The issue is that with a threadsafe hwaccel and multiple enabled
frame threads, hwaccel->uninit() is never called.
Previously, the function was guaranteed to never have any threads
with hwaccel contexts, so it never bothered to uninit any.
nvenc declares support for these formats, but if hwcontext_cuda doesn't
do that as well, then it's not possible to hwupload them for use in a
possible cuda pipeline before encoding.
I'm not sure why I originally did this, but there's no good reason to
put pointers to the cuda context and stream in the priv struct. They
are directly available in the device context that is already being
stored there.
- Changes in mov_write_video_tag function to handle EVC elementary stream
- Provided structure EVCDecoderConfigurationRecord that specifies the decoder configuration information for ISO/IEC 23094-1 video content
Signed-off-by: Dawid Kozinski <d.kozinski@samsung.com>
This reverts commit 9a245bdf5d.
This commit basically broke all samples with fractional framerates,
rather than fixing them.
I at this point do not understand the original issue anymore, and I'm
not sure how this slipped my initial testing.
All my test samples must have happened to have a simple timebase.
The actual dts values pretty much always are just a simple chain of
1,2,3,4,5,... Or maybe slightly bigger steps. Each increase by one means
an advance in time by one unit of the timebase.
So a fractional framerate/timebase is already not an issue.
So with this patch applied, the calculation might end up substracting
huge values (1001 is a common one) from the dts, which would be an
offset of that many frames, not of that many fractions of a second.
This broke at least muxing into mp4, if the sample happened to have a
fractional framerate.
I do not thing the original issue this patch tried to fix existed in the
first place, so it can be reverted without further consequences.
This flag is used to indicate to the hw frames in the gaps,
vaapi constructs it from a bunch of implicit API information
around surface ids. vulkan should just send it explicitly.
Reviewed-by: Lynne <dev@lynne.ee>
Enable the checked bitreader to avoid overread.
Also add a few checks in loops and between blocks so we exit instead of continued
execution.
Alternatively we could add manual checks so that no overread can happen. This would be
slightly faster but a bit more work and a bit more fragile
Fixes: Out of array accesses
Fixes: 59640/clusterfuzz-testcase-minimized-ffmpeg_dem_JPEGXL_ANIM_fuzzer-6584117345779712
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The tag comes from samples/ffmpeg/mov/unrecognized/bartjones.mov
really looks like some random data. Now the random tag matched
another file, which isn't a mov.
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>
This also fixed a warning: implicit conversion from enumeration
type 'TF_DataType' (aka 'enum TF_DataType') to different
enumeration type 'DNNDataType'.
Signed-off-by: Zhao Zhili <zhilizhao@tencent.com>
Determined experimentally, on various videos and hardware.
On Intel, using less resources in-flight is around 15% faster,
with similar results on Nvidia hardware.
This reduces memory needed dramatically, as unneeded resources
can be immediately returned to the pool.
Although waitforfences is threadsafe, we add a mutex wait around
it, as the mutex fence in combination with waitforfences assures
us that no other thread will reset the fence in the meanwhile
whilst the mutex is locked. This allows is to call
ff_vk_exec_discard_deps.
It was introduced for Vulkan, but it is equivalent to
short_term_ref_pic_set_size when !short_term_ref_pic_set_sps_flag,
and when !!short_term_ref_pic_set_sps_flag, Vulkan hardcodes a zero
anyway.
Lynne
Philip Langdale
Paul B Mahol
James Almer
Dawid Kozinski
Gyan Doshi
Timo Rothenpieler
Niklas Haas
Dave Airlie
Zhao Zhili
Michael Niedermayer
Martin Storsjö
Marvin Scholz