cuda_runtime.h as well as dynlink_loader.h used nonstandard inclusion
guards with an AV_ prefix, although these files are not in an libav*/
path. So change the inclusion guards and adapt the ref file of the
source fate test accordingly.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
This avoids using the CUDA SDK at all; instead, we provide a minimal
reimplementation of the basic functionality that lavfi actually uses.
It generates very similar code to what NVCC produces.
The header contains no implementation code derived from the SDK.
The function and type declarations are derived from the SDK only to the
extent required to build a compatible implementation. This is generally
accepted to qualify as fair use.
Because this option does not require the proprietary SDK, it does not require
the "--enable-nonfree" flag in configure.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Recently we encountered an issue when decoding a h.263 file:
FFmpeg will freeze when decoding h.263 video with NVDEC. Turns out this is not directly related to NVDEC but is a problem that shows with several other HW decoders like VDPAU, though the exact kind of error is different (either error messages or freezing[1]). The root cause is that ff_thread_finish_setup() is called twice per frame from ff_h263_decode_frame(). This is not supported by ff_thread_finish_setup() and specifically checked for and warned against in the functions code. The issue is also specific to hw accelerated decoding only as the second call to ff_thread_finish_setup() is only issued when hw acceleration is on. The fix is simple: add a check that the first call is only send when hw acceleration is off, and the second call only when hw acceleration is on (see attached patch). This works fine as far as I was able to test with vdpau and nvdec/nvcuvid hw decoding. The patch also adds NVDEC to the hw config list if available.
I also noticed a secondary issue when browsing through the code which is that, according to documentation, ff_thread_finish_setup() should only be called if the codec implements update_thread_context(), which h263dec does not. The patch does not address this and I'm not sure any action needs to be taken here at all.
[1] This is depending on whether or not the hw decoder sets the HWACCEL_CAPS_ASYNC_SAFE flag
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
This is unneeded for interplay video
Fixes: memleak
Fixes: 15562/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_INTERPLAY_VIDEO_fuzzer-5162268645392384
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -1241665686 + -1340629419 cannot be represented in type 'int'
Fixes: 15922/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-5692826442006528
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: left shift of 1 by 31 places cannot be represented in type 'int'
Fixes: 15631/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CLEARVIDEO_fuzzer-5690110605000704
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array access
Fixes: 15365/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-5716153105645568
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
From https://aomediacodec.github.io/av1-isobmff/#codecsparam, the parameters
sample entry 4CC, profile, level, tier, and bitDepth are all mandatory fields.
All the other fields are optional, mutually inclusive (all or none).
Fixes ticket #8049
Signed-off-by: James Almer <jamrial@gmail.com>
codecpar->extradata is not going to change between packets. New extradata
is instead propagated using packet side data.
Use ff_alloc_extradata() as well.
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes: Timeout (too long -> 1sec)
Fixes: 15232/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VP3_fuzzer-5769583086010368
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Peter Ross <pross@xvid.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Theora is forced to be non zero if it is zero and a sample
is asked for, as suggested by reimar
Fixes: Timeout (2min -> 600ms)
Fixes: 15366/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_THEORA_fuzzer-5737849938247680
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Simply moves and templates the actual transforms to support an
additional data type.
Unlike the float version, which is equal or better than libfftw3f,
double precision output is bit identical with libfftw3.
Fixes: left shift of negative value -1
Fixes: 15531/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VC1IMAGE_fuzzer-5759556258365440
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 32796 * 65536 cannot be represented in type 'int'
Fixes: 15430/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VC1IMAGE_fuzzer-5735424087031808
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: 15524/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TRUEMOTION2_fuzzer-5173148372172800
Fixes: signed integer overflow: 13701388 - -2134868270 cannot be represented in type 'int'
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Changing the type to an unsigned one to avoid many casts was suggested
This may be inadequate for fixing the UB on ILP64
Fixes: signed integer overflow: -1418162611 * 383 cannot be represented in type 'int'
Fixes: 15547/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5691384901664768
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 755176387 + 1515360583 cannot be represented in type 'int'
Fixes: 15506/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5706859232624640
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: left shift of negative value -4
Fixes: signed integer overflow: -15091694 * 167 cannot be represented in type 'int'
Fixes: signed integer overflow: 1898547155 + 453967445 cannot be represented in type 'int'
Fixes: 15258/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5759095564402688
Fixes: signed integer overflow: 962196438 * 31 cannot be represented in type 'int'
Fixes: 15364/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5718799845687296
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Behave like we do for V4L2_BUF_FLAG_ERROR, implemented in commit 28f20d2ff4 .
For some devices (probably also related to the V4L driver implementation)
it happens that when invoking the ioctl DQBUF, the returned buffer is not
of the expected size. Here are two examples for such occurrences:
[video4linux2,v4l2 @ 0x258b440] Dequeued v4l2 buffer contains 609596 bytes, but 614400 were expected. Flags: 0x00000001.
/dev/video1: Invalid data found when processing input
[video4linux2,v4l2 @ 0x225f440] Dequeued v4l2 buffer contains 609508 bytes, but 614400 were expected. Flags: 0x00000001.
/dev/video1: Invalid data found when processing input
For the ffmpeg CLI tool this means it will stop capturing and exit.
The described behaviour was observed at least with one OmniVision USB
web cam and with some stk1160 devices.
If you search the web for the error message, you will find quite a few
instances of this problem. Some of them experienced on other devices.
Probably fixes ticket #4795
Signed-off-by: Alexander Strasser <eclipse7@gmx.net>
There is one way to find out if avpriv_find_start_code has found a start
code or not: One has to check whether the state variable contains a
start code, i.e. whether the three most significant bytes are 0x00 00 01.
Checking for whether the return value is the end of the designated
buffer is not enough: If the last four bytes constitute a start code,
the return value is also the end of the buffer. This happens with
sequence_end_codes which have been ignored for exactly this reason,
although e.g. all three files used for fate tests of cbs_mpeg2 contain
sequence_end_codes.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
1. Currently, cbs_mpeg2_split_fragment uses essentially three variables
to hold the start code values found by avpriv_find_start_code. By
rearranging the code, one of them can be omitted.
2. The return value of avpriv_find_start_code points to the byte after
the byte containing the start code identifier (or to the byte after the
last byte of the fragment's data if no start code was found), but
cbs_mpeg2_split_fragment needs to work with the pointer to the byte
containing the start code identifier; it already did this, but in a
clumsy way. This has been changed.
3. Also use the correct type for the variable holding the
CodedBitstreamUnitType.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Sequence End units (or actually, sequence_end_codes) have up until now
not been decomposed; in fact due to a bug in cbs_mpeg2_split_fragment they
have mostly been treated as part of the preceding unit. So implement
decomposing them as preparation for fixing said bug.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Up until now, if an SEI messages was to be added to a fragment, it was
tried to add said SEI message to the first SEI NAL unit of the fragment
and if this SEI NAL unit already contained H264_NAL_SEI SEI messages (an
arbitrary limit imposed by cbs_h264), adding failed; if there was no SEI
NAL unit, a new one has been added.
With this commit, the fragment is searched for further NAL units to add
the SEI messages to. If all of them are full, a new SEI NAL unit is added.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
cbs is currently inconsistent regarding the opaque field that can be
used as a special argument to av_buffer_create in order to be used
during freeing the buffer: ff_cbs_alloc_unit_content and all the free
functions used name this parameter as if it should contain a pointer to
the unit whose content is about to be created; but both
ff_cbs_alloc_unit_content as well as ff_cbs_h264_add_sei_message
actually use a pointer to the CodedBitstreamContext as opaque. It should
actually be neither, because it is unneeded (as is evidenced by the fact
that none of the free functions use this pointer at all) and because it
ties the unit's content to the lifetime of other objects, although a
refcounted buffer is supposed to have its own lifetime that only ends
when its reference count reaches zero. This problem manifests itself in
the pointer becoming dangling.
The pointer to the unit can become dangling if another unit is added to
the fragment later as happens in the bitstream filters; in this case,
the pointer can point to the wrong unit (if the fragment's unit array
needn't be relocated) or it can point to where the array was earlier.
It can also become dangling if the unit's content is meant to survive
the resetting of the fragment it was originally read with. This applies
to the extradata of H.264 and HEVC.
The pointer to the context can become dangling if the context is closed
before the content is freed. Although this doesn't seem to happen right
now, it could happen, in particular if one uses different
CodedBitstreamContexts for in- and output.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The minimum header is 8 bytes, the smallest bitstream that is passed to
the MB decode code is 4 bytes
Fixes: Timeout (35sec -> 18sec)
Fixes: 15800/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EATQI_fuzzer-5684154517159936
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The minimal size which unpack() will not fail on is 5 bytes
Fixes: Timeout (14sec -> 77ms) (testcase 15508)
Fixes: 15508/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EATGV_fuzzer-5700053513011200
Fixes: 15996/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EATGV_fuzzer-5751353223151616
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
since tf.pad is enabled, the conv2d(valid) changes back to its original behavior.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
the reason to add this layer first is that vf_sr uses it in its
tensorflow model, and the next plan is to update the python script
to convert tf.pad into native model.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
