Martin Storsjö [Tue, 22 Nov 2016 20:58:35 +0000 (22:58 +0200)]
aarch64: vp9itxfm: Do separate functions for half/quarter idct16 and idct32
This work is sponsored by, and copyright, Google.
This avoids loading and calculating coefficients that we know will
be zero, and avoids filling the temp buffer with zeros in places
where we know the second pass won't read.
This gives a pretty substantial speedup for the smaller subpartitions.
The code size increases from 14740 bytes to 24292 bytes.
The idct16/32_end macros are moved above the individual functions; the
instructions themselves are unchanged, but since new functions are added
at the same place where the code is moved from, the diff looks rather
messy.
Martin Storsjö [Tue, 22 Nov 2016 09:07:38 +0000 (11:07 +0200)]
arm: vp9itxfm: Do a simpler half/quarter idct16/idct32 when possible
This work is sponsored by, and copyright, Google.
This avoids loading and calculating coefficients that we know will
be zero, and avoids filling the temp buffer with zeros in places
where we know the second pass won't read.
This gives a pretty substantial speedup for the smaller subpartitions.
The code size increases from 12388 bytes to 19784 bytes.
The idct16/32_end macros are moved above the individual functions; the
instructions themselves are unchanged, but since new functions are added
at the same place where the code is moved from, the diff looks rather
messy.
Martin Storsjö [Wed, 23 Nov 2016 08:56:12 +0000 (10:56 +0200)]
arm: vp9itxfm: Make the larger core transforms standalone functions
This work is sponsored by, and copyright, Google.
This reduces the code size of libavcodec/arm/vp9itxfm_neon.o from
15324 to 12388 bytes.
This gives a small slowdown of a couple tens of cycles, up to around
150 cycles for the full case of the largest transform, but makes
it more feasible to add more optimized versions of these transforms.
Moritz Barsnick [Thu, 9 Mar 2017 09:23:28 +0000 (10:23 +0100)]
libavfilter/avf_showwaves: make sqrt and cbrt scale option values available to showwavespic by name
The 'sqrt' and 'cbrt' scalers were added in commit 80262d8c86e94ff9a4bb3a9e3c2d734e04ccb399, but their symbolic option values
only made available to the showwaves filter, not showwavespic, despite
the scalers working properly by their numerical option values.
avcodec/vp56: Reset have_undamaged_frame on resolution changes
Fixes: timeout in 758/clusterfuzz-testcase-4720832028868608 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Aaron Boxer [Fri, 10 Mar 2017 04:01:48 +0000 (23:01 -0500)]
libavcodec/libopenjpegenc: enable lossless option, remove layer option, and improve defaults
1. limit to single layer, as there is no current support for setting distortion/quality of multiple layers
2. encoder mode should be kept at default setting (0)
3. remove fixed_alloc parameter from context : seldom if ever used, and no way of properly configuring at the moment
4. add irreversible setting, to allow for lossless encoding. Set to OpenJPEG default (enabled)
5. set numresolution max to 33, which is the maximum number of allowed resolutions according the J2K spec
Signed-off-by: Michael Bradshaw <mjbshaw@google.com>
wm4 [Mon, 6 Mar 2017 10:34:20 +0000 (11:34 +0100)]
avcodec: clarify some decoding/encoding API details
Make it clear that there is no timing-dependent behavior. In particular,
there is no state in which both input and output are denied, and where
you have to wait for a while yourself to make progress (apparently some
hardware decoders like to do this).
Avoid wording that makes references to time. It shouldn't be mistaken
for some kind of asynchronous API (like POSIX read() can return EAGAIN
if there is no new input yet). It's a state machine, so try to use
appropriate terms.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Merges Libav commit 8a60bba0ae.
wm4 [Wed, 8 Mar 2017 11:36:05 +0000 (12:36 +0100)]
aacdec: do not mutate input packet metadata
Apparently the demuxer outputs the wrong padding for HE-AAC (based on
the raw sample rate, or so). aacdec contains a hack to adjust the muxer
padding accordingly before it's used to trim the decoder output. This
modified the packet side data, which in combination with the old
decoding API would change the packet the user passed to the decoder.
This is clearly not allowed, and it breaks running some gapless fate
tests with "-fflags +keepside" applied (without keepside, the packet
metadata is typically newly allocated, essentially making a copy and not
modifying the user's input packet).
This should probably be fixed in the demuxer (and consequently also the
muxer), but for now only fix the immediate problem.
Thomas Turner [Wed, 8 Mar 2017 00:36:15 +0000 (16:36 -0800)]
avutil/tests/lfg.c: added proper normality test
The Chen-Shapiro(CS) test was used to test normality for
Lagged Fibonacci PRNG.
Normality Hypothesis Test:
The null hypothesis formally tests if the population
the sample represents is normally-distributed. For
CS, when the normality hypothesis is True, the
distribution of QH will have a mean close to 1.
This change became superfluous when support for C11 atomics was introduced.
Reverting it will make the removal of this implementation in an upcoming
merge conflict free.
Reviewed-by: wm4 <nfxjfg@googlemail.com> Signed-off-by: James Almer <jamrial@gmail.com>
avcodec/vp8: Check for the bitstream end per MB in decode_mb_row_no_filter()
Fixes: timeout in 730/clusterfuzz-testcase-5265113739165696 (part 2 of 2) Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg Reviewed-by: BBB Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
avcodec/vp568: Check that there is enough data for ff_vp56_init_range_decoder()
Fixes: timeout in 730/clusterfuzz-testcase-5265113739165696 (part 1 of 2) Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg Reviewed-by: BBB Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Aman Gupta [Tue, 21 Feb 2017 18:48:37 +0000 (10:48 -0800)]
avcodec/h264, videotoolbox: fix crash after VT decoder fails
The way videotoolbox hooks in as a hwaccel is pretty hacky. The VT decode
API is not invoked until end_frame(), so alloc_frame() returns a dummy
frame with a 1-byte buffer. When end_frame() is eventually called, the
dummy buffer is replaced with the actual decoded data from
VTDecompressionSessionDecodeFrame().
When the VT decoder fails, the frame returned to the h264 decoder from
alloc_frame() remains invalid and should not be used. Before 9747219958060d8c4f697df62e7f172c2a77e6c7, it was accidentally being
returned all the way up to the API user. After that commit, the dummy
frame was unref'd so the user received an error.
However, since that commit, VT hwaccel failures started causing random
segfaults in the h264 decoder. This happened more often on iOS where the
VT implementation is more likely to throw errors on bitstream anomolies.
A recent report of this issue can be see in
http://ffmpeg.org/pipermail/libav-user/2016-November/009831.html
The issue here is that the dummy frame is still referenced internally by the
h264 decoder, as part of the reflist and cur_pic_ptr. Deallocating the
frame causes assertions like this one to trip later on during decoding:
Assertion h->cur_pic_ptr->f->buf[0] failed at src/libavcodec/h264_slice.c:1340
With this commit, we leave the dummy 1-byte frame intact, but avoid returning it
to the user.
wm4 [Thu, 2 Mar 2017 10:00:40 +0000 (11:00 +0100)]
avformat/flvdec: remove meaningless warning
Ever since the codecpar changes, this has been always printed when
opening a flv file. This is because the codecpar changes made all
streams to be added lazily as read_packet is called.
wm4 [Thu, 2 Mar 2017 09:37:26 +0000 (10:37 +0100)]
avcodec: consider an error during decoder draining as EOF
There is no reason that draining couldn't return an error or two. But
some decoders don't handle this very well, and might always return an
error. This can lead to API users getting into an infinite loop and
burning CPU, because no progress is made and EOF is never returned.
In fact, ffmpeg.c contains a hack against such a case. It is made
unnecessary with this commit, and removed with the next one. (This
particular error case seems to have been fixed since the hack was
added, though.)
This might lose frames if decoding returns errors during draining.
Fixes: undefined behavior in 717/clusterfuzz-testcase-5434924129583104 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
avcodec/mpeg4videodec: Fix runtime error: signed integer overflow: 134527392 * 16 cannot be represented in type 'int'
This checks the sprite delta intermediates for overflow Fixes: 716/clusterfuzz-testcase-4890287480504320 Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
avcodec/mpeg4videodec: Improve the overflow checks in mpeg4_decode_sprite_trajectory()
Also clear the state on errors
Fixes integer overflows in 701/clusterfuzz-testcase-6594719951880192
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
wm4 [Thu, 2 Mar 2017 18:14:58 +0000 (19:14 +0100)]
ffmpeg: properly cleanup filter graph on init failure
The filter field is often used to check whether a filter is
configured. If configuring the filter actually fails somewhere in
the middle of it, these fields could still be set to non-NULL, which
lead to other code accessing the half-configured filter graph, which
in turn could lead to crashes within libavfilter.
Solve this by properly resetting all fields.
This was triggered by a fuzzed sample after the recent changes. It's
unknown whether this behavior could be triggered before that.
wm4 [Thu, 2 Mar 2017 15:01:01 +0000 (16:01 +0100)]
ffmpeg: delay processing of subtitles before filters are initialized
If a subtitle packet came before the first video frame could be fully
decoded, the subtitle packet would get discarded. This puts the subtitle
into a queue instead, and processes it once the attached filter graph is
initialized.
Anton Khirnov [Mon, 27 Jun 2016 17:03:42 +0000 (19:03 +0200)]
ffmpeg: restructure sending EOF to filters
Be more careful when an input stream encounters EOF when its filtergraph
has not been configured yet. The current code would immediately mark the
corresponding output streams as finished, while there may still be
buffered frames waiting for frames to appear on other filtergraph
inputs.
Anton Khirnov [Fri, 27 May 2016 10:14:33 +0000 (12:14 +0200)]
ffmpeg: init filtergraphs only after we have a frame on each input
This makes sure the actual stream parameters are used, which is
important mainly for hardware decoding+filtering cases, which would
previously require various weird workarounds to handle the fact that a
fake software graph has to be constructed, but never used.
This should also improve behaviour in rare cases where
avformat_find_stream_info() does not provide accurate information.
This merges Libav commit a3a0230. It was previously skipped.
The code in flush_encoders() which sets up a "fake" format wasn't in
Libav. I'm not sure if it's a good idea, but it tends to give
behavior closer to the old one in certain corner cases.
The vp8-size-change gives different result, because now the size of
the first frame is used. libavformat reported the size of the largest
frame for some reason.
The exr tests now use the sample aspect ratio of the first frame. For
some reason libavformat determines 0/1 as aspect ratio, while the
decoder returns the correct one.
The ffm and mxf tests change the field_order values. I'm assuming
another libavformat/decoding mismatch.
Anton Khirnov [Fri, 27 May 2016 10:04:29 +0000 (12:04 +0200)]
ffmpeg: do packet ts rescaling in write_packet()
This will be useful in the following commit, after which the muxer
timebase is not always available when encoding.
This merges Libav commit 3e265ca. It was previously skipped.
There are some changes with how/when the mux_timebase field is set,
because the Libav approach often causes a too imprecise time base
to be set. This is hard, because the muxer's write_header function
can readjust the timebase, at which point we might already have
encoded packets buffered. (It might be better to buffer them after
the encoder, instead of after all the timestamp handling logic
before muxing.)
The two FATE tests change because the output time base is raised
for subtitles. (Needed to avoid certain rounding issues in other
cases.)
Includes a minor merge fix by Mark Thompson, and
avconv: Move rescale to stream timebase before monotonisation
wm4 [Wed, 8 Feb 2017 08:53:26 +0000 (09:53 +0100)]
ffmpeg: make sure packets put into the muxing FIFO are refcounted
Some callers (like do_subtitle_out()) call this with an AVPacket that is
not refcounted. This can cause undefined behavior.
Calling av_packet_move_ref() does not make a packet refcounted if it
isn't yet. (And it can't be made to, because it always succeeds,
and can't return ENOMEM.)
Call av_packet_ref() instead to make sure it's refcounted.
I couldn't find a case that is fixed by this with the current code. But
it will fix the fate-pva-demux test with the later patches applied.