swscale: aarch64: Add a NEON implementation of interleaveBytes

This allows speeding up format conversions from yuv420 to nv12.

                             Cortex A53      A72      A73
interleave_bytes_c:             86077.5  51433.0  66972.0
interleave_bytes_neon:          19701.7  23019.2  15859.2
interleave_bytes_aligned_c:     86603.0  52017.2  67484.2
interleave_bytes_aligned_neon:   9061.0   7623.0   6309.0

Signed-off-by: Martin Storsjö <martin@martin.st>

checkasm: sw_rgb: Add a test for interleaveBytes

Signed-off-by: Martin Storsjö <martin@martin.st>

libavcodec: aarch64: Add a NEON implementation of pixblockdsp

                        Cortex A53    A72    A73
get_pixels_c:                140.7   87.7   72.5
get_pixels_neon:              46.0   20.0   19.5
get_pixels_unaligned_c:      140.7   87.7   73.0
get_pixels_unaligned_neon:    49.2   20.2   26.2
diff_pixels_c:               209.7  133.7  138.7
diff_pixels_neon:             54.2   31.7   23.5
diff_pixels_unaligned_c:     209.7  134.2  139.0
diff_pixels_unaligned_neon:   68.0   27.7   41.7

Signed-off-by: Martin Storsjö <martin@martin.st>

libavcodec: arm: Add a NEON implementation of pixblockdsp

                         Cortex A7     A8     A9    A53   A72
get_pixels_c:                144.7  146.0  143.0  137.7   69.0
get_pixels_armv6:            112.0  106.7   90.2   95.0   72.5
get_pixels_neon:              69.0   29.7   68.7   40.2   19.0
get_pixels_unaligned_c:      144.7  146.2  143.0  137.7   69.0
get_pixels_unaligned_neon:    77.0   36.5   72.5   48.5   19.0
diff_pixels_c:               376.7  319.7  265.5  307.7  148.0
diff_pixels_armv6:           179.0  159.5  205.5  139.0  142.0
diff_pixels_neon:             69.0   40.2   77.5   53.2   26.0
diff_pixels_unaligned_c:     376.7  319.7  265.5  307.7  148.0
diff_pixels_unaligned_neon:   85.0   54.5   93.5   66.7   26.0

Signed-off-by: Martin Storsjö <martin@martin.st>

checkasm: pixblockdsp: Add tests for get_pixels_unaligned and diff_pixels_unaligned

Signed-off-by: Martin Storsjö <martin@martin.st>

swscale: arm: fix NEON hscale init

The NEON hscale function only supports X8 filter sizes and should only
be selected when these are being used. At the moment filterAlign is
set to 8 but in the future when extra NEON assembly for specific sizes is
added they will need to have checks here too.

The immediate usecase for this change is making the hscale checkasm
test easier and without NEON specific edge-cases (x86 already has these
guards).

This applies the same fix from 718c8f9aa59751bb490e2688acf2b5cb68fd5ad1
on the 32 bit arm version of the function, fixing fate-checkasm-sw_scale
there.

Signed-off-by: Martin Storsjö <martin@martin.st>

avformat/mov: Read the QT Metadata Keys only once

If you have a file with multiple Metadata Keys, the second time you parse
the keys, you will re-alloc c->meta_keys without freeing the old one.
This change will avoid parsing all the consecutive Metadata keys.

Reviewed-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avutil/common: Add saturated add/sub operations for int64_t.

Many places are using their own custom code for handling overflow
around timestamps or other int64_t values. There are enough of these
now that having some common saturated math functions seems sound.

Signed-off-by: Dale Curtis <dalecurtis@chromium.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/mov: Don't allow negative sample sizes.

Signed-off-by: Dale Curtis <dalecurtis@chromium.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

checkasm: aarch64: Check for stack overflows

Also fill x8-x17 with garbage before calling the function.

Figure out the number of stack parameters and make sure that the
value on the stack after those is untouched.

Signed-off-by: Martin Storsjö <martin@martin.st>

checkasm: arm: Check for stack overflows

Figure out the number of stack parameters and make sure that the
value on the stack after those is untouched.

Signed-off-by: Martin Storsjö <martin@martin.st>

checkasm: arm: Don't use blx to call checkasm_fail_func

We should just use a normal bl here, and the linker will add the 'x'
bit if necessary.

This fixes calling the checkasm_fail_func on windows, where the
code is built in thumb mode (and the linker doesn't clear the 'x'
bit in the blx instruction).

Signed-off-by: Martin Storsjö <martin@martin.st>

checkasm: arm: Make the indentation consistent with other files

This makes it easier to share code with e.g. the dav1d implementation
of checkasm.

Signed-off-by: Martin Storsjö <martin@martin.st>

aarch64: Explicitly forbid using the x18 register

On windows and darwin (and modern android), the x18 register is reserved
and shouldn't be modified by user code, while it is freely available on
linux. Strictly avoid it, to keep the assembly code portable.

This would have helped catch the issue fixed in 872790b1f9be6
immediately.

Signed-off-by: Martin Storsjö <martin@martin.st>

Revert "avfilter/vf_framerate: if metadata lavfi.scd.mafd exists, we'll use it first"

This reverts commit 339593ca90cb3e05d659ec99a1479904ec742294.

Fixes null pointer dereference.

Signed-off-by: Marton Balint <cus@passwd.hu>

Revert "avfilter/vf_minterpolate: if metadata lavfi.scd.mafd exists, we'll use it first"

This reverts commit d88e1c9838dbcfe29d7835f2705ffc9ee6a36bf3.

Fixes null pointer dereference.

Signed-off-by: Marton Balint <cus@passwd.hu>

checkasm: add hscale test

This tests the hscale 8bpp to 14/18bpp functions with different filter
sizes.

Signed-off-by: Josh de Kock <josh@itanimul.li>

checkasm: add function to check and diff memory

This was ported from dav1d (c950e7101bdf5f7117bfca816984a21e550509f0).

Signed-off-by: Josh de Kock <josh@itanimul.li>

swscale: fix NEON hscale init

The NEON hscale function only supports X8 filter sizes and should only
be selected when these are being used. At the moment filterAlign is
set to 8 but in the future when extra NEON assembly for specific sizes is
added they will need to have checks here too.

The immediate usecase for this change is making the hscale checkasm
test easier and without NEON specific edge-cases (x86 already has these
guards).

Signed-off-by: Josh de Kock <josh@itanimul.li>

avcodec/nvenc: use framerate if available

The h264_nvenc and hevc_nvenc encoders aren't respecting the framerate in the codec context.
Instead it was using the timebase which in our use-case was 1/1000 so the encoder was behaving
as if we wanted 1000fps. This resulted in poor encoding results due to an extremely low bitrate.

Both the amf and qsv encoders already contain similar logic to first check the framerate before
falling back to the timebase.

Signed-off-by: Zachariah Brown <zachariah@renewedvision.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>

avcodec/option_table: mark venc_params as a video decoder flag opt type

It's not meant for audio or subtitles, or for encoders of any kind.

Reviewed-by: mypopy@gmail.com <mypopy@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

hwcontext_vulkan: fix incorrect print argument

FATE: add fate test for minterpolate filter

have tested on linux x86_32/64, mingw32/64 arm & mips qemu

Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_minterpolate: change the default threshold to get better scene change detect result

 ./ffmpeg -loglevel debug -i ../fate-suite/svq3/Vertical400kbit.sorenson3.mov -vf
 minterpolate=fps=60:mi_mode=blend -an -f null -
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 1600
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 4120
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 5780
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 6700
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 8140
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 9740
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 14060
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 15680
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 18480
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 20020
 [Parsed_minterpolate_0 @ 0x7fe7f3e193c0] scene changed, input pts 21740

 The results are consistent with tests/ref/fate/filter-metadata-scenedetect

 For the master, it'll detect more than 20 scene change for the same source.

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_minterpolate: correct the mafd calculation

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

fate: add scdet metadata test

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_minterpolate: if metadata lavfi.scd.mafd exists, we'll use it first

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_framerate: if metadata lavfi.scd.mafd exists, we'll use it first

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_scdet: add filter to detect scene change

Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avformat/mpegtsenc: Remove two duplicated fields

ts->{tsid,onid} stores the values of ts->{transport_stream_id,original_network_id}

Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Andriy Gelman <andriy.gelman@gmail.com>

avformat/mpegtsenc: Don't use heap allocated array to store pids

A temporary heap array currently stores pids from all streams.  It is
used to make sure there are no duplicated pids. However, this array is
not needed because the pids from past streams are stored in the
MpegTSWriteStream structs.

Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Andriy Gelman <andriy.gelman@gmail.com>

FATE: add a test for colorbalance

Tested on x86-32/64, mingw32/64, arm & mips qemu

Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_colorbalance:: Fix for framecrc bitexact for 32bit and 64bit system

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avcodec/mv30: fix warning: suggest braces around initialization of subobject [-Wmissing-braces]

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avformat/tls_schannel: immediately return decrypted data if available

Until now, we would have only attempted to utilize already decrypted
data if it was enough to fill the size of buffer requested, that could
very well be up to 32 kilobytes.

With keep-alive connections this would just lead to recv blocking
until rw_timeout had been reached, as the connection would not be
officially closed after each transfer. This would also lead to a
loop, as such timed out I/O request would just be attempted again.

By just returning the available decrypted data, keep-alive based
connectivity such as HLS playback is fixed with schannel.

avformat/tls_schannel: always decrypt all received data

The dec_buf seems to be properly managed between read calls,
and we have no logic to decrypt before attempting socket I/O.
Thus - until now - such data would not be decrypted in case of
connections such as HTTP keep-alive, as the recv call would
always get executed first, block until rw_timeout, and then get
retried by retry_transfer_wrapper.

Thus - if data is received - decrypt all of it right away. This way
it is available for the following requests in case they can be
satisfied with it.

pixblockdsp, avdct: Add get_pixels_unaligned

Use this in vf_spp.c, where the get_pixels operation is done on
unaligned source addresses.

Hook up the x86 (mmx and sse) versions of get_pixels to this
function pointer, as those implementations seem to support unaligned
use.

This fixes fate-filter-spp on armv7.

Signed-off-by: Martin Storsjö <martin@martin.st>

avcodec/adpcm: Fix integer overflow in ADPCM THP

The reference (thp.txt) uses floats so wrap around would seem incorrect.

Fixes: signed integer overflow: 1073741824 + 1073741824 cannot be represented in type 'int'
Fixes: 20658/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ADPCM_THP_fuzzer-5646302555930624
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

lavc/libopenh264enc: Add coder option to replace cabac

Set DEPRECATED flag to option cabac, replace with coder. The
priority logic is:
1. s->coder; then
2. avctx->coder_type; then
3. s->cabac.

Change the default option to -1 and allow the default cabac to be
determined by profile.

Add FF_API_OPENH264_CABAC macro for cabac to remove this option after
LIBAVCODEC_VERSION_MAJOR = 59.

Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Linjie Fu <linjie.fu@intel.com>

lavc/libopenh264enc: Allow specifying the profile through AVCodecContext

And determine the profile with following priority:
1. s->profile; then
2. avctx->profile; then
3. s->cabac; then
4. a default profile.

This seems more natural in case user somehow sets both avctx->profile and
s->profile.

Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Linjie Fu <linjie.fu@intel.com>

lavc/libopenh264enc: Rewrite profile handling

Support the profiles "constrained_baseline" and "high" for libopenh264
version >= 1.8, support "constrained_baseline" and "main" for earlier
version.

If option not supported with current version, convert to constrained
baseline with a warning for users.

Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Linjie Fu <linjie.fu@intel.com>

hwcontext_vulkan: don't add the optional VK_KHR_surface extension by default

Both API and CLI users can enable any extension they'd like using the options.

hwcontext_vulkan: don't error on unavailable user-specified extensions

Only warn instead. API users can find out which extensions were unavailable
by using the enabled_inst_extensions and enabled_dev_extensions fields.
This eliminates having to trial-and-error to find which extensions were missing.

hwcontext_vulkan: use the maximum amount of queues for each family

Due to our AVHWDevice infrastructure, where API users are offered a way
to derive contexts rather than always create new one, our filterchains,
being supported by a single hardware device context, can grow to considerable
size.
Hence, in such situations, using the maximum amount of queues the device offers
can be benefitial to eliminating bottlenecks where queue submissions on the
same family have to wait for the previous one to finish.

hwcontext_vulkan: update prepare_frame() for multiple semaphores when exporting

avcodec/ralf: Check num_blocks before use

Fixes: out of array access
Fixes: 20659/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_RALF_fuzzer-5739471895265280
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/iff: Test video_size being non zero

Fixes: Out of array access
Fixes: 20659/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_IFF_ILBM_fuzzer-5658548592967680
Fixes: 20659/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_IFF_ILBM_fuzzer-5723561177382912
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/cdtoons: Check sprite_offset is within the packet

Fixes: out of array read
Fixes: 20659/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CDTOONS_fuzzer-5754518731227136
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>

libavcodec/decode: Mark decode_simple_internal() as inline

This was suggested in
https://github.com/google/oss-fuzz/issues/3787
to reduce the grouping errors by oss-fuzz

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avformat/hlsenc: free options fix memleak in hls_write_trailer

Signed-off-by: Steven Liu <liuqi05@kuaishou.com>

avformat/hlsenc: fix filename memleak in hls_write_packet

Signed-off-by: Steven Liu <liuqi05@kuaishou.com>

avformat/hlsenc: move number out of hls_start

Signed-off-by: Steven Liu <liuqi05@kuaishou.com>

vp9dec: support exporting QP tables through the AVVideoEncParams API

vp9dec: factorise freeing per-tile allocated data

lavc: add a flag for exporting AVVideoEncParams from decoders

vf_showinfo: support AV_FRAME_DATA_VIDEO_ENC_PARAMS

libavutil: add API for exporting video frame quantizers

This is intended to replace the deprecated the AV_FRAME_DATA_QP_TABLE*
API and extend it to a wider range of codecs.

In the future, it may also be extended to support other encoding
parameters such as motion vectors.

Additional changes by Anton Khirnov <anton@khirnov.net> with suggestions
by Lynne <dev@lynne.ee>.

Signed-off-by: Juan De León <juandl@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Anton Khirnov <anton@khirnov.net>

doc/examples/muxing: convert to new encoding API

doc/examples/demuxing_decoding: convert to new decoding API

doc/examples/demuxing_decoding: drop -refcount

Non-refcounted frames are deprecated and there is no good reason to use
them.

avcodec/utvideodec: Fix integer overflow in decode_plane()

Fixes: signed integer overflow: 2147483594 + 142 cannot be represented in type 'int'
Fixes: 20492/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_UTVIDEO_fuzzer-5658568101724160
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/ttadsp: Fix several integer overflows in tta_filter_process_c()

Fixes: signed integer overflow: 1931744255 + 252497024 cannot be represented in type 'int'
Fixes: 20492/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TTA_fuzzer-5763348114440192
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/ralf: Fix integer overflow in decode_block()

Fixes: signed integer overflow: 289082077 - -2003141111 cannot be represented in type 'int'
Fixes: 20492/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_RALF_fuzzer-5196077752123392
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/nuv: widen buf_size type

Fixes: signed integer overflow: 65312 * 65312 cannot be represented in type 'int'
Fixes: 20492/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_NUV_fuzzer-5740176118906880
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libswscale: fix for floating point formats, require full chroma

upon more floating point testing, looks like I missed adding this bit.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

test: Add codecview test

tests: Add pp7, spp tests based on pp test

avfilter/vf_spp: Fix endian-dependance in add_block()

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avfilter/vf_spp: Remove unused AVCodecContext

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Revert "hwcontext_vulkan: only use one semaphore per image"

This reverts commit 97b526c192add6f252b327245fd9223546867352.
It broke the API, and assumed no other APIs used multiple semaphores.
This also disallowed certain optimizations to happen.

Dealing with APIs that give or expect single semaphores is easier when
we use per-image semaphores.

avformat/ivfenc: move bsf insertion to the init function

Reviewed-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avformat/ivfenc: add an AVOutputFormat.init() function

Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/iff: Fix several integer overflows

Fixes: negation of -2147483648 cannot be represented in type 'int32_t' (aka 'int'); cast to an unsigned type to negate this value to itself
Fixes: signed integer overflow: -2147483648 - 1 cannot be represented in type 'int32_t' (aka 'int')
Fixes: 20492/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_IFF_ILBM_fuzzer-5764066459254784
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/g729postfilter: Clip gain before scaling with AGC_FAC1

The fixed point integer reference specifies the multiplication used
to have 16bit input and clips so we need to clip the input
The floating point implementation does not seem to do that.

Fixes: signed integer overflow: 6317568 * 410 cannot be represented in type 'int'
Fixes: 20492/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_G729_fuzzer-5700189272932352
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/alac: Fix integer overflow with 24/20bps samples

Fixes: signed integer overflow: 1020048 * 4096 cannot be represented in type 'int'
Fixes: 20492/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALAC_fuzzer-5753877751660544
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

avcodec/adpcm: Clip predictor for IMA_APM

Fixes: signed integer overflow: -2147483647 - 61436 cannot be represented in type 'int'
Fixes: 20492/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ADPCM_IMA_APM_fuzzer-5092176004644864
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

hwcontext_vulkan: convert to general layout and transfer queue when exporting

The specs note that images should be in the GENERAL layout when exporting
for maximum compatibility.
CUDA exported images are handled differently, and the queue is the same,
so we don't need to do that there.

hwcontext_vulkan: create all images with concurrent sharing mode

As it turns out, we were already assuming and treating all images as if they had
concurrent access mode. This just changes the flag to CONCURRENT, which has less
restrictions than EXCLUSIVE, and fixed validation messages on machines with
multiple queues.
The validation layer didn't pick this up because the machine I was testing on
had only a single queue.

hwcontext_vulkan: fix inverted condition when exporting images to drm_prime

Calling vkGetImageSubresourceLayout is only legal for linear and drm images.

hwcontext_vulkan: update debugging layer name

hwcontext_vulkan: remove unused internal REQUIRED extension flag

This is a leftover from an old version which used the 1.0 Vulkan API
with the maintenance extensions being required.

hwcontext_vulkan: expose enabled device and instance extensions

This solves a huge oversight - it lets users reliably use their own
AVVulkanDeviceContext. Otherwise, the extensions supplied and enabled
are not discoverable by anything outside of hwcontext_vulkan.
Also clarifies that any user-supplied VkInstance must be at least 1.1.

hwcontext_vulkan: let users enable device and instance extensions using options

Also documents all options supported by the hwdevice.
This lets users enable all extensions they need without writing their own
instance initialization code.

avformat/hlsenc: Simplify setting base_output_dirname

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>

avformat/hlsenc: Simplify setting basename with av_asprintf()

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>

avformat/hlsenc: Simplify setting subtitle basename with av_asprintf

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>

avformat/hlsenc: Don't cast const away

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>

avformat/hlsenc: Remove redundant initializations

For every variantstream vs, vs->packets_written is set to one, only to be
set to zero a few lines below. Given that the relevant structure has
been zeroed during the allocation, this commit removes both assignments.
A redundant initialization for vs->init_range_length has been removed as
well a few lines below. Given that the relevant structure has been
zeroed during the allocation, this commit removes both assignments. A
redundant initialization for vs->init_range_length has been removed as
well.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>

avformat/hlsenc: Don't reset AVIOContext pointer manually a second time

ff_format_io_close() already does it for us.

Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>

avfilter/vf_signalstats: reindent after last commit

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_signalstats: make the alloc array used for any bit depth

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_signalstats: add SignalstatsContext.maxsize variable

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avfilter/vf_signalstats: rename config_props -> config_output

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

fate: add signalstats 8bit and 10bit test

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avformat/http: Fix for invalid use of av_strtok

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avformat/ftp: Fix for invalid use of av_strtok

By the av_strtok() description:
 * On the first call to av_strtok(), s should point to the string to
 * parse, and the value of saveptr is ignored. In subsequent calls, s
 * should be NULL, and saveptr should be unchanged since the previous
 * call.

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avformat/wavenc: simplify, use av_rescale_q() instead

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

avformat/vividas: simplify, use av_rescale_q() instead

note it'll cause a small difference in accuracy for the pts, please see the testing result below:
 $ wget http://samples.ffmpeg.org/archive/all/unknown+unknown+unknown+unknown+5029_paramount_en_1250.viv
 $ ./ffmpeg -t 0.04 -i ./unknown+unknown+unknown+unknown+5029_paramount_en_1250.viv -f null -
 old:
 pts: 522
 pts: 1044
 pts: 1567
 pts: 3918
 pts: 8097
 pts: 12277
 pts: 16457
 ...

 new:
 pts: 522
 pts: 1045
 pts: 1567
 pts: 3918
 pts: 8098
 pts: 12278
 pts: 16457
 ...

Signed-off-by: Limin Wang <lance.lmwang@gmail.com>

hwcontext_vulkan: optionally enable the VK_KHR_surface extension if available

This allows any phys_device derived to be used as a display rendering device.

avcodec/dstdec: Check sample rate

Fixes: out of array access
Fixes: 20659/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DST_fuzzer-5735812071424000
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>