This avoids allocations and error checks etc. as well
as duplicate pointer lists in the CodedBitstreamFooContexts.
It also avoids casting const away for use as opaque,
as the RefStruct API supports const opaques.
The fact that some of the units are not refcounted
(i.e. they are sometimes part of an encoding context
like VAAPIEncodeH264Context) meant that CodedBitstreamUnit
still contains two pointers, one to the content
and another ownership pointer, replacing the AVBufferRef* pointer.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It avoids allocations and the corresponding error checks.
Also avoids casts and indirections.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It avoids allocations and the corresponding error checks.
It also avoids indirections and casts.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Avoids allocations and error checks when syncing the buffers.
Also avoids indirections.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Avoids allocations and error checks for these allocations;
e.g. syncing buffers across threads can't fail any more
and needn't be checked. It also gets rid of casts and
indirections.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Avoids allocations and error checks for these allocations;
e.g. syncing buffers across threads can't fail any more
and needn't be checked. It also avoids having to keep
H264ParamSets.pps and H264ParamSets.pps_ref and PPS.sps
and PPS.sps_ref in sync and gets rid of casts and indirections.
(The removal of these checks and the syncing code even more
than offset the additional code for RefStruct.)
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Avoids allocations and frees and error checks for said allocations;
also avoids a few indirections and casts.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
For now, this API is supposed to replace all the internal uses
of reference counted objects in libavcodec; "internal" here
means that the object is created in libavcodec and is never
put directly in the hands of anyone outside of it.
It is intended to be made public eventually, but for now
I enjoy the ability to modify it freely.
Several shortcomings of the AVBuffer API motivated this API:
a) The unnecessary allocations (and ensuing error checks)
when using the API. Besides the need for runtime checks it
imposes upon the developer the burden of thinking through
what happens in case an error happens. Furthermore, these
error paths are typically not covered by FATE.
b) The AVBuffer API is designed with buffers and not with
objects in mind: The type for the actual buffers used
is uint8_t*; it pretends to be able to make buffers
writable, but this is wrong in case the buffer is not a POD.
Another instance of this thinking is the lack of a reset
callback in the AVBufferPool API.
c) The AVBuffer API incurs unnecessary indirections by
going through the AVBufferRef.data pointer. In case the user
tries to avoid this indirection and stores a pointer to
AVBuffer.data separately (which also allows to use the correct
type), the user has to keep these two pointers in sync
in case they can change (and in any case has two pointers
occupying space in the containing context). See the following
commit using this API for H.264 parameter sets for an example
of the removal of such syncing code as well as the casts
involved in the parts where only the AVBufferRef* pointer
was stored.
d) Given that the AVBuffer API allows custom allocators,
creating refcounted objects with dedicated free functions
often involves a lot of boilerplate like this:
obj = av_mallocz(sizeof(*obj));
ref = av_buffer_create((uint8_t*)obj, sizeof(*obj), free_func, opaque, 0);
if (!ref) {
av_free(obj);
return AVERROR(ENOMEM);
}
(There is also a corresponding av_free() at the end of free_func().)
This is now just
obj = ff_refstruct_alloc_ext(sizeof(*obj), 0, opaque, free_func);
if (!obj)
return AVERROR(ENOMEM);
See the subsequent patch for the framepool (i.e. get_buffer.c)
for an example.
This API does things differently; it is designed to be lightweight*
as well as geared to the common case where the allocator of the
underlying object does not matter as long as it is big enough and
suitably aligned. This allows to allocate the user data together
with the API's bookkeeping data which avoids an allocation as well
as the need for separate pointers to the user data and the API's
bookkeeping data. This entails that the actual allocation of the
object is performed by RefStruct, not the user. This is responsible
for avoiding the boilerplate code mentioned in d).
As a downside, custom allocators are not supported, but it will
become apparent in subsequent commits that there are enough
usecases to make it worthwhile.
Another advantage of this API is that one only needs to include
the relevant header if one uses the API and not when one includes
the header or some other component that uses it. This is because there
is no RefStruct type analog of AVBufferRef. This brings with it
one further downside: It is not apparent from the pointer itself
whether the underlying object is managed by the RefStruct API
or whether this pointer is a reference to it (or merely a pointer
to it).
Finally, this API supports const-qualified opaque pointees;
this will allow to avoid casting const away by the CBS code.
*: Basically the only exception to the you-only-pay-for-what-you-use
rule is that it always uses atomics for the refcount.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is only used by encoders; this unfortunately necessitated
to add separate allocations to the SVQ1 encoder which uses
motion estimation without being a full member of mpegvideo.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This is no longer needed as the side data is available for decoders in the
AVCodecContext.
The tests affected reflect the removal of useless CPB and Stereo 3D side
data in packets.
Signed-off-by: James Almer <jamrial@gmail.com>
The changed references for fate-hevc-dv-rpu fate-mov-zombie happen because,
unlike ffmpeg and ffplay, ffprobe never injected packet side data, so the
display matrix side data at the container level is now present in the output
frames.
Signed-off-by: James Almer <jamrial@gmail.com>
Deprecate AVStream.side_data and its helpers in favor of the AVStream's
codecpar.coded_side_data.
This will considerably simplify the propagation of global side data to decoders
and from encoders. Instead of having to do it inside packets, it will be
available during init().
Global and frame specific side data will therefore be distinct.
Signed-off-by: James Almer <jamrial@gmail.com>
This will simplify the propagation of side data to decoders and from encoders.
Global side data will now reside in the AVCodecContext, thus be available
during init(), removing the need to propagate it inside packets.
Global and frame specific side data will therefore be distinct.
Signed-off-by: James Almer <jamrial@gmail.com>
Handling AVPacketSideData directly, which can used on structs other than
AVPacket.
This will be useful in the following commits.
Signed-off-by: James Almer <jamrial@gmail.com>
Drop reference to constants removed in 94eed68ace.
In particular, rename me_method to me_quality and add description for
supported values.
Address trac issue:
http://trac.ffmpeg.org/ticket/10003
In particular, clarify that it can either be set as a field of
AVCodecContext or as an x264 option, using a different specification.
Should address trac issue:
http://trac.ffmpeg.org/ticket/3947
The spec caps the prefix alphabet size to 32768 (i.e. 1 << 15) so we
should check for that and reject alphabets that are too large, in order
to prevent over-allocating.
Additionally, there's no need to allocate buffers that are as large as
the maximum alphabet size as these aren't stack-allocated, they're heap
allocated and thus can be variable size.
Added an overflow check as well, which fixes leaking the buffer, and
capping the alphabet size fixes two potential overruns as well.
Fixes: out of array access
Fixes: 62089/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-
5437089094959104.fuzz
Found-by: continuous fuzzing process
https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Found-by: Hardik Shah of Vehere (Dawn Treaders team)
Co-authored-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Leo Izen <leo.izen@gmail.com>
This patch will cause the parser to abort if it detects an icc profile
with an invalid size. This is particularly important if the icc profile
is entropy-encoded with zero bits per symbol, as it can prevent a
seemingly infinite loop during parsing.
Fixes: infinite loop
Fixes: 62374/clusterfuzz-testcase-minimized-ffmpeg_IO_DEMUXER_fuzzer
-5551878085410816
Found-by: continuous fuzzing process
https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reported-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Leo Izen <leo.izen@gmail.com>
Also, avoid spurious end-of-line after side data entries, and improve
rendering of compact output, by adding an indication of the side data
type for each entry.
Also fixes issue:
http://trac.ffmpeg.org/ticket/9266
Do not use put_sbits() where only unsigned is stored.
Reduce size of data_check_present field.
Reduce size of table of codebook_extremes[].
Avoid anonymously typedeffed structs.
Use encoder private context to store parameters.
Fix wrapping when calculating offsets.
Restructure arrays in encoder private context so to keep
arrays belonging to same subblock into separate structure.
Disable matrix coefficients as they are sometimes
producing wrong results.
Fixes multiplane support on Nvidia.
Also, remove the ENCODE usage, even if the driver signals it as supported.
Currently, it's not used, and when it is used, it'll be gated behind
two extension checks.
Some clips (i.e. SLIST_B_Sony_9) will use PPS 0 and 8, before PPS 1-7.
vulkan_hevc expects {sps,pps,vps}_list to be filled in order, which
causes PPS 8 to not be added to the Vulkan session params when it is
being used by a picture.
This removes the expectation that these lists are filled in order. The
indicies into vps_list are saved since there are multiple usages of it.
This also fixes a bug with some clips (i.e. PPS_A_qualcomm_7) which use
all 64 available PPS slots, causing the old loop to think there are more
than 64 PPS-es.
SVT-AV1 does not support requesting keyframes at arbitrary points
by setting pic_type to EB_AV1_KEY_PICTURE. So set force_key_frames
to 1 only when gop_size == 1.
Please see the comments in
https://gitlab.com/AOMediaCodec/SVT-AV1/-/issues/2076 for a bit more
details.
Signed-off-by: Vignesh Venkatasubramanian <vigneshv@google.com>
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
These flags will be overwritten later in ff_mpv_frame_start().
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It has currently not been done for H263, H263P and MPEG4.
Doing so avoids having to initialize the IDCT permutation
lateron when decoding packets in order to be able to parse
a quant matrix; it means that every mpegvideo decoder always
has an initialized IDCTDSPContext after init.
Initializing is done generically in ff_mpv_decode_init().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, ff_mpeg_update_thread_context() zeroes
the context to initialize on initialization failure.
This has been added in e1d7d4bd13.
Just as now, ff_mpeg_update_thread_context() simply
copied the src MpegEncContext over the dst MpegEncContext
to initialize it, but clear_context() was only added in
b160fc290c, so that cleaning up
on init failure was a minefield if performed.
It was not always performed, namely not before the first
allocation needed to be freed. In the fuzzer sample that
led to e1d7d4bd13, the call
to av_image_check_size() failed and before said commit,
the context contained lots of pointers from the src context,
leading to assert violations lateron.
Of course, the proper fix for this is resetting the pointers
(or even better, not copying them in the first place), so
this zeroing is unnecessary since commit
b160fc290c. It is also harmful,
because it makes initializing something only once during init
more complicated; See the h264chroma handling in the diff
for an example. Therefore it is removed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Before 998c9f15d1, the IDCTDSPContext
has only been initialized in ff_mpv_common_init() which is deferred
until immediately before decoding a picture; to nevertheless parse
the quant matrices in sequence headers or quant matrix extensions,
a dummy (identity) permutation has been stored in the codec's init
function; after ff_mpv_common_init() which could change the permutation
the matrices were repermutated.
Yet since said commit, the IDCTDSPContext is initialized during init
and does not change afterwards (unless the user forces different CPU
flags), so there is no need to reinitialize it; the repermutation code
can be removed as well.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Andreas Rheinhardt
Rémi Denis-Courmont
Paul B Mahol
James Almer
Stefano Sabatini
Leo Izen
Lynne
Benjamin Cheng
Vignesh Venkatasubramanian via ffmpeg-devel