It was done on a whim because of the FATE header check and was actually
meant to be removed before pushing.
Also, nobody in review spotted it.
Reviewed-by: wm4
libavcodec now automatically serializes decoding for hwaccels which
are not thread-safe. This means API users, which rely on the libavcodec
native software fallback mechanism, can now simply enable threading
without running into problems.
Certain hardware decoding APIs are not guaranteed to be thread-safe, so
having the user access decoded hardware surfaces while the decoder is
running in another thread can cause failures (this is mainly known to
happen with DXVA2).
For such hwaccels, only allow the decoding thread to run while the user
is inside a lavc decode call (avcodec_send_packet/receive_frame).
Merges Libav commit d4a91e65.
Signed-off-by: wm4 <nfxjfg@googlemail.com>
Tested-by: Michael Niedermayer <michael@niedermayer.cc>
This improves commit 59c7022740.
In ff_thread_report_progress(), the fast code path can load
progress[field] with the relaxed memory order, and the slow code path
can store progress[field] with the release memory order. These changes
are mainly intended to avoid confusion when one inspects the source code.
They are unlikely to have measurable performance improvement.
ff_thread_report_progress() and ff_thread_await_progress() form a pair.
ff_thread_await_progress() reads progress[field] with the acquire memory
order (in the fast code path). Therefore, one expects to see
ff_thread_report_progress() write progress[field] with the matching
release memory order.
In the fast code path in ff_thread_report_progress(), the atomic load of
progress[field] doesn't need the acquire memory order because the
calling thread is trying to make the data it just decoded visible to the
other threads, rather than trying to read the data decoded by other
threads.
In ff_thread_get_buffer(), initialize progress[0] and progress[1] using
atomic_init().
Signed-off-by: Wan-Teh Chang <wtc@google.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Merges Libav commit 343e2833.
Signed-off-by: wm4 <nfxjfg@googlemail.com>
When decoding with threads enabled, the get_format callback will be
called with one of the per-thread codec contexts rather than with the
outer context. If a hwaccel is in use too, this will add a reference
to the hardware frames context on that codec context, which will then
propagate to all of the other per-thread contexts for decoding. Once
the decoder finishes, however, the per-thread contexts are not freed
normally, so these references leak.
Merges Libav commit fd0fae60.
Signed-off-by: wm4 <nfxjfg@googlemail.com>
This patch deprecates anything that has to do with merging/splitting
side data. Automatic side data merging (and splitting), as well as all
API symbols involved in it, are removed completely.
Two FF_API_ defines are dedicated to deprecating API symbols related to
this: FF_API_MERGE_SD_API removes av_packet_split/merge_side_data in
libavcodec, and FF_API_LAVF_KEEPSIDE_FLAG deprecates
AVFMT_FLAG_KEEP_SIDE_DATA in libavformat.
Since it was claimed that changing the default from merging side data to
not doing it is an ABI change, there are two additional FF_API_ defines,
which stop using the side data merging/splitting by default (and remove
any code in avformat/avcodec doing this): FF_API_MERGE_SD in libavcodec,
and FF_API_LAVF_MERGE_SD in libavformat.
It is very much intended that FF_API_MERGE_SD and FF_API_LAVF_MERGE_SD
are quickly defined to 0 in the next ABI bump, while the API symbols are
retained for a longer time for the sake of compatibility.
AVFMT_FLAG_KEEP_SIDE_DATA will (very much intentionally) do nothing for
most of the time it will still be defined. Keep in mind that no code
exists that actually tries to unset this flag for any reason, nor does
such code need to exist. Code setting this flag explicitly will work as
before. Thus it's ok for AVFMT_FLAG_KEEP_SIDE_DATA to do nothing once
side data merging has been removed from libavformat.
In order to avoid that anyone in the future does this incorrectly, here
is a small guide how to update the internal code on bumps:
- next ABI bump (probably soon):
- define FF_API_LAVF_MERGE_SD to 0, and remove all code covered by it
- define FF_API_MERGE_SD to 0, and remove all code covered by it
- next API bump (typically two years in the future or so):
- define FF_API_LAVF_KEEPSIDE_FLAG to 0, and remove all code covered
by it
- define FF_API_MERGE_SD_API to 0, and remove all code covered by it
This forces anyone who actually wants packet side data to temporarily
use deprecated API to get it all. If you ask me, this is batshit fucked
up crazy, but it's how we roll. Making AVFMT_FLAG_KEEP_SIDE_DATA to be
set by default was rejected as an ABI change, so I'm going all the way
to get rid of this once and for all.
Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
* commit '12004a9a7f':
audiodsp/x86: yasmify vector_clipf_sse
audiodsp: reorder arguments for vector_clipf
Merged the version from Libav after a discussion with James Almer on
IRC:
19:22 <ubitux> jamrial: opinion on 12004a9a7f?
19:23 <ubitux> it was apparently yasmified differently
19:23 <ubitux> (it depends on the previous commit arg shuffle)
19:24 <ubitux> i don't see the magic movsxdifnidn in your port btw
19:24 <ubitux> it's a port from 1d36defe94
19:25 <jamrial> seems better thanks to said arg shuffle
19:25 <jamrial> the loop is the same, but init is simpler
19:25 <jamrial> probably worth merging
19:25 <ubitux> OK
19:25 <ubitux> thanks
19:26 <jamrial> curious they didn't make len ptrdiff_t after the previous bunch of commits, heh
19:26 <ubitux> yeah indeed
Both commits are merged at the same time to prevent a conflict with our
existing yasmified ff_vector_clipf_sse.
Merged-by: Clément Bœsch <u@pkh.me>
* commit '1d6c76e11f':
audiodsp/x86: fix ff_vector_clip_int32_sse2
No functionnal changes, only cosmetics. This issue was fixed in
9a9e2f1c8a.
Merged-by: Clément Bœsch <u@pkh.me>
* commit '6ce93757ee':
ppc: Update #endif comments
This commit is mostly a noop as we seem to support PPC LE (see
902ce2a6c4). Only the h264 chunks are
updated.
Merged-by: Clément Bœsch <u@pkh.me>
* commit '52730e0f86':
iir_filter: Change type of array stride parameters to ptrdiff_t
The merge also updates the MIPS code and drop the extra log.h include.
Merged-by: Clément Bœsch <u@pkh.me>
Fixes timeout with 847/clusterfuzz-testcase-5291877358108672
Fixes timeout with 850/clusterfuzz-testcase-5721296509861888
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: 864/clusterfuzz-testcase-4774385942528000
See: [FFmpeg-devel] [PATCH 1/2] avcodec/h264_direct: Fix runtime error: signed integer overflow: 2147483647 - -14133 cannot be represented in type 'int'
See: [FFmpeg-devel] [PATCH 2/2] avcodec/h264_direct: Fix runtime error: signed integer overflow: -9 - 2147483647 cannot be represented in type 'int'
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/targets/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
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 21512 bytes to 31400 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.
Before:
vp9_inv_dct_dct_16x16_sub1_add_10_neon: 284.6
vp9_inv_dct_dct_16x16_sub2_add_10_neon: 1902.7
vp9_inv_dct_dct_16x16_sub4_add_10_neon: 1903.0
vp9_inv_dct_dct_16x16_sub8_add_10_neon: 2201.1
vp9_inv_dct_dct_16x16_sub12_add_10_neon: 2510.0
vp9_inv_dct_dct_16x16_sub16_add_10_neon: 2821.3
vp9_inv_dct_dct_32x32_sub1_add_10_neon: 1011.6
vp9_inv_dct_dct_32x32_sub2_add_10_neon: 9716.5
vp9_inv_dct_dct_32x32_sub4_add_10_neon: 9704.9
vp9_inv_dct_dct_32x32_sub8_add_10_neon: 10641.7
vp9_inv_dct_dct_32x32_sub12_add_10_neon: 11555.7
vp9_inv_dct_dct_32x32_sub16_add_10_neon: 12499.8
vp9_inv_dct_dct_32x32_sub20_add_10_neon: 13403.7
vp9_inv_dct_dct_32x32_sub24_add_10_neon: 14335.8
vp9_inv_dct_dct_32x32_sub28_add_10_neon: 15253.6
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 16179.5
After:
vp9_inv_dct_dct_16x16_sub1_add_10_neon: 282.8
vp9_inv_dct_dct_16x16_sub2_add_10_neon: 1142.4
vp9_inv_dct_dct_16x16_sub4_add_10_neon: 1139.0
vp9_inv_dct_dct_16x16_sub8_add_10_neon: 1772.9
vp9_inv_dct_dct_16x16_sub12_add_10_neon: 2515.2
vp9_inv_dct_dct_16x16_sub16_add_10_neon: 2823.5
vp9_inv_dct_dct_32x32_sub1_add_10_neon: 1012.7
vp9_inv_dct_dct_32x32_sub2_add_10_neon: 6944.4
vp9_inv_dct_dct_32x32_sub4_add_10_neon: 6944.2
vp9_inv_dct_dct_32x32_sub8_add_10_neon: 7609.8
vp9_inv_dct_dct_32x32_sub12_add_10_neon: 9953.4
vp9_inv_dct_dct_32x32_sub16_add_10_neon: 10770.1
vp9_inv_dct_dct_32x32_sub20_add_10_neon: 13418.8
vp9_inv_dct_dct_32x32_sub24_add_10_neon: 14330.7
vp9_inv_dct_dct_32x32_sub28_add_10_neon: 15257.1
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 16190.6
Signed-off-by: Martin Storsjö <martin@martin.st>
This work is sponsored by, and copyright, Google.
This reduces the code size of libavcodec/aarch64/vp9itxfm_16bpp_neon.o from
26288 to 21512 bytes.
This gives a small slowdown of a couple of tens of cycles, but makes
it more feasible to add more optimized versions of these transforms.
Before:
vp9_inv_dct_dct_16x16_sub4_add_10_neon: 1887.4
vp9_inv_dct_dct_16x16_sub16_add_10_neon: 2801.5
vp9_inv_dct_dct_32x32_sub4_add_10_neon: 9691.4
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 16154.9
After:
vp9_inv_dct_dct_16x16_sub4_add_10_neon: 1899.5
vp9_inv_dct_dct_16x16_sub16_add_10_neon: 2827.2
vp9_inv_dct_dct_32x32_sub4_add_10_neon: 9714.7
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 16175.9
Signed-off-by: Martin Storsjö <martin@martin.st>
This work is sponsored by, and copyright, Google.
This reduces the code size of libavcodec/arm/vp9itxfm_16bpp_neon.o from
17500 to 14516 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.
Before: Cortex A7 A8 A9 A53
vp9_inv_dct_dct_16x16_sub4_add_10_neon: 4237.4 3561.5 3971.8 2525.3
vp9_inv_dct_dct_16x16_sub16_add_10_neon: 6371.9 5452.0 5779.3 3910.5
vp9_inv_dct_dct_32x32_sub4_add_10_neon: 22068.8 17867.5 19555.2 13871.6
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 37268.9 38684.2 32314.2 23969.0
After:
vp9_inv_dct_dct_16x16_sub4_add_10_neon: 4375.1 3571.9 4283.8 2567.2
vp9_inv_dct_dct_16x16_sub16_add_10_neon: 6415.6 5578.9 5844.6 3948.3
vp9_inv_dct_dct_32x32_sub4_add_10_neon: 22653.7 18079.7 19603.7 13905.3
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 37593.2 38862.2 32235.8 24070.9
Signed-off-by: Martin Storsjö <martin@martin.st>
Keep the idct32 coefficients in narrow form in q6-q7, and idct16
coefficients in lengthened 32 bit form in q0-q3. Avoid clobbering
q0-q3 in the pass1 function, and squeeze the idct16 coefficients
into q0-q1 in the pass2 function to avoid reloading them.
The idct16 coefficients are clobbered and reloaded within idct32_odd
though, since that turns out to be faster than narrowing them and
swapping them into q6-q7.
Before: Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_sub4_add_10_neon: 22653.8 18268.4 19598.0 14079.0
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 37699.0 38665.2 32542.3 24472.2
After:
vp9_inv_dct_dct_32x32_sub4_add_10_neon: 22270.8 18159.3 19531.0 13865.0
vp9_inv_dct_dct_32x32_sub32_add_10_neon: 37523.3 37731.6 32181.7 24071.2
Signed-off-by: Martin Storsjö <martin@martin.st>
