av_ts_make_time_string() used "%.6g" format, but this format was losing
precision even when the timestamp to be printed was not that large. For example
for 3 hours (10800) seconds, only 1 decimal digit was printed, which made this
format inaccurate when it was used in e.g. the silencedetect filter. Other
detection filters printing timestamps had similar issues. Also time base
parameter of the function was *AVRational instead of AVRational.

Resolve these problems by introducing a new function, av_ts_make_time_string2().

We change the used format to "%.*f", use a precision of 6, except when printing
values near 0, in which case we calculate the precision dynamically to aim for
a similar precision in normal form as with %.6g.  No longer using scientific
representation can make parsing the timestamp easier for the users, we can
safely do this because the theoretical maximum of INT64_MAX*INT32_MAX still
fits into the string buffer in normal form.

We somewhat imitate %g by trimming ending zeroes and the potential decimal
point characters. In order not to trim "inf" as well, we assume that the
decimal point string does not contain the letter "f". Note that depending on
printf %f implementation, we might trim "infinity" to "inf".

Thanks for Allan Cady for bringing up this issue.
Signed-off-by: Marton Balint <cus@passwd.hu>

Broken in afa471d0. It just happened
to work before due to x86inc.asm previously performing XMM spills in
INIT_MMX mode which was more of a bug than an intentional feature.

Fixes many -Wenum-conversion warnings with Clang caused by
e6c2c870.
See e.g.
https://fate.ffmpeg.org/log.cgi?time=20240325033545&slot=aarch64-linux-clang-10&log=compileReviewed-by: Henrik Gramner <henrik@gramner.com>
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Forgotten in ecdc94b9.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Otherwise it would not be lossless.
Signed-off-by: James Almer <jamrial@gmail.com>

Reviewed-by: Xiang, Haihao <haihao.xiang@intel.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Co-Authored-by: Dave Airlie <airlied@redhat.com>

This is wanted by the Vulkan decoder.

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

- Properly initialize all the mappings to -1 by default.
- Make sure every output channel is assigned exactly once
- Autodetect a native layout when only native channels are present
- Always honor the user specified layout, but make sure the mapping is
  compatible with it

The last item is a regression from 4af412be.
Signed-off-by: Marton Balint <cus@passwd.hu>

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

MAX_CH is 64, therefore the maximum index is 63.
Signed-off-by: Marton Balint <cus@passwd.hu>

In this case in_channel_idx was never set and the default 0 was used.
Suprisingly no one noticed that the respective fate test output was wrong.
Signed-off-by: Marton Balint <cus@passwd.hu>

FL channel ID is 0, so for an unset value we must check for ID < 0.

Regression since 1f96db95.
Signed-off-by: Marton Balint <cus@passwd.hu>

Previously we always assumed that the channels are in native order, even if
they were not. The new channel layout API allows us to signal the proper
channel order, so let's do so.

Fixes ticket #98.

C11 provides static assertions via _Static_assert and
provides static_assert as a convenience define for this
in assert.h. Our codebase uses the latter, as _Static_assert
has actually already been deprecated in C23.

Not all toolchains that declare support for C11 actually
support it; e.g. MSVC 19.27 does not support _Static_assert,
but somehow supports static_assert. MSVC 19.27 admits to be
a "preview implementation of the ISO C11 standard",
so this is not surprising (MSVC 19.28 does not come with
this caveat).

Furthermore some FATE boxes [1] use old GCC toolchains (with
only experimental support for C11) where _Static_assert is
supported, but assert.h does not provide the fallback define.
They are broken since the first usage of static_assert.

This commit therefore checks whether static_assert and
_Static_assert work with assert.h included; if not,
configure errors out.

This intentionally drops support for MSVC 19.27. Users like
the old FATE boxes above can still add -Dstatic_assert=_Static_assert
to cflags as a workaround if desired.

[1]: https://fate.ffmpeg.org/report.cgi?time=20240321123620&slot=sh4-debian-qemu-gcc-4.7Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

Otherwise these statements might be torn apart by
av_logs from other threads.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

It has been moved after "st  = s->streams[pkt->stream_index]"
in b140b833.
Deduplicate ff_read_packet() and ff_buffer_packet()
while fixing this.
This also fixes shadowing in ff_read_packet().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>

The continue statement will break out of the do/while loop, not the
outer loop as intended. This is one (compound) statement anyway, so we
can remove the do/while entirely.
Signed-off-by: Leo Izen <leo.izen@gmail.com>

These variables are never read from, so they trigger -Wunused-variables
Signed-off-by: Leo Izen <leo.izen@gmail.com>

Make things up-to-date with upstream.

https://code.videolan.org/videolan/x86inc.asm

x86inc.asm adds defines for <argument_name>{b,w,d,q} which clashes with
the nasm d{b,w,d,q} pseudo-instructions for writing initialized data.

Following the usual logic for H.274 film grain.

I restricted this SEI to HEVC for now, until I see a H.264 sample.

Based on the AOMedia Film Grain Synthesis 1 (AFGS1) spec:
  https://aomediacodec.github.io/afgs1-spec/

The parsing has been changed substantially relative to the AV1 film
grain OBU. In particular:

1. There is the possibility of maintaining multiple independent film
   grain parameter sets, and decoders/players are recommended to pick
   the one most appropriate for the intended display resolution. This
   could also be used to e.g. switch between different grain profiles
   without having to re-signal the appropriate coefficients.

2. Supporting this, it's possible to *predict* the grain coefficients
   from previously signalled parameter sets, transmitting only the
   residual.

3. When not predicting, the parameter sets are now stored as a series of
   increments, rather than being directly transmitted.

4. There are several new AFGS1-exclusive fields.

I placed this parser in its own file, rather than h2645_sei.c, since
nothing in the generic AFGS1 film grain payload is specific to T.35, and
to compartmentalize the code base.

Implementation copied wholesale from dav1d, sans SIMD, under permissive
license. This implementation was extensively verified to be bit-exact,
so it serves as a much better starting point than trying to re-engineer
this from scratch for no reason. (I also authored the original
implementation in dav1d, so any "clean room" implementation would end up
looking much the same, anyway)

The notable changes I had to make while adapting this from the dav1d
code-base to the FFmpeg codebase include:

- reordering variable declarations to avoid triggering warnings
- replacing several inline helpers by avutil equivalents
- changing code that accesses frame metadata
- replacing raw plane copying logic by av_image_copy_plane

Apart from this, the implementation is basically unmodified.

Common utility function that can be used by all codecs to select the
right (any valid) film grain parameter set. In particular, this is
useful for AFGS1, which has support for multiple parameters.

However, it also performs parameter validation for H274.

To allow for AFGS1 usage, which can expose multiple parameter sets for
a single frame.

Not directly signalled by AV1, but we should still set this accordingly
so that users will know what the original intended video characteristics
and chroma resolution were.

Not directly signalled by AV1, but we should still set this accordingly
so that users will know what the original intended video characteristics
and chroma resolution were.

H.274 specifies that film grain parameters are signalled as intended for
4:4:4 frames, so we always signal this, regardless of the frame's actual
subsampling.

Follow the establish convention of printing the bit depth metadata
per-component.

This is needed for AV1 film grain as well, when using AFGS1 streams.
Also add extra width/height and subsampling information, which AFGS1
cares about, as part of the same API bump. (And in principle, H274
should also expose this information, since it is needed downstream to
correctly adjust the chroma grain frequency to the subsampling ratio)

Deprecate the equivalent H274-exclusive fields. To avoid breaking ABI,
add the new fields after the union; but with enough of a paper trail to
hopefully re-order them on the next bump.

small cleanup for style, redundant semicolons, goto labels,
in FFmpeg, we put goto labels at brace level.
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>

Fixes: out of array access
Fixes: 67021/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-4883576579489792

Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpegSigned-off-by: Michael Niedermayer <michael@niedermayer.cc>

Fixes fate-ffmpeg-loopback-decoding with THREADS=random*

avpkt->pos needs to be set for generic indexing or features such as the
stream_loop option will not work.
Co-authored-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Signed-off-by: Leo Izen <leo.izen@gmail.com>