This is done a second time for 5.0 because master was
merged into 5.0 so that it contains the recent DOVI additions.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, we had a PacketList structure which is actually
a PacketListEntry; a proper PacketList did not exist
and all the related functions just passed pointers to pointers
to the head and tail elements around. All these pointers were
actually consecutive elements of their containing structs,
i.e. the users already treated them as if they were a struct.
So add a proper PacketList struct and rename the current PacketList
to PacketListEntry; also make the functions use this structure
instead of the pair of pointers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This function is quite small (96B with GCC 11.2 on x64 Ubuntu 21.10
at -O3), making it more economical to duplicate it into libavformat
instead of exporting it as avpriv: Doing so saves 2x24B in .dynsim,
2x16B in .dynstr, 2x2B .gnu.version, 24B in .rela.plt, 16B in .plt,
16B in .plt.sec (if enabled), 4B .gnu.hash; besides the actual
duplicated code this also adds 2x8B .eh_frame_hdr and 24B .eh_frame.
In other words: Duplicating is neutral size-wise (it is also presumed
neutral for other systems). Given that it avoids the runtime
overhead of dynamic symbols, it is advantageouos to duplicate the
function.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
These arrays have a size of 180 resp. six bytes. This does not
make it worthwhile to export them due to the overhead this occurs;
for x64 Elf/Linux/GNU: 2x2B version, 2x24B .dynsym, 24B .rela.dyn,
8B .got, 4B hash + twice the size of the name (here 20+23B).
Therefore these symbols are unavprived and duplicated for shared
builds.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
avpriv_mpeg4audio_sample_rates has a size of 64B and it is currently
avpriv; a clone of it exists in aacenctab.h and from there it is inlined
in aacenc.c (which also uses the avpriv version) and in the FLV muxer.
This means that despite it being avpriv both libavformat as well as
libavcodec have copies already.
This situation is clearly suboptimal. Given the overhead of exporting
symbols (for x64 Elf/Linux/GNU: 2x2B version, 2x24B .dynsym, 24B .rela.dyn,
8B .got, 4B hash + twice the size of the name (here 31B)) the object is
unavprived, i.e. duplicated into libavformat when creating a shared
build; but the duplicates in the AAC encoder and FLV muxer are removed.
This involves splitting of the sample rate table into a file of its own;
this allowed to break some spurious dependencies (e.g. both the AAC
encoder as well as the Matroska demuxer actually don't need the
mpeg4audio_get_config stuff).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This group is mainly for the users of *_mpeg4audio_get_config2();
it is not for those who only use avpriv_mpeg4audio_sample_rates.
This is in preparation for splitting the latter into a file of its own;
if there were no CONFIG_EXTRA group for *_mpeg4audio_get_config2()
users, one would have to add a dependency to the new file for all
these users on top of the existing dependency on mpeg4audio.o.
Adding a new CONFIG_EXTRA group only takes effect after a reconfigure;
so in order to force a reconfigure some unnecessary headers from
libavdevice/alldevices.c have been removed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There are seven MJPEG-tables, five small (1x12, 4x17) and two
not small (2x162). These are all avpriv, despite this not being
worthwhile due to the overhead of exporting a symbol: The total
overhead for each symbol consists of two entries in .dynsym (24B each),
one entry in the importing library's .rela.dyn (24B) and one in .got
(8B) as well as 2x2B for symbol versions and 4B for symbol hashes
in the exporting library; in addition to that, the name of the symbol
is included in both exporting and importing libraries, using 2x210 bytes
in this case.
(The above numbers are for a x64 Elf/Linux/GNU system. Other platforms
will give different numbers.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Said table is 64 bytes long and exported so that it can be used both
in libavcodec and libavformat. This commit stops doing so and instead
duplicates it for shared builds, because the overhead of exporting the
symbol is bigger than 64 bytes. It consists of the length of the name of
the symbol (2x24 bytes), two entries in .dynsym (2x24 bytes), two
entries for symbol version (2x2 bytes), one hash value in the exporting
library (4 bytes) in addition to one entry in the importing library's
.got and .rela.dyn (8 + 24 bytes).
(The above numbers are for a Linux/GNU/Elf system; the numbers for other
platforms may be different.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is small (16 B) and therefore the overhead of exporting it more
than outweighs the size savings from not having duplicated symbols:
When the symbol is no longer avpriv, one saves twice the size of
the string containing the symbols name (2x30 byte), two entries
in .dynsym (24 bytes each on x64), one entry in the importing libraries
.got and .rela.dyn (8 + 24 bytes on x64) and two entries for the
symbol version (2 bytes each) and one hash value in the exporting
library (4 bytes).
(The exact numbers are of course different for other platforms
(e.g. when using dlls), but given that the strings saved alone
more than outweigh the array size it can be presumed that this
is beneficial for all platforms.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
libavcodec currently exports four avpriv symbols that deal with
PixelFormatTags: avpriv_get_raw_pix_fmt_tags, avpriv_find_pix_fmt,
avpriv_pix_fmt_bps_avi and avpriv_pix_fmt_bps_mov. The latter two are
lists of PixelFormatTags, the former returns such a list and the second
searches a list for a pixel format that matches a given fourcc; only
one of the aforementioned three lists is ever searched.
Yet for avpriv_pix_fmt_bps_avi, avpriv_pix_fmt_bps_mov and
avpriv_find_pix_fmt the overhead of exporting these functions actually
exceeds the size of said objects (at least for ELF; the following numbers
are for x64 Ubuntu 20.10):
The code size of avpriv_find_pix_fmt is small (GCC 10.2 37B, Clang 11 41B),
yet exporting it adds a 20B string for the name alone to the exporting
as well as to each importing library; there is more: Four bytes in the
exporting libraries .gnu.hash; two bytes each for the exporting as well
as each importing libraries .gnu.version; 24B in the exporting as well
as each importing libraries .dynsym; 16B+24B for an entry in .plt as
well as the accompanying relocation entry in .rela.plt for each
importing library.
The overhead for the lists is similar: The strings are 23B and the
.plt+.rela.plt pair is replaced by 8B+24B for an entry in .got and
a relocation entry in .rela.dyn. These lists have a size of 80 resp.
72 bytes.
Yet for ff_raw_pix_fmt_tags, exporting it is advantageous compared to
duplicating it into libavformat and potentially libavdevice. Therefore
this commit replaces all library uses of the four symbols with a single
function that is exported for shared builds. It has an enum parameter
to choose the desired list besides the parameter for the fourcc. New
lists can be supported with new enum values.
Unfortunately, avpriv_get_raw_pix_fmt_tags could not be removed, as the
fourcc2pixfmt tool uses the table of raw pix fmts. No other user of this
function remains.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes regression in a1c4929f: there apparently are devices out there
that expose a pin default format that has parameters outside the
capabilities of any of the formats exposed on the pin (sic?). The
VirtualCam plugin (v 2.0.5) of OBS-Studio (v 27.1.3) is such a device.
Now when a default format was found, but not selected when iterating all
formats, fall back to directly setting the default format.
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
fix regression introduced in 911ba8417e.
Removal of WINAPI decoration from function signatures caused crashed
when using dshow on x86.
Fixes#9568
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
Cleanup was missing for when the show_analog_tv_tuner_audio_dialog is
true.
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
And expose the parsed values as frame side data. Update FATE results to
match.
It's worth documenting that this relies on the dovi configuration record
being present on the first AVPacket fed to the decoder, which in
practice is the case if if the API user has called something like
av_format_inject_global_side_data, which is unfortunately not the
default.
This commit is not the time and place to change that behavior, though.
Signed-off-by: Niklas Haas <git@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
To avoid the ref for this growing to a very large size when attaching
the parsed RPU side data. Since this sample does not have any dynamic
metadata, two frames will serve just as well as 100.
Signed-off-by: Niklas Haas <git@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Based on a mixture of guesswork, partial documentation in patents, and
reverse engineering of real-world samples. Confirmed working for all the
samples I've thrown at it.
Contains some annoying machinery to persist these values in between
frames, which is needed in theory even though I've never actually seen a
sample that relies on it in practice. May or may not work.
Since the distinction matters greatly for parsing the color matrix
values, this includes a small helper function to guess the right profile
from the RPU itself in case the user has forgotten to forward the dovi
configuration record to the decoder. (Which in practice, only ffmpeg.c
and ffplay do..)
Notable omissions / deviations:
- CRC32 verification. This is based on the MPEG2 CRC32 type, which is
similar to IEEE CRC32 but apparently different in subtle enough ways
that I could not get it to pass verification no matter what parameters
I fed to av_crc. It's possible the code needs some changes.
- Linear interpolation support. Nothing documents this (beyond its
existence) and no samples use it, so impossible to implement.
- All of the extension metadata blocks, but these contain values that
seem largely congruent with ST2094, HDR10, or other existing forms of
side data, so I will defer parsing/attaching them to a future commit.
- The patent describes a mechanism for predicting coefficients from
previous RPUs, but the bit for the flag whether to use the
prediction deltas or signal entirely new coefficients does not seem to
be present in actual RPUs, so we ignore this subsystem entirely.
- In the patent's spec, the NLQ subsystem also loops over
num_nlq_pivots, but even in the patent the number is hard-coded to one
iteration rather than signalled. So we only store one set of coefs.
Heavily influenced by https://github.com/quietvoid/dovi_tool
Documentation drawn from US Patent 10,701,399 B2 and ETSI GS CCM 001
Signed-off-by: Niklas Haas <git@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In order to be able to extend this struct later (as the Dolby Vision RPU
evolves), all of the 'container' structs are considered extensible, and
the individual constituent fields must instead be accessed via offsets.
The precedent for this style of access is set in
<libavutil/detection_bbox.h>
Signed-off-by: Niklas Haas <git@haasn.dev>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Improves readability and slightly decreases codesize.
While just at it, also remove a check whether the packet list is
nonempty before freeing it, as freeing an empty list is fine
and basically a no-op.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This removes one of the last usages of sizeof(AVPacket)
in the generic muxing code.
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In case of shared builds, some object files containing tables
are currently duplicated into other libraries: log2_tab.c,
golomb.c, reverse.c. The check for whether this is duplicated
is simply whether CONFIG_SHARED is true. Yet this is crude:
E.g. libavdevice includes reverse.c for shared builds, but only
needs it for the decklink input device, which given that decklink
is not enabled by default will be unused in most libavdevice.so.
This commit changes this by making it more explicit about what
to duplicate from other libraries. To do this, two new Makefile
variables were added: SHLIBOBJS and STLIBOBJS. SHLIBOBJS contains
the objects that are duplicated from other libraries in case of
shared builds; STLIBOBJS contains stuff that a library has to
provide for other libraries in case of static builds. These new
variables provide a way to enable/disable with a finer granularity
than just whether shared builds are enabled or not. E.g. lavd's
Makefile now contains: SHLIBOBJS-$(CONFIG_DECKLINK_INDEV) += reverse.o
Another example is provided by the golomb tables. These are provided
by lavc for static builds, even if one uses a build configuration
that makes only lavf use them. Therefore lavc's Makefile contains
STLIBOBJS-$(CONFIG_MXF_MUXER) += golomb.o, whereas lavf's Makefile
has a corresponding SHLIBOBJS-$(CONFIG_MXF_MUXER) += golomb_tab.o.
E.g. in case the MXF muxer is the only component needing these tables
only libavformat.so will contain them for shared builds; currently
libavcodec.so does so, too.
(There is currently a CONFIG_EXTRA group for golomb. But actually
one would need two groups (golomb_avcodec and golomb_avformat) in
order to know when and where to include these tables. Therefore
this commit uses a Makefile-based approach for this and stops
using these groups for the users in libavformat.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The packet given to muxers is not used afterwards; it is always
unreferenced by libavformat. Ergo muxers are allowed to keep
the references in the packets and e.g. move the ownership to
a packet list. This is what this commit does.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Improves code legibility by not using bit shifts.
Also avoids duplicating the dvcC/dvvC ISOM box writing code.
Signed-off-by: quietvoid <tcChlisop0@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Adds handling of dvcC/dvvC block addition mappings.
The parsing creates AVDOVIDecoderConfigurationRecord side data.
The configuration block is written when muxing into Matroska,
if DOVI side data is present for the track.
Most of the Matroska element parsing is based on Plex's FFmpeg source code.
Signed-off-by: quietvoid <tcChlisop0@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Both parse/write implementations are based on mov/movenc.
This only adds support for the "Dolby Vision configuration box".
Other configuration boxes, such as
"Dolby Vision enhancement layer configuration box" are not supported.
The new functions will be used to implement parsing/writing the DOVI config
for Matroska, as well as to refactor both mov/movenc to use dovi_isom functions.
Signed-off-by: quietvoid <tcChlisop0@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
'*src' and '*avctx' point to the same memory. It is enough to keep one of them.
Signed-off-by: Yu Yang <yuyang14@kuaishou.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
On 64 bit Operating System, sizeof(ScaleVulkanContext):
reduce from 2400 to 2392 on Linux
reduce from 2416 to 2408 on Windows
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
The following command is on how to apply passthrough option:
ffmpeg -init_hw_device vulkan -i input.264 -vf hwupload=extra_hw_frames=16,transpose_vulkan=passthrough=landscape,hwdownload,format=yuv420p output.264
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
We don't use it. Was copied from libplacebo's recommended defaults.
Creates problems with validation on Intel devices, where the driver
still advertizes it, even though it's not usable without a swapchain.
No speex samples with non default frame sizes are known (to me)
the official speexenc seems to only generate the 3 default ones.
Thus it may be that the fuzzer samples where the first non default
values encountered by the decoder.
Possibly the "<" should be "!="
Fixes: out of array access
Fixes: 42821/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SPEEX_fuzzer-5640695772217344
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
added "cache_redirect" option to http.
when enabled, requests issued after seek will use the latest redirected
url.
when disabled, each call to seek will revert to the original url that
was sent to http_open.
currently, the default remains 'enabled', until the next major
libavformat bump, where it will change to 'disabled'.
Very high stts sample deltas may occasionally be intended but usually
they are written in error or used to store a negative value for dts correction
when treated as signed 32-bit integers.
This option lets the user set an upper limit, beyond which the delta is clamped to 1.
Values greater than the limit if negative when cast to int32 are used to adjust onward dts.
Unit is the track time scale. Default is UINT_MAX - 48000*10 which
allows upto a 10 second dts correction for 48 kHz audio streams while
accommodating 99.9% of uint32 range.
Signed-off-by: Gyan Doshi <ffmpeg@gyani.pro>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
We do not support this as we multiply by 1000
Fixes: signed integer overflow: -45318575073853696 * 1000 cannot be represented in type 'long'
Fixes: 42804/clusterfuzz-testcase-minimized-ffmpeg_dem_LIVE_FLV_fuzzer-4630325425209344
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The whole concept is just not correct, also as it seems not to be needed
at all, all dng files i have decode without this.
Fixes: various crashes
Fixes: 42937/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TIFF_fuzzer-4625073334517760
Fixes: 42938/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TIFF_fuzzer-4643368217477120
Fixes: 42939/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TIFF_fuzzer-4925325908246528
Fixes: 42940/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TIFF_fuzzer-4925378806808576
Fixes: 42941/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TIFF_fuzzer-6202009265504256
Fixes: 42944/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TIFF_fuzzer-6076860998483968
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This value is later passed to MediaCodec and checked at decoder init.
Notably decoding of 10-bit streams before this commit would "work" without
returning errors but only return garbage output (on most Android devices).
The V4L2M2M API operates asynchronously, so multiple packets can
be enqueued before getting a batch of frames back. Since it was
only possible to receive a frame by submitting another packet,
there wasn't a way to drain those excess output frames from when
avcodec_receive_frame() returned AVERROR(EAGAIN).
Reviewed-by: Andriy Gelman <andriy.gelman@gmail.com>
Signed-off-by: Cameron Gutman <aicommander@gmail.com>
Most of user data unregistered SEIs are privated data which defined by user/
encoder. currently, the user data unregistered SEIs found in input are forwarded
as side-data to encoders directly, it'll cause the reencoded output including some
useless UDU SEIs.
I prefer to add one option to enable/disable it and default is off after I saw
the patch by Andreas Rheinhardt:
https://patchwork.ffmpeg.org/project/ffmpeg/patch/AM7PR03MB66607C2DB65E1AD49D975CF18F7B9@AM7PR03MB6660.eurprd03.prod.outlook.com/
How to test by cli:
ffmpeg -y -f lavfi -i testsrc -c:v libx264 -frames:v 1 a.ts
ffmpeg -y -i a.ts -c:v libx264 -udu_sei 1 b.ts
ffmpeg -y -i a.ts -c:v libx264 -udu_sei 0 c.ts
# check the user data unregistered SEIs, you'll see two UDU SEIs for b.ts.
# and mediainfo will show with wrong encoding setting info
ffmpeg -i b.ts -vf showinfo -f null -
ffmpeg -i c.ts -vf showinfo -f null -
This fixes tickets #9500 and #9557.
Reviewed-by: "zhilizhao(赵志立)" <quinkblack@foxmail.com>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
This patch will fix following command:
ffmpeg -hwaccel vaapi -hwaccel_output_format vaapi -i input.264 \
-vf 'scale_vaapi=w=7680:h=4096' -c:v vp9_vaapi output.ivf
Max width of a vp9 tile is 4096. If the source frame > 4096, we need split to multiple tiles.
Reviewed-by: Limin Wang <lance.lmwang@gmail.com>
Signed-off-by: Zhang yuankun <yuankunx.zhang@intel.com>
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
Some DirectShow devices (Logitech C920 webcam) expose each DirectShow
format they support twice, once without and once with extended color
information. During format selection, both match, this patch ensures
that the format with extended color information is selected if it is
available, else it falls back to a matching format without such
information. This also necessitated a new code path taken for default
formats of a device (when user didn't request any specific video size,
etc), because the default format may be one without extended color
information when a twin with extended color information is also
available. Getting the extended color information when available is
important as it allows setting the color space, range, primaries,
transfer characteristics and chroma location of the stream provided by
dshow, enabling users to get more correct color automatically out of
their device.
Closes: #9271
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
Enabled discovering a DirectShow device's color range, space, primaries,
transfer characteristics and chroma location, if the device exposes that
information. Sets them in the stream's codecpars.
Co-authored-by: Valerii Zapodovnikov <val.zapod.vz@gmail.com>
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
fftools now print info about what media type(s), if any, are provided by
sink and source avdevices.
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
The list returned by get_device_list now contains info about what media
type(s), if any, can be provided by each device.
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
An avdevice, regardless of whether its category says its an audio or
video device, may provide access to devices providing different media
types, or even single devices providing multiple media types. Also, some
devices may provide no media types. dshow is an example encompassing all
of these cases. Users should be provided with this information, so
AVDeviceInfo is extended to provide it.
Bump avdevice version
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
the list_devices option of dshow didn't indicate whether a specific
device provides audio or video output. This patch iterates through all
media formats of all pins exposed by the device to see what types it
provides for capture, and prints this to the console for each device.
Importantly, this now allows to find devices that provide both audio and
video, and devices that provide neither.
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
avdevice/dshow is a realtime device and as such does not support
seeking. Therefore, its demuxer format should define the
AVFMT_NOBINSEARCH, AVFMT_NOGENSEARCH and AVFMT_NO_BYTE_SEEK flags.
With these flags set, attempting to seek (with, e.g.,
avformat_seek_file()) correctly yields -1 (operation not permitted)
instead of -22 (invalid argument).
This actually seems to apply to many other devices, at least the
gdigrab, v4l2, vfwcap, x11grab, fbdev, kmsgrab and android_camera
devices, from reading the source.
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
GetTime may return an error indication that the sample has not
timestamps, or may return a NULL start time. In those cases, fall back
to graph time. Emit log when that happens.
Improve logging in the frame receive function: now logged against
correct avclass instead of NULL.
Better debug message in case sample dropped: could now be audio or
video frame.
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
No need to query twice, use value we've already unconditionally got.
Improve variable names
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
The dshow avdevice ignores timestamps for video frames provided by the
DirectShow device, instead using wallclock time, apparently because the
implementer of this code had a device that provided unreliable
timestamps. Me (and others) would like to use the device's timestamps.
The new use_video_device_timestamps option for dshow device enables them
to do so. Since the majority of video devices out there probably provide
fine timestamps, this patch sets the default to using the device
timestamps, which means best fidelity timestamps are used by default.
Using the new option, the user can switch this off and revert to the old
behavior, so a fall back remains available in case the device provides
broken timestamps.
add use_video_device_timestamps to docs.
Closes: #8620
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
list_options true would crash when both a video and an audio device were
specified as input. Crash would occur on line 784 because
ctx->device_unique_name[otherDevType] would be NULL
Signed-off-by: Diederick Niehorster <dcnieho@gmail.com>
Reviewed-by: Roger Pack <rogerdpack2@gmail.com>
The earlier code did not account for the fact that
av_display_rotation_set() wants the angle in the anticlockwise
direction (despite what its documentation stated for a long time);
furthermore, the H.2645 spec wants the flips applied first,
whereas our code did it the other way around. This can be fixed
by negating the angle once for every flip.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The transpose filter has modes equivalent to "rotation by 90°/270°"
followed by horizontal flips.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In case of an orthogonal transformation av_display_rotation_get()
returns the (anticlockwise) degree that the unit vector in x-direction
gets rotated by; get_rotation in cmdutils.c makes a clockwise degree
out of this. So if one inserts a transpose filter corresponding to
this degree, then the x-vector gets mapped correctly and there are
two possibilities for image of the y-vector, namely the two unit
vectors orthogonal to the image of the x-vector.
E.g. if the x-vector gets rotated by 90° clockwise, then the two
possibilities for the y-vector are the unit vector in x direction
or its opposite. The latter case is a simple 90° rotation for both
vectors* whereas the former is a simple 90° clockwise rotation followed
by a horizontal flip. These two cases can be distinguished by looking
at the x-coordinate of the image of the y-vector, i.e. by looking
at displaymatrix[3]. Similarly for the case of a 270° clockwise
rotation.
These two cases were previously wrong (they were made to match
wrongly parsed exif rotation tag values).
*: For display matrices, the y-axis points downward.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The cases in which there was flipping together with a rotation
that is not a multiple of the identity were wrong.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
sets coded_width / coded_height too to keep them consistent with
width / height
Fixes: OOM
Fixes: 42263/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TIFF_fuzzer-5653333619113984
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fix#7830
When we upload a frame that is not padded as MSDK requires, we create a
new AVFrame to copy data. The frame's padding data is uninitialized so
it brings run to run problem. For example, If we run the following
command serveral times we will get different outputs.
ffmpeg -init_hw_device qsv=qsv:hw -qsv_device /dev/dri/renderD128 \
-filter_hw_device qsv -f rawvideo -s 192x200 -pix_fmt p010 \
-i 192x200_P010.yuv -vf "format=nv12,hwupload=extra_hw_frames=16" \
-c:v hevc_qsv output.265
According to https://github.com/Intel-Media-SDK/MediaSDK/blob/master/doc/mediasdk-man.md#encoding-procedures
"Note: It is the application's responsibility to fill pixels outside
of crop window when it is smaller than frame to be encoded. Especially
in cases when crops are not aligned to minimum coding block size (16
for AVC, 8 for HEVC and VP9)"
I add a function to fill padding area with border pixel to fix this
run2run problem, and also move the new AVFrame to global structure
to reduce redundant allocation operation to increase preformance.
Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
- Ensure the yadif .metal compiles when targeting any Metal runtime version
- Use some preprocessor awkwardness to ensure Core Video's Metal-specific
functionality is exposed regardless of our deployment target (this works
around what seems to be an SDK header bug, filed as FB9816002)
- Ensure all direct references to Metal functions and classes are gated
behind runtime version checks (this satisfies clang's deployment-target
violation warnings provided by -Wunguarded-availability).
Apparently Metal.framework is included with the command line tools
(and thus may be present without Xcode), but the Metal compiler is only
included as part of Xcode.
On some encoders, this defaults to true, which can result in encode speed
being _limited_ to only slightly above realtime (as a power-saving measure),
so we need a way to disable it.
VideoToolbox internally sets all the colorspace parameters to BT709,
regardless of what the bitstream actually indicates, so we need to
replace that with what we've parsed.
Adds support for concat demuxer to copy the side data information
from the input file to the resulting file. It will behave like the
metadata copy, where the metadata of the first file is kept in the
the output file.
Extract the current code that already performs the stream side_data
copy into a separate method and reuse the method in the concat demuxer.
Signed-off-by: Gerard Sole <g.sole.ca@gmail.com>
Check for the patch version as well as the major+minor version.
The VK_API_VERSION macros are not usable in preprocessor code due
to casts.
The patch (header) version is meant to linearly increment and
not be reset, however it's better to trust, but verify.
The check here is meant to check for whether avcintra-class option
(default value -1) has been set; yet it checks for the x264_param_t
value where 0 is the default value (treated as "no avcintra-mode"
by x264). This meant that in-band extradata has been added unnecessarily
when using global headers; furthermore, the first output packet
had two x264 SEIs.
Reviewed-by: Limin Wang <lance.lmwang@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Resolves a warning that duration is being innaccurately estimated based
on bitrate.
Signed-off-by: John-Paul Stewart <jpstewart@personalprojects.net>
Reviewed-by: Peter Ross <pross@xvid.org>
Prior to this patch, for version 2 of the file format the frame rate was
hard-coded at 15 fps. This uses the 64-bit floating-point value from
the data stream, similar to what is already done for version 3 of the
file format (around line 206).
Signed-off-by: John-Paul Stewart <jpstewart@personalprojects.net>
Reviewed-by: Peter Ross <pross@xvid.org>
Current listed maintainers for vaapi plugin are
not reponsive and/or currently active in the
ffmpeg community. Thus, vaapi plugin patches
(and qsv plugin) have generally gone ignored or
lost in the ether for too long.
Remove Gwenole Beauchesne from vaapi maintainer
who has not been active since 2016.
Current alternative maintainer for vaapi is Mark
Thompson whom has not been active since
March/April 2021.
Therefore, add Haihao Xiang to vaapi maintainer
who's primary role is FFmpeg development with a
focus on the vaapi and qsv plugins. Haihao has
over a decade of media experience and many years
of FFmpeg development experience, amongst other
media frameworks.
The additional patch for adding Haihao as qsv
plugin maintainer has been submitted previously:
https://patchwork.ffmpeg.org/project/ffmpeg/patch/20210608141134.27448-1-zhongli_dev@126.com/
This will help FFmpeg to continue to be the leading
multimedia framework by allowing these plugins to be
actively improved, enhanced, and maintained for existing
and future HW platforms.
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
The SDK defines HEVC, VP9 and AV1 profiles in the same values
e.g.
MFX_PROFILE_HEVC_MAIN =1,
MFX_PROFILE_VP9_0 =1,
MFX_PROFILE_AV1_MAIN =1,
To avoid potential errors when adding VP9, AV1 profiles later,
this patch defines profile array per codec.
Signed-off-by: Zhong Li <zhongli_dev@126.com>
The SDK doesn't support VC1 encoding. In addition, both
MFX_PROFILE_VC1_SIMPLE and MFX_PROFILE_HEVC_MAIN are 1 in the SDK, HEVC
main profile is recognized as simple profile in the verbose output if
don't remove VC1 profiles.
$ ffmpeg -v verbose -qsv_device /dev/dri/renderD129 -f lavfi -i
yuvtestsrc -c:v hevc_qsv -f null -
[hevc_qsv @ 0x55bdf7eb4eb0] profile: simple; level: 21
Signed-off-by: Zhong Li <zhongli_dev@126.com>
Every modification of the data that is copied in update_thread_context()
is a data race if it happens after ff_thread_finish_setup. ffv1dec's
update_thread_context() simply uses memcpy for updating the new context,
so that every modification of the src's context is a race.
Some of these modifications are unnecessary: picture_number is write-only
for the decoder and cur will be reset when decoding the next frame anyway.
So remove them. And while just at it, also don't set cur for the slice
contexts as this variable is write-only.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The previous implementation swapped the two halves of the plaintext. The
existing tests only decrypted data with a plaintext of all zeroes, which is
not affected by swapping the halves. Tests which detect the old buggy behavior
have been added.
Signed-off-by: Sebastian Kirmayer <ffmpeg@kirmayer.eu>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
They test libavfilter internal API, so they should be libavfilter
test programs (which implies: linked statically to libavfilter
to access internal APIs and linked normally (statically or dynamically
depending upon the build configuration) against all the other libs).
Right now, they are always linked statically against all libs,
which is a significant size waste compared to shared libs as all
of libavcodec has been pulled in despite not being really used.
This also leads to linking failures on systems for which av_export_avutil
is intended: libavcodec does not expect to be linked statically
against the library providing avpriv_(cga|vga16)_font in this case.
This is fixed by this commit.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There is no mxfenc dependency any more since commit
b9a26b9d55.
Also remove a dnxhddata.h inclusion in mxfenc that was forgotten
in the very same commit.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
deinterlaces CVPixelBuffers, i.e. AV_PIX_FMT_VIDEOTOOLBOX frames
for example, an interlaced mpeg2 video can be decoded by avcodec,
uploaded into a CVPixelBuffer, deinterlaced by Metal, and then
encoded to h264 by VideoToolbox as follows:
ffmpeg \
-init_hw_device videotoolbox \
-i interlaced.ts \
-vf hwupload,yadif_videotoolbox \
-c:v h264_videotoolbox \
-b:v 2000k \
-c:a copy \
-y progressive.ts
(note that uploading AVFrame into CVPixelBuffer via hwupload
requires 504c60660d)
this work is sponsored by Fancy Bits LLC
Reviewed-by: Ridley Combs <rcombs@rcombs.me>
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Aman Karmani <aman@tmm1.net>
Fixes: signed integer overflow: -16777216 * 131 cannot be represented in type 'int'
Fixes: 23835/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALAC_fuzzer-5669943160078336
Fixes: 41101/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALAC_fuzzer-4636330705944576
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Return error codes when constructing a stream config fails, rather than
just disregarding the failure and continuing.
Propagate the error codes from av_sdp_create().
Fixes: signed integer overflow: 9223372036200463215 + 1109914409 cannot be represented in type 'long'
Fixes: 41480/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-6553086177443840
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: division by zero
Fixes: 42198/clusterfuzz-testcase-minimized-ffmpeg_dem_MV_fuzzer-5054366405492736.fuzz
Fixes: 42222/clusterfuzz-testcase-minimized-ffmpeg_dem_MV_fuzzer-4561249331970048
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Peter Ross <pross@xvid.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Because the s->buffer has been freed by av_freep in avio_closep.
It should not av_freep the buffer in label fail after avio_closep.
Then just move the av_freep before avio_closep and remove the label fail.
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Reviewed-by: Zhao Zhili <zhilizhao@tencent.com>
Signed-off-by: Steven Liu <liuqi05@kuaishou.com>
This is possible by incrementing the counter of allocated rects
directly after said allocation succeeded.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Do this by actually incrementing the counter for the number
of rects at the right time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, various subtitle decoders have not cleaned up
the AVSubtitle on error; this task must not be left to the user
because the documentation explicitly states that the AVSubtitle
"must be freed with avsubtitle_free if *got_sub_ptr is set"
(which it isn't on error).
Leaks happen upon failure in ff_ass_add_rect() or in
ass_decode_frame(); freeing generically also allows to remove
now redundant freeing code in pgssubdec and dvbsubdec.
While just at it, also reset got_sub_ptr generically on error.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The caller of display_end_segment() frees the AVSubtitle on error
in case ENOMEM is returned or err_recognition is set to explode,
so display_end_segment() doesn't have to.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This piece of code has been added as FFmpeg's answer to
infinite loops in try_decode_frame() in commit
6072a19b4f. There is no loop
around try_decode_frame() any more, so this code can be removed.
This code is only triggered in case a) the codec parameter could
not be determined, b) the decode delay could not be guessed or
c) no packet was ever encountered and the encoder has the
AV_CODEC_CAP_CHANNEL_CONF. In these cases the new code will
no longer emit a "decoding for stream %d failed" message, which is
prima facie false. In case a) an additional "Could not find codec
parameters" message is (and will be) emitted. No warning will be
emitted any more in case b) (this happens e.g. with some
h264-conformance FATE-files).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When flushing, try_decode_frame() itself loops until the desired
properties have been found or the decoder is drained.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This puts it directly near the NALU type which is more natural
and furthermore reduces the size of the structure because it
can be placed in padding (on 64-bit systems).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Forgotten in 03b82b3ab9.
(Moving data to the front is only done to make existing
initializations like H2645NAL nal = { NULL } not emit int->pointer
conversion warnings.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes so that fate under 64 bit Windows passes.
These functions replace all ff_hscale8to15_*_ssse3 when avx2 is available.
Signed-off-by: James Almer <jamrial@gmail.com>
This was renamed upstream quite a while ago (v3.112.0). Rename the
option name as well for consistency (and expand the description just
slightly).
Signed-off-by: Niklas Haas <git@haasn.dev>
Support for mapping/unmapping hardware frames has been added into
libplacebo itself, so we can scrap this code in favor of using the new
functions. This has the additional benefit of being forwards-compatible
as support for more complicated frame-related state management is added
to libplacebo (e.g. mapping dolby vision metadata).
It's worth pointing out that, technically, this would also allow
`vf_libplacebo` to accept, practically unmodified, other frame types
(e.g. vaapi or drm), or even software input formats. (Although we still
need a vulkan *device* to be available)
To keep things simple, though, retain the current restriction to vulkan
frames. It's possible we could rethink this in a future commit, but for
now I don't want to introduce any more potentially breaking changes.
LSX and LASX is loongarch SIMD extention.
They are enabled by default if compiler support it, and can be disabled
with '--disable-lsx' '--disable-lasx'.
Change-Id: Ie2608ea61dbd9b7fffadbf0ec2348bad6c124476
Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Reviewed-by: guxiwei <guxiwei-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Because the hls_ts_options will be misunderstand by user,
and then user can use hls_segment_options instead of hls_ts_options.
Signed-off-by: Steven Liu <liuqi05@kuaishou.com>
Because the hls_ts_options will be misunderstand by user that only can
be used in mpegts segments option. So add this option for segments.
Signed-off-by: Steven Liu <liuqi05@kuaishou.com>
ffmal_add_packet() basically duplicated the logic in
av_packet_make_refcounted() with the added twist that it always
created a reference even if one is already available.
This commit stops doing this.
Tested-by: Cameron Gutman <aicommander@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
extra_data and normal packets (from ff_decode_get_packet) processing do
not overlap, thus we can re-use the spare AVPacket to send to
ffmmal_add_packet.
Furthermore, this removes allocation of AVPacket on the stack and stops
using deprecated av_init_packet.
Tested-by: Cameron Gutman <aicommander@gmail.com>
Signed-off-by: Ho Ming Shun <cyph1984@gmail.com>
MMAL is an fundamentally an asynchronous decoder, which was a bad fit
for the legacy dataflow API. Often multiple packets are enqueued before
a flood of frames are returned from MMAL.
The previous lockstep dataflow meant that any delay in returning packets
from the VPU would cause ctx->queue_decoded_frames to grow with no way
of draining the queue.
Testing this with mpv streaming from a live RTSP source visibly reduced
latency introduced by frames waiting in queue_decoded_frames from
roughly 2s to 0.
Tested-by: Cameron Gutman <aicommander@gmail.com>
Signed-off-by: Ho Ming Shun <cyph1984@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Otherwise there is no way to detect an error returned by avio_close() because
ff_format_io_close cannot get the return value.
Checking the return value of the close function is important in order to check
if all data was successfully written and the underlying close() operation was
successful.
It can also be useful even for read mode because it can return any pending
AVIOContext error, so the user don't have to manually check AVIOContext->error.
In order to still support if the user overrides io_close, the generic code only
uses io_close2 if io_close is either NULL or the default io_close callback.
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes crashes if the font name is NULL (which it is if a \fn tag
is not followed by a font name).
Signed-off-by: Charlie Monroe <charlie@charliemonroe.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Improves readability and avoids a redundant index variable
that was mistakenly called "tracksize".
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The current checks just check whether the boxes fit into the remaining
size of the packet instead of whether they actually fit into the box
size. This has been changed; part of this change is to pass the size of
the box (minus the box header) as parameter instead of a pointer to
the AVPacket by which the box parsing function is supposed to
recalculate whether enough data is available.
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The base size of a box refers to the size the box has in a file,
not in memory; so size_t is not their natural type. Therefore use
a plain unsigned which is smaller on 64bit systems and still big
enough to represent any conceivable base size.
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The following command is on how to apply cclock option:
ffmpeg -init_hw_device vulkan -i input.264 -vf \
hwupload=extra_hw_frames=16,transpose_vulkan=dir=cclock,hwdownload,format=yuv420p \
output.264
The following command is on how to apply clock_flip option:
ffmpeg -init_hw_device vulkan -i input.264 -vf \
hwupload=extra_hw_frames=16,transpose_vulkan=dir=clock_flip,hwdownload,format=yuv420p \
output.264
The following command is on how to apply clock option:
ffmpeg -init_hw_device vulkan -i input.264 -vf \
hwupload=extra_hw_frames=16,transpose_vulkan=dir=clock,hwdownload,format=yuv420p \
output.264
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
Always require one semaphore per sw_format plane. This is what
the implementation uses and relies upon throughout. This was
a leftover from an earlier revision that was never needed.
When vulkan image exports to drm, the tilling need to be
VK_IMAGE_TILING_DRM_FORMAT_MODIFIER_EXT. Now add code to create vulkan
image using this format.
Now the following command line works:
ffmpeg -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 -hwaccel_output_format \
vaapi -i input_1080p.264 -vf "hwmap=derive_device=vulkan,format=vulkan, \
scale_vulkan=1920:1080,hwmap=derive_device=vaapi,format=vaapi" -c:v h264_vaapi output.264
Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
Further-modifications-by: Lynne <dev@lynne.ee>
Add support to map vulkan frames to software frames when
using contiguous_planes flag.
Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
Further-modifications-by: Lynne <dev@lynne.ee>
VAAPI on Intel can import external frame, but the planes of the external
frames should be in the same drm object. A new option "contiguous_planes"
is added to device. This flag tells device to allocate places in one
memory. When device is derived from vaapi this flag will be enabled.
A new flag frame_flag is also added to AVVulkanFramesContext. User
can use this flag to force enable or disable this behaviour.
A new variable "offset "is added to AVVKFrame. It describe describe the
offset from the memory currently bound to the VkImage.
Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
Further-modifications-by: Lynne <dev@lynne.ee>
This way we can pass explicit modifiers in. Sometimes the
modifier matters for the number of memory planes that
libva accepts, in particular when dealing with
driver-compressed textures. Furthermore the driver might
not actually be able to determine the implicit modifier
if all the buffer-passing has used explicit modifier.
All these issues should be resolved by passing in the
modifier, and for that we switch to using the PRIME_2
memory type.
Tested with experimental radeonsi patches for modifiers
and kmsgrab. Also tested with radeonsi without the
patches to double-check it works without PRIME_2 support.
v2:
Cache PRIME_2 support to avoid doing two calls every time on
libva drivers that do not support it.
v3:
Remove prime2_vas usage.
Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
The following command is on how to apply transpose_vulkan filter:
ffmpeg -init_hw_device vulkan -i input.264 -vf \
hwupload=extra_hw_frames=16,transpose_vulkan,hwdownload,format=yuv420p output.264
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
If memory allocation fails, ERROR(ENOMEM) '-12' will be returned.
When resample() is done first, the negative size param would cause buffer-overflow and SEGV in swri_rematrix().
When swri_rematrix() is run first, resample() would not cause an error but Err num as a wrong parameter passing.
Err num should be returned immediately. And remove an unneeded term from an assert.
coredump info:
#0 0x499517 in posix_memalign (/home/r1/ffmpeg/ffmpeg_4.4.1+0x499517)
#1 0x6c1f0b4 in av_malloc /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavutil/mem.c:86:9
#2 0x6c208fe in av_mallocz /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavutil/mem.c:239:17
#3 0x6c207ad in av_mallocz_array /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavutil/mem.c:195:12
#4 0x654b2e5 in swri_realloc_audio /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libswresample/swresample.c:418:14
#5 0x654f9a1 in swr_convert_internal /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libswresample/swresample.c:601:17
#6 0x654d2c0 in swr_convert /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libswresample/swresample.c:766:19
#7 0x186cf56 in flush_frame /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/af_aresample.c:251:13
#8 0x186a454 in request_frame /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/af_aresample.c:288:20
#9 0x787d9c in ff_request_frame_to_filter /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/avfilter.c:459:15
#10 0x7877f1 in forward_status_change /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/avfilter.c:1257:19
#11 0x77ed7e in ff_filter_activate_default /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/avfilter.c:1288:20
#12 0x77e4e1 in ff_filter_activate /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/avfilter.c:1441:11
#13 0x793b3f in ff_filter_graph_run_once /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/avfiltergraph.c:1403:12
#14 0x7a7bee in get_frame_internal /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/buffersink.c:131:19
#15 0x7a7287 in av_buffersink_get_frame_flags /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/buffersink.c:142:12
#16 0x792888 in avfilter_graph_request_oldest /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/libavfilter/avfiltergraph.c:1356:17
#17 0x5d07df in transcode_from_filter /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/fftools/ffmpeg.c:4639:11
#18 0x59e557 in transcode_step /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/fftools/ffmpeg.c:4729:20
#19 0x593970 in transcode /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/fftools/ffmpeg.c:4805:15
#20 0x58f7a4 in main /home/r1/ffmpeg/ffmpeg-4.4.1/build/src/fftools/ffmpeg.c:5010:9
#21 0x7f6fd2dee0b2 in __libc_start_main /build/glibc-eX1tMB/glibc-2.31/csu/../csu/libc-start.c:308:16
SUMMARY: AddressSanitizer: negative-size-param (/home/r1/ffmpeg/ffmpeg_4.4.1+0x497e67) in __asan_memcpy
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
There are three types of style entries which are redundant:
a) Entries with length zero. They are already discarded.
b) Entries that are equivalent to the default style:
They can be safely discarded.
c) Entries that are equivalent to the immediately preceding style
if the start of the current style coincides with the end of the
preceding style. In this case the styles can be merged.
This commit implements discarding/merging in cases b) and c).
This fixes ticket #9548. In said ticket each packet contained
exactly one style entry that covered the complete packet with
the exception of the last character (probably created by a tool
that didn't know that the style's end is exclusive). Said style
coincided with the default style, leading to a superfluous reset,
which is now gone.
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Both TextSampleEntry and TextSample can contain StyleRecords;
yet both the code as well as the structures for them were duplicated.
This commit changes this.
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Giving elements of a structure called StyleBox names like
"style_start" or "style_end" is redundant, especially given
that the relevant variables are also called style.
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Don't use different src and dst in av_tea_crypt(); use in-place
modifications instead. Also let av_tea_crypt() encrypt all three
blocks in one call.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
We use ECB, not CBC mode here, so one does not need to reinitialize
the context; for the same reason, one can also just let av_tea_crypt()
loop over the blocks, avoiding a loop here.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, the packets have been read in blocks of at most
eight bytes at a time; then these blocks have been decrypted
and copied into a buffer on the stack (that was double the size
needed...). From there they have been copied to the dst packet.
This commit changes this: The data is read in one go; and
the decryption avoids temporary buffers, too, by making
use of the fact that src and dst of av_tea_crypt() can coincide.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Due to this bush.aa (from the FATE suite) exported garbage metadata
with key "_040930".
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
av_image_copy() expects an array of four pointers according to its
declaration; although it currently only touches pointers that
are actually in use (depending upon the pixel format) this might
change at any time (as has already happened for the linesizes
in d7bc52bf45).
This fixes a -Wstringop-overflow= warning with GCC 11.2.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The concatf protocol returns an opaque error on open if
concatf list file contains trailing newlines.
Signed-off-by: Gyan Doshi <ffmpeg@gyani.pro>
Reviewed-by: James Almer <jamrial@gmail.com>
If the input stream framerate is known, it will be configured on the
relevant filtergraph input and get propagated to the output stream in
the above line. That makes these assignments redundant.
Instead of storing the protocol pointer in the opaque iteration state,
store just the index of the next protocol, similarly to how
ff_urlcontext_child_class_iterate() works.
Silences e.g. the following warning in gcc 10:
src/libavformat/ftp.c: In function ‘ftp_move’:
src/libavformat/ftp.c:1122:46: warning: ‘%s’ directive output may be truncated writing up to 4095 bytes into a region of size 4091 [-Wformat-truncation=]
1122 | snprintf(command, sizeof(command), "RNTO %s\r\n", path);
| ^~ ~~~~
src/libavformat/ftp.c:1122:5: note: ‘snprintf’ output between 8 and 4103 bytes into a destination of size 4096
1122 | snprintf(command, sizeof(command), "RNTO %s\r\n", path);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Silences the following warning with gcc 10:
src/libavdevice/v4l2.c: In function ‘v4l2_get_device_list’:
src/libavdevice/v4l2.c:1042:64: warning: ‘%s’ directive output may be truncated writing up to 255 bytes into a region of size 251 [-Wformat-truncation=]
1042 | ret = snprintf(device_name, sizeof(device_name), "/dev/%s", entry->d_name);
| ^~
src/libavdevice/v4l2.c:1042:15: note: ‘snprintf’ output between 6 and 261 bytes into a destination of size 256
1042 | ret = snprintf(device_name, sizeof(device_name), "/dev/%s", entry->d_name);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Previous patches intending to silence it have proposed increasing the
buffer size, but doing that correctly seems to be tricky. Failing on
truncation is simpler and just as effective (as excessively long device
names are unlikely).
device and cap are local to the loop iteration, there is no need for
them to retain their values. Especially for device it may be dangerous,
since it points to av_malloc'ed data.
The FD opened here is local to the loop iteration, there is no reason to
store it in the context. Since read_header() may have already been
called, this may ovewrite an existing valid FD.
Maximum output size with a 32-bit int is 17 bytes, or 26 with a 64-bit
int.
Silences the following gcc 10 warning:
src/libavdevice/jack.c: In function ‘audio_read_header’:
src/libavdevice/jack.c:171:45: warning: ‘snprintf’ output may be truncated before the last format character [-Wformat-truncation=]
171 | snprintf(str, sizeof(str), "input_%d", i + 1);
| ^
src/libavdevice/jack.c:171:9: note: ‘snprintf’ output between 8 and 17 bytes into a destination of size 16
171 | snprintf(str, sizeof(str), "input_%d", i + 1);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Do this by switching from the dynamic buffer API to the AVBPrint API;
the former has no defined way to check for errors.
This also avoids allocating an AVIOContext.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Do this by switching from the dynamic buffer API to the AVBPrint API;
the former has no defined way to check for errors.
This also avoids allocating an AVIOContext.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
av_image_copy() expects an array of four pointers and linesizes
according to its declaration; it currently only pointers that are
actually in use (depending upon the pixel format), but this might
change at any time. It has already happened for the linesizes in
d7bc52bf45 and so increasing their
array fixes a stack-buffer overread.
This fixes a -Wstringop-overflow= and -Wstringop-overread warning
from GCC 11.2.
Reviewed-by: Linjie Fu <linjie.justin.fu@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This reduces codesize because the offsets of commonly used elements
are now smaller and thus need less bytes to encode in ptr+offset
addressing modes (with GCC 11.2 on x64: 0x1b8b -> 0x1a7b).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently, it also tests whether extended_data points to something
different than the AVFrame's data array and frees extended_data
if it is different. Yet this is only necessary for one of its three
callers, namely av_frame_unref(); meanwhile the other two callers
took measures to avoid this (or rather, to make it to an av_free(NULL)).
This commit moves this chunk to av_frame_unref() (so that
get_frame_defaults() now treats its input as uninitialized)
and removes the now superfluous code in the other two callers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The only caller of do_video_out() doesn't need the frame afterwards,
ergo one can replace an av_frame_ref() by av_frame_move_ref().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently, adding a (separately allocated) element to a list of pointers
works by first reallocating the array of pointers and (on success)
incrementing its size and only then allocating the new element.
If the latter allocation fails, the size is inconsistent, i.e.
array[nb_array_elems - 1] is NULL. Our cleanup code crashes in such
scenarios.
Fix this by adding an auxiliary function that atomically allocates
and adds a new element to a list of pointers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The data stored in data[3] in VAAPI AVFrame is VASurfaceID while
the data stored in pair->first is the pointer of VASurfaceID, so
we need to do cast to make following commandline works:
ffmpeg -hwaccel vaapi -hwaccel_device /dev/dri/renderD128 \
-hwaccel_output_format vaapi -i input.264 \
-vf "hwmap=derive_device=qsv,format=qsv" -c:v h264_qsv output.264
Signed-off-by: nyanmisaka <nst799610810@gmail.com>
Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
The documentation states that here 0 should be used for read-only and
1 for a writable buffer. AVIO_FLAG_WRITE however is 2, while it works
due to the way the flag is handled internally, it is still wrong
according to the documentation.
Additionally it makes it seem as if the AVIO_FLAG_* values could be used
here, which is actually not true, as when AVIO_FLAG_READ would be used
here it would create a writable buffer as AVIO_FLAG_READ is defined as 1.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Old system is:
OSX version: 10.11.6
Apple LLVM version 8.0.0 (clang-800.0.42.1)
Target: x86_64-apple-darwin15.6.0
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
this allows getting rid of the hardcoded max size of SDP.
Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
The length of this list is a compile-time constant, so there is
no need to calculate it again at runtime.
(This also avoids an implicit requirement of -1 == AV_SAMPLE_FMT_NONE.)
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
An AVBPrint's internal string is always already zero-terminated;
writing another '\0' is unnecessary as long as one treats
the string only as a C-string.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
choose_pix_fmts() used the dynamic buffer API to write strings;
as is common among uses of this API, only opening the dynamic buffer
was checked, but not the end result, leading to crashes in case
of allocation failure.
Furthermore, some static strings were duplicated; the allocations
performed here were not properly checked: Allocation failure would
be treated as "could not determine pixel format".
The first issue is fixed by switching to the AVBPrint API which allows
to easily perform checks at the end. Furthermore, the internal buffer
avoids almost all allocations in case the AVBPrint is used.
The AVBPrint also allows to solve the second issue in an elegant way,
because it allows to return the static strings directly.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is not really natural, it requires internal allocations
of its own and its error handling is horrible (i.e. the implicit
(re)allocations here are unchecked).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, the h264_redundant_pps_bsf stored the initial value
of pic_init_qp_minus26 of the most recently encountered PPS;
it also modified the slices based upon to assumption that
the most recent PPS is the PPS the slice belongs to.
Yet this assumption is flawed, as there can be several PPS
with different IDs that are visible at any given time.
If these have different pic_init_qp_minus26 values,
the output can be invalid.
Fix this by directly using the pic_init_qp_minus26 value of
the input PPS.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
extradata_pic_init_qp is unset since
fa75e43875
(and resetting current_pic_init_qp to the value it had in extradata
never made much sense).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is not documented that freeing the last (and only) entry of
an AVDictionary frees the dictionary.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
SWR_CH_MAX is internal only and the arrays are therefore not required
to have that many elements (and they typically don't do it). So remove
this potentially confusing hint.
(Newer versions of GCC emit -Warray-parameter= warnings for this,
because the definition with explicit size differs from the declaration
(which leaves the size unspecified); this is IMO a false-positive,
because definition and declaration didn't conflict, but anyway it is
fixed by this commit.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The number of audio channels is stored after the magic number
identifying the audio format. Prior to this patch the code has been
reading it earlier, causing files with only one audio channel to be
handled incorrectly.
Reviewed-by: Peter Ross <pross@xvid.org>
The mpeg4 encoder is slice-threaded and its output depends upon
the number of threads used. Therefore all tests of this encoder
use a hardcoded number of threads (ENC_OPTS in fate-run.sh contains
"-threads 1"; only the vsynth%-mpeg4-thread tests override this
for the mpeg4 encoder, but they also use a hardcoded value to
be consistent across different systems); only the new shortest
and copy-shortest[12] (implicitly due to the sample used) tests
don't and this leads to FATE-failures.
Fix this by explicitly setting the thread count.
Also switch the shortest test to framecrc, because hashing side data
is itchy even though the side data used here (AV_PKT_DATA_QUALITY_STATS)
has a defined endianness.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This filter flips the input video both horizontally and vertically
in one compute pipeline, and it's no need to use two pipelines for
hflip_vulkan,vflip_vulkan anymore.
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
The format of a mov_text (3GPP Timed Text) sample is:
uint16_t text_length;
uint8_t text[text_length];
TextSampleModifierBox text_modifier;
Yet in case our encoder receives an AVSubtitle with multiple
ASS AVSubtitleRects, it creates something like this:
uint16_t text_length;
uint8_t text[text_length_1];
TextSampleModifierBox text_modifier_1;
uint8_t text[text_length_2];
TextSampleModifierBox text_modifier_2;
...
where text_length is the sum of all the text_length_*.
This commit fixes this by writing the TextSampleModifierBoxes only
after all the rects have been written.
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This avoids abusing a variable called length for the return value
and ensures that the AVBPrint is always reset before using it;
previously this has been forgotten in some error paths.
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Don't mark all streams as finished, instead make sync_opts keep track of the
stream's duration, and set recording_time to it, same as in transcoding paths.
Fixes tickets #9512 and #9513.
Signed-off-by: James Almer <jamrial@gmail.com>
Adds demuxer for Square Enux SCD files.
Based off [1] and personal investigation.
This has only been tested against Drakengard 3 (PS3) *_SCD.XXX files
(big-endian). As it is highly likely that FFXIV (PC) files are little-endian,
this demuxer is marked as experimental until this can be confirmed.
[1]: http://ffxivexplorer.fragmenterworks.com/research/scd%20files.txt
Reviewed-by: Peter Ross <pross@xvid.org>
Reviewed-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Signed-off-by: Zane van Iperen <zane@zanevaniperen.com>
Currently they are ordered as-written (i.e. by increasing position);
in case av_interleaved_write_frame() is used, this is (mostly)
the same as ordered by increasing dts.
Yet the Matroska specification strongly recommends (SHOULD) that
the CuePoints be sorted by CueTime. mkvalidator warns when they are
not. Therefore this commit sorts them accordingly.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Freeing the new H.264 specific fields has been forgotten.
(This leak only appears in case the encoder has not been completely
drained.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, it has only been freed when the QSVFrame is reused,
so that the last one contained in it leaked at the end.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Since commit 3bbe0c210b, the Payloads
array of every QSVFrame leaks as soon as the frame is reused;
the leak is small and not very noticeable, but if there is an attempt
to use said array the ensuing crash is much more noticeable.
This happens when encoding H.264 with A53 CC side data.
Furthermore, if said array can not be allocated at all, an AVFrame
leaks.
Fix all of this by not allocating the array separately at all; put it
in QSVFrame instead and restore the Payloads array upon reusing the
frame.
Finally, use av_freep() instead of av_free() to free the payload
entries.
Reviewed-by: Xiang, Haihao <haihao.xiang@intel.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
If fifo_thread_recover() succeeds immediately after
fifo_thread_dispatch_message() fails, the dts of the packet is scaled
twice, causing cur_dts to be abnormally large and "Application provided
invalid, non monotonically increasing dts to muxer in stream" to occur
repeatedly.
Steps to reproduce:
1. ffmpeg -f lavfi -i testsrc -c:v libx264 -map 0:v -flags +global_header -f fifo -fifo_format flv -attempt_recovery 1 -recover_any_error 1 rtmp://example.com/livekey
2. set a breakpoint on fifo_thread_recover
3. force disconnect from the rtmp server
4. wait for break
5. reconnect to the rtmp server
6. resume execution of ffmpeg
Signed-off-by: Ryoji Gyoda <gy.cft4@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
FLV AMF tags have a 24bit field for timestamps plus an 8bit for extended
timestamps.
All FLV AMF tags except when we write metadata handle this correctly
using the put_timestamp function.
Until now when writing metadata we were only using the first
24 bits and thus the timestamp value was wraping around 4 hours 40
minutes (16,800,000 ms, max 24 bit value 16,777,216) of playback.
This commit fixes this applying this same function put_timestamp
for the metadata FLV tag.
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes: signed integer overflow: -776522110086937600 * 16 cannot be represented in type 'long'
Fixes: 40563/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-6644829447127040
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array access
Fixes: Timeout
Fixes: 40481/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VQA_fuzzer-6502647583080448
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This was almost completely redundant. The only functionality that's no longer
available after this removal is the videotoolbox_pixfmt arg, which has been
obsolete for several years.
In order to know that the earlier code did not use uninitialized
values one needs to know that the lowest four bits of each used
value of pframe_block4x4_coefficients_tab do not vanish identically.
E.g. Coverity did not get this and warned about it in ticket #1466632.
Fix this by slightly rewriting the code.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Due to reasons, mpv doesn't pass filename when probing. mpv also sets
default probescore threshold to 26. Since the current jpeg_probe
implementation returns 25 until EOI, it means that the whole image needs
to be probed to succeed. Worse, the whole image is not passed at once;
increasingly large buffers are tried before that. Adding it up together,
if many demuxers are enabled, moderately large JPEG files (few MB) can
take several seconds to open, despite taking less than 1 second to
actually decode.
Therefore, adjust the heuristic to be more optimistic if proper JFIF or
Exif segments are found. While not strictly required, the vast majority
of JPEG-ish files have one or the other or both.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The check for m->size >= 0xF000 is intended to avoid skipping too much
garbage data between JPEG frames in test_roman (thus missing next SOI),
but it erroneously also skips valid markers between SOI and SOS. Instead
of this, we should simply skip parsing markers other than SOI after EOI.
That way, we will not accidentally skip over SOI due to some garbage
between frames. There is still a small risk of encountering FFD8 in the
garbage data, but the chance of this is fairly low.
Fixes: https://trac.ffmpeg.org/ticket/8967
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
For most check_bitstream() functions this just avoids having
to dereference s->streams[pkt->stream_index] themselves; but for
meta-muxers it will allow to forward the packet to stream with
a different stream_index (belonging to a different AVFormatContext)
without using a spare packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The types used by the AVFifo API are inconsistent:
av_fifo_(space|size)() returns an int; av_fifo_alloc() takes an
unsigned, other parts use size_t. This commit therefore ensures
that the size of the muxing_queue FIFO never exceeds INT_MAX.
While just at it, also make sure not to call av_fifo_size()
unnecessarily often.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This reverts commit 5258f64a14.
The premise of said commit (that conversions from pointer to int
are ok) is wrong: C99/C11 6.3.2.3 5: "Any pointer type may be converted
to an integer type. [...] If the result cannot be represented in the
integer type, the behavior is undefined." (C90 6.3.4 contains a similar
restriction.) So don't disable -Wpointer-to-int-cast.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
h263_get_motion_length() forgot to take an absolute value;
as a consequence, a negative index was used to access an array.
This leads to potential crashes, but mostly it just accesses what
is to the left of ff_mvtab (unless one uses ASAN), thereby defeating
the purpose of the AV_CODEC_FLAG2_NO_OUTPUT because the sizes of
the returned packets differ from the sizes the encoder would actually
have produced.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Validation layer is an indispensable part of developing on Vulkan.
The following commands is on how to enable validation layers:
ffmpeg -init_hw_device vulkan=0,debug=1,validation_layers=VK_LAYER_LUNARG_monitor+VK_LAYER_LUNARG_api_dump
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
path1 and sanitized_path are both MAX_URL_SIZE bytes long, yet the latter is
copied from the former with the addition of one extra character.
Should fix a -Wformat-truncation warning.
Signed-off-by: James Almer <jamrial@gmail.com>
As per 8.6.1.2.2 of ISO/IEC 14496-12:2015(E), STTS sample offsets
are to be always stored as uint32_t. So far, they have been signed ints
which led to desync in files with very large offsets.
The MOVStts struct was used to store CTTS offsets as well. These can be
negative in version 1. So a new struct MOVCtts was created and all
declarations for CTTS usage changed to MOVCtts.
C99/C11 6.3.2.3 5: "Any pointer type may be converted to an integer
type. [...] If the result cannot be represented in the integer type,
the behavior is undefined." So stop casting pointers to int; use
uintptr_t instead.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This context is transient, so putting it on the stack is more natural.
Also reduces codesize: 24E6->2296 B with GCC 10 and -O3.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
"All commands that are allowed on a queue that supports transfer
operations are also allowed on a queue that supports either
graphics or compute operations. Thus, if the capabilities of a
queue family include VK_QUEUE_GRAPHICS_BIT or VK_QUEUE_COMPUTE_BIT,
then reporting the VK_QUEUE_TRANSFER_BIT capability separately for
that queue family is optional."
What happens on startup is that ffmpeg.c initializes the filter,
then frees it without feeding a single frame through. With no
input frame, the filter lacks a hardware device. The rest of the
uninit code checks if Vulkan objects exist, which they must if there's
a hardware device, but vk->DeviceWaitIdle does not require an object.
So, add a check for it.
send_frame_to_filters() sends a frame to all the filters that
need said frame; for every filter except the last one this involves
creating a reference to the frame, because
av_buffersrc_add_frame_flags() by default takes ownership of
the supplied references. Yet said function has a flag which
changes its behaviour to create a reference itself.
This commit uses this flag and stops creating the references itself;
this allows to remove the spare AVFrame holding the temporary
references; it also avoids unreferencing said frame.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The actual frame_size is no longer used since commit
3d38e45eb85c7a2420cb48a9cd45625c28644b2e; and the check for
"< 0" is equivalent to the CID being valid. But this is already
checked by avpriv_dnxhd_get_interlaced() (and is actually already
ensured by mxf_dnxhd_codec_uls containing this CID).
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is just a flag per supported CID. So there is no reason to use
an avpriv function for this purpose.
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
As well as the custom get_buffer2() implementation which would become a
redundant wrapper for avcodec_default_get_buffer2() after this
Signed-off-by: James Almer <jamrial@gmail.com>
It avoids the overhead of the packet list; furthermore, using
ff_interleave_packet_per_dts() is wrong for the null muxer anyway,
because said muxer accepts packets without timestamps, which
ff_interleave_packet_per_dts() can't handle.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It avoids branches lateron and will allow to easily avoid the overhead
of the linked list currently in use in case there is only one stream.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The following command is on how to apply vflip_vulkan filter:
ffmpeg -init_hw_device vulkan -i input.264 -vf hwupload=extra_hw_frames=16,vflip_vulkan,hwdownload,format=yuv420p output.264
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
The following command is on how to apply hflip_vulkan filter:
ffmpeg -init_hw_device vulkan -i input.264 -vf hwupload=extra_hw_frames=16,hflip_vulkan,hwdownload,format=yuv420p output.264
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
It's got a much better API that's actually maintained, it eliminates
race conditions, it comes with a pkg-config file by default, and
unfortunately isn't currently packaged by Debian or other large
distributions.
The issue is that libavfilter depends on libavcodec, and when doing a
static build, if libavcodec also includes "libavfilter/vulkan.c", then
during link-time, compiling programs will fail as there would be multiple
definitions of the same symbols in both libavfilter and libavcodec's
object files.
Linkers are, however, more permitting if both files that include
a common file that's used as a template are one-to-one identical.
Hence, to make both files the same in the future, export all avfilter
specific functions to a separate file.
There is some work in progress to make templated files like this be
compiled only once, so this is not a long-term solution.
This also removes a macro that could be used to toggle SPIRV compilation
capability on #include-time, as this could cause the files to be different.
It has already been checked immediately before that said
AVDictionaryEntry exists; checking again is redundant.
Furthermore, av_hwdevice_find_type_by_name() requires its argument
to be non-NULL, so adding a codepath that automatically calls it
with that parameter is nonsense. The same goes for the argument
corresponding to %s.
Fixes Coverity issue 1491394.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This av_buffer_create() does nothing but leak an AVBuffer and an
AVBufferRef (except on allocation error).
Fixes Coverity issue 1491393.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Treat values returned from av_dict_get() as const, since they are
internal to AVDictionary.
Signed-off-by: Chad Fraleigh <chadf@triularity.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Treat values returned from av_dict_get() as const, since they are
internal to AVDictionary.
Signed-off-by: Chad Fraleigh <chadf@triularity.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Besides being unused it should not be used at all:
The order of options of bitstream filters is not guaranteed
to be stable at all.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The MXF muxers only write the header after they have received
a packet; the actual write_header function does not write anything.
So make an init function out of it.
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
None of the muxers here has the AVFMT_NOSTREAMS flag set,
so it is checked generically that there are streams.
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This is especially important in case avio_write() would be switched
to an unsigned type like size_t, then a potential error from avio_read()
(with negative return value) would no longer be handled gracefully by
avio_write().
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The memset here is both unnecessary (avio_read() ignores the previous
content of the destination buffer) as well as nonsense (for a char
buf[BUFSIZE] sizeof(buf) and sizeof(BUFSIZE) are not the same; the
latter is sizeof(int)).
Fixes Coverity issue #1465863.
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In 45bfe8b838, short_seek_threshold was removed
from the public AVIO struct. Although this option was private and not intended
to be used by public API users, it was nonetheless, because it provided functionality
that could otherwise not be gained via public API.
This was especially important for networked I/O like HTTP, where the internal
size for lavf could be way to small depending on the specifics of a user's
usecase, such as reading interlavd media files from cloud storage.
Add an AVOption to make this functionality accessible to the HTTP client.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Whether failed or not, the block of codes labeled with fail should
be always run to ensure the av_free(kernel_def) is executed.
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
Add Branch Target Identifiers (BTIs) to all functions defined in
AArch64 assembly files. Most of the BTI landing pads are added
automatically by the 'function' macro.
BTI support is turned on or off at compile time based on the presence
of the __ARM_FEATURE_BTI_DEFAULT feature macro.
A binary compiled with BTI support can be executed on an Armv8-A
processor without BTI support because the instructions are defined in
NOP space.
Signed-off-by: Jonathan Wright <jonathan.wright@arm.com>
Signed-off-by: Elijah Ahmad <elijah.ahmad@arm.com>
Signed-off-by: Martin Storsjö <martin@martin.st>
Change AArch64 assembly code to use:
ret x<n>
instead of:
br x<n>
"ret x<n>" is already used in a lot of places so this patch makes it
consistent across the code base. This does not change behavior or
performance.
In addition, this change reduces the number of landing pads needed in
a subsequent patch to support the Armv8.5-A Branch Target
Identification (BTI) security feature.
Signed-off-by: Jonathan Wright <jonathan.wright@arm.com>
Signed-off-by: Martin Storsjö <martin@martin.st>
Currently, the code doing this is spread over several places and may
behave in unexpected ways. E.g. automatic 'default' marking is only done
for streams fed by complex filtergraphs. It is also applied in the order
in which the output streams are initialized, which is effectively
random.
Move processing the dispositions at the end of open_output_file(), when
we already have all the necessary information.
Apply the automatic default marking only if no explicit -disposition
options were supplied by the user, and apply it to the first stream of
each type (excluding attached pics) when there is more than one stream
of that type and no default markings were copied from the input streams.
Explicitly document the new behavior.
Changes the results of some tests, where the output file gets a default
disposition, while it previously did not.
This commit adds a powerful and customizable gblur Vulkan filter,
which provides a maximum 127x127 kernel size of Gaussian Filter.
The size could be adjusted by requirements on quality or performance.
The following command is on how to apply gblur_vulkan filter:
ffmpeg -init_hw_device vulkan -i input.264 -vf hwupload=extra_hw_frames=16,gblur_vulkan,hwdownload,format=yuv420p output.264
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
Changes av_clipf to return amin if a is nan.
Before if a is nan av_clipf_c returned nan and
av_clipf_sse would return amax. Now the both
should behave the same.
This works because nan > amin is false.
The max(nan, amin) will be amin.
Signed-off-by: James Almer <jamrial@gmail.com>
There is no reason to wrap them in #ifndef guards, they should only be
defined here and nowhere else. The define guards just add the
possibility to accidentally use the same FF_API name in different
libraries.
NEEDS_UNSAFE has the same value as NEEDS_FILE,
causing "duration not allowed if safe" error
while duration directive doesn't require unsafe mode.
Signed-off-by: Googleplex <yyoung2001@gmail.com>
Previously this was hardcoded to 2500000 bytes, so probing of the stream codecs
was always limited by this, and not probesize.
Also keep track of the actual size of packets in raw_packet_buffer and not the
remaining size for simplicity.
Fixes ticket #5860.
Signed-off-by: Marton Balint <cus@passwd.hu>
Unfortunately pad_len and pad_dur behaviour was different if 0 was specified,
pad_dur handled 0 duration as infinity, for pad_len, infinity was -1.
Let's make the behaviour consistent by handling 0 duration for pad_dur and
whole_dur as indeed 0 duration. This somewhat changes the behaviour of the
filter if 0 was explicitly specified, but deprecating the old option and adding
a new for the corrected behaviour seemed a bit overkill. So let's document the
change instead.
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes: signed integer overflow: 2105344 * 539033345 cannot be represented in type 'int'
Fixes: out of array write
Fixes: 39956/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SPEEX_fuzzer-4766419250708480
Fixes: 40293/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SPEEX_fuzzer-5219910217760768
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is ment to be a cosmetic change
old timings:
42780 UNITS in grayf32le, 1 runs, 0 skips
56720 UNITS in grayf32le, 2 runs, 0 skips
67265 UNITS in grayf32le, 4 runs, 0 skips
58082 UNITS in grayf32le, 8 runs, 0 skips
63512 UNITS in grayf32le, 16 runs, 0 skips
52720 UNITS in grayf32le, 32 runs, 0 skips
46491 UNITS in grayf32le, 64 runs, 0 skips
68500 UNITS in grayf32be, 1 runs, 0 skips
66930 UNITS in grayf32be, 2 runs, 0 skips
62305 UNITS in grayf32be, 4 runs, 0 skips
55510 UNITS in grayf32be, 8 runs, 0 skips
50216 UNITS in grayf32be, 16 runs, 0 skips
44480 UNITS in grayf32be, 32 runs, 0 skips
42394 UNITS in grayf32be, 64 runs, 0 skips
new timings:
46660 UNITS in grayf32le, 1 runs, 0 skips
51830 UNITS in grayf32le, 2 runs, 0 skips
53390 UNITS in grayf32le, 4 runs, 0 skips
50910 UNITS in grayf32le, 8 runs, 0 skips
44968 UNITS in grayf32le, 16 runs, 0 skips
40349 UNITS in grayf32le, 32 runs, 0 skips
38330 UNITS in grayf32le, 64 runs, 0 skips
39980 UNITS in grayf32be, 1 runs, 0 skips
49630 UNITS in grayf32be, 2 runs, 0 skips
53540 UNITS in grayf32be, 4 runs, 0 skips
59767 UNITS in grayf32be, 8 runs, 0 skips
51206 UNITS in grayf32be, 16 runs, 0 skips
44743 UNITS in grayf32be, 32 runs, 0 skips
41468 UNITS in grayf32be, 64 runs, 0 skips
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Since texinfo 6.8, there's no longer an INLINE_CONTENTS variable.
makeinfo: warning: set_from_init_file: unknown variable INLINE_CONTENTS
texinfo commit 62a6adfb33b006e187483779974bbd45f0f782b1 replaced
INLINE_CONTENTS with OUTPUT_CONTENTS_LOCATION.
texinfo commit 41f8ed4eb42bf6daa7df7007afd946875597452d replaced
OUTPUT_CONTENTS_LOCATION with CONTENTS_OUTPUT_LOCATION.
With texinfo 6.8 and above, the same as INLINE_CONTENTS=1 could be
achieved by CONTENTS_OUTPUT_LOCATION=inline.
https://www.gnu.org/software/texinfo/manual/texinfo/html_node/HTML-Customization-Variables.html
Since texinfo commit 6a5ceab6a48a4f052baad9f3474d741428409fd7, the
formatting functions, in particular begin_file, program_string and
end_file, are prefixed with format_, i.e. format_begin_file, etc.
This patch fixes building the documentation when texinfo 6.8, or
above, is used:
Unknown formatting type begin_file
at /usr/bin/makeinfo line 415.
Unknown formatting type program_string
at /usr/bin/makeinfo line 415.
Unknown formatting type end_file
at /usr/bin/makeinfo line 415.
This makes output consistent with a similar warning just few
lines above where this flag is checked in the same way.
Signed-off-by: softworkz <softworkz@hotmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
When viewing logs, it's sometimes useful to be able to see whether
execution was ended via q command.
Signed-off-by: softworkz <softworkz@hotmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Before:
overlay AVOptions:
x <string> ..FV....... set the x expression (default "0")
y <string> ..FV....... set the y expression (default "0")
eof_action <int> ..FV....... Action to take when encountering EOF from secondary input (from 0 to 2) (default repeat)
repeat 0 ..FV....... Repeat the previous frame.
endall 1 ..FV....... End both streams.
pass 2 ..FV....... Pass through the main input.
eval <int> ..FV....... specify when to evaluate expressions (from 0 to 1) (default frame)
After:
a
overlay AVOptions:
x <string> ..FV....... set the x expression (default "0")
y <string> ..FV....... set the y expression (default "0")
eof_action <int> ..FV....... Action to take when encountering EOF from secondary input (from 0 to 2) (default repeat)
repeat 0 ..FV....... Repeat the previous frame.
endall 1 ..FV....... End both streams.
pass 2 ..FV....... Pass through the main input.
eval <int> ..FV....... specify when to evaluate expressions (from 0 to 1) (default frame)
Signed-off-by: softworkz <softworkz@hotmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
In particular, allows users to go all the way up to PL_LOG_TRACE if
desired. (While also avoiding some potentially unnecessary callbacks for
filtered messages, including e.g. the CPU cost of printing out shader
sources)
Response to runtime log level changes by updating it once per
filter_frame(), which should hopefully be often enough.
This filter conceptually maps the libplacebo `pl_renderer` API into
libavfilter, which is a high-level image rendering API designed to work
with an RGB pipeline internally. As such, there's no way to avoid e.g.
chroma interpolation with this filter, although new versions of
libplacebo support outputting back to subsampled YCbCr after processing
is done.
That being said, `pl_renderer` supports automatic integration of the
majority of libplacebo's shaders, ranging from debanding to tone
mapping, and also supports loading custom mpv-style user shaders, making
this API a natural candidate for getting a lot of functionality out of
relatively little code.
In the future, I may approach this problem either by rewriting this
filter to also support a non-renderer codepath, or by upgrading
libplacebo's renderer to support a full YCbCr pipeline.
This unfortunately requires a very new version of libplacebo (unreleased
at time of writing) for timeline semaphore support. But the amount of
boilerplate needed to hack in backwards compatibility would have been
very unreasonable.
Include windows.h to fix it. Normally, it'd be better to include it in
vulkan_functions.h, but I'm reasonably confident nothing else that uses
the Vulkan code will need to include Windows functions and not windows.h.
Finally, this is as close to usable as it gets for glslang.
Much faster to compile as well, and eliminates the need for a C++
compiler, which is great.
Also, changes to the resource limits won't break users, as we
can use designated initializers in C90.
This simplifies and makes queue family picking simpler and more robust.
The requirements on the device context are relaxed. They made no sense
in the first place.
The video encode/decode extension is still in beta, at least on paper,
but I really doubt they'd change needing a separate queue family.
Make get_int/set_int symetric. The int64_t to double to int64_t
conversion is unprecise for large value.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
x264.h: "the payloads of all output NALs are guaranteed to be
sequential in memory." Therefore we can omit the loop.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In function vtenc_populate_extradata(), there is a manually created
pixel buffer that has not been released. So we should use CVPixelBufferRelease
to release this pixel buffer at the end, otherwise will cause a memory leak.
Signed-off-by: Rick Kern <kernrj@gmail.com>
Decoders implementing the receive_frame API currently mostly use
stack packets to temporarily hold the packet they receive from
ff_decode_get_packet(). This role directly parallels the role of
in_pkt, the spare packet used in decode_simple_internal for the
decoders implementing the traditional decoding API. Said packet
is unused by the generic code for the decoders implementing the
receive_frame API, so allow them to use it to fulfill the function
it already fulfills for the traditional API for both APIs.
There is only one caveat in this: The packet is automatically
unreferenced in avcodec_flush_buffers(). But this is actually
positive as it means the decoders don't have to do this themselves
(in case the packet is preserved between receive_frame calls).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
From ETSI EN 300 472 V1.3.1 (2003-05) Specification for conveying ITU-R System
B Teletext in DVB bitstreams:
4.1 Transport Stream (TS) packet format
The standard TS packet syntax and semantics are followed, noting the following
constraint:
- adaptation_field_control only the values "01" and "10" are permitted.
Some set top boxes (Motorola, Arris, Zyxel) refuse non-conforming packets.
Signed-off-by: Alex Shumsky <alexthreed@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Commit 6f36eb0da7 claim it fixes endless loop on
package generation if muxrate specified and copyts used. But actually it does
not work properly if -mpegts_copyts 1 is specified:
ffmpeg -y -copyts -i loewe.ts -c:v libx264 -x264opts nal-hrd=cbr:force-cfr=1 -b:v 3500k -minrate 3500k -maxrate 3500k -bufsize 1000k -c:a mp2 -f mpegts -mpegts_copyts 1 -muxrate 4500k -vframes 1000 test.ts
ffmpeg generate huge file until it reach zero-based pcr value equal to first dts.
Attached patch fixes it.
Reviewed-by: Limin Wang <lance.lmwang@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Try to make the feature easier to use, especially since the user
have enabled -strict experimental manually. The user shouldn't
be surprised that hls_playlist is enabled for lhls automatically,
so change the log level from warning to info for that.
There is a little chance that user specified contradicted options
like -streaming 0 -ldash 1, however, it's more likely that user
didn't know or forgot to enable streaming for ldash. So enabling
streaming automatically to make the feature easier to use, similar
like enable FF_MOV_FLAG_FRAGMENT/EMPTY_MOOV/DEFAULT_BASE_MOOF and
so on for FF_MOV_FLAG_CMAF.
Fixes -Wstringop-overflow warnings with libaom >= 2.0.0, where the unused alpha
plane was removed from aom_image.
Reviewed-by: James Zern <jzern@google.com>
Signed-off-by: James Almer <jamrial@gmail.com>
This fixes linking errors where variables cannot be correctly linked in from an external shared library such as with msvc (requires dllimport which is not used by libaom). Instead just call the function that returns the same variable.
Signed-off-by: Matt Oliver <protogonoi@gmail.com>
OpenBSD only supports riscv64 but this is an attempt at adding
some of the initial bits for RISC-V support.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If original packet is corrupted, then parsed packet is probably corrupted too.
Let the application decide what to do.
Signed-off-by: Alex Shumsky <alexthreed@gmail.com>
Prevents desktop stutters caused by the change (specifically on KDE).
We're not a game, we don't actually need it disabled.
Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Zane van Iperen <zane@zanevaniperen.com>
The packets delivered to this decoder are often decoded to more than
one frame and if the internal buffer packet is not unreferenced,
the decoder will still output frames derived from the old packet (from
before the flush).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This decoder may output multiple AVFrames for every AVPacket
passed to it, but after it has returned AVERROR(EAGAIN),
it is completely drained and there is no reason to flush it
at the end with a NULL packet. Furthermore, there is also no
delay in the common sense of the word.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Such fields can be seen as generally useful in cases where the
API user is not implementing custom AVIO callbacks, but still would
like to know if data is being read or written out, such as in case
data is being read from input but no AVPacket has been received yet.
Originally added as a private entry in commit
3f75e5116b, but its grouping with
the comment noting its private state was missed during merging of
the field from Libav (most likely due to an already existing field
in between).
Looking at 3f75e5116b, the field
was supposed to be private, but during merging the field and the
group that had the comment about it got separated.
Thus, move the actual privately utilized state of this variable
into the private FFIOContext. Additionally, name the private field
somewhat better, so that it does not get confused with the amount
of bytes written out.
bit_rate is not a critical field, and we shouln't hard fail if we
can't caluclate it due to a large timebase - it needlessly breaks
valid files.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Fixes: signed integer overflow: 822841647 + 1647055738 cannot be represented in type 'int'
Fixes: 39935/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TTA_fuzzer-4592657142251520
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
summary for the adjustments:
1, remove the extra "," in the ,}
...{0.2004,0.3001,0.4008,0.5005,0.6002,0.7009,0.8006,0.9013,}
to
...{0.2004,0.3001,0.4008,0.5005,0.6002,0.7009,0.8006,0.9013}
2, add "," between the } and new field
} fraction_bright_pixels
to
}, fraction_bright_pixels
3, remove the extra space between "} }"
...{0.2004,0.3001,0.4008,0.5005,0.6002,0.7009,0.8006,0.9013,} }
to
...{0.2004,0.3001,0.4008,0.5005,0.6002,0.7009,0.8006,0.9013,}}
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
GB/T 17975.1
Information technology-Generic coding of moving pictures and associated audio
information-Part 1:Systems
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
GB/T 17975.1
Information technology-Generic coding of moving pictures and associated audio
information-Part 1:Systems
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
Reduces the risk of finding false frames that happens to have valid values and CRC.
Fixes ticket #9185 ffmpeg flac decoder incorrectly finds junk frame
https://trac.ffmpeg.org/ticket/9185
If a decoding error happens before frame side data is allocated, this assert may be
triggered. And since applying film grain is not enforced (we just warn it wasn't
applied and move on), we can just do that in such scenarios.
Fixes: Assertion failure
Fixes: clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_H264_fuzzer-5528650032742400
Signed-off-by: James Almer <jamrial@gmail.com>
MISB ST 0604 and ST 2101 require user data unregistered SEI messages
(precision timestamps and sensor identifiers) to be included. That
currently isn't supported for libx264. This patch adds support
for user data unregistered SEI messages in accordance with ISO/IEC
14496-10:2020(E) section D.1.7 (syntax) and D.2.7 (semantics).
This code is based on a similar change for libx265 (commit
1f58503013).
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Passing an uninitialized variable as argument to a function is
undefined behaviour (UB). The compiler can assume that UB does not
happen.
Hence, the compiler can assume that the variables are never
uninitialized when passed as argument, which means that the codepaths
that initializes them must be taken.
In ff_seek_frame_binary, this means that the compiler can assume
that the codepaths that initialize pos_min and pos_max are taken,
which means that the conditions "if (sti->index_entries)" and
"if (index >= 0)" can be optimized out.
Current Clang git versions (upcoming Clang 14) enabled an optimization
that does this, which broke the current version of this function
(which intentionally left the variables uninitialized, but silencing
warnings about being uninitialized). See [1] for discussion on
the matter.
[1] https://reviews.llvm.org/D105169#3069555
Signed-off-by: Martin Storsjö <martin@martin.st>
Don't use the loaded registers directly, avoiding stalls on in
order cores. Use vrhadd.u8 with q registers where easily possible.
Signed-off-by: Martin Storsjö <martin@martin.st>
VAAPI needs 2 output surface for film grain frame. One used for
reference and the other used for applying film grain and pushing
to downstream.
Signed-off-by: Fei Wang <fei.w.wang@intel.com>
For film grain clip, vaapi_av1 decoder will cache additional 8
surfaces that will be used to store frames which apply film grain.
So increase the pool size by plus 8 to avoid leak of surface.
Signed-off-by: Fei Wang <fei.w.wang@intel.com>
Defined in spec 5.9.24/5.9.25. Since function void
global_motion_params(AV1DecContext *s) already updates
gm type/params, the wminvalid parameter only need to get
the value from cur_frame.gm_invalid.
Signed-off-by: Tong Wu <tong1.wu@intel.com>
Since order_hint_bits_minus_1 range is 0~7, cur_frame_hint can be
most 128. And similar return value for cbs_av1_get_relative_dist.
So if plus them and use int8_t for the result may lose its precision.
Signed-off-by: Fei Wang <fei.w.wang@intel.com>
The condition (pos < len) is always true and the
rest of the OpenCL program code would not be read, while
the maximum number of "rb" is "len - pos - 1", and then, the
maximum number of the "pos" is "len - 1".
Fixes: trac.ffmpeg.org/ticket/9217
Similar in spirit and design to 66845cffc3, but slightly simpler due
to the lack of interlaced frames in HEVC. See that commit for more
details.
For the seed value, since no specification for this appears to exist, I
semi-arbitrarily decided to base it off the POC id alone, since there's
no analog of the idr_pic_id in HEVC's I-frames. This design is stable
across remuxes and seeks, but changes for adjacent frames with a period
that's typically long enough not to be noticeable, which makes it
satisfy all of the requirements that a film grain seed should have.
Tested with and without threading, using a patch to insert film grain
metadata artificially (for lack of real files containing film grain).
Preparation for metadata changes in the following patches. Saves
having to create an extra buffer.
Signed-off-by: Zane van Iperen <zane@zanevaniperen.com>
The refactoring in 844890b1bc caused
fate-source to point out that this could be av_clip_uintp2 (or
rather av_clip_uint8).
Signed-off-by: Martin Storsjö <martin@martin.st>
High color 15-bit VQA3 video streams contain high level chunks with
only codebook updates that shouldn't be considered new frames. Now
the demuxer stores a reference to such VQFL chunks and returns them
later along with a VQFR chunk with full frame data.
This reverts commit 723c37d3b7.
Said commit was in preparation for auto-inserting the idet filter.
This has never happened; even if it did, the code is wrong, because
it segfaults if the filter instance doesn't have a name (having one
is not mandatory). Furthermore, it is documented for libavfilter to
not assign any semantics to the name, which this check violates.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
I spotted an interesting pattern that I didn't see before that leads to the implementation being faster.
The bit shifting table I was using before is no longer needed, and was able to remove quite a few lines.
I also add use of FMA on the AVX2 version.
f32 1920x1080 1 thread with prelut
c impl
1434012700 UNITS in lut3d->interp, 1 runs, 0 skips
1434035335 UNITS in lut3d->interp, 2 runs, 0 skips
1423615347 UNITS in lut3d->interp, 4 runs, 0 skips
1426268863 UNITS in lut3d->interp, 8 runs, 0 skips
sse2
905484420 UNITS in lut3d->interp, 1 runs, 0 skips
905659010 UNITS in lut3d->interp, 2 runs, 0 skips
915167140 UNITS in lut3d->interp, 4 runs, 0 skips
915834222 UNITS in lut3d->interp, 8 runs, 0 skips
avx
574794860 UNITS in lut3d->interp, 1 runs, 0 skips
581035090 UNITS in lut3d->interp, 2 runs, 0 skips
584116720 UNITS in lut3d->interp, 4 runs, 0 skips
581460290 UNITS in lut3d->interp, 8 runs, 0 skips
avx2
301698880 UNITS in lut3d->interp, 1 runs, 0 skips
301982880 UNITS in lut3d->interp, 2 runs, 0 skips
306962430 UNITS in lut3d->interp, 4 runs, 0 skips
305472025 UNITS in lut3d->interp, 8 runs, 0 skips
gbrap16 1920x1080 1 thread with prelut
c impl
1480894840 UNITS in lut3d->interp, 1 runs, 0 skips
1502922990 UNITS in lut3d->interp, 2 runs, 0 skips
1496114307 UNITS in lut3d->interp, 4 runs, 0 skips
1492554551 UNITS in lut3d->interp, 8 runs, 0 skips
sse2
980777180 UNITS in lut3d->interp, 1 runs, 0 skips
986121520 UNITS in lut3d->interp, 2 runs, 0 skips
986489840 UNITS in lut3d->interp, 4 runs, 0 skips
998832248 UNITS in lut3d->interp, 8 runs, 0 skips
avx
622212360 UNITS in lut3d->interp, 1 runs, 0 skips
622981160 UNITS in lut3d->interp, 2 runs, 0 skips
645396315 UNITS in lut3d->interp, 4 runs, 0 skips
641057075 UNITS in lut3d->interp, 8 runs, 0 skips
avx2
321336400 UNITS in lut3d->interp, 1 runs, 0 skips
321268920 UNITS in lut3d->interp, 2 runs, 0 skips
323459895 UNITS in lut3d->interp, 4 runs, 0 skips
324949967 UNITS in lut3d->interp, 8 runs, 0 skips
This gets rid of of rist_receiver_data_read, rist_receiver_data_block_free and rist_parse_address
these functions have been deprecated since librist release v0.2.1 and are replaced with functions
suffixed with 2.
I added a version macro check at the top of the file to ensure ffmpeg can still be compiled against
older versions.
Signed-off-by: Gijs Peskens <gijs@peskens.net>
Signed-off-by: Marton Balint <cus@passwd.hu>
The maximum allowed useful PES payload data was set to PES_packet_length, but
it is in fact smaller by the length of the PES header.
This changes how corrupt streams are packetized:
- If PES header length is bigger than PES_packet_length then the PES packet
payload will be handled as an unbound packet
- PES packets with payload across multiple MPEGTS packets will always be
splitted if with the next chunk of data the payload should exceed
PES_packet_length, previously a PES_header_length amount of excess was
allowed.
Fixes ticket #9355.
Signed-off-by: Marton Balint <cus@passwd.hu>
This renames PESContext->total_size to PESContext->PES_packet_length and keeps
it 0 for unbound packets, so its name and semantics will match the standard.
There should be no change in functionality.
Signed-off-by: Marton Balint <cus@passwd.hu>
(Inside a function a stray ';' is an empty statement; outside of
a function it is actually invalid, but compilers happen to accept
it without complaint (unless e.g. using -pedantic).)
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The voice registration system in libflite is broken: It is not
thread-safe and also not based on internal counters; instead
any call to unregister a voice frees said voice even if there are still
many other users of said voice who have also registered said voice.
While there is no way to guard against another library unregistering
voices behind our back, we can at least be correct in the absence of
other users of libflite. The current code already tried this by using
a reference count of our own for each voice; but the implementation
of this is not thread-safe at all.
Fix this by using a mutex to guard all of libavfilter's libflite
registration and unregistration calls, thereby being thread-safe
in the absence of other libflite users.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When an flite filter instance is uninitialized and the refcount
of the corresponding voice_entry reaches zero, the voice is
unregistered, yet the voice_entry's pointer to the voice is not reset.
(Whereas some other pointers are needlessly reset.)
Because of this a new flite filter instance will believe said voice
to already be registered, leading to use-after-frees.
Fix this by resetting the right pointer instead of the wrong ones.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Could also happen if initializing flite failed* or if an unknown voice
has been selected or if registering the voice failed.
*: which it currently can't, because it is a no-op.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes segfaults with filters that either return AVERROR(EAGAIN)
(or another error) or that do not set everything and rely on
filter_query_formats() to set the rest.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The calling code does not handle failures and will fail with assertion failures later.
Seeking can always fail even when the position was previously read.
Fixes: Assertion failure
Fixes: 35253/clusterfuzz-testcase-minimized-ffmpeg_dem_MATROSKA_fuzzer-4693059982983168
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Some packages may not define custom cflags, in which case a simple
"pkg-config --cflags" call will return an empty string.
This change will be useful to get a valid include path that can be
used in library checks.
Reviewed-by: Haihao Xiang <haihao.xiang@intel.com>
Signed-off-by: James Almer <jamrial@gmail.com>
av_image_copy() expects an array of four pointers according to its
declaration; although it currently only touches pointers that
are actually in use (depending upon the pixel format) this might
change at any time (as has already happened for the linesizes
in d7bc52bf45).
This fixes ticket #9264 as well as a warning from GCC 11.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
(It is actually UB if a declaration and its definition differ wrt
their types like they do in this case (the declaration in allfilters
is const).)
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now setting the input and output devices lists is guarded
by a mutex. This prevents data races emanating from multiple concurrent
calls to avpriv_register_devices() (triggered by multiple concurrent
calls to avdevice_register_all()). Yet reading the lists pointers was
done without any lock and with nonatomic variables. This means that
there are data races in case of concurrent calls to
av_(de)muxer_iterate() and avdevice_register_all() (but only if the
iteration in av_(de)muxer_iterate exhausts the non-device (de)muxers).
This commit fixes this by putting said pointers into atomic objects.
Due to the unavailability of _Atomic the object is an atomic_uintptr,
leading to ugly casts. Switching to atomics also allowed to remove
the mutex currently used in avpriv_register_devices().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The output stream's packet may not have been allocated
at that point. This happens when quitting in the following command line:
$ ./ffmpeg -lavfi abuffer=sample_fmt=u8:sample_rate=48000:channel_layout=stereo -f null -
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When inserting an auto-resampler, it may be that the configuration
of the filters that the auto-resampler is supposed to connect is
already partially merged, i.e. converter->inputs[0].incfg.foo and
converter->outputs[0].outcfg.foo (where foo is one of formats,
samplerates, channel_layouts) can coincide. Therefore merging
the converter filter's input link might modify the outcfg of the
converter' outlink. Yet the current code in avfiltergraph.c used
pointers from before merging the inlink for merging the outlink,
leading to a use-after-free in command lines like:
$ ffmpeg -f lavfi -i anullsrc=cl=stereo -lavfi channelsplit,axcorrelate -f null -
Fix this by not using outdated values when merging the outlink.
This is a regression since 85a6404d7e.
Found-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: runtime error: signed integer overflow: 727298502 * 3 cannot be represented in type 'int'
Fixes: 39172/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-638602483033702
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -2145648640 - 3357696 cannot be represented in type 'int'
Fixes: 38899/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5358815017566208
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Also covers muxing and demuxing of nonstandard FLAC channel layouts
and the multi-dim-quant option of the FLAC encoder
(all of which was hitherto uncovered).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This brings the FLAC demuxer in line with all the other demuxers.
Furthermore, if it is not done and the FLAC decoder is disabled,
the FLAC parser will overwrite the channel layout with the standard
channel layout for that number of channels.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This allows nicer tests by having a greater range of inputs available
(without requiring adding further samples to the fate-suite).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Provides coverage for the muxer.
(Thanks to tresh for modifying the whitespace commit hook
to allow to push this ref file with tabs.)
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Uninit crashed if an array containing frames could not be allocated
or config_props() has never been called.
Reviewed-by: Timo Rothenpieler <timo@rothenpieler.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fix this by switching to av_dynarray_add_nofree() which is more
natural anyway because the entries of the array are pointers.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It uses the test-lrc.lrc sample which was added years ago, but never
used until now.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In most contexts, arrays are automatically converted to a pointer
to their first element; taking the address of the array just yields
a pointer to an array of fixed-size arrays, which is not intended here.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In this case it also allows to avoid a runtime check to which filter
an AVFilterContext belongs to.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In this case it also allows to avoid a runtime check to which filter
an AVFilterContext belongs to.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In this case it also allows to avoid a runtime check to which filter
an AVFilterContext belongs to.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In this case switching to .formats.samples even allows to avoid
the runtime check for which filter is currently used.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
If one looks at the many query_formats callbacks in existence,
one will immediately recognize that there is one type of default
callback for video and a slightly different default callback for
audio: It is "return ff_set_common_formats_from_list(ctx, pix_fmts);"
for video with a filter-specific pix_fmts list. For audio, it is
the same with a filter-specific sample_fmts list together with
ff_set_common_all_samplerates() and ff_set_common_all_channel_counts().
This commit allows to remove the boilerplate query_formats callbacks
by replacing said callback with a union consisting the old callback
and pointers for pixel and sample format arrays. For the not uncommon
case in which these lists only contain a single entry (besides the
sentinel) enum AVPixelFormat and enum AVSampleFormat fields are also
added to the union to store them directly in the AVFilter,
thereby avoiding a relocation.
The state of said union will be contained in a new, dedicated AVFilter
field (the nb_inputs and nb_outputs fields have been shrunk to uint8_t
in order to create a hole for this new field; this is no problem, as
the maximum of all the nb_inputs is four; for nb_outputs it is only
two).
The state's default value coincides with the earlier default of
query_formats being unset, namely that the filter accepts all formats
(and also sample rates and channel counts/layouts for audio)
provided that these properties agree coincide for all inputs and
outputs.
By using different union members for audio and video filters
the type-unsafety of using the same functions for audio and video
lists will furthermore be more confined to formats.c than before.
When the new fields are used, they will also avoid allocations:
Currently something nearly equivalent to ff_default_query_formats()
is called after every successful call to a query_formats callback;
yet in the common case that the newly allocated AVFilterFormats
are not used at all (namely if there are no free links) these newly
allocated AVFilterFormats are freed again without ever being used.
Filters no longer using the callback will not exhibit this any more.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Most instances ff_add_formats() actually only ever add one format;
this function can be used to simplify those callers.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
A filter's formats.query callback is only called after all
the inputs and outputs have already been created.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The current code used a pointer to an array (of arrays) that
is offset relative to the start of the actually allocated buffer.
Yet offsetting the pointer is only done on success, whereas the
freeing code believes it to have happened even on error.
So if any of the subarrays (or the subarrays' subarrays) can't
be successfully allocated, one gets a bad free in free_lut().
Furthermore, said offsetting is only permissible in case the
offsetted pointer points in the allocated buffer (here: in case
the LUT's min_r is <= 0), as pointer arithmetic is undefined
in case it exceeds the allocated object.
Moreover, in case one of the subarrays couldn't be allocated,
the code nevertheless tried to free the subarray's subarrays;
and in case one of the subarray's subarrays could not be allocated
successfully, there will be an invalid free, too, because the
pointers for the subarrays' subarrays are also offset compared
to the base pointer.
This commit fixes all of this, by using the actually allocated
pointer for freeing and by adding appropriate checks before
freeing the subarrays. The former also allows to distinguish
the cases in which the lut is currently only half-allocated due to
an error in an earlier allocation attempt from the success case.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This information is coded in a standard MP4 KindBox and utilizes the
scheme and values as per the DASH role scheme defined in MPEG-DASH.
Other schemes are technically allowed, but where multiple schemes
define the same concepts, the DASH scheme should be utilized.
Such flagging is additionally utilized by the DASH-IF CMAF ingest
specification, enabling an encoder to inform the following component
of the roles of the incoming media streams.
A test is added for this functionality in a similar manner to the
matroska test.
Signed-off-by: Jan Ekström <jan.ekstrom@24i.com>
This is especially useful when reading things such as null-terminated
strings from MOV/MP4-likes, where the size of the box is known, but
not the exact size of the string.
Signed-off-by: Jan Ekström <jan.ekstrom@24i.com>
For now, same as ff_read_line_to_bprint_overwrite, but reads until
the end of a null-terminated string.
Signed-off-by: Jan Ekström <jan.ekstrom@24i.com>
Additionally:
* rename it to read_string_to_bprint
* split most of ff_read_line_to_bprint_overwrite into an internal
function which can then be utilized to implement other
functionality without duplicating code.
Signed-off-by: Jan Ekström <jan.ekstrom@24i.com>
It allows compilers to inline the one and only call to this function
in its caller or even to optimize it away completely (this function
is empty in case TRACE is not defined).
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Given that the packet sent to av_bsf_send_packet() is always
already refcounted, it is doubtful whether the error can even
be triggered currently.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In this case it means replacing a packet in the muxer's context by
a pointer to an AVPacket, namely AVFormatInternal.pkt.
Because this packet is freed generically, one can remove the muxer's
deinit function.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The documentation of said packet ("Every user has to ensure that
this packet is blank after using it") perfectly fits how we use said
packet in the generic muxing code. Better than the documentation of pkt.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently the interleave_packet functions use a packet for
a new packet to be interleaved (may be NULL if there is none) and
a packet for output; said packet is always a stack packet in
interleaved_write_packet(). But all the interleave_packet functions
in use first move the packet to the packet list and then check whether
a packet can be returned, i.e. the effective lifetime of the new packet
ends before the packet for output is touched.
So one can use one packet both for input and output by adding a new
parameter that indicates whether there is a packet to add to the packet
list; there is just one complication: In case the muxer is flushed,
there is no packet available. This can be solved by reusing one of
the packets from AVFormatInternal. They are currently unused when
flushing in av_interleaved_write_frame().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The earlier documentation claimed that av_interleaved_write_frame()
always orders by dts, which is not necessarily true when using muxers
with custom interleavement functions or the audio_preload option.
Furthermore, the documentation stated that libavformat takes ownership
of the reference of the provided packet (if it is refcounted) and that
the caller may not access the data through this reference after the
function returns. This suggests that the returned packet is not blank,
but instead still contains some set, but invalid fields, which implies
that it would be dangerous to unreference this packet again.
But this is not true: av_interleaved_write_frame()'s actual behaviour
is to always output blank packet (even on error). This commit documents
this fact so that callers know that they can directly reuse this packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The BSF API treats such packets as signalling EOF and therefore
such a packet might corrupt the BSF state. In such a case,
the guarantee that av_interleaved_write_frame() always frees
the packet is not upheld.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It claims to always zero-terminate its buffer like snprintf(),
yet it does it not on EOF. Because of this the mcc demuxer
used uninitialized values when reading an empty input file.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
ff_subtitles_queue_insert() does not require its events to be
zero-terminated as it has a parameter for the length.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, the scc demuxer not only read the line that it intends
to process, but also the next line, in order to be able to calculate
the duration of the current line. This approach leads to unnecessary
complexity and also to bugs: For the last line, the timing of the
next subtitle is not only logically indeterminate, but also
uninitialized and the same applies to the duration of the last packet
derived from it.* Worse yet, in case of e.g. an empty file, it is not
only the duration that is uninitialized, but the whole timing as well
as the line buffer itself.** The latter is used in av_strtok(), which
could lead to crashes. Furthermore, the current code always outputs
at least one packet, even for empty files.
This commit fixes all of this: It stops using two lines at a time;
instead only the current line is dealt with and in case there is
a packet after that, the duration of the last packet is fixed up
after having already parsed it; consequently the duration of the
last packet is left in its default state (meaning "unknown/up until
the next subtitle"). If no further line could be read, processing
is stopped; in particular, no packet is output for an empty file.
*: Due to stack reuse it seems to be zero quite often; for the same
reason Valgrind does not report any errors for a normal input file.
**: While ff_subtitles_read_line() claims to always zero-terminate
the buffer like snprintf(), it doesn't do so if it didn't read anything.
And even if it did, it would not necessarily help here: The current
code jumps over 12 bytes that it deems to have read even when it
hasn't.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Said function did exactly what ff_default_query_formats() does
for audio; so just remove it, so that ff_default_query_formats()
will be called.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
By reallocating the array of pointers to the AVFilterContexts
before allocating the new AVFilterContext one can avoid freeing
the new AVFilterContext in case the array could not be reallocated.
Also switch to av_realloc_array() while just at it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Mixing unsigned and signed often leads to unexpected arithmetic results.
Fixes: out of array write
Found-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
SVT-AV1 seems to have switched their default from CQP to CRF in February,
so enforce the controlling option accordingly.
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
MSVC's headers include function-like macros min and max which
collide with function pointers in vf_morpho.c, leading to
compilation failures. Fix this by renaming said function pointers.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This patch increases several stack buffers in order to fix
stack-buffer-overflows (e.g. in put_hevc_qpel_uni_hv_9 in
line 814 of hevcdsp_template.c) detected with ASAN in the hevc_pel
checkasm test.
The buffers are increased by the minimal amount necessary
in order not to mask potential future bugs.
Reviewed-by: Martin Storsjö <martin@martin.st>
Reviewed-by: "zhilizhao(赵志立)" <quinkblack@foxmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
avutil_version() currently performs several checks before
just returning the version. There is a static int that aims
to ensure that these tests are run only once. The reason is that
there used to be a slightly expensive check, but it has been removed
in 92e3a6fdac. Today running only
once is unnecessary and can be counterproductive: GCC 10 optimizes
all the actual checks away, but the checks_done variable and the code
setting it has been kept. Given that this check is inherently racy
(it uses non-atomic variables), it is best to just remove it.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Before adts_aac_resync would always bail out after probesize amount
of bytes had been progressed from the start of the input.
Now just query the current position when entering resync, and at most
advance probesize amount of data from that start position.
Fixes#9433
It is now set generically for all those encoders whose corresponding
AVCodecDescriptor has the AV_CODEC_PROP_INTRA_ONLY.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently, the AV_PKT_FLAG_KEY is automatically set for audio encoders;
yet this is wrong, as both MLP and TrueHD have non-keyframes. So set it
based upon AV_CODEC_PROP_INTRA_ONLY (from the corresponding
AVCodecDescriptor) instead. This also sets it for some video codecs,
which is intended.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
TrueHD/MLP is one of the audio formats with keyframes. Currently,
the generic encoding code just sets the keyframe flag for all
returned packets, yet this is wrong for these encoders and will
be changed in a future commit. So set the flag here for those
packets that ought to have it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Sets vtctx->has_b_frames to 0 if the VideoToolbox compression
session will not emit B-frames (and, in consequence, no valid
DTSs). Required for the handling of invalid DTSs in
'vtenc_cm_to_avpacket' (line 2018ff) to work correctly and not
abort encoding with "DTS is invalid." when no B-frames are
generated.
Signed-off-by: NoHalfBits <ffmpeg-devel@fluthaus.com>
Signed-off-by: Rick Kern <kernrj@gmail.com>
They already uncovered an uninitialized-value bug in the ATRAC3 code
in the demuxer; and provide coverage for ID3v2.3.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There can only be a maximum of 255 entries in a tfrf tag, so using
more makes no sense; moreover, several size computations can overflow
in this case. Fix this by limiting it to 255.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Do this by using the AVStream's priv_data for the buffer holding
the packet size data.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
If an array for the packet sizes could not be successfully reallocated
when writing a packet, the CAF muxer frees said array, but does not
reset the number of valid bytes. As a result, when the trailer is
written later, avio_write tries to read that many bytes from NULL,
which segfaults.
Fix this by not freeing the array in case of error; also, postpone
writing the packet data after having successfully (re)allocated the
array, so that even on allocation error the file can be correctly
finalized.
Also remove an unnecessary resetting of the number of size entries
used at the end.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
(As long as avio_write() only accepts an int, it makes no sense
to try to support sizes that don't fit into an int.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: signed integer overflow: 2145649668 + 3956526 cannot be represented in type 'int'
Fixes: 38351/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-4647077926273024
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 9223372036854775360 + 536870912 cannot be represented in type 'long'
Fixes: 37940/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-6095637855207424
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The new format (given in big/little endian forms) matches the
existing X2RGB10 format, except with B and R channels switched.
AV_PIX_FMT_X2BGR10 data often is created by OpenGL programs
whose buffers use the GL_RGB10 internal format.
Signed-off-by: Manuel Stoeckl <code@mstoeckl.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This resolves a problem where conversions from YUV to X2RGB10LE
would produce color values a factor 4 too small, because an 8-bit
value was placed in a 10-bit channel.
Signed-off-by: Manuel Stoeckl <code@mstoeckl.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Just check for the existence of the bsf. This is equivalent to
the old criterion of the AVCodecContext being a decoder.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is now checked by FATE that no encoder capable of flushing
uses frame threads, so this now redundant runtime check can
be removed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
If an AVCodec has a private class, its priv_data_size must be > 0
and at the end of a successful call to avcodec_open2()
the AVCodecContext's priv_data must exist and its first element
must be a pointer to said AVClass. This should not be conditional
on priv_data_size being > 0 (which is tested by FATE) or
on the private context having been successfully allocated
(which has to have happened at that point). So remove these
preconditions to make the test stricter.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The current name comes from a time in which libavcodec/utils.c
contained the whole core of libavcodec.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, the list of pixfmts is reallocated every time an entry
is added to it; there are currently 196 pixel formats, so this matters:
It causes 5541704 calls to av_realloc_array() in a typical FATE run,
which is the majority for said function (8095768 calls) and even
a large chunk of the calls to av_realloc() itself (12589508 calls).
Fix this by using ff_formats_pixdesc_filter() instead.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, it has returned the AVFilterFormats list via
an AVFilterFormats** parameter; the actual return value was an int
that was always AVERROR(ENOMEM) on error. The AVFilterFormats**
argument was a pure output parameter which was only documented
by naming the parameter rfmts. Yet nevertheless all callers
initialized the underlying AVFilterFormats* to NULL.
This commit changes this to return a pointer to AVFilterFormats
directly. This is more in line with the API in general, as it
allows to avoid checks for intermediate values.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This avoids a -Wstringop-truncation warning from GCC which takes
issue with the fact that the destination might not be NUL-terminated.
Reviewed-by: Zane van Iperen <zane@zanevaniperen.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When a color indexing transform with 16 or fewer colors is used,
WebP uses "pixel packing", i.e. storing several pixels in one byte,
which virtually reduces the width of the image (see WebPContext's
reduced_width field). This reduced_width should always be used when
reading and applying subsequent transforms.
Updated patch with added fate test.
The source image dual_transform.webp can be downloaded by cloning
https://chromium.googlesource.com/webm/libwebp-test-data/
Fixes: 9368
Signed-off-by: James Zern <jzern@google.com>
This muxer was untested up until now; had it been tested, it would
have been obvious that it has been broken for years.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The terminating '\0' is no longer included in the size of
the extradata output by the demuxer since commit
36e61e24e7.
E.g. if one remuxes the JACOsub sample JACOsub_capability_tester.jss
from the FATE suite, one receives a file not recognized as JACOsub
before this patch.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Dead since commit 93016f5d1d
which ensured that the packets received by encoders are always blank.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Command below failed.
ffmpeg -v verbose -init_hw_device vaapi=va:/dev/dri/renderD128
-init_hw_device qsv=qs@va -hwaccel qsv -hwaccel_device qs
-filter_hw_device va -c:v h264_qsv
-i 1080P.264 -vf "hwmap,format=vaapi" -c:v h264_vaapi output.264
Cause: Assign pair->first directly to data[3] in vaapi frame.
pair->first is *VASurfaceID while data[3] in vaapi frame is
VASurfaceID. I fix this line of code. Now the command above works.
Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
This improves performance: For msvideo1, the performance improved by
4.8% when encoding the sample from the fate-vsynth1-msvideo1 test;
when encoding the sample from fate-vsynth1-cinepak, performance
improved by 2%. The compiler user was GCC 10 and the calls to encode2
have been timed.
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This is currently unused and it is only added to enable changes
while maintaining ABI compatibility. The type is uintptr_t in order
to potentially accept a pointer argument.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This is possible because the number of elements needed in each
recursion step decreases geometrically, so the geometric series
provides an upper bound for the sum of number of elements of
the needed buffers.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Every filter exposing the framesync options via its child_next
callback already calls framesync_preinit() in its preinit callback.
So the filter is already preinited whenever its child_next is called.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This filter uses ff_set_common_all_samplerates().
Also don't overwrite outlink->sample_rate in config_output;
it is harmless, because it is overwritten with the value it already
had, but it is an API violation.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This filter uses ff_set_common_all_samplerates().
Also don't overwrite outlink->sample_rate in config_output;
it is harmless, because it is overwritten with the value it already
had, but it is an API violation.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This filter uses ff_set_common_all_samplerates().
Also don't overwrite outlink->sample_rate in config_output;
it is harmless, because it is overwritten with the value it already
had, but it is an API violation.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Just like the sidechaingate filter, the sidechaincompress filter
overwrote the channel layout and channel count of its output in
its config_output callback to match the channel layout of its main
input instead of linking the main input and its output together
in its query_formats callback.
This is an API violation that can lead to segfaults, as in the
following filtergraph, where stereotools rightly expects stereo,
yet receives only mono:
[in]aformat=channel_layouts=mono,aformat=channel_layouts=stereo|mono[out];\
[out][in2]sidechaincompress,stereotools
Fix this by linking the channel layouts of the main input and the output
in query_formats and remove the code overwriting it in config_output.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The sidechaingate filter wants its main input and its (only) output
to have the same channel layout and number of channels; yet it does
not link them in its query_formats callback. Instead it sets the
outlink to only accept the first offered choice for the main input's
channel layout and then sets both inputs to independently accept
any channel counts. The config_output callback then overwrote the
outlink's channel layout and channels properties with the main input's,
even though they may differ in case the first offered choice for
the main input's channel layout turns out not to be the final one.
Consider e.g. the following filtergraph:
[in]aformat=channel_layouts=mono,aformat=channel_layouts=stereo|mono[out];\
[out][in2]sidechaingate,stereotools
The two aformats ensure that the first offered channel layout (stereo)
will not be chosen for the input; yet it is the only offered channel
layout for the output of sidechaingate and will therefore be chosen
by the query_formats framework. Because the sidechaingate outputs
interleaved doubles which stereotools expects the output of
sidechaingate appears to be suitable as input for stereotools without
further conversions. Yet stereotools actually only receives a mono frame
and therefore overreads its input buffer which leads to segfaults;
it can also lead to heap corruption because there can be writes beyond
the end of the buffer, too.
Fix this by linking the channel layouts of the main input and the output
in query_formats and remove the code overwriting it in config_output.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The acrossfade filter uses the ff_set_common_* functions in its
query_formats(), so that the formats, the sample rates as well as
the channel layouts and counts of all links coincide.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The transpose, rotate, hflip, and vflip filters don't support them.
Fixes ticket #9432.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: James Almer <jamrial@gmail.com>
If no --cpu= option was passed to configure, we detect what the
compiler defaults to. This detected value was then fed back to the
rest of the configure logic, as if it was an explicit choice.
This breaks on Ubuntu 21.10 with GCC 11.1.
Since GCC 8, it's possible to add configure extra features via the
-march option, like e.g. -march=armv7-a+neon. If the -mfpu= option
is configured to default to 'auto', the fpu setting gets taken
from the -march option.
GCC 11.1 in Ubuntu seems to be configured to use -mfpu=auto. This
has the effect of breaking any compilation command that specifies
-march=armv7-a, because the driver implicitly also adds -mfloat-abi=hard,
and that combination results in this error:
cc1: error: ‘-mfloat-abi=hard’: selected processor lacks an FPU
One can compile successfully by passing e.g. -march=armv7-a+fp.
Therefore, restructure configure. If no specific preference was set
(and the 'cpu' configure variable was set as the output of
probe_arm_arch), the value we tried to set via -march= was the same
value that we just tried to detect as the compiler default.
So instead, just try to detect what the compiler defaults to, with
to allow setting other configure settings (such as 'fast_unaligned'),
but don't try to spell out the compiler's default via the -march flag.
Signed-off-by: Martin Storsjö <martin@martin.st>
Up until now, each call to avpriv_elbg_do() would result
in at least six allocations. And this function is called a lot:
A typical FATE run results in 52213653 calls to av_malloc; of these,
34974671 originate from av_malloc_array and from these 34783679
originate from avpriv_elbg_do; the msvideo1 encoder tests are behind
most of these.
This commit changes this by keeping the buffers and only reallocating
them when needed. E.g. for the encoding part of fate-vsynth1-msvideo1
total heap usage went down from 11,407,939 allocs and frees with
468,106,207 bytes allocated to 3,149 allocs and frees with 13,181,847
bytes allocated. The time for one encode2-call went down by 69%.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This affects all the arguments that don't change during a call
to avpriv_elbg_do(); doing so makes it easily recognizable which
arguments change upon recursive calls.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It will be used in future commits to avoid having to allocate and free
all the buffers used.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is in line with our naming conventions for types.
Also change numCB to num_cb for the same reason.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
These functions are always called directly after another with
the exact same arguments. This avoids exporting a symbol;
it also avoids having to perform two calls for every caller.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It does the same as av_calloc(), so one of them should be removed.
Given that av_calloc() has the shorter name, it is retained.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
silk_lsp2poly()'s declaration contained arrays with array sizes;
yet these array sizes exceeded the number of actually accessed array
elements (which is related to another parameter) and this leads to
-Wstringop-overflow= warnings from GCC 11, because the arrays provided
by callers are only large enough for the actually used elements.
So replace the incorrect array sizes with comments containing
the correct array sizes. Given that these sizes are not compile-time
constants, they can only be communicated via a comment.
Reported by Paul B Mahol.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is supported only by a few decoders (h263, h263p, mpeg(1|2|)video
and mpeg4) and is entirely redundant with parsers. Furthermore, using
it leads to missing frames, as flushing the decoder at the end does not
work properly.
Co-authored-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
SSE2 is x86 specific, yet due to the call to av_get_cpu_flags()
compilers were unable to optimize the checks (and the call) away
on other arches.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In this case the current code tries to warn once; to do so, it uses
ordinary static ints to store whether the warning has already been
emitted. This is both a data race (and therefore undefined behaviour)
as well as a race condition, because it is really possible for multiple
threads to be the one thread to emit the warning. This is actually
common since the introduction of the new multithreaded scaling API.
This commit fixes this by using atomic integers for the state;
furthermore, these are not static anymore, but rather contained
in the user-facing SwsContext (i.e. the parent SwsContext in case
of slice-threading).
Given that these atomic variables are not intended for synchronization
at all (but only for atomicity, i.e. only to output the warning once),
the atomic operations use memory_order_relaxed.
This affected the nv12, nv21, yuv420, yuv420p10, yuv422, yuv422p10 and
yuv444 filter-overlay FATE-tests.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This allows to associate log messages from slice contexts to
the user-visible SwsContext.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
By using preinit, the AVDCT already exists directly after
allocating the filter, so that the filter's AVClass's child_next
becomes usable for setting options with the AV_OPT_SEARCH_CHILDREN
search flag. This means that it is no longer necessary to use
the init_dict callback for this filter.
Furthermore, the earlier code did not abide by the documentation
of the init_dict callback at all: Instead of only returning the
options that have not been recognized it always returned all options
on any av_opt_set() error and errored out in this case, even if it
is just an unrecognized option. This behaviour has been inherited by
avfilter_init_dict(), contradicting its documentation. This is also
fixed in this commit.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
By using preinit, the SwrContext already exists directly after
allocating the filter, so that the filter's AVClass's child_next
becomes usable for setting options with the AV_OPT_SEARCH_CHILDREN
search flag. This means that it is no longer necessary to use
the init_dict callback for this filter.
Furthermore, the earlier code did not abide by the documentation
of the init_dict callback at all: Instead of only returning the
options that have not been recognized it always returned all options
on any av_opt_set() error and errored out in this case; yet if
the error was just caused by an unrecognized option, it should not
error out at all and instead return said option.
This behaviour has been inherited by avfilter_init_dict(),
contradicting its documentation. This is also fixed by this commit.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Do this by putting an AVBuffer structure into BufferPoolEntry and
reuse it for all subsequent uses of said BufferPoolEntry.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
zimg's color range enum values are off-by-one compared to ours;
therefore the code just adds one when converting from theirs to ours.
Yet this is not how one should deal with enums; use a switch instead.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
SRTSOCKET is an abstraction designed by libsrt, it's not guaranteed
to be a real file descriptor. Even if it is, it should not be
operated directly outside of libsrt.
Signed-off-by: Marton Balint <cus@passwd.hu>
It's useful for test client which pass streamid to ffmpeg/ffplay.
For example, use ffmpeg to test streamid support in VLC:
./ffmpeg -v info -re -i foo.mp4 -c copy -f mpegts -mode listener srt://127.0.0.1:9000
./vlc srt://127.0.0.1:9000?streamid=foobar
Signed-off-by: Marton Balint <cus@passwd.hu>
When a possible overflow was detected, there was a break to exit the while
loop. However, it should have already substracted 2 bytes from
program_info_length (descriptor ID + length).
Fixes ticket #9422.
Signed-off-by: Marton Balint <cus@passwd.hu>
The problem was caused by if the width of the processed block
minus 1 is a multiple of the aligned number the instruction
jle .bscale_scalar would skip the Optimized Loop Step, which
will lead to an incorrect sampling when specifying steps more
than 1. Move the Optimized Loop Step after .bscale_scalar to
ensure the loop step is enabled.
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
This filter chain was supposed to convert from narrow range
to full range yuv444p, but didn't:
buffer=width=1280:height=720:pix_fmt=yuv444p:frame_rate=25/1:\
time_base=1/25:sar=1/1,zscale=min=709:rin=limited:pin=709:\
tin=709:t=linear,format=gbrpf32le,zscale=tin=linear:p=709:m=709:\
r=full:t=709,format=pix_fmts=yuv444p,buffersink
Fixes: error: 1.66789e+11 is outside the range of representable values of type 'int'
Fixes: Ticket8201
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Using ff_set_common_formats_from_list() avoids using different functions
depending upon how many inputs the filter has.
Reviewed-by: Thilo Borgmann <thilo.borgmann@mail.de>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently an error from init could be overwritten by successfully
setting the enable expression.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
libavformat/utils.c has over 4800 lines and is supposed to contain
"various utility functions for use within FFmpeg". In reality it
contains all that and the whole demuxing core of libavformat.
This is especially bad, because said file includes the FFMPEG_VERSION
(the git commit sha) so that it is rebuilt whenever the commit HEAD
points to changes. Therefore this commit makes it smaller by moving
the demuxing code out to a new file, demux.c (in analogy to mux.c
for the muxing code).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
libavformat/utils.c has over 5500 lines and is supposed to contain
"various utility functions for use within FFmpeg". In reality it
contains all that and the whole demuxing+seeking core of libavformat.
This is especially bad, because said file includes the FFMPEG_VERSION
(the git commit sha) so that it is rebuilt whenever the commit HEAD
points to changes. Therefore this commit starts making it smaller
by factoring the seeking code out.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Do this by allocating AVStream together with the data that is
currently in AVStreamInternal; or rather: Put AVStream at the
beginning of a new structure called FFStream (which encompasses
more than just the internal fields and is a proper context in its own
right, hence the name) and remove AVStreamInternal altogether.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This gets rid of ugly "->internal" and is in preparation for removing
AVStreamInternal altogether.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Do this by allocating AVFormatContext together with the data that is
currently in AVFormatInternal; or rather: Put AVFormatContext at the
beginning of a new structure called FFFormatContext (which encompasses
more than just the internal fields and is a proper context in its own
right, hence the name) and remove AVFormatInternal altogether.
The biggest simplifications occured in avformat_alloc_context(), where
one can now simply call avformat_free_context() in case of errors.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This gets rid of ugly "->internal" and is in preparation for removing
AVFormatInternal altogether.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: signed integer overflow: 4611686025627387904 + 4611686025627387904 cannot be represented in type 'long'
Fixes: 35489/clusterfuzz-testcase-minimized-ffmpeg_dem_SBG_fuzzer-4862678601433088
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: runtime error: signed integer overflow: 65312 * 65535 cannot be represented in type 'int'
Fixes: 32832/clusterfuzz-testcase-minimized-ffmpeg_dem_RM_fuzzer-4817710040088576
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 2788626175500000000 + 7118941284000000000 cannot be represented in type 'long'
Fixes: 35215/clusterfuzz-testcase-minimized-ffmpeg_dem_SBG_fuzzer-6123272247836672
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Defined in spec 5.9.8. When superres is enabled, SuperresDenom equals
"coded_denom + SUPERRES_DENOM_MIN" instead of coded_denom.
Signed-off-by: Tong Wu <tong1.wu@intel.com>
Signed-off-by: Hendrik Leppkes <h.leppkes@gmail.com>
Fixes: signed integer overflow: 7020950083487072256 * 2 cannot be represented in type 'long long'
Fixes: 37523/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EXR_fuzzer-5133634955771904
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -5994697211974418462 + -3255307777713450286 cannot be represented in type 'long'
Fixes: 35332/clusterfuzz-testcase-minimized-ffmpeg_dem_MATROSKA_fuzzer-5868035117285376
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The documentation does not require the packet to be blank in this case
(i.e. it can now contain opaque_ref), but it does contain that the
contents will be reset upon success.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When an empty list bsf is used for passthrough, there is a check
for every packet in bsf_list_filter() before ff_bsf_get_packet_ref()
is called. Directly using the null bsf avoids that.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: negation of -2147483648 cannot be represented in type 'int'; cast to an unsigned type to negate this value to itself
Fixes: 34651/clusterfuzz-testcase-minimized-ffmpeg_dem_JACOSUB_fuzzer-5157941012463616
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This macro will allow to share options between AVClasses without
having to redefine the option name (as is currently done) and will
also allow to share the AVClasses itself (which is possible now
that AVClass.child_class_next is gone).
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes trac issue #7473.
Removes encoder delay (skip samples) and writes remaining frame samples after EOF to get correct sample count.
Output is now accurate vs players that use Microsoft's codecs (Windows Media Format Runtime).
Tested vs encode>decode WMAv2 with MS's codecs and most sample rate/bit rate/channel/mode combinations in ASF/XWMA.
WMAv1 appears to use the same delay, from FFmpeg samples.
Signed-off-by: bnnm <bananaman255@gmail.com>
When a device is derived from a source device, there are at least 2
devices, and usually the derived device is the expected device, so let's
pick the last device if user doesn't specify the filter device with
filter_hw_device option
After applying this patch, the command below can work:
$> ffmpeg -init_hw_device vaapi=va:/dev/dri/renderD128 -init_hw_device
qsv=hw@va -f lavfi -i yuvtestsrc -vf
format=nv12,hwupload=extra_hw_frames=64 -c:v h264_qsv -y out.h264
Reviewed-by: Soft Works <softworkz@hotmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
This bug flew under the radar because, in practice, these values are
0-initialized for the first invocation. But for subsequent invocations
(with different h/v values), reading from the uninitialized parts of
`out` is undefined behavior.
Avoid this by simply adjusting the iteration range of the following
loops. Has the added benefit of being a minor speedup.
Signed-off-by: James Almer <jamrial@gmail.com>
fixes#8857
If we do not clear the enc_ctrl, we will reuse previous frames' data like FrameType.
Reviewed-by: Xiang, Haihao <haihao.xiang@intel.com>
Signed-off-by: James Almer <jamrial@gmail.com>
When encoding yuva420 (alpha) frames, the vpx encoder uses a second
vpx_codec_ctx to encode the alpha stream. However, codec options were
only being applied to the primary encoder. This patch updates
codecctl_int and codecctl_intp to also apply codec options to the alpha
codec context when encoding frames with alpha.
This is necessary to take advantage of libvpx speed optimizations
such as 'row-mt' and 'cpu-used' when encoding videos with alpha.
Without this patch, the speed optimizations are only applied to the
primary stream encoding, and the overall encoding is just as slow
as it would be without the options specified.
Signed-off-by: Adam Chelminski <chelminski.adam@gmail.com>
Signed-off-by: James Zern <jzern@google.com>
It is a special BSF that is only available via the av_bsf_list-API;
it is not part of the list generated from the declarations in
bitstream_filters.c and therefore needn't have external linkage.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Adding DX11 relevant device type checks and adjusting callback with
proper MediaSDK pair type support.
Signed-off-by: Artem Galin <artem.galin@intel.com>
Adding DX11 relevant device type checks and adjusting callback with
proper MediaSDK pair type support.
Signed-off-by: Artem Galin <artem.galin@intel.com>
Adding DX11 relevant device type checks and adjusting callback with
proper MediaSDK pair type support.
Signed-off-by: Artem Galin <artem.galin@intel.com>
Microsoft VideoProcessor requires texture with D3DUSAGE_RENDERTARGET flag as output.
There is no way to allocate array of textures with D3D11_BIND_RENDER_TARGET flag
and .ArraySize > 2 by ID3D11Device_CreateTexture2D due to the Microsoft limitation.
Adding AVD3D11FrameDescriptors array to store array of single textures
instead of texture with multiple slices resolves this.
Signed-off-by: Artem Galin <artem.galin@intel.com>
UPD: Rebase of last patch set over current master and use DX9 as default device type.
Makes selection of dxva2/DX9 device type by default as before with explicit d3d11va/DX11 usage to cover more HW configurations.
Added warning message to expect changing default device type in the future.
Fixes TGL / AV1 decode as requires DX11 with explicit DX11 type
selection.
Add headless/multi adapter support and fixes:
https://trac.ffmpeg.org/ticket/7511https://trac.ffmpeg.org/ticket/6827http://ffmpeg.org/pipermail/ffmpeg-trac/2017-November/041901.htmlhttps://trac.ffmpeg.org/ticket/7933338fbcd5bbhttps://github.com/jellyfin/jellyfin/issues/2626#issuecomment-602153952
Any other fixes are welcome including OpenCL interop patch since I don't have proper setup to validate this use case
Decoding, encoding, transcoding have been validated.
child_device_type option is responsible for d3d11va/dxva2 device selection
Usage examples:
DirectX 11:
-init_hw_device qsv:hw,child_device_type=d3d11va
-init_hw_device qsv:hw,child_device_type=d3d11va,child_device=0
OR
-init_hw_device d3d11va=dx -init_hw_device qsv@dx
DirectX 9 is still supported but requires explicit selection:
-init_hw_device qsv:hw,child_device_type=dxva2
OR
-init_hw_device dxva2=dx -init_hw_device qsv@dx
Signed-off-by: Artem Galin <artem.galin@intel.com>
This enables usage of non-powered/headless GPU, better HDR support.
Pool of resources is allocated as one texture with array of slices.
Signed-off-by: Artem Galin <artem.galin@intel.com>
Adding DX11 relevant device type checks and adjusting callbacks with
proper MediaSDK pair type support.
Extending structure for proper MediaSDK pair type support.
Signed-off-by: Artem Galin <artem.galin@intel.com>
It is supposed to be used with different bit depth and/or sample rates
per each substream, but such currently not implemented feature is not
important and current state causes problems when implementing variable
restart interval to fix decoding with sample rates not multiple of 40.
This reverts commit 628a73f8f3.
At the time of said commit there was talk of removing the audio bitrate
"ab" option to bring FFmpeg in line with what Libav has done in 2012 in
commit 041cd5a0c5. By having different
option flags for the "ab" and the ordinay bitrate "b" option is is
possible to have different default bitrates for audio and video. In
order to maintain this behaviour and not break user scripts the commit
to be reverted added code to ffmpeg.c that set the bitrate value to the
audio default for audio codecs, but only if AVCodec.defaults didn't
exist (as in this case the default would be codec-default and not
affected by the "ab" removal).
This had the downside of being an API violation, because
AVCodec.defaults is not a public field. Given that the "ab" option
and its audio-specific default value have never been removed,
said API violation can be simply fixed by reverting said commit.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This way the CLI accepts for "filter_threads" the same values as for the
libavcodec specific option "threads".
Fixes FATE with THREADS=auto which was broken in bdc1bdf3f5.
Signed-off-by: James Almer <jamrial@gmail.com>
The earlier code did not properly check these initializations:
It only recorded whether the part of init where these initializations
are has been reached, but it did not check whether the initializations
were successful, although destroying them would be undefined behaviour
if they had not been initialized successfully.
Furthermore cleanup() always locked a mutex regardless of whether there
was any attempt to initialize these mutexes at all.
Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Now that the mutexes and conditions are only initialized and destroyed
once, said function only had one purpose: free the entries array.
Given that vp9_alloc_entries() already does this if the array is already
allocated it is unnecessary to call vp9_free_entries() anywhere except
when closing. And then one can just inline the one free into
vp9_decode_free().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Otherwise the context would be in an inconsistent state
if vp9_alloc_entries() failed (and if this would be checked).
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
We have more mutexes/condition variables whose initialization is
unchecked.
Also use a proper namespace for these functions.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Also do not destroy and reinitialize mutexes and conditions when
certain input parameters change. Given that the decoder did not
create these variables at all during init, uninitialized mutexes
and conditions are destroyed before the very first initialization.
This is undefined behaviour and certain threading implementations
like pthreadGC2 crash when it is attempted.
Fix this by initializing these objects once during init and freeing
them in close.
Reported-by: Steve Lhomme <robux4@ycbcr.xyz>
Reviewed-by: Steve Lhomme <robux4@ycbcr.xyz>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
H.264 and H.265 levels' names are usually of the form "x" or "x.y"
with x and y being single digits; the one exception are the H.264 1b
levels. All of those levels' names fit into a char[4] and it is likely
that this future levels will do so, too.
Therefore this commit changes the H26(4|5)LevelDescriptor structures
to use such a char [4] instead of a pointer to a const char. This makes
the structures smaller (when sizeof(char*) == 8) and avoids relocations,
thereby moving the corresponding arrays from .data.rel.ro into .rodata.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
They were done in preparation for an upcoming 1.0 release.
Keep supporting previous releases for the time being.
Reviewed-by: BBB
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, ff_write_chained() copied the packet (manually, not with
av_packet_move_ref()) from a packet given to it to a stack packet whose
timing and stream_index is then modified before being sent to another
muxer via av_(interleaved_)write_frame(). Afterwards it is intended to
sync the fields of the packet relevant to freeing again; yet this only
encompasses buf, side_data and side_data_elems and not the newly added
opaque_ref. The other fields are not synced so that the returned packet
can have a size > 0 and data != NULL despite its buf being NULL (this
always happens in the interleaved codepath; before commit
fe251f77c8 it could also happen in the
noninterleaved one). This leads to double-frees if the interleaved
codepath is used and opaque_ref is set.
This commit therefore changes this by directly reusing the packet
instead of a spare packet. Given that av_write_frame() does not
change the packet given to it, one only needs to restore the timing
information to return it as it was; for the interleaved codepath
it is not possible to do likewise*, because av_interleaved_write_frame()
takes ownership of the packets given to it and returns blank packets.
But precisely because of this users of the interleaved codepath
have no legitimate expectation that their packet will be returned
unchanged. In line with av_interleaved_write_frame() ff_write_chained()
therefore returns blank packets when using the interleaved codepath.
Making the only user of said codepath compatible with this was trivial.
*: Unless one wanted to create a full new reference.
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
These were intended to pass options to auto-inserted avresample
resampling filters. Yet FFmpeg uses swresample for this purpose
(with its own AVDictionary swr_opts similar to resample_opts).
Therefore said options were not forwarded any more since commit
911417f0b34e611bf084319c5b5a4e4e630da940; moreover since commit
420cedd497 avresample options are
not even recognized and ignored any more. Yet there are still
remnants of all of this. This commit gets rid of them.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
ff_pnm_parser and ff_vp3_parser already hit the current limit;
an addition to the former (to handle pfm) is planned.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Unlike libx264, libx265 does not handle the chroma format check
on its own side, so in order to not write out values which are
supposed to be ignored according to the specification, we limit
the writing out of chroma sample location to 4:2:0 only.
Fixes invalid reports of bad lossless crc.
While here make end of stream message into debug level as it is
not really important to user.
Also wait for new major sync frame as invalid concating of files
may produce invalid files, which cause various errors.
It is unnecessary and also ill-defined: av_malloc() returns a 1-byte
block of memory in this case, but this is not documented.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The chapters are independently allocated, so that comparing
the pointers is undefined behaviour. Furthermore, its result
is not platform-independent (and may not even be deterministic
on a particular platform). So compare the chapters' ids instead.
(avpriv_new_chapter() ensures that there are no duplicate ids.)
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This structure is only used for demuxers (mostly in
avformat_find_stream_info()), so only allocate it for them.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It has been allocated and initialized in avformat_find_stream_info()
until fd0368e7ca when the structure
was moved to AVStreamInternal and its allocation to avformat_new_stream.
In order to also initialize the struct for new streams that only get
created during avformat_find_stream_info() said the initialization has
been added to avformat_new_stream() later. Due to the Libav-FFmpeg split
this has been done twice: In 4cda8aa1c5
and in 30c26c2442. The initialization in
avformat_find_stream_info() has not been removed at all despite being
redundant. This commit removes it and the duplicated initialization in
avformat_new_stream().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The packet queue is already flushed in avformat_free_context() which
is called a few lines below.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
An AVStream's internal AVCodecContext is pretty much unused for muxing:
The only place where any of its fields are set is
avformat_transfer_internal_stream_timing_info() where its time base is
set based upon the desired output format. The max_b_frames field is
never set at all, so don't read it in mux.c.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Since 1c0885334d ff_compute_frame_duration
is only called from within utils.c and only for demuxers. So make it
static and remove the code in it that deals with muxers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
AVFormatContext.internal is already allocated by
avformat_alloc_context() on success; and on error,
avformat_alloc_context() cleans up manually without
avformat_free_context().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The WebM DASH Manifest demuxer creates a comma-delimited list of
all the timestamps of index entries. It allocates 20 bytes per
timestamp; yet the largest 64bit numbers have 20 decimal digits
(for int64_t it can be '-'+ 19 digits), so that one needs 21B
per entry because of the comma (resp. the final NUL).
The code uses snprintf, but snprintf returns the strlen of the string
that would have been written had the supplied buffer been big enough.
And if this is 21, then the next entry is written at an offset of 21
from the current position. So if enough such entries exist, the buffer
won't suffice.
This commit fixes this by replacing the allocation of buffer for
the supposedly worst-case with dynamic allocations by using an AVBPrint.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This reverts commit b3a0548a98.
This breaks the usage of swscale options, scale_sws_opts should be
passed to auto-inserted scale-filters.
Signed-off-by: Linjie Fu <linjie.justin.fu@gmail.com>
Fixes: signed integer overflow: 9223372034248226491 + 3275247799 cannot be represented in type 'long'
Fixes: clusterfuzz-testcase-minimized-audio_decoder_fuzzer-4538729166077952
Reported-by: Matt Wolenetz <wolenetz@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
We introduced a ff_horiz_slice_avx2/512() implemented on a new algorithm.
In a nutshell, the new algorithm does three things, gathering data from
8/16 rows, blurring data, and scattering data back to the image buffer.
Here we used a customized transpose 8x8/16x16 to avoid the huge overhead
brought by gather and scatter instructions, which is dependent on the
temporary buffer called localbuf added newly.
Performance data:
ff_horiz_slice_avx2(old): 109.89
ff_horiz_slice_avx2(new): 666.67
ff_horiz_slice_avx512: 1000
Co-authored-by: Cheng Yanfei <yanfei.cheng@intel.com>
Co-authored-by: Jin Jun <jun.i.jin@intel.com>
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
The new vertical slice with AVX2/512 acceleration can significantly
improve the performance of Gaussian Filter 2D.
Performance data:
ff_verti_slice_c: 32.57
ff_verti_slice_avx2: 476.19
ff_verti_slice_avx512: 833.33
Co-authored-by: Cheng Yanfei <yanfei.cheng@intel.com>
Co-authored-by: Jin Jun <jun.i.jin@intel.com>
Signed-off-by: Wu Jianhua <jianhua.wu@intel.com>
EINVAL is the wrong error code here, since the arguments passed to the
function are valid. The error is that the function is not implemented in
the build, which corresponds to ENOSYS.
When both -filter_threads and -threads are specified, the latter takes
effect. Since -threads is an encoder option and -filter_threads is a
filter option, it makes sense for the -filter_threads to take
precedence.
Implements a gray world color correction algorithm
using a log scale LAB colorspace.
Signed-off-by: Paul Buxton <paulbuxton.mail@googlemail.com>
Signed-off-by: Paul B Mahol <onemda@gmail.com>
Don't allocate the buffer for the title ourselves, leave it to
av_dict_set(). This simplifies freeing.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
A single VorbisComment consists of a length field and a
non-NUL-terminated string of the form "key=value". Up until now,
when parsing such a VorbisComment, zero-terminated duplicates of
key and value would be created. This is wasteful if these duplicates
are freed shortly afterwards, as happens in particular in case of
attached pictures: In this case value is base64 encoded and only
needed to decode the actual data.
Therefore this commit changes this: The buffer is temporarily modified
so that both key and value are zero-terminated. Then the data is used
in-place and restored to its original state afterwards.
This requires that the buffer has at least one byte of padding. All
buffers currently have AV_INPUT_BUFFER_PADDING_SIZE bytes padding,
so this is ok.
Finally, this also fixes weird behaviour from ogm_chapter():
It sometimes freed given to it, leaving the caller with dangling
pointers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This results in warnings on compilers which don't support it,
objections were raised during the review process about it but went unnoticed,
and the speed benefit is highly compiler and version specific, and
also not very critical.
We generally hand-write assembly to optimize loops like that, rather
than use compiler magic, and for 40% best case scenario, it's simply
not worth it.
Plus, tree vectorization is still problematic with GCC and disabled by default
for a good reason, so enabling it locally is sketchy.
This patch renames the InferenceItem to LastLevelTaskItem in the
three backends to avoid confusion among the meanings of these structs.
The following are the renames done in this patch:
1. extract_inference_from_task -> extract_lltask_from_task
2. InferenceItem -> LastLevelTaskItem
3. inference_queue -> lltask_queue
4. inference -> lltask
5. inference_count -> lltask_count
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
Remove async flag from filter's perspective after the unification
of async and sync modes in the DNN backend.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit removes the unused sync mode specific code from the DNN
filters since the sync and async mode are now unified from the
filters' perspective.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit unifies the async and sync mode from the DNN filters'
perspective. As of this commit, the Native backend only supports
synchronous execution mode.
Now the user can switch between async and sync mode by using the
'async' option in the backend_configs. The values can be 1 for
async and 0 for sync mode of execution.
This commit affects the following filters:
1. vf_dnn_classify
2. vf_dnn_detect
3. vf_dnn_processing
4. vf_sr
5. vf_derain
This commit also updates the filters vf_dnn_detect and vf_dnn_classify
to send only the input frame and send NULL as output frame instead of
input frame to the DNN backends.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
As the second argument for init_get_bits(avctx and buf) can be crafted,
a return value check for this function call is necessary,
so replace init_get_bits with init_get_bits8 and add return value check.
Consider data as invalid if ff_wma_run_level_decode
gets out with an error.
It avoids an unpleasant sound distorsion.
See http://trac.ffmpeg.org/ticket/9358
As the second argument for init_get_bits(buf) can be crafted,
a return value check for this function call is necessary.
Also replace init_get_bits with init_get_bits8.
Signed-off-by: Peter Ross <pross@xvid.org>
Do this by allocating AVBSFContext together with the data that is
currently in AVBSFInternal; or rather: Put AVBSFContext at the beginning
of a new structure called FFBSFContext (which encompasses more than just
the internal fields and is a proper context in its own right, hence the
name) and remove the AVBSFInternal altogether.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This can be achieved by allocating the AVIOContext and
the dynamic buffer's opaque and internal write buffer together.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently AVIOContext's private fields are all over AVIOContext.
This commit moves them into a new structure in avio_internal.h instead.
Said structure contains the public AVIOContext as its first element
in order to avoid having to allocate a separate AVIOContextInternal
which is costly for those use cases where one just wants to access
an already existing buffer via the AVIOContext-API.
For these cases ffio_init_context() can't fail and always returned zero,
which was typically not checked. Therefore it has been made to not
return anything.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is the more natural place for it given that it only deals with I/O;
in fact, the function already has the ffio prefix and its declaration
already is in avio_internal.h.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This unbreaks the fate-checkasm-hevc_pel test on arm targets.
The assembly assumed that the width passed to the DSP functions is
a multiple of 8, while the checkasm test used other widths too.
This wasn't noticed before, because the hevc_pel checkasm tests
(that were added in 9c513edb79 in
January) weren't run as part of fate until in
b492cacffd in August.
As this hasn't been an issue in practice with actual full decoding
tests, it seems like the actual decoder doesn't call these functions
with such widths. Therefore, we could alternatively fix the test
to only test things that the real decoder does, and this modification
could be reverted.
Signed-off-by: Martin Storsjö <martin@martin.st>
subtitles.mak's fate-sub tests utilize a more strict comparator
("rawdiff"), which causes the tests fail in case of white space
differences, such as CRLF vs LF. This in turn causes these
ffprobe-using TTML-in-MP4 tests to fail on non-LF systems such as
Windows or wine.
Includes basic support for both the ISMV ('dfxp') and MP4 ('stpp')
methods. This initial version also foregoes fragmentation support
in case the built-in sample squashing is to be utilized, as this
eases the initial review.
Additionally, add basic tests for both muxing modes in MP4.
Signed-off-by: Jan Ekström <jan.ekstrom@24i.com>
If slice_type is > 9, the access to ff_h264_golomb_to_pict_type is
out-of-bounds. Fix this by simply setting the slice_type to 0 in this
case.
This is completely inconsequential because the value is only being used
to being used as an offset in the calculation of the film grain seed
value, a corruption of which is practically invisible.
Fixes coverity ticket #1490802
Signed-off-by: James Almer <jamrial@gmail.com>
Some frame threaded decoders set it, but this information never reached the
caller in frame threading scenarios.
Signed-off-by: James Almer <jamrial@gmail.com>
Because we need access to ref frames without film grain applied, we have
to add an extra AVFrame to H264Picture to avoid messing with the
original. This requires some amount of overhead to make the reference
moves work out, but it allows us to benefit from frame multithreading
for film grain application "for free".
Unfortunately, this approach requires twice as much RAM to be constantly
allocated for ref frames, due to the need for an extra buffer per
H264Picture. In theory, we could get away with freeing up this memory as
soon as it's no longer needed (since ref frames do not need film grain
buffers any longer), but trying to call ff_thread_release_buffer() from
output_frame() conflicts with possible later accesses to that same frame
and I'm not sure how to synchronize that well.
Tested on all three cases of (no fg), (fg present but exported) and (fg
present and not exported), with and without threading.
Co-authored-by: James Almer <jamrial@gmail.com>
Signed-off-by: Niklas Haas <git@haasn.dev>
Signed-off-by: James Almer <jamrial@gmail.com>
This could arguably also be a vf, but I decided to put it here since
decoders are technically required to apply film grain during the output
step, and I would rather want to avoid requiring users insert the
correct film grain synthesis filter on their own.
The code, while in C, is written in a way that unrolls/vectorizes fairly
well under -O3, and is reasonably cache friendly. On my CPU, a single
thread pushes about 400 FPS at 1080p.
Apart from hand-written assembly, one possible avenue of improvement
would be to change the access order to compute the grain row-by-row
rather than in 8x8 blocks. This requires some redundant PRNG calls, but
would make the algorithm more cache-oblivious.
The implementation has been written to the wording of SMPTE RDD 5-2006
as faithfully as I can manage. However, apart from passing a visual
inspection, no guarantee of correctness can be made due to the lack of
any publicly available reference implementation against which to
compare it.
Signed-off-by: Niklas Haas <git@haasn.dev>
Signed-off-by: James Almer <jamrial@gmail.com>
From SMPTE RDD 5-2006, the grain seed is to be computed from the
following definition of `pic_offset`:
> When decoding H.264 | MPEG-4 AVC bitstreams, pic_offset is defined as
> follows:
> - pic_offset = PicOrderCnt(CurrPic) + (PicOrderCnt_offset << 5)
> where:
> - PicOrderCnt(CurrPic) is the picture order count of the current frame,
> which shall be derived from [the video stream].
>
> - PicOrderCnt_offset is set to idr_pic_id on IDR frames. idr_pic_id
> shall be read from the slice header of [the video stream]. On non-IDR I
> frames, PicOrderCnt_offset is set to 0. A frame shall be classified as I
> frame when all its slices are I slices, which may be optionally
> designated by setting primary_pic_type to 0 in the access delimiter NAL
> unit. Otherwise, PicOrderCnt_offset it not changed. PicOrderCnt_offset is
> updated in decoding order.
Co-authored-by: James Almer <jamrial@gmail.com>
Signed-off-by: Niklas Haas <git@haasn.dev>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, the Matroska muxer did not use the dispositions it is
given as-is; instead it by default overrode the disposition of the first
track of a kind (audio, video, subtitles) if no track of this kind has
the default disposition set. And up until recently, it also enforced
by default that no more than one track of each kind be marked as
default.
The rationale for the former is that there are lots of containers which
lack the concept of default streams, so that it is not uncommon for no
stream to be marked as default at all; the rationale for the latter was
that up until recently, it was dubious whether the Matroska specification
allowed more than one default stream for track type (e.g. mkvmerge
disallowed it). It was this point which led to the implementation of
the above mentioned behaviour inspired by mkvmerge.
Yet the Matroska specifications have changed and now explicitly allow
to set more than one track of each type as default, so that the main
reason of not using the dispositions as-is was rendered moot. Therefore
this commit changes the default to pass the disposition through.
The matroska-mpegts-remux FATE-test has been updated to still use the
old "infer" mode so that it is still covered by FATE; the
matroska-zero-length-block test has also been updated to cover
the infer_no_subs mode. The references for lots of other FATE tests
needed to be updated because of a newly added FlagDefault element with
value zero (whereas a FlagDefault with value 1 needn't be coded at all,
as it coincided with the default value of said element).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The Matroska specifications have evolved and now allow to mark
multiple tracks of the same kind as default (whether this was legal or
not before was dubious; e.g. mkvmerge disallowed it). Yet when the
Matroska muxer is set to infer default dispositions if absent, it also
enforced the now outdated restriction. So update this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
this prevents some mismatches in config values for realtime and all
intra modes, avoiding failures like:
[libaom-av1 @ ...] Failed to initialize encoder: Invalid parameter
[libaom-av1 @ ...] Additional information: g_lag_in_frames out of
range [..0]
Signed-off-by: James Zern <jzern@google.com>
The low overhead OBU format provides no means to resync after performing
a byte-based seek; in other words: Byte based seeking is just not
supported.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The av1_merge_frame BSF outputs its cached data when it sees the
beginning of a new frame, i.e. when it sees a temporal delimiter OBU.
Therefore it typically has a temporal delimiter OBU cached after
outputting a packet.
This implies that the OBU demuxer must flush its BSF upon seeking
because otherwise the first frame returned after a seek consists
of an old temporal delimiter OBU only.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It allows demuxers to perform certain tasks after
a successful generic seek.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This can be enabled/disabled on a per-pad basis by setting
the AVFILTERPAD_FLAG_FREE_NAME flag; variants of ff_append_(in|out)pads
that do this for you have been added and will be put to use in the
following commits.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The MOV muxer can store streamids as track ids but they aren't
visible when probing the result via lavf/dump or ffprobe due to
lack of this flag in the demuxer.
transpose_4x8H was declared in vp9lpf_16bpp_neon, however this macro is
not unique to vp9 and could be used elsewhere.
Signed-off-by: Mikhail Nitenko <mnitenko@gmail.com>
Signed-off-by: Martin Storsjö <martin@martin.st>
These checks emit warnings in case the channel layouts lists are
inconsistent; yet since 69f5f6ea37
a function that is called earlier errors out if they are inconsistent.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This currently happens by accident in a few filters that use
ff_set_common_(samplerates|channel_layouts) like afir (if the response
option is set) or agraphmonitor (due to the default code in
avfiltergraph.c). So change those functions to make sure it does no
longer happen.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The ff_set_common_(formats|channel_layouts|samplerates) have to free
their list in case it doesn't have an owner; therefore they tracked
whether they attached it to an owner. But the list's refcount already
contains such a counter, so we don't have to keep track of whether we
have attached the list to an owner.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is unnecessary as the number of static inputs and outputs can now
be directly read via AVFilter.nb_(in|out)puts.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is intended as replacement for avfilter_pad_count(). In contrast to
the latter, it avoids a loop.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, an AVFilter's lists of input and output AVFilterPads
were terminated by a sentinel and the only way to get the length
of these lists was by using avfilter_pad_count(). This has two
drawbacks: first, sizeof(AVFilterPad) is not negligible
(i.e. 64B on 64bit systems); second, getting the size involves
a function call instead of just reading the data.
This commit therefore changes this. The sentinels are removed and new
private fields nb_inputs and nb_outputs are added to AVFilter that
contain the number of elements of the respective AVFilterPad array.
Given that AVFilter.(in|out)puts are the only arrays of zero-terminated
AVFilterPads an API user has access to (AVFilterContext.(in|out)put_pads
are not zero-terminated and they already have a size field) the argument
to avfilter_pad_count() is always one of these lists, so it just has to
find the filter the list belongs to and read said number. This is slower
than before, but a replacement function that just reads the internal numbers
that users are expected to switch to will be added soon; and furthermore,
avfilter_pad_count() is probably never called in hot loops anyway.
This saves about 49KiB from the binary; notice that these sentinels are
not in .bss despite being zeroed: they are in .data.rel.ro due to the
non-sentinels.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Besides being nicer code this also has the advantage of not making
assumptions about the internal implementation: While it is documented
that the AVFilter.inputs and AVFilter.outputs arrays are terminated
by a zeroed sentinel, one is not allowed to infer that one can just
check avfilter_pad_get_name(padarray, i) to see whether one has reached
the sentinel:
It could be that the pointer to the string is contained
in a different structure than AVFilterPad that needs to be accessed
first: return pad->struct->string.
It could be that for small strings an internal buffer in
AVFilterPad is used (to avoid a relocation) whereas for longer strings
an external string is used; this is useful to avoid relocations:
return pad->string_ptr ? pad->string_ptr : pad->interal_string
Or it could be that the name has a default value:
return pad->name ? pad->name : "default"
(This actually made sense for us because the name of most of our
AVFilterPads is just "default"; doing so would save lots of relocations.)
The only thing one is allowed to infer from the existence of the
sentinel is that one is allowed to use avfilter_pad_count() to get
the number of pads. Therefore it is used.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Use an uint32_t for the NAL unit size of an AVC H.264 NAL unit instead
of an int as a left shift of a signed value is undefined behaviour
if the result doesn't fit into the target type.
Also make the log message never output negative lengths.
Fixes: left shift of 16711968 by 8 places cannot be represented in type 'int'
Fixes: 36601/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_H264_fuzzer-6581933285965824
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
because subtitles streams will be written to webvtt m3u8 list
so the stream index should minus subtitles streams count when subtitle
between audio and video streams.
testcase:
before patch:
ffmpeg -i input -map 0:a:0 -map 0:s:0 -map 0:v:0 -f hls aaaa.m3u8
will EXC_BAD_ACCESS
after patch:
ffmpeg -i input -map 0:a:0 -map 0:s:0 -map 0:v:0 -f hls aaaa.m3u8
will ok
Signed-off-by: Steven Liu <liuqi05@kuaishou.com>
"frag_start" is redundant, and every occurance can be replaced with cluster[0].dts - start_dts
The proof of no behaviour changes: (All line number below is based on commit bff7d662d7)
"frag_start" is read at 4 place (with all possible call stacks):
mov_write_packet
...
mov_flush_fragment
mov_write_moof_tag
mov_write_moof_tag_internal
mov_write_traf_tag
mov_write_tfxd_tag (#1)
mov_write_tfdt_tag (#2)
mov_add_tfra_entries (#3)
mov_write_sidx_tags
mov_write_sidx_tag (#4)
mov_write_trailer
mov_auto_flush_fragment
mov_flush_fragment
... (#1#2#3#4)
mov_write_sidx_tags
mov_write_sidx_tag (#4)
shift_data
compute_sidx_size
get_sidx_size
mov_write_sidx_tags
mov_write_sidx_tag (#4)
All read happens in "mov_write_trailer" and "mov_write_moof_tag". So we need to prove no behaviour change in these two
functions.
Condition 1: for every track that have "trk->entry == 0", trk->frag_start == trk->track_duration.
Condition 2: for every track that have "trk->entry > 0", trk->frag_start == trk->cluster[0].dts - trk->start_dts.
Definition 1: "Before flush" means just before the invocation of "mov_flush_fragment", except for the auto-flush case in
"mov_write_single_packet", which means before L5934.
Lemma 1: If Condition 1 & 2 is true before flush, Condition 1 & 2 is still true after "mov_flush_fragment" returns.
Proof:
No update to the tracks that have "trk->entry == 0" before flushing, so we only consider tracks that have "trk->entry > 0":
Case 1: !moov_written and moov will be written in this iteration
trk->entry = 0 L5366
trk->frag_start == trk->cluster[0].dts - trk->start_dts Lemma condition
trk->frag_start += trk->start_dts + trk->track_duration - trk->cluster[0].dts; L5363
So trk->entry == 0 && trk->frag_start == trk->track_duration
Case 2: !moov_written and moov will NOT be written in this iteration
nothing changed
Case 3: moov_written
trk->entry = 0 L5445
trk->frag_start == trk->cluster[0].dts - trk->start_dts Lemma condition
trk->frag_start += trk->start_dts + trk->track_duration - trk->cluster[0].dts; L5444
So trk->entry == 0 && trk->frag_start == trk->track_duration
Note that trk->track_duration may be updated for the tracks that have "trk->entry > 0" (mov_write_moov_tag will
update track_duration of "tmcd" track, but it must have 1 entry). But in all case, trk->frag_start is also updated
to consider the new value.
Lemma 2: If Condition 1 & 2 is true before "ff_mov_write_packet" invocation, Condition 1 & 2 is still true after it returns.
Proof:
Only the track corresponding to the pkt is updated, and no update to relevant variables if trk->entry > 0 before invocation.
So we only need to prove "trk->frag_start == trk->cluster[0].dts - trk->start_dts" after trk->entry increase from 0 to 1.
Case 1: trk->start_dts == AV_NOPTS_VALUE
Case 1.1: trk->frag_discont && use_editlist
trk->cluster[0].dts = pkt->dts at L5741
trk->frag_start = pkt->pts at L5785
trk->start_dts = pkt->dts - pkt->pts at L5786
So trk->frag_start == trk->cluster[0].dts - trk->start_dts
Case 1.2: trk->frag_discont && !use_editlist
trk->cluster[0].dts = pkt->dts at L5741
trk->frag_start = pkt->dts at L5790
trk->start_dts = 0 at L5791
So trk->frag_start == trk->cluster[0].dts - trk->start_dts
Case 1.3: !trk->frag_discont
trk->cluster[0].dts = pkt->dts at L5741
trk->frag_start = 0 init
trk->start_dts = pkt->dts at L5779
So trk->frag_start == trk->cluster[0].dts - trk->start_dts
Case 2: trk->start_dts != AV_NOPTS_VALUE
Case 2.1: trk->frag_discont
trk->cluster[0].dts = pkt->dts at L5741
trk->frag_start = pkt->dts - trk->start_dts at L5763
So trk->frag_start == trk->cluster[0].dts - trk->start_dts
Case 2.2: !trk->frag_discont
trk->cluster[0].dts = trk->start_dts + trk->track_duration at L5749
trk->track_duration == trk->frag_start Lemma condition
So trk->frag_start == trk->cluster[0].dts - trk->start_dts
Lemma 3: Condition 1 & 2 is true in all case before and after "ff_mov_write_packet" invocation, before flush and after
"mov_flush_fragment" returns.
Proof: All updates to relevant variable happen either in "ff_mov_write_packet", or during flush. And Condition 1 & 2
is true initially. So with lemma 1 & 2, we can prove this use induction.
Noticed that all read of "frag_start" only happen in "trk->entry > 0" branch. Now we need to prove Condition 2 is true
before each read.
Because no update to variables relevant to Condition 2 between "before flush" and "mov_write_moof_tag" invocation, we
can conclude Condition 2 is true before every invocation of "mov_write_moof_tag". No behaviour change in
"mov_write_moof_tag" is proved.
In "mov_write_trailer", No update to relevant variables after the last flush and before the invocation of
"mov_write_sidx_tag". So no behaviour change to "mov_write_trailer" is proved.
Q.E.D.
Signed-off-by: Hu Weiwen <sehuww@mail.scut.edu.cn>
Signed-off-by: Martin Storsjö <martin@martin.st>
track->mdat_buf can be not NULL while the track is still empty if the
last packet write failed.
Signed-off-by: Hu Weiwen <sehuww@mail.scut.edu.cn>
Signed-off-by: Martin Storsjö <martin@martin.st>
Unlike libx264, libx265 does not have a separate "unspecified"/"auto"
default for color range, so we do always have to specify it.
Thus, we are required to handle the RGB case on the libavcodec
side to enable the correct value to be written out in in case
of RGB content with unspecified color range being received.
In other words:
1. If the user has set color range specifically, follow that.
2. If the user has not set color range specifically, set full
range by default in case of RGB and YUVJ pixel formats.
By default the x264 full range flag is set to -1. By not setting
it to something else, we can let libx264 handle the RGB case.
Additionally, change the preference order to user-specified range
first, and then any fall-back logic left for the YUVJ pix_fmts.
Fixes the capture part of #9374
These fields are mutually exclusive, so putting them in a union
is possible and makes AVFilterPad smaller.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In particular, document that av_opt_copy() always disentangles
allocated options even on error; this guarantee is needed to e.g.
properly free duplicated thread contexts in libavcodec on error.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently user may use '-init_hw_device type=name' to initialize a hw
device, however the key parameter is ignored when use '-init_hw_device
type=name,key=value'. After applying this patch, user may set key
parameter if needed.
Reviewed-by: Soft Works <softworkz@hotmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
The last init_opaque callback has been removed in commit
07ffdedf784e86b88074d8d3e08e55752869562a; the opaque argument has been
always NULL since 0acf7e268b.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The code for inserting inpads can't be reached by ff_vsrc_openclsrc
(unsurprising given that it is a source filter), so it didn't get
the flag.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
AVFrame.metadata is always owned by its AVFrame, it is not shared
in the first place, so one does not need to make the frame writable
to modify it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The current way of doing it involves writing the ctx parameter twice.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The current code reads the wrong number of bits for `fg_model_id`, which
causes all of the values downstream of this to contain corrupt values.
Fixes: corrupt SEI values
Fixes: 4ff73add5d
Signed-off-by: Niklas Haas <git@haasn.dev>
Certain mov/mp4 files have parameter sets out of band, and when required for a
sample it may be propagated within the relevant packet's side data.
This fixes parsing said files if the SPS and/or PPS in the side data is
different than the one in extradata.
Signed-off-by: James Almer <jamrial@gmail.com>
Using ff_cbs_read() on the raw buffer will not parse it as extradata,
resulting in parsing errors for example when handling ISOBMFF avcC.
This helper works around that.
Signed-off-by: James Almer <jamrial@gmail.com>
In 1c42fd9323 the ipcm identifier was
added in order to demux additional raw audio from Sony MP4 files.
Unfortunately, it was not noticed that this same list is utilized
for muxing as well, thus causing ipcm to get preferred compared
to the identifier officially specified in QTFF documentation.
This fixes the order of preference for 24bit PCM, where ipcm is
still allowed, but in24 is the first match - thus being preferred.
Fixes fate-acodec-pcm-s24be.
Partition struct may be reallocated, so let's store the score directly in order
to avoid use-after-free.
Also mxf->current_partition might be null when reading some local tags.
Signed-off-by: Marton Balint <cus@passwd.hu>
The reason why the generic av_image_copy_uc_from() doesn't really
fit in the case for Vulkan is because some planes may be copied via
other methods (such as mapping GPU memory), and if they don't satisfy
the strict alignment requirements, a gpu image->gpu buffer->cpu ram
copy is performed.
We need this for hwcontext_vulkan, and I think this will also be
useful to API users like libplacebo who would rather not write
a custom SIMD memcpy.
Several combinations of functions happen quite often in query_format
functions; e.g. ff_set_common_formats(ctx, ff_make_format_list(sample_fmts))
is very common. This commit therefore adds functions that are equivalent
to commonly used function combinations in order to reduce code
duplication.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The SDK supports LowPower and non-LowPower modes, but some features are
available only under one of the two modes. Currently non-LowPower mode
is always chosen in FFmpeg if the mode is not set to LowPower
explicitly. User will experience some SDK errors if a LowPower related
feature is specified but the mode is not set to LowPower. With this
patch, the mode is set to unknown by default in FFmpeg, the SDK is able
to choose a workable mode for the specified features.
Reviewed-by: Soft Works <softworkz@hotmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Since 580e168a94, av_size_mult() is no
longer inlined; on systems where interposing is a thing, this also
inhibits the compiler from inlining said function into the internal
callers of said function, although inlining such a small function is
typically beneficial: With GCC 10.3 on Ubuntu x64 and -O3 this decreases
the size of av_realloc_array from 91B to 23B, from 129B to 81B for
av_realloc_f and from 77B to 23B for each of av_malloc_array,
av_mallocz_array and av_calloc.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This allows user set hw_device_ctx instead of hw_frames_ctx for QSV
decoders, hence we may remove the ad-hoc libmfx setup code from FFmpeg.
"-hwaccel_output_format format" is applied to QSV decoders after
removing the ad-hoc libmfx code. In order to keep compatibility with old
commandlines, the default format is set to AV_PIX_FMT_QSV, but this
behavior will be removed in the future. Please set "-hwaccel_output_format qsv"
explicitly if AV_PIX_FMT_QSV is expected.
The normal device stuff works for QSV decoders now, user may use
"-init_hw_device args" to initialise device and "-hwaccel_device
devicename" to select a device for QSV decoders.
"-qsv_device device" which was added for workarounding device selection
in the ad-hoc libmfx code still works
For example:
$> ffmpeg -init_hw_device qsv=qsv:hw_any,child_device=/dev/dri/card0
-hwaccel qsv -c:v h264_qsv -i input.h264 -f null -
/dev/dri/renderD128 is actually open for h264_qsv decoder in the above
command without this patch. After applying this patch, /dev/dri/card0
is used.
$> ffmpeg -init_hw_device vaapi=va:/dev/dri/card0 -init_hw_device
qsv=hw@va -hwaccel_device hw -hwaccel qsv -c:v h264_qsv -i input.h264
-f null -
device hw of type qsv is not usable in the above command without this
patch. After applying this patch, this command works as expected.
Reviewed-by: Soft Works <softworkz@hotmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Will remove unnecessary allocations when both src and dst picture contain
references to the same buffers.
Signed-off-by: James Almer <jamrial@gmail.com>
If a slice header fails to parse, and the next one uses different Sequence and
Picture parameter sets, certain values may not be read if they are not coded,
resulting in the previous slice values being used.
Signed-off-by: James Almer <jamrial@gmail.com>
The existing error concealment makes no sense for the image formats, they
use transformed source images which is different from keyframe + MC+difference
for which the error concealment is designed.
Of course feel free to re-enable this if you have a case where it works and
improves vissual results
Fixes: Timeout
Fixes: 36234/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_VC1IMAGE_fuzzer-6300306743885824
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: negation of -2147483648 cannot be represented in type 'int'; cast to an unsigned type to negate this value to itself
Fixes: 35593/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AAC_FIXED_fuzzer-5182217725804544
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This commit adds the case handling if the asynchronous execution
of a request fails by checking the exit status of the thread when
joining before starting another execution. On failure, it does the
cleanup as well.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
The frame allocation and filling the TaskItem with execution
parameters is common in the three backends. This commit shifts
this logic to dnn_backend_common.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
Since requests are running in parallel, there is inconsistency in
the status of the execution. To resolve it, we avoid using mutex
as it would result in single TF_Session running at a time. So add
TF_Status to the TFRequestItem
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This patch adds error handling for cases where the execute_model_tf
fails, clears the used memory in the TFRequestItem and finally pushes
it back to the request queue.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit enables async execution in the TensorFlow backend
and adds function to flush extra frames.
The async execution mechanism executes the TFInferRequests on
a separate thread which is joined before the next execution of
same TFRequestItem/while freeing the model.
The following is the comparison of this mechanism with the existing
sync mechanism on TensorFlow C API 2.5 CPU variant.
Async Mode: 4m32.846s
Sync Mode: 5m17.582s
The above was performed on super resolution filter using SRCNN model.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit adds a function for execution of TFInferRequest and documentation
for functions related to TFInferRequest.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit adds an async execution mechanism for common use
in the TensorFlow and Native backends.
This commit also adds the documentation of typedefs and functions in
the async module for common use in DNN backends.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
After standardizing the use of 'pxor' in commit 'ebedd26', FFmpeg
build failed with upstream compiler, for 'pxor' is not supported
in time. This patch helps to workaround the build failure by
checking whether 'pxor' is supported during configuration, if not,
MMI will be disabled.
Reviewed-by: yinshiyou-hf@loongson.cn
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If autoflushing on a new packet (e.g. due to the frag_every_frame
flag being set), there's no samples to be written in the new fragment,
so we can't overwrite the track duration in order to make it line
up with the next packet to be written.
Signed-off-by: Martin Storsjö <martin@martin.st>
The function ff_qsvvpp_filter_frame should return a FFmpeg error code if
there is an error. However it might return a SDK error code without this
patch.
Reviewed-by: Soft Works <softworkz@hotmail.com>
The flite filter apparently only wanted to declare a struct,
but mistakenly also defined an unused and zero-initialized element
with external linkage.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
hybrid2_re() has a parameter declared as "const INTFLOAT filter[8]",
although the actual argument for said parameter only has seven elements;
the code itself only uses seven elements, so change the parameter.
Fixes a -Wstringop-overread warning with GCC 11.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The function definition used float *data_buf[14], although there are
only 13 elements (and only 13 are used); the declaration used 13.
Given that the type will be converted to float **data_buf anyway,
this is not in violation of the C specs, but nevertheless a bug.
GCC 11 has a new warning for this -Warray-parameter.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Before 3793caa5e2 the code was
"if (...) do { ... } while (...);". After said commit this became
"if (...) av_assert0(...); do { ... } while (...);", i.e. the loop
is always executed. This commit changes the logic to what it was before
said commit. Notice that the condition is always true in FATE, so no
changes are necessary there.
This fixes a -Wmisleading-indentation warning from GCC 11.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In some extrme cases, like with adpcm_ms samples with an extremely high channel
count, get_audio_frame_duration() may return a negative frame duration value.
Don't propagate it, and instead return 0, signaling that a duration could not
be determined.
Fixes ticket #9312
Signed-off-by: James Almer <jamrial@gmail.com>
av_dict_copy() puts the onus on the caller to clean up dst on failure;
it can be nonempty if copying a later entry of src fails after having
successfully copied an earlier entry.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
As the second argument for init_get_bits (buf) can be crafted, a return value check for this function call is necessary.
'buf' is part of 'AVPacket pkt'.
replace init_get_bits with init_get_bits8.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 9223372036854775791 + 18 cannot be represented in type 'long'
Fixes: 36307/clusterfuzz-testcase-minimized-ffmpeg_dem_MPC8_fuzzer-4917863877050368
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -9223372036854775808 * 2 cannot be represented in type 'long long'
Fixes: 36244/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EXR_fuzzer-6090656186499072
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
677a030b26 introduced more printable
side data types in ffprobe, however the Audio Service Type side data
'type' field that was introduced aliases an existing field of the same
name within the side data array, which can lead to JSON output like:
"side_data_list": [
{
"side_data_type": "Audio Service Type",
"type": 0
},
{
"side_data_type": "Stereo 3D",
"type": "side by side",
"inverted": 1
}
]
This, while technically valid JSON, is considered bad practice, since it
forces all downstream users to manually parse it and check all types;
it makes simple deserialization impossible. Worse, in som loosely
type languages, it can lead to silent bugs if exising code assumed
it was a different type.
As such, rename this second "type" field to "service_type".
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
The reasons for including them don't exist any longer: ff_tlog() has
been moved to libavutil/internal.h and FF_QSCALE_TYPE_* has been moved
to qp_table.h.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is also used by libavfilter and it is only natural to define it
alongside ff_dlog().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Said AVCodecContext is only used for logging; it furthermore avoids
an avcodec.h inclusion.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Since the removal of the 16-bit FFT said define is unnecessary as
FFT_FIXED_32 is always !FFT_FLOAT. But one wouldn't believe it when
looking at the code.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
8b83dad825 added another potentially used
video enhancement filter without increasing a define for the number of
such options which is used as the size of stack array. This can lead to
a buffer overrun if all filters are used simultaneously. So increase
said number.
Fixes Coverity ticket #1489775.
Reviewed-by: Linjie Fu <linjie.justin.fu@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
They are not used by the header at all and only used by very few files;
so include the headers in their users instead of in internal.h.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Similar to CVE-2013-0868, here return value check for 'init_vlc' is needed.
crafted DNxHD data can cause unspecified impact.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
MISB ST 0604 and ST 2101 require user data unregistered SEI messages
(precision timestamps and sensor identifiers) to be included. That
currently isn't supported for libx265. This patch adds support
for user data unregistered SEI messages in accordance with
ISO/IEC 23008-2:2020 Section D.2.7
The design is based on nvenc, with support finished up at
57de80673c
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
11d3b03fcb added consideration of default stream disposition for audio
and video when choosing the 'best' stream among all the inputs. This can
lead to video streams with lower resolution or audio streams with fewer
channels being selected.
Stream disposition, however, only sets a priority for a stream
among all other streams in the *same input*. It cannot set a priority
for a stream across all inputs.
This patch sets a middle-way and selects the best stream from each file
with default disposition considered. Then it discards disposition weight
and selects best stream as per the original criteria of highest
resolution for video and most channels for audio.
Fixes: signed integer overflow: 9223372036854775807 + 86400000000 cannot be represented in type 'long'
Fixes: 29102/clusterfuzz-testcase-minimized-ffmpeg_dem_SBG_fuzzer-6731040263634944
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 9223372033098784808 + 4294967072 cannot be represented in type 'long'
Fixes: 29102/clusterfuzz-testcase-minimized-ffmpeg_dem_AVI_fuzzer-6732488912273408
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
For JPEG 2000 essence, the MXF input format module currently uses the value of
byte 14 of the essence container UL to determine whether the J2K essence is
clip- (byte 14 is 0x02) or frame-wrapped (byte 14 is 0x01). Otherwise it
assumes an unknown wrapping.
Additional wrappings are documented in SMPTE ST422:2019:
0x03: Interlaced Frame, 1 field/KLV
0x04: Interlaced Frame, 2 fields/KLV
0x05: Field-wrapped Picture Element
0x06: Frame-wrapped Picture Element
And these should also be handled as frame wrapped content.
Signed-off-by: Pierre-Anthony Lemieux <pal@sandflow.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Instead only include libavutil/version.h; including avutil.h is a
remnant from the time in which the version was in it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, including error.h alone does not make the AVERROR_* defines
usable, because they just expand to something involving MKTAG, but
without the header providing MKTAG. So include macros.h, the header
providing MKTAG.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
common.h currently contains several things: Math macros, UTF-8 macros,
other fundamental macros; furthermore it also contains miscellaneous
math functions and it (directly and indirectly) includes lots of other
headers.
This commit moves the "other fundamental macros" to macros.h which is
a more fitting place.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It only uses an AVIOContext and an AVBPrint.
When doing so, it turned out that several non-users of
ff_read_line_to_bprint_overwrite() and ff_bprint_to_codecpar_extradata()
relied on libavformat/internal.h to include bprint.h or avstring.h
for them. In order to avoid a repeat of this and in order to reduce
unnecessary dependencies, a forward declaration of struct AVBPrint is
used instead of including bprint.h.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Some function had exceed 30 inline assembly register oprands limiation
when using LOONGSON2 version of MMI macros. We can avoid that by take
$at, which is register reserved for assembler, as temporary register.
As none of instructions used in these macros is pseudo, it is safe to
utilize $at here.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Loongson3's extention instructions (prefixed with gs) are widely used
in our MMI codebase. However, these instructions are not avilable on
Loongson-2E/F while MMI code should work on these processors.
Previously we introduced mmiutils marcos to provide backward compactbility
but newly commited code didn't follow that. In this patch I revised the
codebase and converted all these instructions into MMI marcos to get
Loongson2 supproted again.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
9888ffb1ce added checks for EOF
in loops in the mov demuxer as a precaution against timeouts;
yet there is no I/O in the loop when parsing the STSZ atom
as the values are read from an already read buffer. So remove said
checks.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
mov_read_stsz() did not ensure that every bit of a buffer is addressable
by an int as is required by the get_bits API, leading to a crash in
ticket #9344. Fix this by restricting the size more thoroughly.
The file from said ticket will then be considered invalid; in the
future, we might read and process the data in chunks to actually support
such files.
Fixes ticket #9344.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: signed integer overflow: 9223372036854775807 + 1442840321 cannot be represented in type 'long'
Fixes: 33670/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-6644379491106816
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The accepted values for GopOptFlag are MFX_GOP_CLOSED (1) and
MFX_GOP_STRICT (2).
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
Signed-off-by: Zhong Li <zhongli_dev@126.com>
Don't attempt to increase the cursor pointer if it was \0.
Fixes invalid reads.
Reviewed-by: Marton Balint <cus@passwd.hu>
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: James Almer <jamrial@gmail.com>
Having the override before autodetection meant that the overridden
value got overwritten by the autodetected result each time,
effectively disabling the ability to utilize the `-top` option
for override purposes.
Somehow I missed this in fbb44bc51a ,
even though the lines were within the context. Probably the code
originally being after this logic had something to do with it,
but previously it only touched the avformat context's codecpar,
which did not affect the encoder codec context whatsoever.
Fixes#9320Fixes#9339
The magic constants come from the unofficial "ITU-R BS.1770-1 filter
specifications"¹ by Raiden (libebur128) which relies on "Parameter
Quantization in Direct-Form Recursive Audio Filters"² by Brian
Neunaber.
The constants seem to include a quantization bias, for example:
- Vb is supposed to be exactly √Vh in a high shelf filter
- the Pre-filter Gain should likely be 4dB
- Pre Q and RLB Q are respectively very close to √½ and ½
Those are not adjusted to prevent the values from drifting away from
the official specifications.
An alternative to this approach would be to requantize on the fly as
proposed by pbelkner³, where the 48kHz code path would use the exact
specifications constants while derivating constants for other
frequencies.
[1]: https://www.scribd.com/document/49991813/ITU-R-BS-1770-1-filters
[2]: https://www.scribd.com/document/6531763/Direct-Form-Filter-Parameter-Quantization
[3]: https://hydrogenaud.io/index.php?topic=86116.msg740092#msg740092
Use avfilter_graph_alloc() instead of av_mallocz(sizeof(AVFilterGraph))
to allocate an AVFilterGraph; this also properly allocates the graph's
internal. The current code just happened to work because it did not
make any use of said internal.
Also check the allocation; this fixes Coverity #1292528.
Reviewed-by: Jan Ekström <jeebjp@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The code savings more than offset the size of the table
(1936B vs 768B with GCC 10.3 at -O3).
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This has been broken in 25c8507818,
because the hacks for headers that are incompatible with building
for the host in libavcodec/tableprint_vlc.h have not been adjusted.
Moving AV_INPUT_BUFFER_PADDING_SIZE to defs.h which is valid for
both the target as well as the host allowed to remove some of the hacks.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: signed integer overflow: 8511838621821575200 - -3954125146725285889 cannot be represented in type 'long'
Fixes: 33414/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-6610119325515776
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
These have mostly been added because of FF_API_*; yet when these were
removed, removing the header has been forgotten.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
These inclusions are not necessary, as cpu.h is already included
wherever it is needed (via direct inclusion or via the arch-specific
headers).
Also remove other unnecessary cpu.h inclusions from ordinary
non-headers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is not used here at all; instead, add it where it is used without
including it or any of the arch-specific CPU headers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This inclusion has been added before libavutil/error.h was split off
from avcodec.h (in 60c144f700).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
According to the header it is an array of int16_t, yet it is declared as
uint16_t. Fix this by using int16_t troughout and convert the definition
to use values in the range of int16_t.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This change ensures that the linker can drop adpcm_data.o if no decoder
that actually uses anything from there is enabled.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This is to avoid unused variables warnings after the code for
the disabled encoders has been #if'ed away which will happen in
a subsequent commit.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The adpcm_argo encoder does not use the data from adpcm_data.c directly;
instead it shares a function with the adpcm_argo decoder that is in
adpcm.c. When all the ADPCM decoders and the adpcm_argo encoder are
disabled, adpcm.c is not compiled; yet the code in adpcmenc.c calling
said function from adpcm.c is still present, leading to link errors.
Fix this by disabling the code belonging to disabled codecs in
adpcmenc.c.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This is to avoid unused variables warnings if the code for disabled
encoders is #if'ed away which will happen in a subsequent commit.
In case of buf it also avoids shadowing.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In cases where the execution inside the function execute_model_ov fails,
the OVRequestItem must be pushed back to the request_queue before returning
the error. In case pushing back fails, release the allocated memory.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
Fixes: floating point division by 0
Fixes: -nan is outside the range of representable values of type 'int'
Fixes: Ticket8307
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: 1.04064e+10 is outside the range of representable values of type 'int'
Fixes: Ticket 8279
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Read rate enforcement delayed till first decoded frame is obtained, to
speed up init of output streams.
Thanks to Linjie Fu <linjie.justin.fu@gmail.com> for the initial patch.
Since b492fbcc6e, the DSD tables are
always initialized at runtime, so merge the dsd_tablegen.h header
into dsd.c.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It does not modify anything; it only returns a value, so it fulfills
the requirements for av_pure.
The deeper rationale behind this change is that this function is called
quite often inside arguments to FFMIN which may lead to two calls to it;
declaring this function as av_pure allows the compiler to optimize the
second call away.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently every symbol (with external linkage) that starts with "av" is
exported. Yet libaom-av1 has lots of functions that are not meant to be
exported and start with "av1_" (I counted 1236); and libvpx has
average_split_mvs. These functions are exported if one links these
libraries statically into a shared libavcodec.so.
Solve this by tightening the whitelist to "av_", "avcodec_", "avpriv_"
and (as a special-case) "avsubtitle_free".
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Teach AV_HWDEVICE_TYPE_VIDEOTOOLBOX to be able to create AVFrames of type
AV_PIX_FMT_VIDEOTOOLBOX. This can be used to hwupload a regular AVFrame
into its CVPixelBuffer equivalent.
ffmpeg -init_hw_device videotoolbox -f lavfi -i color=black:640x480 -vf hwupload -c:v h264_videotoolbox -f null -y /dev/null
Signed-off-by: Aman Karmani <aman@tmm1.net>
The field is a standard field, yet we were loading it as if it was
a quadword. This worked for forward transforms by chance, but broke
when the transform was inverse.
checkasm couldn't catch that because we only test forward transforms,
which are identical to inverse transforms but with a different revtab.
if input start time is not 0 -t is inaccurate doing stream copy,
will record extra duration according to input start time.
it should base on following cases:
input video start time from 60s, duration is 300s,
1. stream copy:
ffmpeg -ss 40 -t 60 -i in.mp4 -c copy -y out.mp4
open_input_file() will seek to 100 and set ts_offset to -100,
process_input() will offset pkt->pts with ts_offset to make it 0,
so when do_streamcopy() with -t, exits when ist->pts >= recording_time.
2. stream copy with -copyts:
ffmpeg -ss 40 -t 60 -copyts -i in.mp4 -c copy -y out.mp4
open_input_file() will seek to 100 and set ts_offset to 0,
process_input() will keep raw pkt->pts as ts_offset is 0,
so when do_streamcopy() with -t, exits when
ist->pts >= (recording_time+f->start_time+f->ctx->start_time).
3. stream copy with -copyts -start_at_zero:
ffmpeg -ss 40 -t 60 -copyts -start_at_zero -i in.mp4 -c copy -y out.mp4
open_input_file() will seek to 120 and set ts_offset to -60 as start_to_zero option,
process_input() will offset pkt->pts with input file start time,
so when do_streamcopy() with -t, exits when ist->pts >= (recording_time+f->start_time).
0 60 40 60 360
|_______|_____|_______|_______________________|
start -ss -t
This fixes ticket #9141.
Signed-off-by: Shiwang.Xie <shiwang.xie666@outlook.com>
xmllint (silently) replaces the ' with " when fixing and validating the output
of ffprobe in fate-ffprobe_xsd.
Reviewed-by: Tobias Rapp <t.rapp@noa-archive.com>
Signed-off-by: James Almer <jamrial@gmail.com>
When streaming mode is enabled, the DASH manifest is written on the
first packet for the segment so that the segment can be advertised
immediately to clients. It was also still writing the manifest at the
end of the segment leading to two duplicate writes.
Avoids empty "Channel" or "Overall" header lines added to log output
when measurement is restricted to one scope using
"measure_perchannel=none" or "measure_overall=none".
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
Adds schema validation for ffprobe XML output so that updating the
ffprobe.xsd file upon changes to ffprobe is not forgotten. This was
suggested by Marton Balint in:
http://ffmpeg.org/pipermail/ffmpeg-devel/2021-March/278428.html
The schema FATE test is only run if xmllint command is available.
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
Some encoders send GET_PARAMETER requests as a keep-alive mechanism.
If the client doesn't reply with an OK message, the encoder will close
the session. This was encountered with the impath i5110 encoder, when
the RTSP Keep-Alive checkbox is enabled under streaming settings.
Alternatively one may set the X-No-Keepalive: 1 header, but this is more
of a workaround. It's better practice to respond to an encoder's
keep-alive request, than disable the mechanism which may be manufacturer
specific.
Signed-off-by: Hayden Myers <hmyers@skylinenet.net>
Signed-off-by: Martin Storsjö <martin@martin.st>
We already require X264_BUILD >= 118, which includes an unconditional
definition of X264_CSP_BGR in itself, thus making this check
effectively always true.
This makes the libx264rgb check work when pkg-config is utilized
and x264.h is not part of the standard include path (as is often
with cross-compilation, or when you just have a custom prefix in
general in f.ex. your home directory).
The X264_BUILD >= 118 required by configure since 2011 should have
X264_CSP_BGR defined unconditionally (it was added a few X264_BUILD
updates earlier), but as 134cba728b
added this additional check, I have kept it for now.
After fixing AV_PKT_DATA_SKIP_SAMPLES for reading vorbis packets from ogg,
the actual decoded samples become fewer. Three fate tests are failing:
fate-vorbis-20:
The samples in 6.ogg are not frame aligned. 6.pcm file was generated by
ffmpeg before the fix. After the fix, the decoded pcm file does not match
anymore. Ideally the ref file 6.pcm should be updated but it is probably
not worth it including another copy of the same file, only smaller.
SIZE_TOLERANCE is added for this test case.
fate-webm-dash-chapters:
The original vorbis_chapter_extension_demo.ogg is transmuxed to dash-webm.
The ref file webm-dash-chapters needs to be updated.
fate-vorbis-encode:
This exposes another bug in the vorbis encoder that initial_padding is not
correctly set. It is fixed in the previous patch.
Signed-off-by: Guangyu Sun <gsun@roblox.com>
Vorbis has priming samples at the beginning. If the initial_padding is not
set in the encoder, the total sample count will be one frame fewer than it
should be. The result is that we get a truncated version of encoding.
initial_padding should be set to the frame_size used in
vorbis_encode_frame().
Signed-off-by: Guangyu Sun <gsun@roblox.com>
Without end_trimming, the last packet will contain unexpected samples used
for padding.
This commit partially fixes#6367 when the audio length is long enough.
dd if=/dev/zero of=./silence.raw count=20 bs=500
oggenc --raw silence.raw --output=silence.ogg
oggdec --raw --output silence.oggdec.raw silence.ogg
ffmpeg -codec:a libvorbis -i silence.ogg -f s16le -codec:a pcm_s16le silence.libvorbis.ffmpeg.raw
ffmpeg -i silence.ogg -f s16le -codec:a pcm_s16le silence.native.ffmpeg.raw
ls -l *.raw
The original test case in #6367 is still not fixed due to a remaining issue.
The remaining issue is that ogg_stream->private is not kept during
ogg_save()/ogg_restore(). Field final_duration in the private data is
important to calculate end_trimming.
Some common operations such as avformat_open_input() and
avformat_find_stream_info() before reading packet will trigger ogg_save()
and ogg_restore().
Luckily, final_duration will not get updated until the last ogg page. The
save/restore mentioned above will not change final_duration most of the
time. But if the audio length is short, those reads may be performed on
the last ogg page, causing trouble keeping the correct value of
final_duration. We probably need a more complicated patch to address this
issue.
Signed-off-by: Guangyu Sun <gsun@roblox.com>
Frame size of Opus stream was previously presumed here to be 960 samples
(20ms), however sizes of 120, 240, 480, 1920, and 2880 are also allowed.
It can also alter on a per-packet basis and even multiple frames may be
present in a single packet according to the specification, for the sake
of simplicity however, let us assume that this doesn't occur.
Because the mFramesPerPacket field, representing the number of samples
per packet in the ffmpeg terminilogy, is the key factor in calculating
packet durations and all that follows from that (index, bitrate, ...),
it is crucial to get right.
Therefore, if the packet size is not available ahead of time (as it is in
the case of Opus), calculate an average from the stream duration once we
know how many packets there are and update the filed in the header.
This commit adds handling for cases where an error may occur, clearing
the allocated memory resources.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit rearranges the existing code to create a separate function
for the completion callback in execute_model_tf.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit rearranges the existing code to create separate function
for filling request with execution data.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit uses TFRequestItem and the existing sync execution
mechanism to use request-based execution. It will help in adding
async functionality to the TensorFlow backend later.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit introduces a typedef TFInferRequest to store
execution parameters for a single call to the TensorFlow C API.
This typedef is used in the TFRequestItem.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This commit uses the common TaskItem and InferenceItem typedefs
for execution in TensorFlow backend.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
POSIX errno is positive. We have strict_pthread_cond_wait to handle
error code during development.
Signed-off-by: zhilizhao <zhilizhao@tencent.com>
Reviewed-by: Jun Zhao <barryjzhao@tencent.com>
Look at the event flag that signals a new sequence header was found
in the bitstream on supported libdav1d versions for this purpose.
Signed-off-by: James Almer <jamrial@gmail.com>
The main benefit comes from propagating container level metadata like hdr,
which is more commonly used than the relevant Metadata OBUs.
Signed-off-by: James Almer <jamrial@gmail.com>
Dither none is only implemented in full chroma interpolation for these rgb formats
Its also a obscure choice (producing less nice images) that implementing it in the
other code-paths makes no sense
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The child_class_next API relied on different AVCodecs to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different AVCodecs to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different AVCodecs to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different AVCodecs to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different AVCodecs to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different AVCodecs to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different AVCodecs to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Said encoder uses a function in adpcm.c and while it does not use
anything from adpcm_data.c, other parts of both adpcm.c and adpcmenc.c
need it, so adpcm_data.c needs to be enabled anyway.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The beginning of the private contexts of both the FLAC and the TAK
demuxer currently mimick FFRawDemuxerContext: A pointer to an AVClass
followed by the AVOpt-enabled field raw_packet_size. Said field is only
used by the demuxers' read_packet functions via
ff_raw_read_partial_packet(), which treats the private context as an
FFRaawDemuxerContext. Yet this is fragile, so better include a
FFRawDemuxerContext struct at the beginning of said demuxers' private
contexts.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The codec2 muxer has no options and so it needs no AVClass;
and it certainly needs no AVClass of category AV_CLASS_CATEGORY_DEMUXER.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The child_class_next API relied on different (de)muxers to use
different AVClasses; yet this API has been replaced by
child_class_iterate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
These demuxers don't need anything from rawdec; they furthermore only
used rawdec.h to include opt.h. Both of this has been fixed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
by setting the FF_FMT_INIT_CLEANUP flag. Furthermore, also remove
an unnecessary check for NULL before avformat_close_input().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
By default, a demuxer's read_close function is not called automatically
if an error happens when reading the header; instead it is up to the
demuxer to clean up after itself in this case. The mov demuxer did this
by calling its read_close function when it encountered some errors when
reading the header.
This commit changes this by setting the FF_FMT_INIT_CLEANUP flag so that
mov_read_close() is automatically called when an error happens when
reading the header.
(Btw: mov_read_close() is not idempotent: Calling it twice is
dangerouos, because MOVContext.frag_index.item will be av_freep'ed,
yet MOVContext.frag_index.nb_items won't be reset. So the calls to
mov_read_close() have to be removed before the switch to freeing
generically.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
by setting the FF_FMT_INIT_CLEANUP flag.
(Btw: concat_read_close() is not idempotent (it frees cat->files, but
doesn't reset cat->nb_files), so this demuxer was incompatible with
simply calling read_close generically upon read_header failure.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
If reading the header fails, the demuxer's read_close() function (if
existing) is not called automatically; instead several demuxers call it
via "goto fail" in read_header().
This commit intends to change this by adding an internal flag for
demuxers that can be used to set on a per-AVInputFormat basis whether
read_close() should be called generically after an error during
read_header().
The flag controlling this behaviour needs to be added because it might
be unsafe to call read_close() generally (e.g. this might lead to
read_close() being called twice and this might e.g. lead to double-frees
if av_free() is used instead of av_freep(); or a size field has not
been reset after freeing the elements (see the mov demuxer for an
example of this)).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Both AVInputFormat and AVOutputFormat currently lack an equivalent to
AVCodec's caps_internal. E.g. if reading a header fails, each demuxer
is currently required to clean up manually, which often means to just
call the demuxer's read_close function. This could (and will) be done
generically via an equivalent of FF_CODEC_CAP_INIT_CLEANUP.
Because of the unholy ABI-relationship between libavdevice and
libavformat adding such a flag is only possible when the ABI is open
(despite the flag not being part of the public API), such as now.
Therefore such a flag is also added to AVOutputFormat, despite there
being no immediate use for it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The early return caused isses for the "add" mode (got fixed in
c95dfe5cce) and the "select" mode needs a similar
fix. It is probably better to fully remove the check, since all modes work
correctly with NULL metadata.
Signed-off-by: Marton Balint <cus@passwd.hu>
This fixes an issue when multiple cases are fuzzed in a single run and
the limits are adjusted by more than the iteration limit. In that case
the adjusted limit leaked back into the global limit causing the
fuzzer to become ineffective after several iterations, MSS2 was
affected by this for example.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: negation of -9223372036854775808 cannot be represented in type 'int64_t' (aka 'long'); cast to an unsigned type to negate this value to itself
Fixes: 33997/clusterfuzz-testcase-minimized-ffmpeg_dem_WEBM_DASH_MANIFEST_fuzzer-6752039691485184
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
In dd770883e9, support for expressions was added. Among the constants
added were labels of qnstc, qpal, sntsc & spal.
These were added in ba2a8cb40b to represent parameter permutations where
only the resolution is different. They don't have any usage currency and
don't represent any industry standards or convention in terms of framerate.
Currently it is only checked that the rtp port does not exceed rtp_port_max.
Reviewed-by: Martin Storsjö <martin@martin.st>
Signed-off-by: Andriy Gelman <andriy.gelman@gmail.com>
A zero value in the quantization matrix is invalid but in practice will
just set the transform coefficient to zero after inverse quantization.
Try to continue decoding if the AV_EF_EXPLODE flag is not set.
Fixes ticket #9287.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andriy Gelman <andriy.gelman@gmail.com>
92c40ef882 added a listen_timeout option
for sdp. This allowed a user to set variable timeout which was
originally hard coded to 10 seconds.
The commit used the initial_timeout variable to store the value. But
this variable is shared with rtsp where it's used to infer a "listen"
mode. Thus, the timeout value could not be set in rtsp, and the default
value (initial_timeout = -1) would give 100ms timeout.
This was attempted to be fixed in c8101aabee,
which changed the meaning of initial_timeout = -1 to be an infinite
timeout. However, it did not address the issue that the timeout could
still not be set. Being able to set the timeout is useful because it
allows to automatically reconfigure from a udp to tcp connection in the
lower transport.
In this commit this is fixed by using the stimeout variable to
store the timeout value.
Signed-off-by: Andriy Gelman <andriy.gelman@gmail.com>
Before this change, the PREV_OUTPTS and PREV_OUTDTS constants always evaluated
to AV_NOPTS_VALUE.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
In cases where the execution inside the function execute_model_ov fails,
push the RequestItem back to the request_queue before returning the error.
In case pushing back fails, release the allocated memory.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
The recently added setts bsf makes use of the eval API whose
expressions can contain commas. The existing parsing in
av_bsf_list_parse_str() uses av_strtok to naively split
the string at commas, thus preventing the use of setts filter
with expressions containing commas.
av_get_token can work with escaped commas, allowing full use of setts.
Call the scaler function directly rather than through a function
pointer. Drop the now-unused return value from ff_getSwsFunc() and
rename the function to reflect its new role.
This will be useful in the following commits, where it will become
important that the amount of output is different for scaled vs unscaled
case.
The SDK supports NalHrdConformance, RecoveryPointSEI and AUDelimiter for
hevc encoder, so we may allow user to set these coding options like as
what we did for h264_qsv encoder.
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
Signed-off-by: Zhong Li <zhongli_dev@126.com>
Call ff_sws_rgb2rgb_init via ff_thread_once instead of checking one of the
variables it updates.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 3530839700044513368 + 8386093932303352321 cannot be represented in type 'long long'
Fixes: 35182/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EXR_fuzzer-5398383270428672
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
These bits are reserved in earlier versions of the H.264 spec, and
some poor hardware decoders require they are zero. Thus, it is useful
to be able to zero these on streams that may have them set. The result
is still a valid H.264 bitstream.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Support single input for guided filter by adding guidance mode.
If the guidance mode is off, single input is required. And
edge-preserving smoothing is conducted. If the mode is on, two
inputs are needed. The second input serves as the guidance. For
this mode, more tasks are supported, such as detail enhancement,
dehazing and so on.
Signed-off-by: Xuewei Meng <xwmeng96@gmail.com>
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Remove some incorrect (or at least misleading) statements, such as the
formats being ordered by quality, or the first format being the native
one. Neither of those are true for hardware acceleration, which is the
main use of this callback.
HDR10+ metadata is stored in the bit stream for HEVC. The story is
different for VP9 and cannot store the metadata in the bit stream.
HDR10+ should be passed to packet side data an stored in the container
(mkv) for VP9.
This CL is taking HDR10+ from AVFrame side data in libvpxenc and is
passing it to the AVPacket side data.
Reviewed-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Signed-off-by: James Zern <jzern@google.com>
Decoders like cuviddec ignore and overwrite all the properties set by the generic
code as derived from AVCodecInternal.last_pkt_props. This flag ensures libavcodec
will not store and potentially queue input packets that ultimately will not be used.
Signed-off-by: James Almer <jamrial@gmail.com>
Since commit 89ffcd1, the status pts of the output link is set to a
value in the input link time base, not in the output link time base when
EOF is reached. Usually this pst value is larger than the required one
because the output link time base is more greater than the input link
time base. When "-vf vpp_qsv,fps" is used, user has to wait a long time
for the ending of the pipeline because fps filter output a huge number
of frames until the wrong status pts is hit.
The issue can be triggered with the command below (use a clip with 1000
frames in this case):
$> time ffmpeg -hwaccel qsv -c:v hevc_qsv -i input.h265 -vf
"vpp_qsv=w=1920:h=1080,fps=fps=30" -f null -
...
[out_0_0 @ 0x564ccd27e020] 10000000 buffers queued in out_0_0, something
may be wrong.
frame=40119596 fps=88080 q=-0.0 Lsize=N/A time=371:28:39.96 bitrate=N/A
speed=2.94e+03x
video:17238889kB audio:0kB subtitle:0kB other streams:0kB global
headers:0kB muxing overhead: unknown
real 9m7.451s
user 2m34.102s
sys 0m39.734s
In order to avoid the above issue, the same time base for input and
ouput links is used in this patch.
Fixes ticket #9286
Signed-off-by: Zhong Li <zhongli_dev@126.com>
Un-hardcode the 200ms minimum latency between emitting subtitle events
so that those that wish to receive a subtitle event for every screen
change could do so.
The problem with delaying realtime output by any amount is that it is
unknown when the next byte pair that would trigger output will happen.
It may be within 200ms, or it may be several seconds later -- that's
not realtime at all.
Maybe such large values could be disallowed earlier and closer to where
they are set.
Fixes: signed integer overflow: 538976288 * 8224 cannot be represented in type 'int'
Fixes: 29102/clusterfuzz-testcase-minimized-ffmpeg_dem_RM_fuzzer-6704350354341888
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 5404200000 - -9223372031709351616 cannot be represented in type 'long'
Fixes: 29102/clusterfuzz-testcase-minimized-ffmpeg_dem_REALTEXT_fuzzer-6737340551790592
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 0 - -9223372036854775808 cannot be represented in type 'long'
Fixes: 29102/clusterfuzz-testcase-minimized-ffmpeg_dem_MPL2_fuzzer-6747053545881600
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 15 + 9223372036854775796 cannot be represented in type 'long'
Fixes: 29102/clusterfuzz-testcase-minimized-ffmpeg_dem_MPC8_fuzzer-6723520756318208
Fixes: 29102/clusterfuzz-testcase-minimized-ffmpeg_dem_MPC8_fuzzer-6739833034768384
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
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: assertion failure
Fixes: 29102/clusterfuzz-testcase-minimized-ffmpeg_dem_DXA_fuzzer-6744985740378112
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: null pointer passed as argument 1, which is declared to never be null
Fixes: 33791/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_IFF_ILBM_fuzzer-5107575256383488.fuzz
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fix memory leak for RequestItem upon error while pushing to the
request_queue in the completion callback.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
Instead use --preprocessor-arg; in binutils 2.36, the --preprocessor
flag was changed so that it no longer accepts a string containing
multiple arguments, but the whole --preprocessor argument is
treated as the path to the preprocessor executable (where the path
can contain spaces).
It's currently unclear whether this behaviour will stay or if it
is going to be reverted in the future, see discussion at [1]. Just
to be safe, avoid using the --preprocessor argument. Don't redeclare
the full preprocessing command, but just add the $(CC_DEPFLAGS) options.
Based on a patch by Kyle Schwartz.
[1] https://sourceware.org/bugzilla/show_bug.cgi?id=27594
Signed-off-by: Martin Storsjö <martin@martin.st>
When streaming mode is enabled with fMP4/CMAF for DASH output, the
segment files are available to read by players as soon as the first byte
is written instead of only after the file is fully written. The DASH
manifest currently only gets written when the final write to the segment
file occurs. This means that players cannot stream the first segment
while it is being written.
When -lhls is enabled with MP4 segments the HLS manifest is written
immediately to advertise the in-flight segments. This change adds the
same behavior for the DASH manifest so players can stream it
immediately.
AV_OPT_TYPE_VIDEO_RATE AVOption types are parsed as expressions, but in a
limited way. For example, name constants can only be parsed alone and not as
part of a longer expression.
This change allows usage like
ffmpeg -i IN -vf fps="if(eq(source_fps\,film)\,ntsc_film\,source_fps)" OUT
Suggested-by: ffmpeg@fb.com
Signed-off-by: James Almer <jamrial@gmail.com>
Again. 240M matrix is different from BT.601! And 170M is the same
as BT.601. It is primaries that are the same in 240M and 170M, as
for jp2k_rsiz see page 17 of ST 422:2019. IT WAS THERE since 2006.
This wrong jp2k_rsiz is a copy-paste of header_open_partition_key.
Add () to avoid undefined behavior
Fixes: signed integer overflow: 9223372036854775790 + 57 cannot be represented in type 'long'
Fixes: 34983/clusterfuzz-testcase-minimized-ffmpeg_dem_RPL_fuzzer-5765822923538432
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 9223126845747118112 - -2594073385365397472 cannot be represented in type 'long'
Fixes: 34936/clusterfuzz-testcase-minimized-ffmpeg_dem_MATROSKA_fuzzer-6739888002170880
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 6854513951393103890 + 3427256975738527712 cannot be represented in type 'long'
Fixes: 32936/clusterfuzz-testcase-minimized-ffmpeg_dem_R3D_fuzzer-5236914752978944
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: -nan is outside the range of representable values of type 'long'
Fixes: signed integer overflow: 1000 * -9223372036854775808 cannot be represented in type 'long'
Fixes: 34890/clusterfuzz-testcase-minimized-ffmpeg_dem_FLV_fuzzer-5334208657620992
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
These properties have values either 0 or 1, so using uint8_t
is a better option as compared to int.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
Convert output_name to char **output_names in TaskItem and use it as
a pointer to array of output names in the DNN backend.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
Extract TaskItem and InferenceItem from OpenVino backend and convert
ov_model to void in TaskItem.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
Broken in 753930bc73, as the path to
Doxyfile passed to doxy-wrapper.sh is relative to the build dir, while
the recipe cd's to the source dir before invoking the wrapper.
Fixes#9283
This fixes setting of 'key_frame' flag in AVFrame when input h264 packets represents individual fields of interlaced video.
In this case, pairs of two consecutive fields represents a single decoded picture and have identical 'CurrPicIdx', however, only
the first field is entirely intra-coded and has the flag 'intra_pic_flag' set and the second field was resetting the flag before
it was even read in the function 'cuvid_output_frame'.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
It can be useful to library users, and is currently being used by ffmpeg.c
Suggested-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Writes a general ARIB stream identifier descriptor, as well
as a data component descriptor which also includes a
pre-defined additional_arib_caption_info structure.
Signed-off-by: zheng qian <xqq@xqq.im>
CID 1485004: Uninitialized variables (UNINIT)
Using uninitialized value "x" when calling "*pixel_belongs_to_region".
Signed-off-by: Ting Fu <ting.fu@intel.com>
There is no good use case for out of order delivery of data. For live
streaming with TSBPD enabled by default, the receiver get data in order
based on the timestamps. However, if TSBPD is disabled, the data can
be delivered out of order.
Signed-off-by: Marton Balint <cus@passwd.hu>
This reverts commit d6d407d2d7.
Hack not needed after a2b1dd0ce3.
Will fix#7480 and #8904.
This will include e.g. CODECS="hvc1.2.4.L123.B0" into m3u8.
Signed-off-by: Valerii Zapodovnikov <val.zapod.vz@gmail.com>
Fixes: floating point division by 0
Fixes: undefined behavior in handling NaN
Fixes: Ticket 8268
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: CID1476303 Bad bit shift operation
Fixes: 34871/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DPX_fuzzer-6331163028357120
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Instead return the dictionary in the state it is at the time the error
occurred. This is more in line with the description of this parameter
and allows to notify the user of unrecognized options if an error
happens lateron (which might very well be due to e.g. misspelled
options).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Besides being unnecessary it is also safer: If the error for an
unrecognized option were triggered (which seems to be impossible right
now), it might be that the stream whose codecpar is accessed is NULL.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The user should not rely on all options always being recognized
(in particular not on error).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is no longer necessary now that ff_frame_thread_encoder_init()
no longer receives an options dictionary.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In case the underlying AVCodec has no private class, the private data
of both the main as well as each worker AVCodecContext is just zeroed
(the codec's init function has not been called on any of them and
without a private class there is no way to legitimately set anything
before the aforementioned init function).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
avcodec_open2() allows to provide options via an AVDictionary;
but it is also allowed to set options by simply setting the value
of the AVCodecContext or via the AVOptions API if the codec has
a private class. Any options provided via an AVDictionary have already
been applied before ff_frame_thread_init(), so in order to copy
all the options from the main AVCodecContext and its private context,
it is enough to av_opt_copy() these options.
The current code does this, but it does more: It also copies the
user-provided AVDictionary and uses it for the initialization of
each of the worker-AVCodecContexts. This is completely unnecessary,
because said options have already been copied from the main context.
Furthermore, these options were also examined to decide if frame
threading should be used for huffman encoding in case this would incur
nondeterminism. This is wrong, because options not set via
an AVDictionary are ignored. Instead inspect the values stored in the
contexts directly. (In order to maintain the current behaviour, the
default value of the "non_deterministic" option has been changed to false,
because the absence of an entry with said key in the AVDictionary
had the consequence of disallowing nondeterminism.)
Finally, the AVDictionary has been removed from the signature of
ff_frame_thread_encoder_init().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, ff_alloc_packet2() has a min_size parameter:
It is supposed to be a lower bound on the final size of the packet
to allocate. If it is not too far from the upper bound (namely,
if it is at least half the upper bound), then ff_alloc_packet2()
already allocates the final, already refcounted packet; if it is
not, then the packet is not refcounted and its data only points to
a buffer owned by the AVCodecContext (in this case, the packet will
be made refcounted in encode_simple_internal() in libavcodec/encode.c).
The goal of this was to avoid data copies and intermediate buffers
if one has a precise lower bound.
Yet those encoders for which precise lower bounds exist have recently
been switched to ff_get_encode_buffer() (which automatically allocates
final buffers), leaving only two encoders to actually set the min_size
to something else than zero (namely aliaspixenc and hapenc). Both of
these encoders use a very low lower bound that is not helpful in any
nontrivial case.
This commit therefore removes the min_size parameter as well as the
codepath in ff_alloc_packet2() for the allocation of final buffers.
Furthermore, the function has been renamed to ff_alloc_packet() and
moved to encode.h alongside ff_get_encode_buffer().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Also combine two if blocks that check for the same condition
and don't check had_partial if we already have a complete packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There are no preallocated buffer packets any more; this feature only
worked with the old encode API and only until said API was turned into
a wrapper for the new API in 93016f5d1d.
So remove its remnants.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data; this also makes it easy
to allow user-supplied buffers. Only one thing needed to be changed:
The earlier code relied on the buffer having been initially zeroed
by av_fast_padded_malloc(), so one now needs to zero the packet at
first.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
fix problem when set x to odd number in nv12 by cuda
test step:
1. ffmpeg -f lavfi testsrc2=s=176x144 -pix_fmt nv12 -t 1 output_overlay.yuv
2. ffmpeg -f lavfi testsrc2=s=352x288 -pix_fmt nv12 -t 1 output_main.yuv
before this patch:
overlay_cuda=x=0:y=0 will right,
overlay_cuda=x=3:y=0 will wrong,
both will right after patch.
Signed-off-by: Steven Liu <liuqi05@kuaishou.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
With these triggering a lot of crashes recently, an option to globally
disable all of them is added as a tool to work around those crashes in
case the SEI data is not needed by the user.
Also re-enables s12m for hevc_nvenc, since the issue is not specifically
with that, but it affects all SEI data.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Clang is more strict on the type of asm operands, float or double
type variable should use constraint 'f', integer variable should
use constraint 'r'.
Signed-off-by: Jin Bo <jinbo@loongson.cn>
Reviewed-by: yinshiyou-hf@loongson.cn
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array access
Fixes: 33960/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_THP_fuzzer-5052852809629696
Fixes: 34163/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_THP_fuzzer-6123678099177472
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This mostly reverts 785bfb1d7b.
But I also added some clarifications so that nobody mixes primaries
with matrix again. SMPTE 240 and 170 primaires are the same, while
matrix coeff. are different, because 240 is derived from 170's new
primaries and white point while 170 uses BT.601 derived from BT.470
System M (yes, with Illuminant C) a.k.a. NTSC 1953. Some nits too.
Reviewed-by: Reto Kromer <lists@reto.ch>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
With audio/video HLS playlists, audio chunklists are treated as
alternative renditions for video chunklists. This is wrong for
audio-only HLS playlists.
fixes: 9252
1.'xor,or,and' to 'pxor,por,pand'. In the case of operating FPR,
gcc supports both of them, clang only supports the second type.
2.'dsrl,srl' to 'ssrld,ssrlw'. In the case of operating FPR, gcc
supports both of them, clang only supports the second type.
Signed-off-by: Jin Bo <jinbo@loongson.cn>
Reviewed-by: yinshiyou-hf@loongson.cn
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This commit corrects the type of pointer of elements from the
inference queue in ff_dnn_free_model_ov.
Signed-off-by: Shubhanshu Saxena <shubhanshu.e01@gmail.com>
This fixes an issue where the yadif filter could cause the timebase denominator to overflow.
Signed-off-by: Tom Boshoven <tom@jwplayer.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The code uses x/ymax + 1 so the maximum is INT_MAX-1
Fixes: signed integer overflow: 2147483647 + 1 cannot be represented in type 'int'
Fixes: 33158/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EXR_fuzzer-5545462457303040
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This feature can be used with dnn detection by setting vf_drawtext's option
text_source=side_data_detection_bboxes, for example:
./ffmpeg -i face.jpeg -vf dnn_detect=dnn_backend=openvino:model=face-detection-adas-0001.xml:\
input=data:output=detection_out:labels=face-detection-adas-0001.label,drawbox=box_source=
side_data_detection_bboxes,drawtext=text_source=side_data_detection_bboxes:fontcolor=green:\
fontsize=40, -y face_detect.jpeg
Please note, the default fontsize of vf_drawtext is 12, which may be too
small to be seen clearly.
Signed-off-by: Ting Fu <ting.fu@intel.com>
This feature can be used with dnn detection by setting vf_drawbox's
option box_source=side_data_detection_bboxes, for example:
./ffmpeg -i face.jpeg -vf dnn_detect=dnn_backend=openvino:model=face-detection-adas-0001.xml:\
input=data:output=detection_out:labels=face-detection-adas-0001.label,\
drawbox=box_source=side_data_detection_bboxes -y face_detect.jpeg
Signed-off-by: Ting Fu <ting.fu@intel.com>
Here the packet size is known before allocating the packet,
so that supporting user-supplied buffers is trivial.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The APNG encoder already uses internal buffers, so that the packet size
is already known before allocating the packet; therefore one can avoid
another (implicit) intermediate buffer by switching to
ff_get_encode_buffer(), thereby also supporting user-supplied buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The FLAC encoder calculates the size in advance, so one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
For all p*m encoders a very sharp upper bound for the size of the
output packets is available before the packet is allocated. This can
be used to avoid an intermediate buffer when encoding by using
ff_get_encode_buffer() instead of ff_alloc_packet2() (without min_size);
this also adds support for user-supplied buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data; also, there is no reason
to add AV_INPUT_BUFFER_MIN_SIZE to the packet size any more, as the
actually needed packet size can be easily calculated: It is three bytes
more than the raw nal size per NALU.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet,
so that supporting user-supplied buffers is trivial.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet because
the encoder provides said information (and works with internal buffers
itself), so one can use this information to avoid the implicit use of
another intermediate buffer for the packet data; and by switching to
ff_get_encode_buffer() one can also allow user-supplied buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet,
so that supporting user-supplied buffers is trivial.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet,
so that supporting user-supplied buffers is trivial.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet because
the encoder provides said information (and works with internal buffers
itself), so one can use this information to avoid the implicit use of
another intermediate buffer for the packet data; and by switching to
ff_get_encode_buffer() one can also allow user-supplied buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: James Zern <jzern@google.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet because
the encoder provides said information (and works with internal buffers
itself), so one can use this information to avoid the implicit use of
another intermediate buffer for the packet data; and by switching to
ff_get_encode_buffer() one can also allow user-supplied buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet because
the encoder provides said information (and works with internal buffers
itself), so one can use this information to avoid the implicit use of
another intermediate buffer for the packet data; and by switching to
ff_get_encode_buffer() one can also allow user-supplied buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The libshine encoder already uses an internal buffer, so that the
packet size is already known before allocating the packet; therefore
one can avoid another (implicit) intermediate buffer by switching
to ff_get_encode_buffer(), thereby also supporting user-supplied
buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet,
so that supporting user-supplied buffers is trivial.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The libmp3lame encoder already uses an internal buffer, so that the
packet size is already known before allocating the packet; therefore
one can avoid another (implicit) intermediate buffer by switching
to ff_get_encode_buffer(), thereby also supporting user-supplied
buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet,
so that supporting user-supplied buffers is trivial.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Here the packet size is known before allocating the packet because
the encoder provides said information (and works with internal buffers
itself), so one can use this information to avoid the implicit use of
another intermediate buffer for the packet data; and by switching to
ff_get_encode_buffer() one can also allow user-supplied buffers.
Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: James Zern <jzern@google.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It has been added in 2016 when this flag made no sense for encoders at
all; now that it makes sense, audiotoolboxenc doesn't support it,
despite claiming to do so.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
ref_frame is owned by the framesync structure and should therefore not
be modified; furthermore, these properties that are copied don't seem to
be used at all, so copying is unnecessary. Finally copying when the
destination frame is NULL gives a guaranteed segfault.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This is marginally slower, but correct for all input values.
The previous implementation failed with certain input seeds, e.g.
"checkasm --test=hevc_idct 98".
Signed-off-by: Martin Storsjö <martin@martin.st>
The twoloop coder is highly loaded with (pseudo-)perceptual metrics,
and the aim of the tests is to piece-wise test each function of the
encoder, for which the 'fast' coder is perfect, since it only decides
on which scalefactors to use, rather than enable or disable encoder
features.
This used to be the default, but was reverted as it was slower than
the 'fast' coder by around 25%.
Since our encoder is still not very good, change back to the twoloop
coder by default. It has much better rate control management as well,
making it closer to CBR, and it sounds much better.
With some minor changes by Marton Balint:
- removed trailing whitespace
- fixed network_descriptors_length
- fixed reserved_future_use flag in the start of the section
- removed unused program variable
- emit first NIT after PAT
- some other cosmetics
Signed-off-by: Ubaldo Porcheddu <ubaldo@eja.it>
Signed-off-by: Marton Balint <cus@passwd.hu>
Also use helper function to set the timestamp. Maybe we could also use
nanosecond precision, but there were some float rounding concerns.
Signed-off-by: Marton Balint <cus@passwd.hu>
Previously, only the size of a given tile was passed, making the
offset and size marked in VASliceParameterBufferAV1 invalid with
multiple tiles.
Signed-off-by: Fei Wang <fei.w.wang@intel.com>
Fixes: CID1398579 Dereference before null check
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Trivial for an encoder that has a good estimate of the size of
the output packet in advance.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Trivial for an encoder that has a very good estimate of the size
of the output packet in advance.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Now that the proper buffer size is calculated (and checked) before
allocating the buffer, it is known that the buffer always suffices.
So use the unchecked PutBit-API; and also use an unchecked bitstream
reader as we check ourselves.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, the JPEG-LS encoder allocated a worst-case-sized packet
at the beginning of each encode2 call; then it wrote the packet header
into its destination buffer and encoded the actual packet data;
said data is written into another worst-case-sized buffer, because it
needs to be escaped before being written into the packet buffer.
Finally, because the packet buffer is worst-case-sized, the generic
code copies the actually used part into a fresh buffer.
This commit changes this: Allocating the packet and writing the header
into it is deferred until the actual data has been encoded and its size
is known. This gives a good upper bound for the needed size of the packet
buffer (the upper bound might be 1/15 too large) and so one can avoid the
implicit intermediate buffer and support user-supplied buffers by using
ff_get_encode_buffer().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This reverts commit b5ca8f2c66.
This commit will make new problem about tickets: 9193,9205
It flush data into file with init file context together,
and it can get keyframe size, maybe need more method to get keyframe
size.
Signed-off-by: Steven Liu <liuqi05@kuaishou.com>
Since a247ac640d, allcodecs.c contained
two lines that matched the regex used by find_filters_extern in
configure; as a result, libx264 appeared twice the list of codecs
(if enabled).
Fix this by using only one matching line by adding a preprocessor define
for the part that differed in the two old lines: The const qualifier.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Regression since b9c5fdf6027010d15ee90a43aa023e45a5189097;
fixes Coverity ID #1484786.
Also remove the check for st->internal->parser as av_parser_close(NULL)
is a no-op.
Reviewed-by: James Almer <jamrial@gmail.com>
Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Otherwise the rate emulation logic in `transcode_step` never gets
hit, and the unavailability flag never gets reset, leading to an
eternal loop with some rate emulation use cases.
This change was missed during the rework of ffmpeg.c, in which
encoder initialization was moved further down the time line in
commit 67be1ce0c6 . Previously,
as the encoder initialization had happened earlier, this state was
not possible (flow getting as far as hitting the rate emulation logic,
yet not having the encoder initialized yet).
Fixes#9160
Two modes are supported in guided filter, basic mode and fast mode.
Basic mode is the initial pushed guided filter without optimization.
Fast mode is implemented based on the basic one by sub-sampling method.
The sub-sampling ratio which can be defined by users controls the
algorithm complexity. The larger the sub-sampling ratio, the lower
the algorithm complexity.
Signed-off-by: Xuewei Meng <xwmeng96@gmail.com>
Reviewed-by: Steven Liu <liuqi05@kuaishou.com>
Export them in UTC, not the local timezone. This way the output is
the same everywhere. The timezone information stored in the file is
still ignored, since there seems to be no simple way to export it
correctly.
Format them according to ISO 8601, which we generally use for exporting
dates.
Fixes fate-flv-demux, which was broken since
958bea5248 on some platforms.
The JPEG-2000 decoder and encoder share common luts; the decoder
initializes them once, guarded by a dedicated AVOnce, whereas
the encoder initializes them always during init. This means that
the decoder is not init-threadsafe; in fact there is a potential
data race because these luts can be initialized while an active
decoder/encoder is using them.
Fix this and make the decoder init-threadsafe by making the
initialization function guard initialization itself with a dedicated
AVOnce.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
mqc currently initializes three arrays at runtime; each of them
has 2 * 47 elements, one is uint16_t, two are uint8_t, so that their
combined size is 8 * 47. The source data for these initializations
is contained in an array of 47 elements of size six. Said array is
only used in order to initialize the other arrays, so the savings
are just 2 * 47B. Yet this is dwarfed by the size of the code for
performing the initializations: It is 109B (GCC 10.2, x64, -O3 albeit
in an av_cold function); this does not even include the size of the
code in the callers. So just hardcode these tables.
This also fixes a data race, because the encoder always initialized
these tables during init, although they might already be used at the
same time by already running encoder/decoder instances.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The Vorbis encoder has an array of a structure containing all
the ingredients for a codebook; this includes a pointer to
the actual codebook and some even have a pointer to an array
containing quant values. Each of these real codebooks is
an array of its own.
These pointers lead to relocations and therefore the array will
be placed in .data.rel.ro and not in .rodata.
This commit avoids the pointers altogether by combining all the actual
codebooks into one big array; the actual codebooks are now accessed
consecutively by incrementing the pointer used to access them by the
length of the actual codebook that has just been dealt with (said length
is contained in the structure describing the codebook). There is
no downside to this given that these codebooks are already only used
once during init.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The Vorbis encoder allocates several arrays destined to contain pointers
to separately allocated arrays; yet these arrays are allocated without
initializing them: They are uninitialized until their final values
are stored in them; so if allocating one of the earlier subarrays fails,
all of the remaining pointers to subarrays are still uninitialized.
But they are used for freeing, resulting in crashes.
Fix this by zero-initializing the arrays with subarrays.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
ff_wma_init() can fail without freeing everything it has allocated;
so add the FF_CODEC_CAP_INIT_CLEANUP to the codecs using it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The address of this variable never leaks, so it cannot be modified
by anyone else at all.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
duplicate ff_hex_to_data() function from avformat and rename it to
hex_to_data() as static function.
Reviewed-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
CID: 1482090
there can return null from av_frame_get_side_data, and will use sd->data
after av_frame_get_side_data, so should check null return value.
Signed-off-by: Steven Liu <liuqi05@kuaishou.com>
Testing model is tensorflow offical model in github repo, please refer
https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf2_detection_zoo.md
to download the detect model as you need.
For example, local testing was carried on with 'ssd_mobilenet_v2_coco_2018_03_29.tar.gz', and
used one image of dog in
https://github.com/tensorflow/models/blob/master/research/object_detection/test_images/image1.jpg
Testing command is:
./ffmpeg -i image1.jpg -vf dnn_detect=dnn_backend=tensorflow:input=image_tensor:output=\
"num_detections&detection_scores&detection_classes&detection_boxes":model=ssd_mobilenet_v2_coco.pb,\
showinfo -f null -
We will see the result similar as below:
[Parsed_showinfo_1 @ 0x33e65f0] side data - detection bounding boxes:
[Parsed_showinfo_1 @ 0x33e65f0] source: ssd_mobilenet_v2_coco.pb
[Parsed_showinfo_1 @ 0x33e65f0] index: 0, region: (382, 60) -> (1005, 593), label: 18, confidence: 9834/10000.
[Parsed_showinfo_1 @ 0x33e65f0] index: 1, region: (12, 8) -> (328, 549), label: 18, confidence: 8555/10000.
[Parsed_showinfo_1 @ 0x33e65f0] index: 2, region: (293, 7) -> (682, 458), label: 1, confidence: 8033/10000.
[Parsed_showinfo_1 @ 0x33e65f0] index: 3, region: (342, 0) -> (690, 325), label: 1, confidence: 5878/10000.
There are two boxes of dog with cores 94.05% & 93.45% and two boxes of person with scores 80.33% & 58.78%.
Signed-off-by: Ting Fu <ting.fu@intel.com>
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Otherwise decoding will crash lateron; e.g. because dct_tokens
is never set or because a VLC that has not been allocated is used.
Reviewed-by: Peter Ross <pross@xvid.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The DNXHD encoder's context contains an array of 32 pointers to
DNXHDEncContexts used in case of slice threading; when trying
to use more than 32 threads with slice threading, the encoder's init
function errors out, but the close function takes avctx->thread_count
at face value and tries to free inexistent elements of the array,
leading to potential crashes.
Fix this by modifying the check used to decide whether the slice
contexts should be freed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In this case this actually fixes a potential data race: The static VLC
tables were reinitialized every time an AVCodecContext has been
initialized; while the mutex in avcodec_open2() ensured that the VLCs
could not be initialized concurrently by multiple threads, nothing
guaranteed that these VLCs are not read concurrently (when decoding a
packet with an already initialized AVCodecContext) while another thread
initializes them. This is undefined behaviour despite the values being
written coinciding with the earlier values.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is not documented to be safe to call inflateEnd() on a z_stream
that has not been successfully initialized via inflateInit(); so
record whether it has been successfully initialized.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The split into vp9_decode_init() and init_frames() is a remnant
of using init_thread_copy() for frame-threading; the latter has
been removed, so there is no reason for init_frames() not be part
of vp9_decode_init().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Add examples on how to use this filter, and improve the code style.
Implement the slice-level parallelism for guided filter.
Add the basic version of guided filter.
Signed-off-by: Xuewei Meng <xwmeng96@gmail.com>
Reviewed-by: Steven Liu <liuqi05@kuaishou.com>
There are no guarantees that all side data types have the same
representation on all platforms.
Tests that change output due to this:
id3v2-priv-remux, cover-art-mp3-id3v2-remux, gapless-mp3: SKIP_SAMPLES,
which is tested by fate-gapless-mp3-side-data
matroska-vp8-alpha-remux: MATROSKA_BLOCKADDITIONAL, which is tested by
remux itself (side data is written into output)
matroska-mastering-display-metadata: MASTERING_DISPLAY_METADATA and
CONTENT_LIGHT_LEVEL, which are tested by ffprobe invocation in the same
test
matroska-spherical-mono-remux: STEREO3D and SPHERICAL, which are tested
by ffprobe invocation in the same test
segment-mp4-to-ts: MPEGTS_STREAM_ID, which is tested by ts remuxing
tests
webm-webvtt-remux: WEBVTT_IDENTIFIER/SETTINGS, which is tested by the
ffprobe invocation in the same test
mxf-d10-user-comments: CPB_PROPERTIES, which is tested by mxf-probe-d10
mov-cover-image: SKIP_SAMPLES, which is tested for mov by
mov-aac-2048-priming
copy-trac3074: AUDIO_SERVICE_TYPE, which is tested by fate-hls-fmp4_ac3
av_mallocz() is superfluous as get_packet_defaults() is called immediately
after it's allocated, which will initialize the entire struct to default
values.
Signed-off-by: James Almer <jamrial@gmail.com>
If a copy callback is provided by the caller, the packet passed to it
was zeroed instead of initialized with default values.
Signed-off-by: James Almer <jamrial@gmail.com>
Libavcodec can now handle the AV1CodecConfigurationRecord structure
as-is when passed as extradata, so the standard behavior of
read-box-into-extradata should suffice, just like with AVC and HEVC.
The SVQ1 decoder does not need mpegvideo or rl.c, but it uses stuff
from h263data.c. But since 61fe481586
h263data.c called ff_rl_init() and this of course led to build errors
when the SVQ1 decoder is enabled and mpegvideo disabled.
Fix this by moving ff_h263_init_rl_inter() to h263.c.
Fixes ticket #9224.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
MSA2 optimizations are attached to MSA macros in generic_macros_msa.h.
It's difficult to do runtime check for them. Remove this part of code
can make it more robust. H264 1080p decoding: 5.13x==>5.12x.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Using mask to avoid judgment, H264 4K decoding speed
improved about 0.1fps tested on 3A4000
Signed-off-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
1. Refined function get_cabac_inline_mips.
2. Optimize function get_cabac_bypass and get_cabac_bypass_sign.
Speed of decoding h264: 4.89x ==> 5.05x(tested on 3A4000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The MSA optimization has been refined in commit 93218c2 and ce0a52e.
It is better than MMI version now.
Speed of decoding H264: 4.83x ==> 4.89x (tested on 3A4000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
It is not documented to be safe to call inflateEnd() on a z_stream
that has not been successfully initialized via inflateInit(); so
record whether it has been successfully initialized.
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This will give us more room to improve the implementation later.
Suggested-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: James Almer <jamrial@gmail.com>
Here the packet size is known before allocating the packet because
the encoder provides said information (and works with internal buffers
itself), so one use this information to avoid the implicit use of another
intermediate buffer for the packet data; and by switching to
ff_get_encode_buffer() one can also allow user-supplied buffers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Signed-off-by: Rick Kern <kernrj@gmail.com>
Said RL VLC is only used by the decoder, ergo don't initialize it for
the encoder and move the whole code and the RL VLC table itself to
dvdec.c.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
It can and therefore we switch from a heap allocated VLC table to
a VLC initialized via the mechanism for static VLCs, but without
an actual static VLC.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
It is init-threadsafe since b9c1ab8907
and except on MIPS even before that due to its use of ff_thread_once()
for static initialization.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
From the comment it's not available on old version. It works now
by testing on macOS 11.2.1. There is no document about since when.
So trying to set the configuration and ignore the error for hevc.
Signed-off-by: Rick Kern <kernrj@gmail.com>
classification is done on every detection bounding box in frame's side data,
which are the results of object detection (filter dnn_detect).
Please refer to commit log of dnn_detect for the material for detection,
and see below for classification.
- download material for classifcation:
wget https://github.com/guoyejun/ffmpeg_dnn/raw/main/models/openvino/2021.1/emotions-recognition-retail-0003.bin
wget https://github.com/guoyejun/ffmpeg_dnn/raw/main/models/openvino/2021.1/emotions-recognition-retail-0003.xml
wget https://github.com/guoyejun/ffmpeg_dnn/raw/main/models/openvino/2021.1/emotions-recognition-retail-0003.label
- run command as:
./ffmpeg -i cici.jpg -vf dnn_detect=dnn_backend=openvino:model=face-detection-adas-0001.xml:input=data:output=detection_out:confidence=0.6:labels=face-detection-adas-0001.label,dnn_classify=dnn_backend=openvino:model=emotions-recognition-retail-0003.xml:input=data:output=prob_emotion:confidence=0.3:labels=emotions-recognition-retail-0003.label:target=face,showinfo -f null -
We'll see the detect&classify result as below:
[Parsed_showinfo_2 @ 0x55b7d25e77c0] side data - detection bounding boxes:
[Parsed_showinfo_2 @ 0x55b7d25e77c0] source: face-detection-adas-0001.xml, emotions-recognition-retail-0003.xml
[Parsed_showinfo_2 @ 0x55b7d25e77c0] index: 0, region: (1005, 813) -> (1086, 905), label: face, confidence: 10000/10000.
[Parsed_showinfo_2 @ 0x55b7d25e77c0] classify: label: happy, confidence: 6757/10000.
[Parsed_showinfo_2 @ 0x55b7d25e77c0] index: 1, region: (888, 839) -> (967, 926), label: face, confidence: 6917/10000.
[Parsed_showinfo_2 @ 0x55b7d25e77c0] classify: label: anger, confidence: 4320/10000.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Different function type of model requires different parameters, for
example, object detection detects lots of objects (cat/dog/...) in
the frame, and classifcation needs to know which object (cat or dog)
it is going to classify.
The current interface needs to add a new function with more parameters
to support new requirement, with this change, we can just add a new
struct (for example DNNExecClassifyParams) based on DNNExecBaseParams,
and so we can continue to use the current interface execute_model just
with params changed.
There's one task item for one function call from dnn interface,
there's one request item for one call to openvino. For classify,
one task might need multiple inference for classification on every
bounding box, so add InferenceItem.
avpriv_set_systematic_pal2() is meant to fill fixed vales for formats that
until recently were tagged as "pseudo pal". This is no longer the case, so
this call is a no-op when used on real PAL formats.
Signed-off-by: James Almer <jamrial@gmail.com>
In particular, document that they initialize different parts of an
RLTable and therefore need not be synchronized.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The SpeedHQ encoder currently reverses the entries of two small tables
and stores them in other tables. These other tables have a size of 48
bytes, yet the code for their initialization takes 135 bytes (GCC 9.3,
x64, O3 albeit in an av_cold function). So remove the runtime
initialization and hardcode the tables.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The SpeedHQ decoder uses and initializes a RLTable's VLC, yet it also
initializes other parts of the RLTable that it does not use. This has
downsides besides being wasteful: Because the SpeedHQ encoder also
initializes these additional fields, there is a potential for data races
(and therefore undefined behaviour). In fact, removing the superfluous
initializations from the decoder automatically makes both the decoder
and the encoder init-threadsafe. This commit does so.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Here the packet size is known before allocating the packet because
the encoder itself works with an internal buffer, so one can use
this information to avoid the implicit use of another intermediate
buffer for the packet data; one can also switch to
ff_get_encode_buffer() and directly use user-supplied buffers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data; this also makes it easy
to allow user-supplied buffers. Only one thing needed to be changed:
One can no longer use a pointer to uint16_t for the destination buffer
because its alignment is unknown.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now, the cljr encoder used buffers that were too big by a
factor of eight (probably bit/byte confusion). This has been fixed.
And because the needed buffer size can be easily calculated in advance,
one can avoid the implicit use of an intermediate buffer and can even
allow user-supplied buffers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data; this also makes it easy
to allow user-supplied buffers. Only one thing needed to be changed:
One can no longer use a pointer to uint16_t for the destination buffer
because its alignment is unknown.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The size of the output buffer is always known in advance and
the code has no alignment requirement (it uses mostly the PutBits API),
so allowing user-supplied buffers is trivial.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data; and one can also use
user-supplied buffers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When the packet size is known in advance like here, one can avoid
an intermediate buffer for the packet data by using
ff_get_encode_buffer() and also set AV_CODEC_CAP_DR1 at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
There are cases where using 1000 as the MP4 timescale is not
accurate enough, for example when one needs sample-accurate audio
handling.
This adds a new AVOption to the MOV/MP4 muxer to override the
movie timescale, but it still defaults to 1000 to maintain current
default behavior.
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
This simply performs a 2nd pass if a LSE is encountered with GRAY8
Fixes: tickets/3933/128.jls
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Use AVSTREAM_EVENT_FLAG_NEW_PACKETS instead, which should provide the
same information in this case.
Finishes removing all uses of this field as started by 87f0c8280c.
Signed-off-by: James Almer <jamrial@gmail.com>
The MLP/TrueHD encoder uses pointers to non-const to access several
static objects that are only initialized at runtime and are therefore
not declared as const. This does not result in compiler warnings, but it
is fragile, as these objects are really not to be modified as they are
not owned by any encoder instance. Therefore this commit adds const to
the pointed to type of the pointers used to access them after their
initialization. One object has even been made const.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The size of ff_qexp is only 32 bytes, but the code to generate it at
runtime takes 47 bytes (GCC 9.3, x64, -O3 in an av_cold function); so
just hardcode it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The Bink video decoder uses VLCs; the longest codes of these VLCs have
different lengths, yet they are all so small that each VLC is read in
one go, so that the number of elements in the VLC table actually used by
each table is 1 << nb_bits, where nb_bits is the length of the longest
code. Yet when determining the size of the VLC table nb_bits has been
overestimated as the number of bits of the longest code in all VLCs,
making said table unnecessary big (2048 vs 976 elements).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
ff_ass_subtitle_header_full() just uses av_asprintf() and is therefore
thread-safe itself.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Note: This decoder uses a static variable in save_display_set() (which
is only enabled if DEBUG is defined); yet said function can't be reached
from the decoder's init function at all, so it is no problem for
setting the FF_CODEC_CAP_INIT_THREADSAFE flag.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
These pointers sometimes point to static storage (namely to
default_clut), so adding const to the pointed-to type is important to
ensure that one does not accidentally modify something that is not owned
by a single AVCodecContext.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
An AVCodecContext's priv_data has already been zeroed generically before
calling the init function.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
For frames decoded with skip_frame == AVDISCARD_ALL, a picture is not allocated
and got_picture is never set to 1 even if a SOF and SOS were parsed.
The existing check in EOI only cares if a SOF was parsed, not if a picture
allocated, so change it and add a new check to explicitly ensure a picture was
allocated when skip_frame != AVDISCARD_ALL.
Fixes probing and decoding when skip_frame is AVDISCARD_ALL.
Signed-off-by: James Almer <jamrial@gmail.com>
While Vulkan itself went more or less the way it was expected to go,
libvulkan didn't quite solve all of the opengl loader issues. It's multi-vendor,
yes, but unfortunately, the code is Google/Khronos QUALITY, so suffers from
big static linking issues (static linking on anything but OSX is unsupported),
has bugs, and due to the prefix system used, there are 3 or so ways to type out
functions.
Just solve all of those problems by dlopening it. We even have nice emulation
for it on Windows.
VkPhysicalDeviceLimits.optimalBufferCopyRowPitchAlignment and
VkPhysicalDeviceExternalMemoryHostPropertiesEXT.minImportedHostPointerAlignment are of type
VkDeviceSize (a typedef uint64_t).
VkPhysicalDeviceLimits.minMemoryMapAlignment is of type size_t.
Signed-off-by: James Almer <jamrial@gmail.com>
Reviewed-by: Lynne <dev@lynne.ee>
please use tools/python/tf_sess_config.py to get the sess_config after that.
note the byte order of session config is in normal order.
bump the MICRO version for the config change.
Signed-off-by: Limin Wang <lance.lmwang@gmail.com>
According to the PES packet definition defined in Table 2-17 of ISO_IEC_13818-1
specification, some fields like PTS/DTS or pes_extension could only appears if
the stream_id meets the condition:
if (stream_id != 0xBC && // program_stream_map
stream_id != 0xBE && // padding_stream
stream_id != 0xBF && // private_stream_2
stream_id != 0xF0 && // ECM
stream_id != 0xF1 && // EMM
stream_id != 0xFF && // program_stream_directory
stream_id != 0xF2 && // DSMCC_stream
stream_id != 0xF8) // ITU-T Rec. H.222.1 type E stream
And the following stream_id types don't have fields like PTS/DTS:
else if ( stream_id == program_stream_map
|| stream_id == private_stream_2
|| stream_id == ECM
|| stream_id == EMM
|| stream_id == program_stream_directory
|| stream_id == DSMCC_stream
|| stream_id == ITU-T Rec. H.222.1 type E stream ) {
for (i = 0; i < PES_packet_length; i++) {
PES_packet_data_byte
}
}
Current implementation skipped the check of stream_id causing some kind of
streams like private_stream_2 to be incorrectly written with actually a
private_stream_1-like PES header with PTS/DTS field. For example, Japan DTV
transmits news and alerts through ARIB superimpose that utilizes
private_stream_2 still could not be remuxed correctly for now.
This patch set fixes the remuxing for private_stream_2 and
other stream_id types.
Signed-off-by: zheng qian <xqq@xqq.im>
Signed-off-by: Marton Balint <cus@passwd.hu>
AVCodecContext.extradata is freed generically by libavcodec for
encoders, so it is unnecessary for an encoder to do it on its own.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Was added in error very early on, passing in only the required fields.
Later, the muxer and demuxer were changed to pass the entire APMState
struct as extradata.
Technically a breaking change, but this was only around for a *very* short
time before it was updated,
Reviewed-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Signed-off-by: Zane van Iperen <zane@zanevaniperen.com>
Compared to the earlier behaviour the following changes:
a) AVCodecInternal.byte_buffer is freed.
b) The last_pkt_props FIFO is emptied before freeing it.
c) If set AVCodecContext.hwaccel is uninitialized and its private data
is freed; hw_frames_ctx and hw_device_ctx are also unreferenced.
d) coded_side_data is freed.
e) active_thread_type is reset.
a), b), d) should be no-ops as the buffer/fifo should be empty and
no coded_side_data should exist at any point of avcodec_open2().
e) is obviously not bad.
c) is in accordance with the documentation of hw_(frames|device)_ctx
which states that libacodec takes over ownership of these references.
At least in the case of VC-1 it is possible for the hw acceleration to
be set during init and in this case freeing it actually fixes a memleak.
avcodec_close() needed only minor adjustments to make it work with
a potentially not fully initialized codec.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Right now all AVCodecContexts except those using frame-threaded decoding
call the codec's init function and expect its close function to be
called. In order to make sure that the close function is not called for
frame-threaded decoding ff_frame_thread_free() resets
AVCodecContext.codec (and because of this it has to free the private
AVOptions of the main AVCodecContext itself). This is not obvious and
potentially fragile. Instead add a field to AVCodecInternal that
indicates whether close should be called for this AVCodecContext.
It is always zero when using frame-threaded decoding, so that resetting
the codec is no longer necessary and has been removed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The frame_thread_encoder has so far not been freed in case an error
happened in avcodec_open2() after ff_frame_thread_encoder_init().
This commit changes this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Given that the AVCodec.next pointer has now been removed, most of the
AVCodecs are not modified at all any more and can therefore be made
const (as this patch does); the only exceptions are the very few codecs
for external libraries that have a init_static_data callback.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
This also allows to exclusively use pointers to const AVCodec in
fftools/ffmpeg_opt.c.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Originally deprecated in 1296b1f6c0631ab79464e22d48a6a1548450b943;
scheduled again for removal in a991526832.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Deprecated in 40cf1bbacc.
(The currently disabled filter vf_mcdeint and vf_uspp were users of
this field; they have not been changed, so that whoever wants to fix
them can see the state of these filters when they were disabled.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Signed-off-by: James Almer <jamrial@gmail.com>
These filters depend on avcodec APIs that are to be removed. Some people
have expressed potential interest in updating these filters, so they are
merely disabled for now instead of being removed.
Signed-off-by: James Almer <jamrial@gmail.com>
avfilter_transform, avfilter_(add|sub|mult)_matrix are not part of the
public API (transform.h is not a public header), yet they are currently
exported because of their name. This commit changes this:
avfilter_transform is renamed to ff_affine_transform; the other
functions are just removed as they have never been used at all.
Found-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Deprecated in ddef3d902f.
(The reference file of the mov-zombie test needed to be updated, because
a rotate metadata tag is no longer exported; the side-data is of course
still present.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Affected function pointers (always NULL) in AVInputFormat,
AVOutputFormat as well as private fields of AVStream.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Announced in 2e8b0446c6.
Two FATE-tests needed to be updated because the checksums of
side data containing an AVCPBProperties struct changed.
buffer_size has also been switched to 64bits because it is a bitsize.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, both the msmpeg4 decoders and encoders initialized several
RLTables common to them (the decoders also initialized the VLCs of these
RLTables). This is an obstacle to making these codecs init-threadsafe.
So move this initialization to ff_msmpeg4_common_init() that already
contains this initialization code. This allows to reuse the AVOnce used
for initializing ff_v2_dc_lum/chroma_table which automatically makes
initializing these RLTables thread-safe.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Up until now the RLTable ff_mpeg4_rl_intra was initialized by both mpeg4
decoder and encoder (except the VLCs that are only used by the decoder).
This is an obstacle to making these codecs init-threadsafe, so move
initializing this to a single function that is guarded by a dedicated
AVOnce.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This already makes several encoders (namely FLV, H.263, H.263+ and
RealVideo 1.0 and 2.0 and SVQ1) that use this init-threadsafe.
It also makes the Snow encoder init-threadsafe; it was already marked
as such since commit d49210788b, because
it was thought to be harmless if one and the same object was
initialized by multiple threads at the same time.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The documentation of the get_encode_buffer() callback does not require
to zero the padding; therefore we do it in ff_get_encode_buffer().
This also constitutes an implicit check for whether the buffer is
actually allocated with padding.
The memset in avcodec_default_get_encode_buffer() is now redundant and
has been removed.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Also improve readability by keeping a pointer to the IVIBandDesc that is
currently freed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This encoder sets the min_size in ff_alloc_packet2(), so it can not rely
on av_packet_make_refcounted() to zero the padding.
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The ASS margins are utilized to generate percentual values, as
the usage of cell-based sizing and offsetting seems to be not too
well supported by renderers.
Signed-off-by: Jan Ekström <jan.ekstrom@24i.com>
Attempts to utilize the TTML cell resolution as a mapping to the
reference resolution, and maps font size to cell size. Additionally
sets the display and text alignment according to the ASS alignment
number.
Signed-off-by: Jan Ekström <jan.ekstrom@24i.com>
This way the encoder may pass on the following values to the muxer:
1) Additional root "tt" element attributes, such as the subtitle
canvas reference size.
2) Anything before the body element of the document, such as regions
in the head element, which can configure styles.
Signed-off-by: Jan Ekström <jan.ekstrom@24i.com>
ADTS frames may contain up to 768 bytes per channel. With 16 channels,
this is 12k, which cannot fit into the maximum 8k buffer.
Signed-off-by: Chris Ribble <chris.ribble@resi.io>
With JPEG-LS PAL8 samples, the JPEG-LS extension parameters signaled with
the LSE marker show up after SOF but before SOS. For those, the pixel format
chosen by get_format() in SOF is GRAY8, and then replaced by PAL8 in LSE.
This has not been an issue given both pixel formats allocate the second data
plane for the palette, but after the upcoming soname bump, GRAY8 will no longer
do that. This will result in segfauls when ff_jpegls_decode_lse() attempts to
write the palette on a buffer originally allocated as a GRAY8 one.
Work around this by calling ff_get_buffer() after the actual pixel format is
known.
Signed-off-by: James Almer <jamrial@gmail.com>
This commit adds a pure x86 assembly SIMD version of the FFT in libavutil/tx.
The design of this pure assembly FFT is pretty unconventional.
On the lowest level, instead of splitting the complex numbers into
real and imaginary parts, we keep complex numbers together but split
them in terms of parity. This saves a number of shuffles in each transform,
but more importantly, it splits each transform into two independent
paths, which we process using separate registers in parallel.
This allows us to keep all units saturated and lets us use all available
registers to avoid dependencies.
Moreover, it allows us to double the granularity of our per-load permutation,
skipping many expensive lookups and allowing us to use just 4 loads per register,
rather than 8, or in case FMA3 (and by extension, AVX2), use the vgatherdpd
instruction, which is at least as fast as 4 separate loads on old hardware,
and quite a bit faster on modern CPUs).
Higher up, we go for a bottom-up construction of large transforms, foregoing
the traditional per-transform call-return recursion chains. Instead, we always
start at the bottom-most basis transform (in this case, a 32-point transform),
and continue constructing larger and larger transforms until we return to the
top-most transform.
This way, we only touch the stack 3 times per a complete target transform:
once for the 1/2 length transform and two times for the 1/4 length transform.
The combination algorithm we use is a standard Split-Radix algorithm,
as used in our C code. Although a version with less operations exists
(Steven G. Johnson and Matteo Frigo's "A modified split-radix FFT with fewer
arithmetic operations", IEEE Trans. Signal Process. 55 (1), 111–119 (2007),
which is the one FFTW uses), it only has 2% less operations and requires at least 4x
the binary code (due to it needing 4 different paths to do a single transform).
That version also has other issues which prevent it from being implemented
with SIMD code as efficiently, which makes it lose the marginal gains it offered,
and cannot be performed bottom-up, requiring many recursive call-return chains,
whose overhead adds up.
We go through a lot of effort to minimize load/stores by keeping as much in
registers in between construcring transforms. This saves us around 32 cycles,
on paper, but in reality a lot more due to load/store aliasing (a load from a
memory location cannot be issued while there's a store pending, and there are
only so many (2 for Zen 3) load/store units in a CPU).
Also, we interleave coefficients during the last stage to save on a store+load
per register.
Each of the smallest, basis transforms (4, 8 and 16-point in our case)
has been extremely optimized. Our 8-point transform is barely 20 instructions
in total, beating our old implementation 8-point transform by 1 instruction.
Our 2x8-point transform is 23 instructions, beating our old implementation by
6 instruction and needing 50% less cycles. Our 16-point transform's combination
code takes slightly more instructions than our old implementation, but makes up
for it by requiring a lot less arithmetic operations.
Overall, the transform was optimized for the timings of Zen 3, which at the
time of writing has the most IPC from all documented CPUs. Shuffles were
preferred over arithmetic operations due to their 1/0.5 latency/throughput.
On average, this code is 30% faster than our old libavcodec implementation.
It's able to trade blows with the previously-untouchable FFTW on small transforms,
and due to its tiny size and better prediction, outdoes FFTW on larger transforms
by 11% on the largest currently supported size.
This sadly required making changes to the code itself,
due to the same context needing to be reused for both versions.
The lookup table had to be duplicated for both versions.
This commit refactors the power-of-two FFT, making it faster and
halving the size of all tables, making the code much smaller on
all systems.
This removes the big/small pass split, because on modern systems
the "big" pass is always faster, and even on older machines there
is no measurable speed difference.
It is only supposed to be freed by libavcodec for decoders, yet
avcodec_open2() always frees it on failure.
Furthermore, avcodec_close() doesn't free it for decoders.
Both of this has been changed.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: signed integer overflow: 9223372036840103978 + 67637280 cannot be represented in type 'long'
Fixes: 33341/clusterfuzz-testcase-minimized-ffmpeg_dem_DSF_fuzzer-6408154041679872
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 9223372036846336888 + 4278255871 cannot be represented in type 'long'
Fixes: 32782/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-6059216516284416
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Before 9b3c46a081 every call to
ff_jpegls_decode_picture() allocated and freed a JLSState. This commit
instead put said structure into the context of the JPEG-LS decoder to
avoid said allocation. But said function can also be called from other
MJPEG-based decoders and their contexts doesn't contain said structure,
leading to segfaults. This commit fixes this: The JLSState is now
allocated on the first call to ff_jpegls_decode_picture() and stored in
the context.
Found-by: Michael Niedermayer <michael@niedermayer.cc>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: signed integer overflow: -1184429040541376544 * 32 cannot be represented in type 'long'
Fixes: 31788/clusterfuzz-testcase-minimized-ffmpeg_dem_CAF_fuzzer-6236746338664448
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
av_set_cpu_flags_mask() has been deprecated in the commit which merged
it: 6df42f98746be06c883ce683563e07c9a2af983f; av_parse_cpu_flags() has
been deprecated in 4b529edff8.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Some files currently rely on libavutil/cpu.h to include it for them;
yet said file won't use include it any more after the currently
deprecated functions are removed, so include attributes.h directly.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The build log:
** Unknown command `@code' (left as is) (in src/doc/muxers.texi l. 2020)
*** '{' without macro. Before: -map} option with the ffmpeg CLI tool. (in src/doc/muxers.texi l. 2020)
*** '}' without opening '{' before: option with the ffmpeg CLI tool. (in src/doc/muxers.texi l. 2020)
Relying on the order of the enum is bad.
It clashes with the new presets having to sit at the end of the list, so
that they can be properly filtered out by the options parser on builds
with older SDKs.
So this refactors nvenc.c to instead rely on the internal NVENC_LOSSLESS
flag. For this, the preset mapping has to happen much earlier, so it's
moved from nvenc_setup_encoder to nvenc_setup_device and thus runs
before the device capability check.
This would only make a difference in case the first attempt to
initialize the encoder failed and the second succeeded. The only
reason I can think of for this to happen is that the options (in
particular the codec whitelist) are not used for the second try
and that obviously implies that we should not even try a second time
to open the decoder.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
ffio_fill() is used when initially writing unknown length elements;
yet it can happen that the amount of bytes written by it is zero in
which case it is of course unnecessary to ever call it. Whether it is
possible to know this during compiletime depends upon how aggressively
the compiler inlines function calls (i.e. if it inlines calls to
start_ebml_master() where the upper bound for the size of the element
implies that the size will be written on one byte) and this depends upon
optimization settings. It is not the aim of this patch to inline all
calls where it is known that ffio_fill() will be unnecessary, but merely
to make compilers that inline such calls aware of the fact that writing
zero bytes with ffio_fill() is unnecessary. To this end
av_builtin_constant_p() is used to check whether the size is a
compiletime constant.
For GCC 10 this made a difference at -O3 only: The size of .text
decreased from 0x747F (with 29 calls to ffio_fill(), eight of which
use size zero) to 0x7337 (with 21 calls to ffio_fill(), zero of which
use size zero).
For Clang 11 it made a difference at -O2 and -O3: At -O2, the size of
.text decreased from 0x879C to 0x871C (with eight calls to ffio_fill()
eliminated); at -O3 the size of .text decreased from 0xAF2F to 0xAEBF.
Once again, eight calls to ffio_fill() with size zero have been
eliminated.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now the cover images will get the stream index 0 in this case,
violating the hardcoded assumption that this is the index of the audio
stream. Fix this by creating the audio stream first; this is also in
line with the expectations of ff_pcm_read_seek() and
ff_spdif_read_packet(). It also simplifies the code to parse the fmt and
xma2 tags.
Fixes#8540; regression since f5aad350d3.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is simpler and more complete (e.g. it copies the framerate
information which allows to write the default duration element).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When parsing ID3v2 tags, special (non-text) metadata is not applied
directly and unconditionally; instead it is stored in a linked list
in which elements are prepended. When traversing the list to add APICs
(or private tags) at the end, the order is reversed. The same also
happens for chapters and therefore the chapter parsing code already
reverses the chapters.
This commit changes this: By keeping pointers to both head and tail
of the linked list one can preserve the order of the entries and
remove the reordering code for chapters. Only the pointer to head
will be exported: No current caller uses a nonempty list, so exporting
both head and tail is unnecessary. This removes the functionality
to combine the lists of special metadata read from different ID3v2 tags,
but that doesn't make really much sense anyway (and would be trivial
to implement if desired) and allows to remove the now unnecessary
initializations performed by the callers.
The FATE-reference for the id3v2-priv test had to be updated
because the order of the tags read into the dict is reversed;
for id3v2-priv-remux only the md5 and not the ffprobe output
of the remuxed file changes because the order of the private tags
has up until now been reversed twice.
The references for the aiff/mp3 cover-art tests needed to be updated,
because the order of the attached pics is reversed upon reading.
It is still not correct, because the muxers write the pics in the order
in which they arrive at the muxer instead of the order given by
pkt->stream_index.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
write_header() already checks that there are only video tracks besides
the one audio track.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Notice that the order of the APIC tracks is currently wrong. This is
a superposition of two bugs: (i) Both muxers write the attached
pictures in the order they arrive in the muxer and not in the
stream_index order, leading to attached pictures that are copied being
written earlier because their timestamp is AV_NOPTS_VALUE, whereas the
timestamp of the encoded pictures is 0. (ii) A bug in the id3v2 parsing
code reverses the order of the parsed pictures.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Don't blindly copy all bytes in extradata past ChannelMappingFamily. Instead
check if ChannelMappingFamily is not 0 and then only write the correct amount
of bytes from ChannelMappingTable, as defined in the spec[1].
Fixes part of ticket #9190.
[1] https://opus-codec.org/docs/opus_in_isobmff.html#4.3.2
Signed-off-by: James Almer <jamrial@gmail.com>
The libwebp_animencoder returns a single packet with the entire animated
stream, as that's what the external library produces. As such, only ensure the
stream was produced by said encoder (or propagated by a demuxer, once support
is added) when attempting to write the requested loop value.
Fixes ticket #9179.
Signed-off-by: James Almer <jamrial@gmail.com>
The only packet produced by this encoder contains the entire animated stream,
so set its pts to the first frame encoded.
Signed-off-by: James Almer <jamrial@gmail.com>
Packets must have at least one of data or side_data. If none are available,
then got_packet must not be signaled.
The generic encode code already discarded these empty packets, but it's better
just not propagating them at all.
Signed-off-by: James Almer <jamrial@gmail.com>
Remove the unneeded wrapping sequence element. Also the
minOccurs/maxOccurs occurrence indicators on the inner element
definitions can be removed.
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
The "packets_and_frames" element has been added to ffprobe.xsd in
0c9f0da0f7 but apparently removing the
check in ffprobe.c has been forgotten.
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
NellyMoserEncodeContext.avctx is only set in init after these checks,
yet it is used by encode_end().
This is a regression since 0a56bfa71f.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
From the ISO/IEC specification for MP4:
The pixel aspect ratio and clean aperture of the video may be specified
using the ‘pasp’ and ‘clap’ sample entry boxes, respectively. These are
both optional; if present, they over-ride the declarations (if any) in
structures specific to the video codec, which structures should be
examined if these boxes are absent. For maximum compatibility, these
boxes should follow, not precede, any boxes defined in or required by
derived specifications.
Fixes trac/#7277.
Move the loop counter decrement further from the branch instruction,
this hides the latency of the decrement.
In loops that first load, then store (the horizontal prediction cases),
do the decrement after the load (where the next instruction would
stall a bit anyway, waiting for the result of the load).
In loops that store twice using the same destination register,
also do the decrement between the two stores (as the second store
would need to wait for the updated destination register from the
first instruction).
In loops that store twice to two different destination registers,
do the decrement before both stores, to do it as soon before the
branch as possible.
This gives minor (1-2 cycle) speedups in most cases (modulo measurement
noise), but the horizontal prediction functions get a rather notable
speedup on the Cortex A53.
Before: Cortex A53 A72 A73
pred8x8_dc_8_neon: 60.7 46.2 39.2
pred8x8_dc_128_8_neon: 30.7 18.0 14.0
pred8x8_horizontal_8_neon: 42.2 29.2 18.5
pred8x8_left_dc_8_neon: 52.7 36.2 32.2
pred8x8_mad_cow_dc_0l0_8_neon: 48.2 27.7 25.7
pred8x8_mad_cow_dc_0lt_8_neon: 52.5 33.2 34.7
pred8x8_mad_cow_dc_l0t_8_neon: 52.5 31.7 33.2
pred8x8_mad_cow_dc_l00_8_neon: 43.2 27.0 25.5
pred8x8_plane_8_neon: 112.2 86.2 88.2
pred8x8_top_dc_8_neon: 40.7 23.0 21.2
pred8x8_vertical_8_neon: 27.2 15.5 14.0
pred16x16_dc_8_neon: 91.0 73.2 70.5
pred16x16_dc_128_8_neon: 43.0 34.7 30.7
pred16x16_horizontal_8_neon: 86.0 49.7 44.7
pred16x16_left_dc_8_neon: 87.0 67.2 67.5
pred16x16_plane_8_neon: 236.0 175.7 173.0
pred16x16_top_dc_8_neon: 53.2 39.0 41.7
pred16x16_vertical_8_neon: 41.7 29.7 31.0
After:
pred8x8_dc_8_neon: 59.0 46.7 42.5
pred8x8_dc_128_8_neon: 28.2 18.0 14.0
pred8x8_horizontal_8_neon: 34.2 29.2 18.5
pred8x8_left_dc_8_neon: 51.0 38.2 32.7
pred8x8_mad_cow_dc_0l0_8_neon: 46.7 28.2 26.2
pred8x8_mad_cow_dc_0lt_8_neon: 55.2 33.7 37.5
pred8x8_mad_cow_dc_l0t_8_neon: 51.2 31.7 37.2
pred8x8_mad_cow_dc_l00_8_neon: 41.7 27.5 26.0
pred8x8_plane_8_neon: 111.5 86.5 89.5
pred8x8_top_dc_8_neon: 39.0 23.2 21.0
pred8x8_vertical_8_neon: 27.2 16.0 14.0
pred16x16_dc_8_neon: 85.0 70.2 70.5
pred16x16_dc_128_8_neon: 42.0 30.0 30.7
pred16x16_horizontal_8_neon: 66.5 49.5 42.5
pred16x16_left_dc_8_neon: 81.0 66.5 67.5
pred16x16_plane_8_neon: 235.0 175.7 173.0
pred16x16_top_dc_8_neon: 52.0 39.0 41.7
pred16x16_vertical_8_neon: 40.2 33.2 31.0
Despite this, a number of these functions still are slower than
what e.g. GCC 7 generates - this shows the relative speedup of the
neon codepaths over the compiler generated ones:
Cortex A53 A72 A73
pred8x8_dc_8_neon: 0.86 0.65 1.04
pred8x8_dc_128_8_neon: 0.59 0.44 0.62
pred8x8_horizontal_8_neon: 1.51 0.58 1.30
pred8x8_left_dc_8_neon: 0.72 0.56 0.89
pred8x8_mad_cow_dc_0l0_8_neon: 0.93 0.93 1.37
pred8x8_mad_cow_dc_0lt_8_neon: 1.37 1.41 1.68
pred8x8_mad_cow_dc_l0t_8_neon: 1.21 1.17 1.32
pred8x8_mad_cow_dc_l00_8_neon: 1.24 1.19 1.60
pred8x8_plane_8_neon: 3.36 3.58 3.76
pred8x8_top_dc_8_neon: 0.97 0.99 1.43
pred8x8_vertical_8_neon: 0.86 0.78 1.18
pred16x16_dc_8_neon: 1.20 1.06 1.49
pred16x16_dc_128_8_neon: 0.83 0.95 0.99
pred16x16_horizontal_8_neon: 1.78 0.96 1.59
pred16x16_left_dc_8_neon: 1.06 0.96 1.32
pred16x16_plane_8_neon: 5.78 6.49 7.19
pred16x16_top_dc_8_neon: 1.48 1.53 1.94
pred16x16_vertical_8_neon: 1.39 1.34 1.98
In particular, on Cortex A72, many of these functions are slower
than the compiler generated code, while they're more beneficial on
e.g. the Cortex A73.
Signed-off-by: Martin Storsjö <martin@martin.st>
Currently, the DASH demuxer omits the final segment for a non-live
stream (using SegmentTemplate) if it is shorter than the other segments.
Correct calc_max_seg_no to round up when calulating the number of
segments instead of rounding down to resolve this issue.
Signed-off-by: Matt Robinson <git@nerdoftheherd.com>
The MJPEG encoder supports some pixel format/color range combinations
only when strictness is set to unofficial or less. Before commit
059fc2d9da said encoder's pix_fmts array
only included the pixel formats supported with default strictness.
When strictness was <= unofficial, fftools/ffmpeg_filter.c used
an extended list of pixel formats instead of the encoder's including
the pixel formats only supported when strictness <= unofficial.
Said commit turned the logic around: The encoder's pix_fmts array now
included all pixel formats and fftools/ffmpeg_filter.c instead used
a small list of all pixel formats supported when strictness is >
unofficial and the encoder's pixel formats instead. In particular,
the codec's pix_fmt is not used when strictness is normal.
This works for the mjpeg encoder; yet it did not work for other
(hardware-based) mjpeg encoders, because the check for whether one is
using the MJPEG encoder is wrong: It just checks the codec id.
So if one used strict unofficial with a hardware-accelerated MJPEG
encoder before commit 059fc2d9da, the unofficial (non-hardware)
pixel formats of the MJPEG encoder would be used; since said commit
the codec's pixel formats are overridden at ordinary strictness
by the ordinary MJPEG pixel formats. This leads to format conversion
errors lateron which were reported in #9186.
The solution to this is to check AVCodec.name instead of its id.
Fixes ticket #9186.
Tested-by: Eoff, Ullysses A <ullysses.a.eoff@intel.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This encoder has AVCodec.pix_fmts set, so ff_encode_preinit() already
checks for this.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The value zero for AVPacket.duration means that the duration is unknown,
which in practice means "play this subtitle until overridden by the next
subtitle". Yet for Matroska a BlockGroup with duration zero means
that the subtitle really has a duration zero. "Display until overridden"
is achieved by not setting a duration on the container level at all and
this is achieved by using a SimpleBlock or a BlockGroup without
duration. This commit implements this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Apparently for various image sequences libavformat/utils.c can
calculate rather fancy r_frame_rate values, such as `186/1921`,
and since ffmpeg.c utilizes r_frame_rate for the filter chain
time base, this can quite deteriorate the output frame timing - even
though the user has requested the image sequence to be interpreted
at a specific, constant frame rate.
Most of the codecs just need everything zeroed. Those that don't
are either handled inline during decode, or pull state from
extradata.
Move state reset/init functionality into adpcm_flush(), and
invoke it from adpcm_decode_init().
Signed-off-by: Zane van Iperen <zane@zanevaniperen.com>
The block size is hardcoded, so the buffer size is always known.
Statically allocate the buffer on the stack.
Signed-off-by: Zane van Iperen <zane@zanevaniperen.com>
MPEG-1/2/4 are the only mpegvideo based encoders that support bframes;
yet even the encoders not supporting bframes have options that only make
sense for an encoder that supports bframes; setting any of these options
for such an encoder has no impact on the encoded outcome (but setting
b_strategy to two slows down encoding considerably). So deprecate these
options.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The MPEG-2 encoder is the only mpegvideo-based encoder that supports
embedding a53 side data.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
mpeg_quant may only be set for MPEG-4 and MPEG-2, yet for the latter
it is no option as the code acts as if it were always set.
So deprecate the option for all codecs for which it makes no sense.
Furthermore, given that the code already errors out if the option is set
for a codec that doesn't support it we can restrict the range of
the option for all these codecs without breaking something. This means
that the checks for whether mpeg_quant is set for these codecs can be
removed as soon as AVCodecContext.mpeg_quant is removed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This has the advantage that one does not waste some allocations
if one errors out because of these checks.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Currently said list contains only the pixel formats that are always
supported irrespective of the range and the value of
strict_std_compliance. This makes the MJPEG encoder an outlier as all
other codecs put all potentially supported pixel formats into said list
and error out if the chosen pixel format is unsupported. This commit
brings it therefore in line with the other encoders.
The behaviour of fftools/ffmpeg_filter.c has been preserved. A more
informed decision would be possible if colour range were available
at this point, but it isn't.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The documentation for AV_PIX_FMT_YUVJ420P reads:
"planar YUV 4:2:0, 12bpp, full scale (JPEG), deprecated in favor of
AV_PIX_FMT_YUV420P and setting color_range"
Yet the LJPEG encoder only accepts full scale yuv420p when strictness is
set to unofficial or lower; with default strictness it emits a nonsense
error message that says that limit range YUV is unofficial. This has
been changed to allow full range yuv420p, yuv422p and yuv444p irrespective
of the level of strictness.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
All encoders using ff_mpv_encode_init() already have pix_fmts set
so that the pixel format is already checked in ff_encode_preinit().
The one exception to this is MJPEG whose check remains.
(Btw: The AVCodec.pix_fmts check for AMV is stricter than the check
in ff_mpv_encode_init().)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Using optimal Huffman tables is not supported for AMV and always
disabled by ff_mpv_encode_init(); therefore one can build
the AMV encoder without mjpegenc_huffman if one adds the necessary
compile-time checks.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Up until now the relevant checks all checked for the existence of the
MJPEG encoder only.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The ProRes encoder allocates huge worst-case buffers just to be safe;
and for huge resolutions (8k in this case) these can be so big that the
number of bits does no longer fit into a (signed 32-bit) int; this means
that one must no longer use the parts of the PutBits API that deal with
bit counters. Yet proresenc_kostya did it, namely for a check about
whether we are already beyond the end. Yet this check is unnecessary
nowadays, because the PutBits API comes with automatic checks (with
a log message and a av_assert2() in put_bits() and an av_assert0() in
flush_put_bits()), so this is unnecessary. So simply remove the check.
Fixes ticket #9173.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Calling av_frame_make_writable() from decoders is tricky, especially
when frame threading is used. It is much simpler and safer to just make
a private copy of the frame.
This is not expected to have a major performance impact, since
APNG_DISPOSE_OP_BACKGROUND is not used often and
av_frame_make_writable() would typically make a copy anyway.
Found-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The (deprecated) field AVCodecContext.mpeg_quant has no range
restriction; MpegEncContext.mpeg_quant is restricted to 0..1.
If the former is set, the latter is overwritten with it without
checking the range. This can trigger an av_assert2() with the MPEG-4
encoder when writing said field.
Fix this by just setting MpegEncContext.mpeg_quant to 1 if
AVCodecContext.mpeg_quant is set.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The pix_fmts of the LJPEG encoder already contain all supported pixel
formats (including the ones only supported when strictness is unofficial
or less); yet the check in ff_encode_preinit() ignored this list in case
strictness is unofficial or less. But the encoder presumed that it is
always applied and blacklists some of the entries in pix_fmts when
strictness is > unofficial. The result is that if one uses an entry not
on that list and sets strictness to unofficial, said entry passes both
checks and this can lead to segfaults lateron (e.g. when using gray).
Fix this by removing the exception for LJPEG in ff_encode_preinit().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
For both the RealMedia as well as the IVR demuxer (which share the same
context) each AVStream's priv_data contains an AVPacket that might
contain data (even when reading the header) and therefore needs to be
unreferenced. Up until now, this has not always been done:
The RealMedia demuxer didn't do it when allocating a new stream's
priv_data failed although there might be other streams with packets to
unreference. (The reason for this was that until recently rm_read_close()
couldn't handle an AVStream without priv_data, so one had to choose
between a potential crash and a memleak.)
The IVR demuxer meanwhile never ever called read_close so that the data
already contained in packets leaks upon error.
This patch fixes both demuxers by adding the appropriate cleanup code.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This makes sure that reading a truncated chunk will never overflow into
the following chunk. It also allows to remove many repeated lines
skipping over the trailing crc checksum.
This data cannot be stored in PNGDecContext.picture, because the
corresponding chunks may be read after the call to
ff_thread_finish_setup(), at which point modifying shared context data
is a race.
Store intermediate state in the context and then write it directly to
the output frame.
Fixes exporting frame metadata after 5663301560Fixes#8972
Found-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Do not store the image buffer pointer/linesize in the context, just
access them directly from the frame.
Stop assuming that linesize is the same for the current and last frame.
ff_vc1_decode_init_alloc_tables() had one error path that forgot to free
already allocated buffers; these would then be overwritten on the next
allocation attempt (or they would just not be freed in case this
happened during init, as the decoders for which it is used do not have
the FF_CODEC_CAP_INIT_CLEANUP set).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Before 998c9f15d1, initializing an
MpegEncContext's IDCT parts occured in ff_mpv_common_init() and this
has been called in h261_decode_frame(), not h261_decode_init().
Yet said commit factored this out of ff_mpv_common_init() and therefore
there is no reason any more not to set this during init as this commit
does.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The RealVideo 3.0 and 4.0 decoders call ff_mpv_common_init() only during
their init function and not during decode_frame(); when the size of the
frame changes, they call ff_mpv_common_frame_size_change(). Yet upon
error, said function calls ff_mpv_common_end() which frees the whole
MpegEncContext and not only those parts that
ff_mpv_common_frame_size_change() reinits. As a result, the context will
never be usable again; worse, because decode_frame() contains no check
for whether the context is initialized or not, it is presumed that it is
initialized, leading to segfaults. Basically the same happens if
rv34_decoder_realloc() fails.
This commit fixes this by only resetting the parts that
ff_mpv_common_frame_size_change() changes upon error and by actually
checking whether the context is in need of reinitialization in
ff_rv34_decode_frame().
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In case of resolution changes rv20_decode_picture_header() closes and
reopens its MpegEncContext; it checks the latter for errors, yet when
an error happens, it might happen that no new attempt at
reinitialization is performed when decoding the next frame; this leads
to crashes lateron.
This commit fixes this by making sure that initialization will always
be attempted if the context is currently not initialized.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When slice-threading is used, ff_mpv_common_init() duplicates
the first MpegEncContext and allocates some buffers for each
MpegEncContext (the first as well as the copies). But the count of
allocated MpegEncContexts is not updated until after everything has
been allocated and if an error happens after the first one has been
allocated, only the first one is freed; the others leak.
This commit fixes this: The count is now set before the copies are
allocated. Furthermore, the copies are now created and initialized
before the first MpegEncContext, so that the buffers exclusively owned
by each MpegEncContext are still NULL in the src MpegEncContext so
that no double-free happens upon allocation failure.
Given that this effectively touches every line of the init code,
it has also been factored out in a function of its own in order to
remove code duplication with the same code in
ff_mpv_common_frame_size_change() (which was never called when using
more than one slice (and if it were, there would be potential
double-frees)).
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This mostly reverts commit 4b2863ff01.
Said commit removed the freeing code from ff_mpv_common_init(),
ff_mpv_common_frame_size_change() and ff_mpeg_framesize_alloc() and
instead added the FF_CODEC_CAP_INIT_CLEANUP to several codecs that use
ff_mpv_common_init(). This introduced several bugs:
a) Several decoders using ff_mpv_common_init() in their init function were
forgotten: This affected FLV, Intel H.263, RealVideo 3.0 and V4.0 as well as
VC-1/WMV3.
b) ff_mpv_common_init() is not only called from the init function of
codecs, it is also called from AVCodec.decode functions. If an error
happens after an allocation has succeeded, it can lead to memleaks;
furthermore, it is now possible for the MpegEncContext to be marked as
initialized even when ff_mpv_common_init() returns an error and this can
lead to segfaults because decoders that call ff_mpv_common_init() when
decoding a frame can mistakenly think that the MpegEncContext has been
properly initialized. This can e.g. happen with H.261 or MPEG-4.
c) Removing code for freeing from ff_mpeg_framesize_alloc() (which can't
be called from any init function) can lead to segfaults because the
check for whether it needs to allocate consists of checking whether the
first of the buffers allocated there has been allocated. This part has
already been fixed in 76cea1d2ce.
d) ff_mpv_common_frame_size_change() can also not be reached from any
AVCodec.init function; yet the changes can e.g. lead to segfaults with
decoders using ff_h263_decode_frame() upon allocation failure, because
the MpegEncContext will upon return be flagged as both initialized and
not in need of reinitialization (granted, the fact that
ff_h263_decode_frame() clears context_reinit before the context has been
reinited is a bug in itself). With the earlier version, the context
would be cleaned upon failure and it would be attempted to initialize
the context again in the next call to ff_h263_decode_frame().
While a) could be fixed by adding the missing FF_CODEC_CAP_INIT_CLEANUP,
keeping the current approach would entail adding cleanup code to several
other places because of b). Therefore ff_mpv_common_init() is again made
to clean up after itself; the changes to the wmv2 decoder and the SVQ1
encoder have not been reverted: The former fixed a memleak, the latter
allowed to remove cleanup code.
Fixes: double free
Fixes: ff_free_picture_tables.mp4
Fixes: ff_mpeg_update_thread_context.mp4
Fixes: decode_colskip.mp4
Fixes: memset.mp4
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
All codec ids on BSF whitelists have a codec descriptor, so one can just
use avcodec_get_name() without worrying about the case of what happens
when no codec descriptor is found.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
av_frame_copy() is allowed to return values >= 0 on success, whereas
the documentation of av_frame_ref() states that the return value is 0 on
success. Ergo the latter must not just return the former's return value.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
All one needs is one byte beyond the end of the normal data; and because
the packet is padded, one already has it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes the following GCC warning:
warning: format ‘%lld’ expects argument of type ‘long long int’,
but argument 4 has type ‘int64_t’ {aka ‘long int’} [-Wformat=]
Reviewed-by: Gyan Doshi <ffmpeg@gyani.pro>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This fixes building with MSVC after
a2a38b1606.
Remove the stray semicolon, and add casts for the input argument
(which is an intptr_t*) to the right type (PVOID volatile *).
Signed-off-by: Martin Storsjö <martin@martin.st>
If only one of the two arrays used for the ICC profile could be
successfully allocated, it might be overwritten and leak when
the next ICC entry is encountered. Fix this by using a common struct,
so that one has only one array to allocate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
This avoids use of uninitialized data
also several checks are inside the band reading code
so it is important that it is run at least once
Fixes: out of array accesses
Fixes: 28209/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CFHD_fuzzer-5684714694377472
Fixes: 32124/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CFHD_fuzzer-5425980681355264
Fixes: 30519/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CFHD_fuzzer-4558757155700736
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Previously the code skipped all security checks when these where encountered but prior data was incorrect.
Also replace an always true condition by an assert
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array accesses
Fixes: 29754/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CFHD_fuzzer-6333598414274560
Fixes: 30519/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CFHD_fuzzer-6298424511168512
Fixes: 30739/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_CFHD_fuzzer-5011292836462592
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
When using external Huffman tables fails during init, the decoder
reverts back to using the default Huffman tables; and when doing so,
the current VLC tables leak because init_default_huffman_tables()
doesn't free them before overwriting them.
Sample:
samples.ffmpeg.org/archive/all/avi+mjpeg+pcm_s16le++mjpeg-interlace.avi
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
render_charset() used static buffers that are always completely
initialized before every use, so that it is unnecessary for the
values in these arrays to be kept after leaving the function.
Given that this is not only unnecessary, but harmful due to the
possibility of data races if several instances of a64multi/a64multi5
run simultaneously these buffers have been replaced by ordinary buffers
on the stack (they are small enough for this).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The current code tries the access the codecpar of a nonexistent
audio stream when seeking. Stop that. Fixes ticket #9121.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
av_cpu_count() intends to emit a debug message containing the number of
logical cores when called the first time. The check currently works with
a static volatile int; yet this does not help at all in case of
concurrent accesses by multiple threads. So replace this with an
atomic_int.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
A buffer may leak in case of YUVA444P10 with dimensions that are not
both divisible by 16.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When allocating a BSF fails, it could happen that the BSF's close
function has been called despite a failure to allocate the private data.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Also don't unnecessarily copy the input data around if it needn't be
reversed; and remove a redundant memset -- av_fast_padded_malloc()
already does this for us.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
The DSS demuxer currently decrements a counter that should be positive
at the beginning of read_packet; should it become negative, it means
that the data to be read can't be read contiguosly, but has to be read
in two parts. In this case the counter is incremented again after the
first read if said read succeeded; if not, the counter stays negative.
This can lead to problems in further read_packet calls; in tickets #9020
and #9023 it led to segfaults if one tries to seek lateron if the seek
failed and generic seek tried to read from the beginning. But it could
also happen when av_new_packet() failed and the user attempted to read
again afterwards.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When encoding E-AC-3, whether coupling is on or not determines whether
an additional frame based coupling exponent strategy element frmcplexpstr
(of size five bits) is present in the bitstream. So just add five to the
number of bits when counting them instead of adding 5*s->cpl_on (the
latter field is currently only 0 or 1, so it doesn't make a difference).
Furthermore, move some parts of the bit allocation that doesn't change
per-frame to count_frame_bits_fixed() (which is only run once during
init).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
AC-3 and EAC-3 are codecs whose packet sizes are known in advance,
so one can use the min_size parameter of ff_alloc_packet2() to
allocate exactly this amount. This avoids a memcpy later in
av_packet_make_refcounted() in encode_simple_internal().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Since the very beginning (since de6d9b6404)
the AC-3 encoder used AC3_MAX_CODED_FRAME_SIZE (namely 3840) for the
size of the output buffer (without any check at all).
This causes problems when encoding EAC-3 for which the maximum is too small,
smaller than the actual size of the buffer: One can run into asserts used
by the PutBits API. Ticket #8513 is about such a case and this commit
fixes it by using the real size of the buffer.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
All instances of adding attached pictures to a stream or adding
a stream and an attached packet to said stream have several things
in common like setting the index and flags of the packet, setting
the stream disposition etc. This commit therefore factors this out.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Before commit f1e17eb446, the qtrle
encoder had undefined pointer arithmetic: Outside of a loop, two
pointers were set to point to the ith element (with index i-1) of
a line of a frame. At the end of each loop iteration, these pointers
were decremented, so that they pointed to the -1th element of the line
after the loop. Furthermore, one of these pointers can be NULL (in which
case all pointer arithmetic is automatically undefined behaviour).
Commit f1e17eb44 added a check in order to ensure that the elements
never point to the -1th element of the array: The pointers are only
decremented if they are bigger than the frame's base pointer
(i.e. AVFrame.data[0]). Yet this check does not work at all in case of
negative linesizes; furthermore in case the pointer that can be NULL is
NULL initializing it still involves undefined pointer arithmetic.
This commit fixes both of these issues: First, non-NULL pointers are
initialized to point to the element after the ith element and
decrementing is moved to the beginning of the loop. Second, if a pointer
is NULL, it is just made to point to the other pointer, as this allows
to avoid checks before decrementing it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
If keeping a reference to an earlier frame failed, the next frame must
be an I frame for lack of reference frame. This commit implements this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Affected ProRes without alpha; affected 32 FATE tests, e.g. prores-422,
prores-422_proxy, prores-422_lt or matroska-prores-header-insertion-bz2.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Print every error in the stack, if more than one, and don't print
bogus errors if there's none logged within OpenSSL.
Retain the underlying IO error code, print an error message out of
it, and pass the error code on to the caller.
Signed-off-by: Martin Storsjö <martin@martin.st>
Fixes: signed integer overflow: -9223372036854775760 - 50 cannot be represented in type 'long'
Fixes: 31673/clusterfuzz-testcase-minimized-ffmpeg_dem_MPC8_fuzzer-580134751869337
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -9223372032574480351 - 4294967296 cannot be represented in type 'long long'
Fixes: 30022/clusterfuzz-testcase-minimized-ffmpeg_dem_KUX_fuzzer-5568610275819520
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Both functions to read attached pictures coincide since
e83f27a21a (save for some log messages
in case av_dict_set failed).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
When allocating the MJpegContext fails (or if the dimensions run afoul
of the 65500x65500 limit), an attempt to free a subbuffer of said
context leads to a segfault in ff_mjpeg_encode_close().
Seems to be a regression since 467d9e27e0.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
In this case it also fixes a potential for compilation failures:
Not all compilers can handle the case in which a function with
a forward declaration declared with an attribute to always inline it
is called before the function body appears. E.g. GCC 4.2.1 on OS X 10.6
doesn't like it.
Reviewed-by: Pavel Koshevoy <pkoshevoy@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Specifically test that the WebVTT flavour is correctly mapped to
the Matroska/WebM CodecID and back; and test that dispositions
unsupported by WebM are discarded even when they would be supported
by Matroska.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Some old DV AVI files have the DSF-Flag of frames set to 0, although it
is PAL (maybe rendered with an old Ulead Media Studio Pro) ... this causes
ffmpeg/VLC-player to produce/play corrupted video (other players/editors
like VirtualDub work fine).
Fixes ticket #8333 and replaces/extends hack for ticket #2177
Signed-off-by: Marton Balint <cus@passwd.hu>
Timestamp difference is available in media timebase (1/90K) where as
rtcp time is in the default microseconds timebase. This patch fixes
the calculated prft wallclock time by rescaling the timestamp delta
to the microseconds timebase.
Signed-off-by: James Almer <jamrial@gmail.com>
A PutBitContext has a field called size_in_bits which is set to the
context's bitsize init_put_bits(); but it isn't used at all (the PutBits
API uses pointers directly and not bit indexes), so remove it (due to
ABI concerns the actual element is only removed at the next bump).
Furthermore, the multiplication inherent in setting this field can lead
to undefined integer overflows. This is particularly true for FFV1,
which uses a very big worst-case buffer (37*4*width*height; even
ordinary 1080p triggers an overflow). Ticket #8350 is about this
overflow which this commit fixes.
This means that the effective range of the PutBits API is no longer
restricted by the /8 as long as one isn't using put_bits_(count|left).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The function to write an ordinary (luma or chroma) plane as well as
the function for writing an alpha plane have some similarities:
They record the initial bitposition (despite said position always being
byte-aligned), flush the PutBitContext themselves and return the amount
of bytes they wrote.
This commit factors this out; it also replaces bitpositions by
bytepositions and it avoids recording the initial byteposition because
said information is already available from the position at the end of
the last plane.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Despite write_huff_codes() receiving an ordinary buffer (not a
PutBitContext), it returned the amount of data written in bits,
not in bytes. This has been changed: There is now no intermediate
bitcount any more.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Several encoders used code like the following to check for the amount of
bytes left in a PutBitContext:
pb->buf_end - pb->buf - (put_bits_count(pb) >> 3)
Besides the fact that using the pointers directly might pose
a maintainence burden in the future this also leads to suboptimal code:
The above code reads all three pointers (buf, buf_ptr and buf_end), but
touching buf is unnecessary and switching to put_bytes_left()
automatically fixes this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Often a caller doesn't want the amount of bits written via a
PutBitContext, but the amount of bytes. This in particular happens
after one has flushed the PutBitContext (e.g. at the end of encoding,
when one wants to know the actual packet size). The current way of doing
this is with put_bits_count(pb)/8 (or (put_bits_count(pb) + 7)/8).
Yet this has some issues: It contains implicit multiplications and
divisions by 8 with a cast in between; it obscurs the intent; and
it restricts the size of the buffer to (currently) INT_MAX/8 (or
to 1/8 of the maximum of whatever put_bits_count() returns), although
said restriction is not really necessary for users that don't need
a bitcount.
Corresponding functions for the amount of bytes left have also been
addded.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The PutBits API checks the available space before every write,
so this check for overread is dead.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
We are already word-aligned here, so one can just as well flush the main
PutBitContext.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Compilation would fail if it were outcommented as it refers to a
nonexistent PutBitContext.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Fixes: signed integer overflow: 136323327 * 281474976710656 cannot be represented in type 'long'
Fixes: 30913/clusterfuzz-testcase-minimized-ffmpeg_dem_IVF_fuzzer-5753392189931520
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This avoids crafted files from consuming excessive resources recomputing the clut after each pixel change
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The check has been added at a time when the code performed the
multiplication itself instead of deferring it to av_malloc_array()
and when our allocation functions used unsigned instead of size_t.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The check has been added at a time when the code performed the
multiplication itself instead of deferring it to av_malloc_array()
and when our allocation functions used unsigned instead of size_t.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This check is outdated because the caller doesn't need to check that
the multiplication overflows when using av_realloc_array() (the code
in question used av_realloc() before that); furthermore, the check
is also a remnant of the time in which our allocation functions
didn't use size_t parameters.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
by keeping the variable uint32_t which in this situation is the natural
type anyway. This affected the FATE-test filter-paletteuse-sierra2_4a.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
by using a multiplication instead. The multiplication can never overflow
an int because the sin-factor is only an int16_t.
Affected the FATE-tests filter-concat and filter-concat-vfr.
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Affects the FATE-tests webm-dash-manifest-unaligned-video-streams,
webm-dash-manifest and webm-dash-manifest-representations.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Fixes: signed integer overflow: -2147471366 - 18638 cannot be represented in type 'int'
Fixes: 30157/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SONIC_fuzzer-5171199746506752
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -1957694447 + -1620425806 cannot be represented in type 'int'
Fixes: 30207/clusterfuzz-testcase-minimized-ffmpeg_dem_JACOSUB_fuzzer-5050791771635712
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The user buffers passed to avcodec_encode_video2() haven't been propagated to
AVCodec.encode2 implementations since 93016f5d1d.
Also, the generic encode code already unrefs the packet if nothing was encoded.
Signed-off-by: James Almer <jamrial@gmail.com>
This is important, for example, for connection timed out events,
when used over a network, returning AVERROR(ETIMEDOUT).
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Fixes: out of array access
Fixes: 31640/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SGA_fuzzer-5630883286614016
Fixes: 31619/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SGA_fuzzer-5176667708456960
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 2314885530818453536 - -9070214327174160352 cannot be represented in type 'long'
Fixes: 31000/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FFWAVESYNTH_fuzzer-6558389742206976
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Up until now, initializing the mutexes/condition variables wasn't
checked by ff_frame_thread_init(). This commit changes this.
Given that it is not documented to be save to destroy a zeroed but
otherwise uninitialized mutex/condition variable, one has to choose
between two approaches: Either one duplicates the code to free them
in ff_frame_thread_init() in case of errors or one records which have
been successfully initialized. This commit takes the latter approach:
For each of the two structures with mutexes/condition variables
an array containing the offsets of the members to initialize is added.
Said array is used both for initializing and freeing and the only thing
that needs to be recorded is how many of these have been successfully
initialized.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
In case an error happened when setting up the child threads,
ff_frame_thread_init() would up until now call ff_frame_thread_free()
to clean up all threads set up so far, including the current, not
properly initialized one.
But a half-allocated context needs special handling which
ff_frame_thread_frame_free() doesn't provide.
Notably, if allocating the AVCodecInternal, the codec's private data
or setting the options fails, the codec's close function will be
called (if there is one); it will also be called if the codec's init
function fails, regardless of whether the FF_CODEC_CAP_INIT_CLEANUP
is set. This is not supported by all codecs; in ticket #9099 it led
to a crash.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Also free the gme_info_t structure immediately after its use.
This simplifies cleanup, because it might be unsafe to call
gme_free_info(NULL) (or even worse, gme_track_info() might even
on error set the pointer to the gme_info_t structure to something
else than NULL).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Commit 003b5c800f introduced seeking in argo_asf,
but this was missed, leading to non-deterministic output.
Signed-off-by: Zane van Iperen <zane@zanevaniperen.com>
Reduces codesize because the offset in pointer+offset addressing
requires less bytes to encode. Reduces the size of .text from 8871B
to 8146B (GCC 10, -O3, x64).
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Up until now, the VC-1 decoders allocated an AVFrame for usage with
sprites during vc1_decode_init(); yet said AVFrame can be freed if
(re)initializing the context (which happens ordinarily during decoding)
fails. The AVFrame does not get allocated again lateron in this case,
leading to segfaults.
Fix this by moving the allocation of said frame immediately before it is
used (this also means that said frame won't be allocated at all any more
in case of a regular (i.e. non-image) stream).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
It automatically records the current length of the string,
whereas the current code contains lots of instances of
snprintf(buf + strlen(buf), buf_size - strlen(buf), ...).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
If the numerical constants for colorspace, transfer characteristics
and color primaries coincide, the current code presumes the
corresponding names to be identical and prints only one of them obtained
via av_get_colorspace_name(). There are two issues with this: The first
is that the underlying assumption is wrong: The names only coincide in
the 0-7 range, they differ for more recent additions. The second is that
av_get_colorspace_name() is outdated itself; it has not been updated
with the names of the newly defined colorspaces.
Fix both of this by using the names from
av_color_(space|primaries|transfer)_name() and comparing them via
strcmp; don't use av_get_colorspace_name() at all.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Our "get name" functions can return NULL for invalid/unknown
arguments. So check for this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When the trailer is never written (or when a stream switches from
non-animation mode to animation mode mid-stream), a cached packet
(if existing) would leak. Fix this by adding a deinit function.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The WebP muxer sometimes caches a packet it receives to write it later;
yet if a cached packet is too small (so small as to be invalid),
it is cached, but not written and not unreferenced. Such a packet leaks,
either by being overwritten by the next packet or because it is never
unreferenced at all.
Fix this by not caching unusable packets at all; and error out on
invalid packets.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Replace it in ipmovie_read_header() by AVFormatInternal.parse_pkt
which is unused when reading the header.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
They will be discarded anyway because this can only happen
for invalid data. This already implies that the pkt won't be used
at all when parsing the very first chunk when reading the header,
so one can use NULL as argument and remove the av_packet_unref()
on error.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When one of these errors happens during ipmovie_read_packet(),
an error is returned and the packet is cleaned up generically.
And since 712d3ac539 the same happens
in ipmovie_read_header().
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Replace it by using AVFormatInternal.parse_pkt which is otherwise unused
when reading a header.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Before 8d78e90a6b the Matroska demuxer
used stack packets to hold temporary packets; now it uses a temporary
packet allocated by the Matroska demuxer. Yet because it used stack
packets the code has always properly reset the packet on error, while
on success these temporary packets were put into a packet list via
avpriv_packet_list_put(), which already resets the source packet.
This means that this code is compatible with just reusing
AVFormatInternal.parse_pkt (which is unused while one is in the
demuxer's read_packet() function). Compared to before 8d78e90a6
this no longer wastes one initialization per AVPacket read
(the resetting of the stack packet performed by av_packet_move_ref()
in avpriv_packet_list_put() was for naught).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Originally added in 12f996edfa
behind #if 0; aebb56e184 then
removed the #if and replaced it by using av_dlog. Then commit
1a3eb042c7 replaced this with av_log
at trace level. Yet the code block always stayed within { }
at an increased level of indentation.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The SVQ3 decoder modifies the input bitstream at two places.
One of them is only reached when the file is watermarked.
Therefore commit 2264c11081
made a copy of all the frame data in this case.
But there is a second possibility for modifying the frame and
therefore Libav commit 1098f5c049
made the decoder always copy the data. This of course makes
the additional copy for watermarked frames redundant, but it hasn't
been removed. This commit does so.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
It takes care of zeroing padding (which has been forgotten here).
Also rename the size variable to indicate that this is not the size
of the current slice.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This makes av_read_frame() return packets with proper timestamps.
As a result, seeking now works in combination with streamcopy.
A FATE-test for this has been added.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Currently, ff_read_packet() sometimes forwards the return value of
AVInputFormat.read_packet() (which should be zero on success, but isn't
for all demuxers) and sometimes it overwrites this with zero.
Furthermore, it uses two variables, one for the read_packet return value
and one for other errors, which is a bit confusing; it is also
unnecessary given that the documentation explicitly states that
ff_read_packet() never returns positive values. Returning a positive
value would lead to leaks with some callers (namely asfrtp_parse_packet
and estimate_timings_from_pts). So always return zero in case of
success.
(This behaviour stems from a time before av_read_packet sanitized
the return value of read_packet at all: It was added in commit
626004690c and was unnecessary since
88b00723906f68b7563214c30333e48888dddf78.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Because the properties of frames returned from ff_get/reget_buffer
are not reset at all, lots of returned frames had palette_has_changed
wrongly set to 1. This has been changed, too.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This field needs to be replaced altogether, not just its type changed.
This will be done in a separate change.
Signed-off-by: James Almer <jamrial@gmail.com>
The test sample has to have no file extension, otherwise probing
happens to work, based off file extension alone, and we want to
test the actual probing function.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
When extended atom size support was added to probing in
fec4a2d232, the buffer
size check was backwards, but probing continued to work
because there was no minimum size check yet, so despite
size being 1 on these atoms, and failing to read the 64-bit
size, the tag was still correctly read.
When 0b78016b2d introduced a
minimum size check, this exposed the bug, and broke probing
any files with extended atom sizes, such as entirely valid
large files that start whith mdat atoms.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Commit 6973df1122 added support
for music tracks by outputting its two containing tracks
together in one packet. But the actual data is not contiguous
in the file and therefore one can't simply use av_get_packet()
(which has been used before) for it. Therefore the packet was
now allocated via av_new_packet() and read via avio_read();
and this is also for non-music files.
This causes problems because one can now longer rely on things
done automatically by av_get_packet(): It automatically freed
the packet in case of errors; this lead to memleaks in several
FATE-tests covering this demuxer. Furthermore, in case the data
read is less than the data desired, the returned packet was not
zero-allocated (the packet's padding was uninitialized);
for music files the actual data could even be uninitialized.
The former problems are fixed by using av_get_packet() for
non-music files; the latter problem is handled by erroring out
unless both tracks could be fully read.
Reviewed-by: Zane van Iperen <zane@zanevaniperen.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Always leaving said packet in a blank state after having used it
allows to avoid having to reset it before one uses it; and it also
allows to use it in more places than just in parse_packet() here.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When flushing, the parser receives a dummy buffer with padding
that lives on the stack of av_parser_parse2(). Certain parsers
(e.g. Dolby E) only analyze the input, but don't repack it. When
flushing, such parsers return a pointer to the stack buffer and
a size of 0. And this is also what av_parser_parse2() returns.
Fix this by always resetting poutbuf in case poutbuf_size is zero.
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Added in dc51a72ba4, yet even back then
the check was always true as the AVCodecContext has already been memset
to zero before that.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This is how it is supposed to happen, yet when using frame threading,
the codec's init function has been called before preinit. This can lead
to crashes when e.g. using unsupported lowres values for decoders
together with frame threading.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
These fields can't be set via AVOptions, ergo one can check them before
having allocated anything.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Add a chapter. @var{id} is an unique identifier, possibly small and
consecutive.
@end table
@subsection Options
@@ -174,7 +207,8 @@ This demuxer accepts the following option:
@table @option
@item safe
If set to 1, reject unsafe file paths. A file path is considered safe if it
If set to 1, reject unsafe file paths and directives.
A file path is considered safe if it
does not contain a protocol specification and is relative and all components
only contain characters from the portable character set (letters, digits,
period, underscore and hyphen) and have no period at the beginning of a
@@ -184,9 +218,6 @@ If set to 0, any file name is accepted.
The default is 1.
-1 is equivalent to 1 if the format was automatically
probed and 0 otherwise.
@item auto_convert
If set to 1, try to perform automatic conversions on packet data to make the
streams concatenable.
@@ -243,11 +274,18 @@ which streams to actually receive.
Each stream mirrors the @code{id} and @code{bandwidth} properties from the
@code{<Representation>} as metadata keys named "id" and "variant_bitrate" respectively.
@section flv, live_flv
@section imf
Interoperable Master Format demuxer.
This demuxer presents audio and video streams found in an IMF Composition.
@section flv, live_flv, kux
Adobe Flash Video Format demuxer.
This demuxer is used to demux FLV files and RTMP network streams. In case of live network streams, if you force format, you may use live_flv option instead of flv to survive timestamp discontinuities.
KUX is a flv variant used on the Youku platform.
@example
ffmpeg -f flv -i myfile.flv ...
@@ -346,6 +384,9 @@ Enabled by default for HTTP/1.1 servers.
@item http_seekable
Use HTTP partial requests for downloading HTTP segments.
Set options for the demuxer of media segments using a list of key=value pairs separated by @code{:}.
@end table
@section image2
@@ -661,6 +702,12 @@ Set mfra timestamps as PTS
Don't use mfra box to set timestamps
@end table
@item use_tfdt
For fragmented input, set fragment's starting timestamp to @code{baseMediaDecodeTime} from the @code{tfdt} box.
Default is enabled, which will prefer to use the @code{tfdt} box to set DTS. Disable to use the @code{earliest_presentation_time} from the @code{sidx} box.
In either case, the timestamp from the @code{mfra} box will be used if it's available and @code{use_mfra_for} is
set to pts or dts.
@item export_all
Export unrecognized boxes within the @var{udta} box as metadata entries. The first four
characters of the box type are set as the key. Default is false.
@@ -679,6 +726,15 @@ specify.
@item decryption_key
16-byte key, in hex, to decrypt files encrypted using ISO Common Encryption (CENC/AES-128 CTR; ISO/IEC 23001-7).
@item max_stts_delta
Very high sample deltas written in a trak's stts box may occasionally be intended but usually they are written in
error or used to store a negative value for dts correction when treated as signed 32-bit integers. This option lets
the user set an upper limit, beyond which the delta is clamped to 1. Values greater than the limit if negative when
cast to int32 are used to adjust onward dts.
Unit is the track time scale. Range is 0 to UINT_MAX. Default is @code{UINT_MAX - 48000*10} which allows upto
a 10 second dts correction for 48 kHz audio streams while accommodating 99.9% of @code{uint32} range.
This is a deprecated option to set the segment length in microseconds, use @var{seg_duration} instead.
@item seg_duration @var{duration}
Set the segment length in seconds (fractional value can be set). The value is
treated as average segment duration when @var{use_template} is enabled and
@@ -337,12 +362,13 @@ Ignore IO errors during open and write. Useful for long-duration runs with netwo
@item lhls @var{lhls}
Enable Low-latency HLS(LHLS). Adds #EXT-X-PREFETCH tag with current segment's URI.
Apple doesn't have an official spec for LHLS. Meanwhile hls.js player folks are
trying to standardize a open LHLS spec. The draft spec is available in https://github.com/video-dev/hlsjs-rfcs/blob/lhls-spec/proposals/0001-lhls.md
This option will also try to comply with the above open spec, till Apple's spec officially supports it.
Applicable only when @var{streaming} and @var{hls_playlist} options are enabled.
hls.js player folks are trying to standardize an open LHLS spec. The draft spec is available in https://github.com/video-dev/hlsjs-rfcs/blob/lhls-spec/proposals/0001-lhls.md
This option tries to comply with the above open spec.
It enables @var{streaming} and @var{hls_playlist} options automatically.
This is an experimental feature.
Note: This is not Apple's version LHLS. See @url{https://datatracker.ietf.org/doc/html/draft-pantos-hls-rfc8216bis}
@item ldash @var{ldash}
Enable Low-latency Dash by constraining the presence and values of some elements.
@@ -380,6 +406,137 @@ adjusting playback latency and buffer occupancy during normal playback by client
@end table
@anchor{fifo}
@section fifo
The fifo pseudo-muxer allows the separation of encoding and muxing by using
first-in-first-out queue and running the actual muxer in a separate thread. This
is especially useful in combination with the @ref{tee} muxer and can be used to
send data to several destinations with different reliability/writing speed/latency.
API users should be aware that callback functions (interrupt_callback,
io_open and io_close) used within its AVFormatContext must be thread-safe.
The behavior of the fifo muxer if the queue fills up or if the output fails is
selectable,
@itemize @bullet
@item
output can be transparently restarted with configurable delay between retries
based on real time or time of the processed stream.
@item
encoding can be blocked during temporary failure, or continue transparently
dropping packets in case fifo queue fills up.
@end itemize
@table @option
@item fifo_format
Specify the format name. Useful if it cannot be guessed from the
output name suffix.
@item queue_size
Specify size of the queue (number of packets). Default value is 60.
@item format_opts
Specify format options for the underlying muxer. Muxer options can be specified
as a list of @var{key}=@var{value} pairs separated by ':'.
@item drop_pkts_on_overflow @var{bool}
If set to 1 (true), in case the fifo queue fills up, packets will be dropped
rather than blocking the encoder. This makes it possible to continue streaming without
delaying the input, at the cost of omitting part of the stream. By default
this option is set to 0 (false), so in such cases the encoder will be blocked
until the muxer processes some of the packets and none of them is lost.
@item attempt_recovery @var{bool}
If failure occurs, attempt to recover the output. This is especially useful
when used with network output, since it makes it possible to restart streaming transparently.
By default this option is set to 0 (false).
@item max_recovery_attempts
Sets maximum number of successive unsuccessful recovery attempts after which
the output fails permanently. By default this option is set to 0 (unlimited).
@item recovery_wait_time @var{duration}
Waiting time before the next recovery attempt after previous unsuccessful
recovery attempt. Default value is 5 seconds.
@item recovery_wait_streamtime @var{bool}
If set to 0 (false), the real time is used when waiting for the recovery
attempt (i.e. the recovery will be attempted after at least
recovery_wait_time seconds).
If set to 1 (true), the time of the processed stream is taken into account
instead (i.e. the recovery will be attempted after at least @var{recovery_wait_time}
seconds of the stream is omitted).
By default, this option is set to 0 (false).
@item recover_any_error @var{bool}
If set to 1 (true), recovery will be attempted regardless of type of the error
causing the failure. By default this option is set to 0 (false) and in case of
certain (usually permanent) errors the recovery is not attempted even when
@var{attempt_recovery} is set to 1.
@item restart_with_keyframe @var{bool}
Specify whether to wait for the keyframe after recovering from
queue overflow or failure. This option is set to 0 (false) by default.
@item timeshift @var{duration}
Buffer the specified amount of packets and delay writing the output. Note that
@var{queue_size} must be big enough to store the packets for timeshift. At the
end of the input the fifo buffer is flushed at realtime speed.
@end table
@subsection Examples
@itemize
@item
Stream something to rtmp server, continue processing the stream at real-time
rate even in case of temporary failure (network outage) and attempt to recover
@@ -2970,8 +2891,6 @@ static int init_input_stream(int ist_index, char *error, int error_len)
av_log(NULL,AV_LOG_WARNING,"Warning using DVB subtitles for filtering and output at the same time is not fully supported, also see -compute_edt [0|1]\n");
// This ismostlythesameas4x32_pass1, butwithoutthetranspose,
@@ -1849,7 +1849,7 @@ function idct32_1d_4x32_pass2\suffix\()_neon
load_acc_storev24.4s, v25.4s, v26.4s, v27.4s, 1
load_acc_storev28.4s, v29.4s, v30.4s, v31.4s, 1
.purgem load_acc_store
br x14
retx14
endfunc
.endm
@@ -1943,7 +1943,7 @@ function vp9_idct_idct_32x32_add_16_neon
ldp d10, d11, [sp], 0x10
ldp d8, d9, [sp], 0x10
br x15
retx15
endfunc
function ff_vp9_idct_idct_32x32_add_10_neon, export=1
@@ -2009,7 +2009,7 @@ function idct32x32_\size\()_add_16_neon
ldp d10, d11, [sp], 0x10
ldp d8, d9, [sp], 0x10
br x15
retx15
endfunc
.endm
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
