavfilter disabled
Mans Rullgard [Sun, 17 Jul 2011 10:55:08 +0000 (11:55 +0100)]
fate: allow testing with libavfilter disabled

This declares dependencies to skip tests using libavfilter
when it is disabled.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agox86: XOP/FMA4 CPU detection support
Jason Garrett-Glaser [Mon, 26 Sep 2011 21:44:47 +0000 (14:44 -0700)]
x86: XOP/FMA4 CPU detection support

13 years agows_snd: misc cosmetic clean-ups
Justin Ruggles [Mon, 12 Sep 2011 14:28:56 +0000 (10:28 -0400)]
ws_snd: misc cosmetic clean-ups

13 years agows_snd: remove the 2-bit ADPCM table and just subtract 2 instead.
Justin Ruggles [Mon, 12 Sep 2011 14:22:31 +0000 (10:22 -0400)]
ws_snd: remove the 2-bit ADPCM table and just subtract 2 instead.

13 years agows_snd: use memcpy() and memset() instead of loops
Justin Ruggles [Mon, 12 Sep 2011 14:04:37 +0000 (10:04 -0400)]
ws_snd: use memcpy() and memset() instead of loops

13 years agows_snd: use samples pointer for loop termination instead of a separate
Justin Ruggles [Mon, 12 Sep 2011 13:59:13 +0000 (09:59 -0400)]
ws_snd: use samples pointer for loop termination instead of a separate
iterator variable.

13 years agows_snd: make sure number of channels is 1
Justin Ruggles [Mon, 12 Sep 2011 13:44:21 +0000 (09:44 -0400)]
ws_snd: make sure number of channels is 1

13 years agows_snd: add some checks to prevent buffer overread or overwrite.
Justin Ruggles [Mon, 12 Sep 2011 13:41:06 +0000 (09:41 -0400)]
ws_snd: add some checks to prevent buffer overread or overwrite.

13 years agows_snd: decode to AV_SAMPLE_FMT_U8 instead of S16.
Justin Ruggles [Mon, 12 Sep 2011 12:55:43 +0000 (08:55 -0400)]
ws_snd: decode to AV_SAMPLE_FMT_U8 instead of S16.

8-bit unsigned is the native sample format.

13 years agoflacdec: fix buffer size checking in get_metadata_size()
Justin Ruggles [Tue, 13 Sep 2011 19:13:44 +0000 (15:13 -0400)]
flacdec: fix buffer size checking in get_metadata_size()

Adds an additional check before reading the next block header and avoids a
potential integer overflow when checking the metadata size against the
remaining buffer size.

13 years agortp: Simplify ff_rtp_get_payload_type
Rafaël Carré [Mon, 26 Sep 2011 15:56:49 +0000 (11:56 -0400)]
rtp: Simplify ff_rtp_get_payload_type

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agortpenc: Add a payload type private option
Rafaël Carré [Mon, 26 Sep 2011 15:56:48 +0000 (11:56 -0400)]
rtpenc: Add a payload type private option

Specifying the payload type is useful when the type number has
already been negotiated before creating the stream, for example
in SIP protocol.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agortp: Correct ff_rtp_get_payload_type documentation
Rafaël Carré [Mon, 26 Sep 2011 15:56:47 +0000 (11:56 -0400)]
rtp: Correct ff_rtp_get_payload_type documentation

Since 0c378ea1f, it can't fail anymore.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agoavconv: replace all fprintf() by av_log().
Anton Khirnov [Tue, 13 Sep 2011 05:31:21 +0000 (07:31 +0200)]
avconv: replace all fprintf() by av_log().

Deprecate -v option, now only -loglevel is enough.

13 years agoavconv: change av_log verbosity from ERROR to FATAL for fatal errors.
Anton Khirnov [Mon, 12 Sep 2011 19:51:02 +0000 (21:51 +0200)]
avconv: change av_log verbosity from ERROR to FATAL for fatal errors.

13 years agocmdutils: replace fprintf() by av_log()
Anton Khirnov [Mon, 26 Sep 2011 06:39:43 +0000 (08:39 +0200)]
cmdutils: replace fprintf() by av_log()

This way, all messages can be easily disabled with -loglevel.

Behavior change -- configuration is printed with AV_LOG_VERBOSE, so it's
not shown by default.

13 years agoavtools: parse loglevel before all the other options.
Anton Khirnov [Mon, 26 Sep 2011 06:15:37 +0000 (08:15 +0200)]
avtools: parse loglevel before all the other options.

This way it can be applied to cmdutils too -- e.g. showing the banner
and printing startup messages.

13 years agooggdec: add support for Xiph's CELT codec
Nicolas George [Mon, 26 Sep 2011 11:19:05 +0000 (13:19 +0200)]
oggdec: add support for Xiph's CELT codec

This patch also introduces CODEC_ID_CELT.

Signed-off-by: Nicolas George <nicolas.george@normalesup.org>
Signed-off-by: Diego Biurrun <diego@biurrun.de>
13 years agosol: return error if av_get_packet() fails.
Justin Ruggles [Sat, 24 Sep 2011 01:43:43 +0000 (21:43 -0400)]
sol: return error if av_get_packet() fails.

This prevents sending a packet with data=NULL size=AVERROR_EOF.

13 years agocosmetics: reindent and pretty-print
Justin Ruggles [Mon, 12 Sep 2011 16:18:00 +0000 (12:18 -0400)]
cosmetics: reindent and pretty-print

13 years agolibgsm: split init and close fuctions for encoder and decoder.
Justin Ruggles [Mon, 12 Sep 2011 16:13:05 +0000 (12:13 -0400)]
libgsm: split init and close fuctions for encoder and decoder.

The decoder shouldn't be allocating coded_frame. It's also cleaner to split
them than to check avctx->codec->encode and avctx->codec->decode.

13 years agoppc: fix some pointer to integer casts
Mans Rullgard [Sun, 25 Sep 2011 17:27:47 +0000 (18:27 +0100)]
ppc: fix some pointer to integer casts

Use uintptr_t instead of plain int.  Without this change, the
comparisons will come out wrong for pointers in certain ranges.
Fixes random failures on ppc64.  Also fixes some compiler warnings.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoppc: fix 32-bit PIC build
Mans Rullgard [Sun, 25 Sep 2011 11:53:44 +0000 (12:53 +0100)]
ppc: fix 32-bit PIC build

On 32-bit ppc, the GOT pointer must be loaded manually.
This adds a "get_got" assembler macro to compute the
GOT address.  The "movrel" macro is updated to take an
additional parameter containing the GOT address since
no register is reserved for this purpose on ppc32.
These changes have no effect on ppc64 builds.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agovmdaudio: fix decoding of 16-bit audio format.
Justin Ruggles [Mon, 12 Sep 2011 00:17:54 +0000 (20:17 -0400)]
vmdaudio: fix decoding of 16-bit audio format.

The initial sample of each block is raw 16-bit PCM, not DPCM.
Fixes decoding of all samples in:
http://streams.videolan.org/samples/game-formats/sierra-vmd/Lighthouse/

13 years agolavf: do not set codec_tag for rawvideo
Mans Rullgard [Sun, 17 Jul 2011 15:02:33 +0000 (16:02 +0100)]
lavf: do not set codec_tag for rawvideo

If the demuxer did not set a codec_tag, there is none and
inventing one makes no sense.  This change stops the rawvideo
"decoder" over-writing user-supplied pixfmt with one derived
from the codec_tag.  The pixfmt-codec_tag-pixfmt round-trip
is lossy since several pixfmts map to the same codec_tag.

This fixes fate-lavf-pixfmt with avfilter disabled.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoh264: check for out of bounds reads in ff_h264_decode_extradata().
Laurent Aimar [Sat, 24 Sep 2011 14:16:39 +0000 (16:16 +0200)]
h264: check for out of bounds reads in ff_h264_decode_extradata().

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agoflvdec: Check for overflow before allocating arrays
Michael Niedermayer [Sat, 24 Sep 2011 15:57:31 +0000 (18:57 +0300)]
flvdec: Check for overflow before allocating arrays

On allocation, the array length is multiplied by sizeof(int64_t),
this prevents the multiplication from overflowing.

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agoavconv: use correct output stream index when checking max_frames
Anton Khirnov [Fri, 23 Sep 2011 12:40:29 +0000 (14:40 +0200)]
avconv: use correct output stream index when checking max_frames

13 years agoavconv: remove fake coded_frame on streamcopy hack
Anton Khirnov [Fri, 23 Sep 2011 10:46:05 +0000 (12:46 +0200)]
avconv: remove fake coded_frame on streamcopy hack

Back in 2002 when it was added, some muxers used information from
coded_frame. It's not true anymore, so this hack is useless and can be
removed.

13 years agoflvdec: Fix invalid pointer deferences when parsing index
Laurent Aimar [Sat, 24 Sep 2011 14:16:38 +0000 (16:16 +0200)]
flvdec: Fix invalid pointer deferences when parsing index

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agoconfigure: disable hardware capabilities ELF section with suncc on Solaris x86
Sean McGovern [Sat, 17 Sep 2011 04:36:37 +0000 (00:36 -0400)]
configure: disable hardware capabilities ELF section with suncc on Solaris x86

When using suncc to build, the Solaris linker will mark
an executable with each instruction set encountered by
the Solaris assembler.  As our libraries contain their own
guards for processor-specific code, instead suppress
generation of the HWCAPS ELF section on Solaris x86 only.

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agoUse explicit struct initializers for AVCodec declarations.
Diego Biurrun [Fri, 23 Sep 2011 19:11:15 +0000 (21:11 +0200)]
Use explicit struct initializers for AVCodec declarations.

13 years agoUse explicit struct initializers for AVOutputFormat/AVInputFormat declarations.
Diego Biurrun [Fri, 23 Sep 2011 18:50:11 +0000 (20:50 +0200)]
Use explicit struct initializers for AVOutputFormat/AVInputFormat declarations.

13 years agoadpcmenc: Set bits_per_coded_sample
Justin Ruggles [Thu, 8 Sep 2011 02:52:06 +0000 (22:52 -0400)]
adpcmenc: Set bits_per_coded_sample

13 years agoadpcmenc: fix QT IMA ADPCM encoder
Baptiste Coudurier [Thu, 8 Sep 2011 02:27:03 +0000 (22:27 -0400)]
adpcmenc: fix QT IMA ADPCM encoder

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
13 years agoadpcmdec: Fix QT IMA ADPCM decoder
Baptiste Coudurier [Sun, 8 May 2011 11:13:17 +0000 (13:13 +0200)]
adpcmdec: Fix QT IMA ADPCM decoder

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
13 years agopermit decoding of multichannel ADPCM_EA_XAS
Peter Ross [Sat, 23 Apr 2011 12:08:48 +0000 (22:08 +1000)]
permit decoding of multichannel ADPCM_EA_XAS

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
13 years agoFix input buffer size check in adpcm_ea decoder.
Reimar Döffinger [Thu, 8 Sep 2011 02:14:07 +0000 (22:14 -0400)]
Fix input buffer size check in adpcm_ea decoder.

Unfortunately the output buffer size check assumes that the
input buffer is never over-consumed, thus this actually
also allowed to write outside the output buffer if "lucky".

Based on:
git.videolan.org/ffmpeg.git
commit 701d0eb185192542c4a17f296e39e37cedf7abc6

13 years agofft: avoid a signed overflow
Sean McGovern [Tue, 20 Sep 2011 01:32:09 +0000 (21:32 -0400)]
fft: avoid a signed overflow

As a signed integer, 1<<31 overflows, so force it to unsigned.

Signed-off-by: Alex Converse <alex.converse@gmail.com>
13 years agompegps: Handle buffer exhaustion when reading packets.
Alex Converse [Fri, 23 Sep 2011 23:28:23 +0000 (16:28 -0700)]
mpegps: Handle buffer exhaustion when reading packets.

13 years agortp: factorize dynamic payload type fallback
Rafaël Carré [Fri, 23 Sep 2011 19:47:56 +0000 (15:47 -0400)]
rtp: factorize  dynamic payload type fallback

Move the identical code in rtp_write_header() and
ff_sdp_write_media() inside ff_rtp_get_payload_type()

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
13 years agoflvdec: Ignore the index if it's from a creator known to be different
Martin Storsjö [Wed, 21 Sep 2011 13:51:13 +0000 (16:51 +0300)]
flvdec: Ignore the index if it's from a creator known to be different

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