Ben Avison [Mon, 15 Jul 2013 17:28:10 +0000 (18:28 +0100)]
arm: Add VFP-accelerated version of int32_to_float_fmul_scalar
Before After
Mean StdDev Mean StdDev Change
This function 1175.0 4.4 366.2 18.3 +220.8%
Overall 19285.5 292.0 18420.5 489.1 +4.7%
Signed-off-by: Martin Storsjö <martin@martin.st>
Ben Avison [Mon, 15 Jul 2013 17:28:09 +0000 (18:28 +0100)]
arm: Add VFP-accelerated version of synth_filter_float
Before After
Mean StdDev Mean StdDev Change
This function 9295.0 114.9 4853.2 83.5 +91.5%
Overall 23699.8 397.6 19285.5 292.0 +22.9%
Signed-off-by: Martin Storsjö <martin@martin.st>
Justin Ruggles [Sat, 13 Jul 2013 15:55:33 +0000 (11:55 -0400)]
libspeexdec: fix detection of final terminator code
Hendrik Leppkes [Fri, 19 Jul 2013 13:44:39 +0000 (15:44 +0200)]
configure: generate full msvc debug symbols when linking in debug mode
Signed-off-by: Martin Storsjö <martin@martin.st>
Derek Buitenhuis [Fri, 19 Jul 2013 18:25:57 +0000 (14:25 -0400)]
libx264: Define X264_API_IMPORTS on MSVC/ICL
libx264 has a few data exports which require X264_API_IMPORTS
to be defined if we link to libx264 dynamically on Windows.
In a similar fashion to how we handle our compat snprintf
implementation, if we define it all the time, the compiler
will first try and link to __imp_x264_symbol_name, and failing
that, as in the case of a static libx264, will attempt to link
to the non-prefixed symbol, which has already been pulled in by
other x264 functions' object files.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Luca Barbato [Fri, 19 Jul 2013 19:34:21 +0000 (21:34 +0200)]
dsicinav: Clip the source size to the expected maximum
A packet larger than cin->bitmap_size does not make sense.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Fri, 19 Jul 2013 19:09:40 +0000 (21:09 +0200)]
dsicinav: Bound-check the source buffer when needed
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Fri, 19 Jul 2013 19:05:44 +0000 (21:05 +0200)]
dsicinav: K&R formatting cosmetics
Vittorio Giovara [Tue, 16 Jul 2013 13:05:01 +0000 (15:05 +0200)]
lavfi: filtfmts: use newer avfilter APIs
Signed-off-by: Martin Storsjö <martin@martin.st>
Ed Torbett [Thu, 27 Jun 2013 07:53:00 +0000 (08:53 +0100)]
rtsp: Support multicast source filters (RFC 4570)
This supports inclusion of one single IP address for now,
at the media level. Specifying the filter at the session level
(instead of at the media level), multiple source addresses,
exclusion, or using FQDNs instead of plain IP addresses is not
supported (yet at least).
Signed-off-by: Martin Storsjö <martin@martin.st>
Martin Storsjö [Thu, 18 Jul 2013 18:12:14 +0000 (21:12 +0300)]
rtpproto: Check the source IP if one single source has been specified
If another peer is sending unicast packets to the same port that
we are listening on, those packets can end up being received despite
using source specific multicast. For those cases, manually check the
source address of received packets against the intended source address.
This only handles the case when the source list is one single IP
address for now, which probably is the most common case.
Based on a patch by Ed Torbett.
Signed-off-by: Martin Storsjö <martin@martin.st>
Ed Torbett [Thu, 18 Jul 2013 17:04:16 +0000 (20:04 +0300)]
rtpproto: Support IGMPv3 source specific multicast inclusion
Blocking/exclusion is not supported yet.
The rtp protocol parameter takes the same form as the existing
sources parameter for the udp protocol.
Signed-off-by: Martin Storsjö <martin@martin.st>
Diego Biurrun [Thu, 18 Jul 2013 13:13:18 +0000 (15:13 +0200)]
compat: Add missing license boilerplates
Diego Biurrun [Thu, 25 Oct 2012 14:49:26 +0000 (16:49 +0200)]
Employ consistent LIBAV_COMPAT_ multiple inclusion guards in compat/
Also fix a comment and an #endif comment.
Martin Storsjö [Wed, 17 Jul 2013 15:24:23 +0000 (18:24 +0300)]
avserver: Remove unused loadable module support
There is no record of this ever being used at all, anywhere,
since the feature was added in
2effd274466.
This gets rid of extra linker tricks just to support a feature
that isn't used, simplifying portability to other platforms.
Signed-off-by: Martin Storsjö <martin@martin.st>
Diego Biurrun [Wed, 17 Jul 2013 18:19:24 +0000 (20:19 +0200)]
Consistently use "cpu_flags" as variable/parameter name for CPU flags
Christophe Gisquet [Thu, 27 Dec 2012 21:33:51 +0000 (22:33 +0100)]
fmtconvert: Explicitly use int32_t instead of int
Signed-off-by: Martin Storsjö <martin@martin.st>
Sean McGovern [Tue, 16 Jul 2013 15:31:23 +0000 (11:31 -0400)]
pthread: Rename thread_init to avoid symbol collision
The AIX threads library exposes a function with the same name.
Signed-off-by: Martin Storsjö <martin@martin.st>
Martin Storsjö [Mon, 15 Jul 2013 13:44:20 +0000 (16:44 +0300)]
lavf: Make sure avg_frame_rate can be calculated without integer overflow
If either of the deltas is too large for the multiplications to
succeed, don't use this for setting the avg frame rate.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Cc: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
Martin Storsjö [Mon, 15 Jul 2013 14:13:54 +0000 (17:13 +0300)]
mov: Do not allow updating the time scale after it has been set
The time scale is set in mdhd, and later validated in the
enclosing trak atom once all of its children have been parsed.
A loose mdhd atom outside of a trak atom could update the time
scale of the last stream without any validation.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Cc: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
Martin Storsjö [Mon, 15 Jul 2013 12:59:50 +0000 (15:59 +0300)]
mov: Seek back if overreading an individual atom
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Cc: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
Martin Storsjö [Mon, 15 Jul 2013 08:28:46 +0000 (11:28 +0300)]
ac3dec: Don't consume more data than the actual input packet size
This was handled properly in the normal return case at the end
of the function, but not in this special case.
Returning a value larger than the input packet size can cause
problems for certain library users.
Returning the actual input buffer size unconditionally, since
it is not guaranteed that frame_size is set to a sensible
value at this point.
Cc: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
Luca Barbato [Sun, 14 Jul 2013 16:16:56 +0000 (18:16 +0200)]
indeo: Reject impossible FRAMETYPE_NULL
A frame marked FRAMETYPE_NULL cannot be scalable and requires a
previous frame successfully decoded.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Sun, 14 Jul 2013 14:49:43 +0000 (16:49 +0200)]
indeo: Do not reference mismatched tiles
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Sun, 14 Jul 2013 13:48:17 +0000 (15:48 +0200)]
indeo: Sanitize ff_ivi_init_planes fail paths
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Sun, 14 Jul 2013 12:41:56 +0000 (14:41 +0200)]
indeo5: return proper error codes
Luca Barbato [Sun, 14 Jul 2013 12:06:16 +0000 (14:06 +0200)]
indeo: Bound-check before applying motion compensation
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Martin Storsjö [Fri, 12 Jul 2013 09:14:12 +0000 (12:14 +0300)]
rtsp: Include an User-Agent header field in all requests
Some rtsp servers like the IP Cam IcyBox IB-CAM2002 need it.
Based on a patch by Carl Eugen Hoyos.
Signed-off-by: Martin Storsjö <martin@martin.st>
Luca Barbato [Fri, 12 Jul 2013 12:33:24 +0000 (14:33 +0200)]
indeo: Bound-check before applying transform
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Fri, 12 Jul 2013 16:10:05 +0000 (18:10 +0200)]
indeo4: Validate scantable dimension
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Fri, 12 Jul 2013 13:02:33 +0000 (15:02 +0200)]
indeo4: Check the quantization matrix index
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Fri, 12 Jul 2013 12:32:03 +0000 (14:32 +0200)]
indeo4: Do not access missing reference MV
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Martin Storsjö [Thu, 11 Jul 2013 13:39:23 +0000 (16:39 +0300)]
ac3dec: Consistently use AC3_BLOCK_SIZE and sizeof
Signed-off-by: Martin Storsjö <martin@martin.st>
Martin Storsjö [Thu, 11 Jul 2013 13:30:18 +0000 (16:30 +0300)]
ac3dec: Increment channel pointers only once per channel
If the channel mapping map multiple output channels to one
input channel, we should only increment the actual pointer once.
Cc: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
Nicolas Bertrand [Sat, 13 Jul 2013 08:10:26 +0000 (10:10 +0200)]
jpeg2000: Remove unused passes array in Jpeg200Cblk structure
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Nicolas Bertrand [Fri, 12 Jul 2013 21:15:43 +0000 (23:15 +0200)]
jpeg2000: Initialize only once mqc arrays
Improves decoding speed.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Tue, 9 Jul 2013 15:57:49 +0000 (17:57 +0200)]
vf_interlace: better handling of odd video size
Signed-off-by: Martin Storsjö <martin@martin.st>
Sean McGovern [Wed, 10 Jul 2013 17:37:05 +0000 (13:37 -0400)]
fate: use current syntax for the scale filter
Signed-off-by: Martin Storsjö <martin@martin.st>
Luca Barbato [Wed, 10 Jul 2013 17:00:15 +0000 (19:00 +0200)]
dca: Respect the current limits in the downmixing capabilities
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Wed, 10 Jul 2013 16:07:45 +0000 (18:07 +0200)]
dca: Error out on missing DSYNC
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Wed, 10 Jul 2013 02:54:49 +0000 (04:54 +0200)]
pcm: always use codec->id instead of codec_id
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Wed, 10 Jul 2013 18:27:03 +0000 (20:27 +0200)]
vdpau: use the correct namespace for the union
Vdp is used by libvdpau, use AVVDPAU as used for the rest.
Reported-by: Alexis Ballier <aballier@gentoo.org>
Luca Barbato [Wed, 10 Jul 2013 02:35:34 +0000 (04:35 +0200)]
mlpdec: Do not set invalid context in read_restart_header
The faulty values rippled further down the codepath causing a
hard-to-track segfault in the assembly code.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Wed, 10 Jul 2013 01:18:14 +0000 (03:18 +0200)]
mlpdsp: x86: Respect cpuflags
Luca Barbato [Sat, 29 Jun 2013 04:37:32 +0000 (06:37 +0200)]
pcx: Do not overread source buffer in pcx_rle_decode
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Sat, 29 Jun 2013 04:25:31 +0000 (06:25 +0200)]
pcx: K&R formatting cosmetics
Martin Storsjö [Tue, 9 Jul 2013 17:51:00 +0000 (20:51 +0300)]
configure: Check for support for labels in the inline assembly
Use this for enabling the ppc timer.h implementation only on
assemblers that support labels in the inline assembly.
Signed-off-by: Martin Storsjö <martin@martin.st>
Luca Barbato [Sat, 29 Jun 2013 03:29:54 +0000 (05:29 +0200)]
wmavoice: conceal clearly corrupted blocks
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Sat, 29 Jun 2013 01:10:59 +0000 (03:10 +0200)]
wmavoice: use init_static_data
Luca Barbato [Sat, 29 Jun 2013 01:02:02 +0000 (03:02 +0200)]
wmavoice: return meaningful error codes
Luca Barbato [Sat, 29 Jun 2013 05:26:48 +0000 (07:26 +0200)]
iff: Do not read over the source buffer
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Sat, 29 Jun 2013 05:05:54 +0000 (07:05 +0200)]
iff: K&R formatting cosmetics
Luca Barbato [Tue, 9 Jul 2013 12:59:33 +0000 (14:59 +0200)]
qdm2: Conceal broken samples
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Tue, 9 Jul 2013 12:44:02 +0000 (14:44 +0200)]
qdm2: refactor joined stereo support
qdm2 does support only two channels. Loop over the run once.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Sun, 7 Jul 2013 10:56:12 +0000 (12:56 +0200)]
adpcm: Write the correct number of samples for ima-dk4
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Tue, 9 Jul 2013 07:18:16 +0000 (09:18 +0200)]
imc: Catch a division by zero
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Tue, 9 Jul 2013 02:44:26 +0000 (04:44 +0200)]
atrac3: Error on impossible encoding/channel combinations
Joint stereo encoded mono is impossible.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Tue, 9 Jul 2013 02:20:23 +0000 (04:20 +0200)]
atrac3: set the getbits context the right buffer_end
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Mon, 8 Jul 2013 23:03:13 +0000 (01:03 +0200)]
atrac3: fix error handling
decode_tonal_components returns a proper AVERROR.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Martin Storsjö [Sun, 7 Jul 2013 14:03:00 +0000 (17:03 +0300)]
movenc: Rename the timestamp parameter to dts in write_hint_packets
This matches its actual intended use better.
Signed-off-by: Martin Storsjö <martin@martin.st>
Martin Storsjö [Sun, 7 Jul 2013 13:55:26 +0000 (16:55 +0300)]
movenc: Make sure the RTP hint tracks have nondecreasing DTS
The RTP timestamps can be decreasing for codecs with B-frames. For
these cases, make sure the timestamps in the MP4 file track itself
are nondecreasing, and add an offset to the RTP packet hint instead
to produce the intended RTP timestamp.
Signed-off-by: Martin Storsjö <martin@martin.st>
Luca Barbato [Thu, 27 Jun 2013 00:50:52 +0000 (02:50 +0200)]
qdm2: check and reset dithering index per channel
Checking per subband would have the index exceed the
dithering noise table size.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Thu, 27 Jun 2013 00:49:15 +0000 (02:49 +0200)]
qdm2: formatting cosmetics
Apply the usual style plus drop few unnecessary return at the end
of void functions.
Luca Barbato [Thu, 27 Jun 2013 00:20:59 +0000 (02:20 +0200)]
qdm2: use init_static_data
Luca Barbato [Thu, 27 Jun 2013 02:30:20 +0000 (04:30 +0200)]
westwood_vqa: do not free extradata on error in read_header
The extradata is already freed by avformat_open_input on
failure.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Thu, 27 Jun 2013 01:19:05 +0000 (03:19 +0200)]
vqavideo: check the version
Prevent out of buffer write.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Sat, 6 Jul 2013 23:52:51 +0000 (01:52 +0200)]
cmdutils: wrap exit explicitly
Some C runtime implementations deadlock when calling threading functions
on the atexit() handler.
Use a simpler wrapper similar to av_log to call the cleanup function
before exit.
Bug-Id: 523
Martin Storsjö [Sun, 7 Jul 2013 11:59:47 +0000 (14:59 +0300)]
movenc: K&R formatting cosmetics
Signed-off-by: Martin Storsjö <martin@martin.st>
Michael Niedermayer [Sun, 7 Jul 2013 09:56:02 +0000 (11:56 +0200)]
Makefile: Remove stray tabs
Signed-off-by: Martin Storsjö <martin@martin.st>
Ronald S. Bultje [Sun, 23 Jun 2013 04:16:11 +0000 (21:16 -0700)]
vp8: Wait for prev_frame to parse segment_map before reading it
This fixes occasional failures of vp8-test-vector-010 with frame-level
multithreading enabled.
Signed-off-by: Martin Storsjö <martin@martin.st>
Ronald S. Bultje [Mon, 24 Jun 2013 23:13:58 +0000 (19:13 -0400)]
yuv4mpeg: Correctly round chroma up for odd luma sizes
Signed-off-by: Martin Storsjö <martin@martin.st>
Michael Niedermayer [Mon, 1 Jul 2013 21:38:08 +0000 (23:38 +0200)]
rmdec: Use the AVIOContext given as parameter in rm_read_metadata()
This fixes crashes when playing back certain RealRTSP streams.
When invoked from the RTP depacketizer, the full realmedia
demuxer isn't invoked, but only certain functions from it, where
a separate AVIOContext is passed in as parameter (for the buffer
containing the data to parse). The functions called from within
those entry points should only be using that parameter, not
s->pb. In the depacketizer case, s is the RTSP context, where ->pb
is null.
Cc: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
Michael Niedermayer [Mon, 24 Jun 2013 12:23:44 +0000 (14:23 +0200)]
avio: Handle AVERROR_EOF in the same way as the return value 0
This makes sure the ffurl_read_complete function actually
returns the number of bytes read, as the documentation of the
function says, even if the underlying protocol uses AVERROR_EOF
instead of 0.
Signed-off-by: Martin Storsjö <martin@martin.st>
Luca Barbato [Mon, 24 Jun 2013 16:12:24 +0000 (18:12 +0200)]
wtv: Mark attachment with a negative stream id
A sid 0 would be mismatched to the attachment.
Prevent NULL pointer dereference.
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Fri, 21 Jun 2013 12:47:01 +0000 (14:47 +0200)]
avprobe: rename the prettyprint structures
Diego Biurrun [Fri, 5 Jul 2013 15:42:37 +0000 (17:42 +0200)]
tools: Wording and formatting cosmetics
Diego Biurrun [Tue, 2 Jul 2013 15:16:18 +0000 (17:16 +0200)]
tests: Only run noproxy test if networking is enabled
Luca Barbato [Sat, 6 Jul 2013 10:05:27 +0000 (12:05 +0200)]
fifo: K&R formatting cosmetics
Luca Barbato [Mon, 1 Jul 2013 01:05:41 +0000 (03:05 +0200)]
kmvc: Clip pixel position to valid range
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Mon, 1 Jul 2013 01:04:15 +0000 (03:04 +0200)]
kmvc: use fixed sized arrays in the context
Avoid some boilerplate code to dynamically allocate and then free the
buffers.
Luca Barbato [Fri, 5 Jul 2013 13:35:30 +0000 (15:35 +0200)]
bitstream_filter: K&R formatting cosmetics
Luca Barbato [Thu, 4 Jul 2013 19:53:08 +0000 (21:53 +0200)]
compat: wrap math.h to avoid AIX-specific clashes
AIX defines a class() function in its math.h header without any
guard.
Luca Barbato [Thu, 27 Jun 2013 01:17:32 +0000 (03:17 +0200)]
log: pass the correct parameters to missing_feature_sample
CC:libav-stable@libav.org
Luca Barbato [Wed, 3 Jul 2013 12:55:50 +0000 (14:55 +0200)]
indeo: reject negative array indexes
Reported-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
CC: libav-stable@libav.org
Luca Barbato [Wed, 3 Jul 2013 12:01:32 +0000 (14:01 +0200)]
indeo: Cosmetic formatting
Trim some overly long lines.
Luca Barbato [Wed, 3 Jul 2013 11:59:16 +0000 (13:59 +0200)]
indeo: Refactor ff_ivi_init_tiles and ivi_decode_blocks
Spin large and mostly self contained blocks into stand alone
functions.
Luca Barbato [Wed, 3 Jul 2013 10:58:40 +0000 (12:58 +0200)]
indeo: Refactor ff_ivi_dec_huff_desc
Spare an indentation level.
Luca Barbato [Wed, 3 Jul 2013 09:18:30 +0000 (11:18 +0200)]
indeo: use a typedef for the mc function pointer
Jason Garrett-Glaser [Mon, 1 Jul 2013 19:06:22 +0000 (12:06 -0700)]
cabac: x86 version of get_cabac_bypass
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Kostya Shishkov [Wed, 3 Jul 2013 18:46:28 +0000 (20:46 +0200)]
aic: use chroma scan tables while decoding luma component in progressive mode
For some unclear reason Apple decided to use the same scan tables for luma and
chroma in the progressive mode while using different ones for luma in the
interlaced mode.
Michael Niedermayer [Mon, 1 Jul 2013 08:01:39 +0000 (10:01 +0200)]
jpeg2000: Use the matching coding style struct for transform selection
Fixes a null pointer dereference.
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:38 +0000 (10:01 +0200)]
jpeg2000: Merge rescaling with interleaving in 9/7 IDWT
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:37 +0000 (10:01 +0200)]
jpeg2000: Optimize output sample conversion
67935 -> 29984 kcycles
Reviewed-by: Nicolas BERTRAND <nicoinattendu@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:36 +0000 (10:01 +0200)]
jpeg2000: Reset s->numX/Ytiles on tile deallocation
Keep the structure fields more consistent after cleanup.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:35 +0000 (10:01 +0200)]
jpeg2000: Optimize dequantization
Float: 4700 -> 2700 cycles
Integer: 4400 -> 2800 cycles
(sandybridge i7)
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:34 +0000 (10:01 +0200)]
jpeg2000: Do not assume a single tile
In preparation of supporting multiple tiles.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:33 +0000 (10:01 +0200)]
jpeg2000: Speed up jpeg2000_decode_tile()
Skip processing bands with dimension set to 0.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:32 +0000 (10:01 +0200)]
jpeg2000: Use EBCOT's CAUSAL and BYPASS mode in decode_cblk()
Speed it up a bit.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:31 +0000 (10:01 +0200)]
jpeg2000: Simplify jpeg2000_decode_packets()
Raise PATCHWELCOME error in case of non-implemented progression order.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Signed-off-by: Nicolas Bertrand <nicoinattendu@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Michael Niedermayer [Mon, 1 Jul 2013 08:01:30 +0000 (10:01 +0200)]
jpeg2000: Do not crash on NULL node in tag_tree_decode
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>