platform/upstream/libav.git
13 years agoalacdec: cosmetics: use 'ch' instead of 'chan' to iterate channels
Justin Ruggles [Sun, 9 Oct 2011 18:21:35 +0000 (14:21 -0400)]
alacdec: cosmetics: use 'ch' instead of 'chan' to iterate channels

13 years agoalacdec: move some declarations to the top of the function
Justin Ruggles [Sun, 9 Oct 2011 18:19:32 +0000 (14:19 -0400)]
alacdec: move some declarations to the top of the function

13 years agoalacdec: always use get_sbits_long() for uncompressed samples
Justin Ruggles [Sun, 9 Oct 2011 18:02:52 +0000 (14:02 -0400)]
alacdec: always use get_sbits_long() for uncompressed samples

13 years agoalacdec: remove unneeded local variable
Justin Ruggles [Sun, 9 Oct 2011 18:02:21 +0000 (14:02 -0400)]
alacdec: remove unneeded local variable

13 years agoalacdec: remove the numchannels parameter from several functions.
Justin Ruggles [Sun, 9 Oct 2011 17:36:01 +0000 (13:36 -0400)]
alacdec: remove the numchannels parameter from several functions.

They only operate on stereo content, so the extra param is not necessary and
also allows for simplifying the code.

13 years agoalacdec: rename 2 functions.
Justin Ruggles [Sun, 9 Oct 2011 17:31:03 +0000 (13:31 -0400)]
alacdec: rename 2 functions.

Now they only do stereo interleaving.

13 years agoalacdec: move appending of extra_bits to a separate function.
Justin Ruggles [Sun, 9 Oct 2011 17:27:16 +0000 (13:27 -0400)]
alacdec: move appending of extra_bits to a separate function.

This should also fix decoding of mono 24-bit.

13 years agoalacdec: split stereo decorrelation into a separate function.
Justin Ruggles [Sun, 9 Oct 2011 17:16:14 +0000 (13:16 -0400)]
alacdec: split stereo decorrelation into a separate function.

It is identical for 16-bit and 24-bit, so there is no need to have duplicate
code.

13 years agoalacdec: cosmetics: rename 'wasted_bits' to 'extra_bits'.
Justin Ruggles [Sun, 9 Oct 2011 17:00:39 +0000 (13:00 -0400)]
alacdec: cosmetics: rename 'wasted_bits' to 'extra_bits'.

The bits are not wasted, they are additional low bits that are added to the
16-bit decompressed samples to increase the output sample depth.

13 years agoalacdec: remove unneeded numsamples checks
Justin Ruggles [Thu, 6 Oct 2011 00:17:32 +0000 (20:17 -0400)]
alacdec: remove unneeded numsamples checks

13 years agoalacdec: check for buffer allocation failure.
Justin Ruggles [Thu, 6 Oct 2011 00:14:48 +0000 (20:14 -0400)]
alacdec: check for buffer allocation failure.

Also rearranges some functions for easier cleanup on failure.

13 years agoalacdec: allocate per-channel buffers based on channel count.
Justin Ruggles [Thu, 6 Oct 2011 00:08:46 +0000 (20:08 -0400)]
alacdec: allocate per-channel buffers based on channel count.

reduces memory usage when the stream has fewer than MAX_CHANNELS

13 years agoalacdec: read/validate number of channels from the extradata.
Justin Ruggles [Thu, 6 Oct 2011 00:07:29 +0000 (20:07 -0400)]
alacdec: read/validate number of channels from the extradata.

check frame header channel count against header/container channel count.

13 years agoalacdec: remove unneeded validation of setinfo_sample_size.
Justin Ruggles [Wed, 5 Oct 2011 23:46:46 +0000 (19:46 -0400)]
alacdec: remove unneeded validation of setinfo_sample_size.

It is already done when using it to set sample_fmt.

13 years agoalacdec: set sample_fmt in alac_decode_init()
Justin Ruggles [Wed, 5 Oct 2011 23:43:49 +0000 (19:43 -0400)]
alacdec: set sample_fmt in alac_decode_init()

13 years agoalacdec: set bytespersample using av_get_bytes_per_sample()
Justin Ruggles [Wed, 5 Oct 2011 23:42:26 +0000 (19:42 -0400)]
alacdec: set bytespersample using av_get_bytes_per_sample()

13 years agothreads: restore has_b_frames in frame_thread_free
Janne Grunau [Wed, 26 Oct 2011 13:28:29 +0000 (15:28 +0200)]
threads: restore has_b_frames in frame_thread_free

Otherwise the delay expressed in has_b_frames increases with every
avcodec_close/avcodec_open.
Fixes fate-ea-dct with more than 1 thread.

13 years agoH.264: Cometics to dsputil_mmx.c
Daniel Kang [Wed, 26 Oct 2011 13:25:48 +0000 (09:25 -0400)]
H.264: Cometics to dsputil_mmx.c

Add whitespace.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
13 years agog722dec: check output buffer size before decoding
Justin Ruggles [Sun, 23 Oct 2011 17:00:33 +0000 (13:00 -0400)]
g722dec: check output buffer size before decoding

13 years agog722dec: cosmetics: reindent/linewrap
Justin Ruggles [Sun, 23 Oct 2011 17:02:04 +0000 (13:02 -0400)]
g722dec: cosmetics: reindent/linewrap

13 years agog722dec: remove the use of lowres for half-rate decoding.
Justin Ruggles [Sat, 22 Oct 2011 22:29:14 +0000 (18:29 -0400)]
g722dec: remove the use of lowres for half-rate decoding.

It is broken because an AVCodecContext can be opened/closed multiple
times, and sample_rate is getting divided by 2 each time that happens.

This removes the only use of lowres for audio.

13 years agotta: check for extradata allocation failure in tta demuxer
Justin Ruggles [Fri, 23 Sep 2011 23:22:06 +0000 (19:22 -0400)]
tta: check for extradata allocation failure in tta demuxer

13 years agotta: check for allocation failure of decode_buffer
Justin Ruggles [Mon, 24 Oct 2011 13:20:07 +0000 (09:20 -0400)]
tta: check for allocation failure of decode_buffer

13 years agotta: use correct frame_length calculation.
Justin Ruggles [Sun, 23 Oct 2011 20:25:02 +0000 (16:25 -0400)]
tta: use correct frame_length calculation.

using a floating-point calculation is not necessary.

13 years agotta: add support for decoding 24-bit sample format
Justin Ruggles [Sun, 23 Oct 2011 20:13:06 +0000 (16:13 -0400)]
tta: add support for decoding 24-bit sample format

Note that this will not work in most cases with avconv and avplay due to the
AVCODEC_MAX_AUDIO_FRAME_SIZE limit, but it will decode correctly if given a
large enough output buffer.

13 years agocosmetics: indentation
Justin Ruggles [Thu, 22 Sep 2011 17:21:38 +0000 (13:21 -0400)]
cosmetics: indentation

13 years agotta: remove pointless braces
Justin Ruggles [Thu, 22 Sep 2011 17:19:36 +0000 (13:19 -0400)]
tta: remove pointless braces

13 years agotta: check output buffer size after adjusting frame length for last frame
Justin Ruggles [Thu, 22 Sep 2011 17:09:05 +0000 (13:09 -0400)]
tta: check output buffer size after adjusting frame length for last frame

13 years agotta: fix reading of format in TTA header.
Justin Ruggles [Thu, 22 Sep 2011 16:54:28 +0000 (12:54 -0400)]
tta: fix reading of format in TTA header.

TTA does not support float at all, and format 2 is encrypted TTA.

13 years agotta: remove useless commented-out lines
Justin Ruggles [Wed, 21 Sep 2011 18:17:56 +0000 (14:17 -0400)]
tta: remove useless commented-out lines

13 years agotta: check remaining bitstream size while reading unary value
Justin Ruggles [Wed, 21 Sep 2011 18:16:24 +0000 (14:16 -0400)]
tta: check remaining bitstream size while reading unary value

13 years agolavf: deprecate AVStream.stream_copy
Anton Khirnov [Sun, 23 Oct 2011 09:22:33 +0000 (11:22 +0200)]
lavf: deprecate AVStream.stream_copy

It's only used in avconv, so it properly belongs to OutputStream struct
there.

13 years agoavconc: split choose_codec() to choose_decoder/choose_encoder.
Anton Khirnov [Sun, 23 Oct 2011 09:10:27 +0000 (11:10 +0200)]
avconc: split choose_codec() to choose_decoder/choose_encoder.

Prevents -c copy from working for input streams and allows to move
stream_copy variable from AVStream to OutputStream.

13 years agolavf: simplify by using FFMAX/FFMIN.
Anton Khirnov [Tue, 18 Oct 2011 07:17:12 +0000 (09:17 +0200)]
lavf: simplify by using FFMAX/FFMIN.

13 years agompegenc: add preload private option.
Anton Khirnov [Sun, 9 Oct 2011 13:04:38 +0000 (15:04 +0200)]
mpegenc: add preload private option.

Deprecate AVFormatContext.preload.

13 years agocosmetics: simplify latm_decode_init
Janne Grunau [Fri, 21 Oct 2011 15:14:58 +0000 (17:14 +0200)]
cosmetics: simplify latm_decode_init

13 years agolatm: avoid unnecessary reinit of the aac decoder
Janne Grunau [Fri, 21 Oct 2011 15:13:03 +0000 (17:13 +0200)]
latm: avoid unnecessary reinit of the aac decoder

13 years agoaacdec: initialize sbr context only in new channel elements
Janne Grunau [Sat, 22 Oct 2011 23:11:44 +0000 (01:11 +0200)]
aacdec: initialize sbr context only in new channel elements

13 years agoresample: reject unhandled conversions
Janne Grunau [Sat, 22 Oct 2011 22:31:16 +0000 (00:31 +0200)]
resample: reject unhandled conversions

audio_resample can not reduce the number of channels

13 years agoresample: remove unused #define
Janne Grunau [Sat, 22 Oct 2011 22:33:44 +0000 (00:33 +0200)]
resample: remove unused #define

13 years agoadpcm: use sign_extend()
Mans Rullgard [Sun, 23 Oct 2011 23:08:23 +0000 (00:08 +0100)]
adpcm: use sign_extend()

This avoids warnings from the overflow checker and simplifies the code.

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agompeg12: fix mpeg_decode_slice context parameter type
Janne Grunau [Sun, 23 Oct 2011 23:05:00 +0000 (01:05 +0200)]
mpeg12: fix mpeg_decode_slice context parameter type

During slice threading only MpegEncContext is passed to
mpeg_decode_slice, remove a wrong cast and change the function
definition to take MpegEncContext pointer.

13 years agoRevert "mpeg12: move full_pel from MpegEncContext to Mpeg1Context"
Janne Grunau [Sun, 23 Oct 2011 22:59:41 +0000 (00:59 +0200)]
Revert "mpeg12: move full_pel from MpegEncContext to Mpeg1Context"

This reverts commit da22ba7df461c13bf0b0eabc953303803a285d91 since it
broke slice threading. Slice threading just duplicates MpegEncContext
so every value used during mpeg_decode_slice has to be in it.
A second patch will fix the illusion that Mpeg1Context is available
in mpeg_decode_slice.

13 years agoMove id3v2 tag writing to a separate file.
Michael Karcher [Fri, 21 Oct 2011 21:32:32 +0000 (23:32 +0200)]
Move id3v2 tag writing to a separate file.

Signed-off-by: Diego Biurrun <diego@biurrun.de>
13 years agoswscale: add missing colons to x86 assembly yuv2planeX.
Ronald S. Bultje [Sun, 23 Oct 2011 16:34:36 +0000 (09:34 -0700)]
swscale: add missing colons to x86 assembly yuv2planeX.

This fixes assembling using "nasm".

13 years agog722: split decoder and encoder into separate files
Justin Ruggles [Sat, 22 Oct 2011 21:49:50 +0000 (17:49 -0400)]
g722: split decoder and encoder into separate files

13 years agocosmetics: remove extra spaces before end-of-statement semi-colons
Justin Ruggles [Sat, 24 Sep 2011 00:00:11 +0000 (20:00 -0400)]
cosmetics: remove extra spaces before end-of-statement semi-colons

13 years agovorbisdec: check output buffer size before writing output
Justin Ruggles [Fri, 23 Sep 2011 23:56:58 +0000 (19:56 -0400)]
vorbisdec: check output buffer size before writing output

13 years agowavpack: calculate bpp using av_get_bytes_per_sample()
Justin Ruggles [Sat, 24 Sep 2011 03:19:53 +0000 (23:19 -0400)]
wavpack: calculate bpp using av_get_bytes_per_sample()

13 years agoac3enc: Set max value for mode options correctly
John Stebbins [Sun, 23 Oct 2011 15:24:26 +0000 (11:24 -0400)]
ac3enc: Set max value for mode options correctly

for example:
Attempting to set ac3 dsur_mode to "on" results in:
[AC-3 Encoder @ 0x7f508f0a3280] Value 2.000000 for parameter 'dsur_mode' out of range
[AC-3 Encoder @ 0x7f508f0a3280] Error setting option dsur_mode to value on.

Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
13 years agolavc: move get_b_cbp() from h263.h to mpeg4videoenc.c
Anton Khirnov [Fri, 21 Oct 2011 14:02:29 +0000 (16:02 +0200)]
lavc: move get_b_cbp() from h263.h to mpeg4videoenc.c

It's not used anywhere else.

13 years agompeg12: move closed_gop from MpegEncContext to Mpeg1Context
Anton Khirnov [Wed, 19 Oct 2011 07:28:45 +0000 (09:28 +0200)]
mpeg12: move closed_gop from MpegEncContext to Mpeg1Context

It's MPEG-1/2 specific.

13 years agompeg12: move full_pel from MpegEncContext to Mpeg1Context
Anton Khirnov [Wed, 19 Oct 2011 07:28:45 +0000 (09:28 +0200)]
mpeg12: move full_pel from MpegEncContext to Mpeg1Context

It's MPEG-1 specific.

13 years agompeg12: move Mpeg1Context from mpeg12.c to mpeg12.h
Anton Khirnov [Wed, 19 Oct 2011 07:25:52 +0000 (09:25 +0200)]
mpeg12: move Mpeg1Context from mpeg12.c to mpeg12.h

It will be used in vdpau code.

13 years agompegvideo: remove some unused variables from MpegEncContext.
Anton Khirnov [Tue, 18 Oct 2011 17:53:25 +0000 (19:53 +0200)]
mpegvideo: remove some unused variables from MpegEncContext.

13 years agoid3v2: fix doxy comment - 'machine byte order' makes no sense on char arrays
Michael Karcher [Sat, 22 Oct 2011 13:41:30 +0000 (15:41 +0200)]
id3v2: fix doxy comment - 'machine byte order' makes no sense on char arrays

Signed-off-by: Diego Biurrun <diego@biurrun.de>
13 years agoVC1: restore mistakenly removed code
Mashiat Sarker Shakkhar [Sat, 22 Oct 2011 14:41:04 +0000 (19:41 +0500)]
VC1: restore mistakenly removed code

The code was mistakenly removed in cad16562c8d76ea2a2a6495f29296c3ff7966946.
It stored some motion vector data for future use in B-pictures.

This fixes Bugzilla bug #57.

Signed-off-by: Diego Biurrun <diego@biurrun.de>
13 years agotwinvq: check output buffer size before decoding
Justin Ruggles [Fri, 23 Sep 2011 23:50:41 +0000 (19:50 -0400)]
twinvq: check output buffer size before decoding

13 years agotwinvq: return an error when the packet size is too small
Justin Ruggles [Fri, 23 Sep 2011 23:25:47 +0000 (19:25 -0400)]
twinvq: return an error when the packet size is too small

13 years agolavf: export some forgotten symbols with non-av prefixes.
Anton Khirnov [Sat, 22 Oct 2011 19:08:31 +0000 (21:08 +0200)]
lavf: export some forgotten symbols with non-av prefixes.

13 years agoswscale: update altivec yuv2planeX asm to new per-plane API.
Ronald S. Bultje [Sat, 15 Oct 2011 23:00:42 +0000 (16:00 -0700)]
swscale: update altivec yuv2planeX asm to new per-plane API.

13 years agoswscale: make yuv2yuvX_10_sse2/avx 8/9/16-bits aware.
Ronald S. Bultje [Sat, 15 Oct 2011 21:19:33 +0000 (14:19 -0700)]
swscale: make yuv2yuvX_10_sse2/avx 8/9/16-bits aware.

Also implement MMX/MMX2 versions and SSE4 versions.

13 years agoyuv2planeX10 SIMD
Kieran Kunhya [Sun, 9 Oct 2011 15:20:48 +0000 (16:20 +0100)]
yuv2planeX10 SIMD

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
13 years agoswscale: decide whether to use yuv2plane1/X on a per-plane basis.
Ronald S. Bultje [Wed, 5 Oct 2011 14:37:24 +0000 (07:37 -0700)]
swscale: decide whether to use yuv2plane1/X on a per-plane basis.

13 years agoswscale: reintroduce full precision in 16-bit output.
Ronald S. Bultje [Wed, 5 Oct 2011 13:27:52 +0000 (06:27 -0700)]
swscale: reintroduce full precision in 16-bit output.

13 years agoSplit up yuv2yuvX functions
Kieran Kunhya [Tue, 4 Oct 2011 11:22:03 +0000 (12:22 +0100)]
Split up yuv2yuvX functions

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
13 years agoSplit out yuv2yuv1 luma and chroma in order to make them generic DSP functions
Kieran Kunhya [Fri, 30 Sep 2011 12:50:49 +0000 (13:50 +0100)]
Split out yuv2yuv1 luma and chroma in order to make them generic DSP functions

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
13 years agolavc: replace references to deprecated AVCodecContext.error_recognition to use AVCode...
Dustin Brody [Sat, 22 Oct 2011 12:44:16 +0000 (08:44 -0400)]
lavc: replace references to deprecated AVCodecContext.error_recognition to use AVCodecContext.err_recognition

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agolavc: translate non-flag-based er options into flag-based ef options at codec open
Dustin Brody [Sat, 22 Oct 2011 12:44:15 +0000 (08:44 -0400)]
lavc: translate non-flag-based er options into flag-based ef options at codec open

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agoadd -err_filter AVOptions to access flag-based error recognition
Dustin Brody [Sat, 22 Oct 2011 12:44:14 +0000 (08:44 -0400)]
add -err_filter AVOptions to access flag-based error recognition

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agoh264_weight: initialize "height" function argument properly.
Ronald S. Bultje [Sat, 22 Oct 2011 01:00:39 +0000 (18:00 -0700)]
h264_weight: initialize "height" function argument properly.

Right now it's not actually initialized on 32-bit, leading to crashes
on win32.

13 years agopresets: spelling error in libvpx 1080p50_60
Lou Logan [Thu, 20 Oct 2011 21:21:31 +0000 (21:21 +0000)]
presets: spelling error in libvpx 1080p50_60

Signed-off-by: Janne Grunau <janne-libav@jannau.net>
13 years agoavplay: fix fullscreen behaviour with SDL 1.2.14 on Mac OS X
Marton Balint [Fri, 21 Oct 2011 12:56:57 +0000 (14:56 +0200)]
avplay: fix fullscreen behaviour with SDL 1.2.14 on Mac OS X

The blue channel went missing when toggling fullscreen on Mac OS X.

13 years agoflvdec: Do not call parse_keyframes_index with a NULL stream
Reimar Döffinger [Wed, 19 Oct 2011 16:41:02 +0000 (18:41 +0200)]
flvdec: Do not call parse_keyframes_index with a NULL stream

Signed-off-by: Martin Storsjö <martin@martin.st>
13 years agolibspeexdec: include system headers before local headers
Justin Ruggles [Fri, 21 Oct 2011 16:13:28 +0000 (12:13 -0400)]
libspeexdec: include system headers before local headers

13 years agolibspeexdec: return meaningful error codes
Justin Ruggles [Fri, 21 Oct 2011 16:13:04 +0000 (12:13 -0400)]
libspeexdec: return meaningful error codes

13 years agolibspeexdec: cosmetics: reindent
Justin Ruggles [Fri, 21 Oct 2011 16:10:35 +0000 (12:10 -0400)]
libspeexdec: cosmetics: reindent

13 years agolibspeexdec: decode one frame at a time.
Justin Ruggles [Fri, 21 Oct 2011 16:07:42 +0000 (12:07 -0400)]
libspeexdec: decode one frame at a time.

This allows for knowing the output size before decoding even when there is no
header (e.g. FLV). Otherwise we would have to do a preliminary full frame
decode to determine the number of frames-per-packet.

13 years agoswscale: fix signed shift overflows in ff_yuv2rgb_c_init_tables()
Mans Rullgard [Fri, 21 Oct 2011 19:36:11 +0000 (20:36 +0100)]
swscale: fix signed shift overflows in ff_yuv2rgb_c_init_tables()

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoMove timefilter code from lavf to lavd.
Anton Khirnov [Fri, 21 Oct 2011 09:47:39 +0000 (11:47 +0200)]
Move timefilter code from lavf to lavd.

It's only used in the JACK device.

Fixes linking shared lavd with JACK enabled.

13 years agomov: add support for hdvd and pgapmetadata atoms
Raivo Hool [Fri, 21 Oct 2011 13:04:15 +0000 (16:04 +0300)]
mov: add support for hdvd and pgapmetadata atoms

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agomov: rename function _stik, some indentation cosmetics
Raivo Hool [Fri, 21 Oct 2011 13:04:14 +0000 (16:04 +0300)]
mov: rename function _stik, some indentation cosmetics

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agomov: rename function _int8 to remove ambiguity, some indentation cosmetics
Raivo Hool [Fri, 21 Oct 2011 13:04:13 +0000 (16:04 +0300)]
mov: rename function _int8 to remove ambiguity, some indentation cosmetics

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agomov: parse the gnre atom
Raivo Hool [Fri, 21 Oct 2011 13:51:33 +0000 (16:51 +0300)]
mov: parse the gnre atom

Signed-off-by: Anton Khirnov <anton@khirnov.net>
13 years agomp3on4: check for allocation failures in decode_init_mp3on4()
Justin Ruggles [Sun, 25 Sep 2011 18:32:42 +0000 (14:32 -0400)]
mp3on4: check for allocation failures in decode_init_mp3on4()

13 years agomp3on4: create a separate flush function for MP3onMP4.
Justin Ruggles [Sun, 25 Sep 2011 17:39:04 +0000 (13:39 -0400)]
mp3on4: create a separate flush function for MP3onMP4.

The correct decoder private context needs to be used.
This fixes mp3on4 playback and seeking in avplay.

13 years agomp3on4: ensure that the frame channel count does not exceed the codec channel
Justin Ruggles [Sun, 25 Sep 2011 17:04:39 +0000 (13:04 -0400)]
mp3on4: ensure that the frame channel count does not exceed the codec channel
count.

This also allows for checking output data size based on the actual
number of channel instead of the maximum number of channels.

13 years agomp3on4: set channel layout
Justin Ruggles [Sun, 25 Sep 2011 16:52:11 +0000 (12:52 -0400)]
mp3on4: set channel layout

13 years agomp3on4: fix the output channel order
Justin Ruggles [Sun, 25 Sep 2011 16:46:54 +0000 (12:46 -0400)]
mp3on4: fix the output channel order

13 years agomp3on4: allocate temp buffer with av_malloc() instead of on the stack.
Justin Ruggles [Sun, 25 Sep 2011 16:30:16 +0000 (12:30 -0400)]
mp3on4: allocate temp buffer with av_malloc() instead of on the stack.

Avoids allocating unnecessary memory and ensures proper alignment.

13 years agomp3on4: copy MPADSPContext from first context to all contexts.
Justin Ruggles [Sun, 25 Sep 2011 16:16:34 +0000 (12:16 -0400)]
mp3on4: copy MPADSPContext from first context to all contexts.

Fixes segfault when decoding multi-channel MP3onMP4 files.

13 years agofmtconvert: port float_to_int16_interleave() 2-channel x86 inline asm to yasm
Justin Ruggles [Mon, 10 Oct 2011 04:43:08 +0000 (00:43 -0400)]
fmtconvert: port float_to_int16_interleave() 2-channel x86 inline asm to yasm

13 years agofmtconvert: port int32_to_float_fmul_scalar() x86 inline asm to yasm
Justin Ruggles [Mon, 10 Oct 2011 03:52:03 +0000 (23:52 -0400)]
fmtconvert: port int32_to_float_fmul_scalar() x86 inline asm to yasm

13 years agofmtconvert: check compile-time x86 instruction set flags
Justin Ruggles [Mon, 10 Oct 2011 00:01:22 +0000 (20:01 -0400)]
fmtconvert: check compile-time x86 instruction set flags

13 years agofmtconvert: port float_to_int16() x86 inline asm to yasm
Justin Ruggles [Sun, 9 Oct 2011 23:12:09 +0000 (19:12 -0400)]
fmtconvert: port float_to_int16() x86 inline asm to yasm

13 years agofmtconvert: fix and extend documentation for float_interleave()
Justin Ruggles [Sun, 9 Oct 2011 20:30:11 +0000 (16:30 -0400)]
fmtconvert: fix and extend documentation for float_interleave()

13 years agoflvenc: store delay and last_ts per-stream.
Justin Ruggles [Thu, 20 Oct 2011 19:08:48 +0000 (15:08 -0400)]
flvenc: store delay and last_ts per-stream.

13 years agoh264: fix ppc/altivec build
Mans Rullgard [Fri, 21 Oct 2011 11:34:08 +0000 (12:34 +0100)]
h264: fix ppc/altivec build

Signed-off-by: Mans Rullgard <mans@mansr.com>
13 years agoEnable multithreding when decoding with libopenjpeg
Jean First [Fri, 30 Sep 2011 07:42:45 +0000 (09:42 +0200)]
Enable multithreding when decoding with libopenjpeg

Enable multithreding when decoding with libopenjpeg

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
13 years agoh264: mark some MC functions with av_always_inline instead of inline.
Ronald S. Bultje [Fri, 21 Oct 2011 06:36:23 +0000 (23:36 -0700)]
h264: mark some MC functions with av_always_inline instead of inline.

This actually causes them to be inlined, leading to a significant
speedup (1-1.5% in my measurements).