MSYS2 converts paths to MinGW-based applications from unix to
pseudo-windows paths on execution time.
Since there was no space between '-include' and the path, MSYS2 doesn't
detect the path properly.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
codecpar->extradata is not going to change between packets. New extradata
is instead propagated using packet side data.
Use ff_alloc_extradata() as well.
Signed-off-by: James Almer <jamrial@gmail.com>
(cherry picked from commit ce6a98e830)
The initialization of the uint32_t plane32_lut matrix uses left shifts
of the form 1 << plane; plane can be as big as 31 which means that this
is undefined behaviour as 1 will be simply an int. So make it unsigned
to avoid this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit f12e662a3d)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: index -1 out of bounds for type 'AV1ReferenceFrameState [8]'
Fixes: 16079/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-5758807440883712
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: James Almer <jamrial@gmail.com>
See: [FFmpeg-devel] [PATCH 05/13] avcodec/cbs_av1_syntax_template: Check ref_frame_idx before use
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit 8174e5c77d)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: left shift of 1073741824 by 1 places cannot be represented in type 'int'
Fixes: signed integer overflow: 2049431315 + 262759074 cannot be represented in type 'int'
Fixes: 16012/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5719016003338240
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit 392c028cd2)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Alternatively the array could be made bigger but the extra values
would not be read without other changes.
Fixes: Out of array access
Fixes: 15658/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ATRAC9_fuzzer-5738260074070016
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Lynne <dev@lynne.ee>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit 208225bd78)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
cuda_runtime.h as well as dynlink_loader.h used nonstandard inclusion
guards with an AV_ prefix, although these files are not in an libav*/
path. So change the inclusion guards and adapt the ref file of the
source fate test accordingly.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Theora is forced to be non zero if it is zero and a sample
is asked for, as suggested by reimar
Fixes: Timeout (2min -> 600ms)
Fixes: 15366/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_THEORA_fuzzer-5737849938247680
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit b4bf7226af)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: left shift of negative value -4
Fixes: signed integer overflow: -15091694 * 167 cannot be represented in type 'int'
Fixes: signed integer overflow: 1898547155 + 453967445 cannot be represented in type 'int'
Fixes: 15258/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5759095564402688
Fixes: signed integer overflow: 962196438 * 31 cannot be represented in type 'int'
Fixes: 15364/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5718799845687296
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit 267eb2ab7f)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The minimum header is 8 bytes, the smallest bitstream that is passed to
the MB decode code is 4 bytes
Fixes: Timeout (35sec -> 18sec)
Fixes: 15800/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EATQI_fuzzer-5684154517159936
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit 5ffb8e8793)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The minimal size which unpack() will not fail on is 5 bytes
Fixes: Timeout (14sec -> 77ms) (testcase 15508)
Fixes: 15508/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EATGV_fuzzer-5700053513011200
Fixes: 15996/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EATGV_fuzzer-5751353223151616
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit 009ec8dc33)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
An incomplete image is not supported prior to this and will
not produce any output. This commit moves the failure before
time consuming operations.
Fixes: Timeout (81sec -> 76ms)
Fixes: 15723/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_BRENDER_PIX_fuzzer-5147265653538816
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit 38b6c48c43)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
avio_read can successfully return even when less than the requested
amount of input was read. wavdec's bext parsing mistakenly assumed a
successful avio_read always read the full amount that was requested.
The result could be dictionary tags populated with partially
uninitialized values.
This change also fixes a broken assertion in wav_parse_bext_string that
was off-by-one, though no known current usage of that method hits that
broken case.
Chromium bug: 987270
Signed-off-by: Matt Wolenetz <wolenetz@chromium.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
(cherry picked from commit 052d41377a)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This avoids using the CUDA SDK at all; instead, we provide a minimal
reimplementation of the basic functionality that lavfi actually uses.
It generates very similar code to what NVCC produces.
The header contains no implementation code derived from the SDK.
The function and type declarations are derived from the SDK only to the
extent required to build a compatible implementation. This is generally
accepted to qualify as fair use.
Because this option does not require the proprietary SDK, it does not require
the "--enable-nonfree" flag in configure.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Recently we encountered an issue when decoding a h.263 file:
FFmpeg will freeze when decoding h.263 video with NVDEC. Turns out this is not directly related to NVDEC but is a problem that shows with several other HW decoders like VDPAU, though the exact kind of error is different (either error messages or freezing[1]). The root cause is that ff_thread_finish_setup() is called twice per frame from ff_h263_decode_frame(). This is not supported by ff_thread_finish_setup() and specifically checked for and warned against in the functions code. The issue is also specific to hw accelerated decoding only as the second call to ff_thread_finish_setup() is only issued when hw acceleration is on. The fix is simple: add a check that the first call is only send when hw acceleration is off, and the second call only when hw acceleration is on (see attached patch). This works fine as far as I was able to test with vdpau and nvdec/nvcuvid hw decoding. The patch also adds NVDEC to the hw config list if available.
I also noticed a secondary issue when browsing through the code which is that, according to documentation, ff_thread_finish_setup() should only be called if the codec implements update_thread_context(), which h263dec does not. The patch does not address this and I'm not sure any action needs to be taken here at all.
[1] This is depending on whether or not the hw decoder sets the HWACCEL_CAPS_ASYNC_SAFE flag
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Fixes: signed integer overflow: 1270564968 + 904828220 cannot be represented in type 'int'
Fixes: 15402/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALS_fuzzer-5755426823471104
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: -538976267 * 31 cannot be represented in type 'int'
Fixes: left shift of 65312 by 16 places cannot be represented in type 'int'
Fixes: 15255/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5718831688843264
Fixes: 15547/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5691384901664768
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: -829262115 + -1410750414 cannot be represented in type 'int'
Fixes: 15251/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_APE_fuzzer-5651742252859392
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: 553590816 - -9223372036315799520 cannot be represented in type 'long'
Fixes: 15743/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FFWAVESYNTH_fuzzer-5705835377852416
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>
The API does not allow it.
Also set poutbuf and poutbuf_size to NULL/0 on error.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>
cbs_h2645_read_more_rbsp_data does not handle malformed input very well:
1. If there were <= 8 bits left in the bitreader, these bits were read
via show_bits. But show_bits requires the number of bits to be read to
be > 0 (internally it shifts by 32 - number of bits to be read which is
undefined behaviour if said number is zero; there is also an assert for
this, but it is only an av_assert2). Furthermore, in this case a shift
by -1 was performed which is of course undefined behaviour, too.
2. If there were > 0 and <= 8 bits left and all of them were zero
(this can only happen for defective input), it was reported that there
was further RBSP data.
This can lead to an infinite loop in H.265's cbs_h265_read_extension_data
corresponding to the [vsp]ps_extension_data_flag syntax elements. If the
relevant flag indicates the (potential) occurence of these syntax elements,
while all bits after this flag are zero, cbs_h2645_read_more_rbsp_data
always returns 1 on x86. Given that a checked bitstream reader is used,
we are also not "saved" by an overflow in the bitstream reader's index.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Fixes: signed integer overflow: 9151595917793558550 + 297519050751678697 cannot be represented in type 'long'
Fixes: 15496/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-5722866475073536
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Replace STnxm_UB and LDnxm_SH with new macros ST_{H/W/D}{1/2/4/8}.
The old macros are difficult to use because they don't follow the same parameter passing rules.
Changing details as following:
1. remove LD4x4_SH.
2. replace ST2x4_UB with ST_H4.
3. replace ST4x2_UB with ST_W2.
4. replace ST4x4_UB with ST_W4.
5. replace ST4x8_UB with ST_W8.
6. replace ST6x4_UB with ST_W2 and ST_H2.
7. replace ST8x1_UB with ST_D1.
8. replace ST8x2_UB with ST_D2.
9. replace ST8x4_UB with ST_D4.
10. replace ST8x8_UB with ST_D8.
11. replace ST12x4_UB with ST_D4 and ST_W4.
Examples of new macro: ST_H4(in, idx0, idx1, idx2, idx3, pdst, stride)
ST_H4 store four half-word elements in vector 'in' to pdst with stride.
About the macro name:
1) 'ST' means store operation.
2) 'H/W/D' means type of vector element is 'half-word/word/double-word'.
3) Number '1/2/4/8' means how many elements will be stored.
About the macro parameter:
1) 'in0, in1...' 128-bits vector.
2) 'idx0, idx1...' elements index.
3) 'pdst' destination pointer to store to
4) 'stride' stride of each store operation.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The typedef used to define EbmlSyntax already includes a const qualifier
so that it is unnecessary to include another const qualifier in future
definitions and declarations. Given that MSVC warns about this, this
commit removes these redundant const qualifiers.
Suggested-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Unknown-length elements end when an element not allowed in them, but
allowed at a higher level is encountered. In order to check for this,
c1abd95a added a pointer to every syntax level's parent to each
EbmlSyntax. Given that the parent must of course also reference the
child in order to be able to enter said child level, one needs to use
forward declarations.
These forward declarations constitute tentative definitions and tentative
definitions with internal linkage (like our syntaxes) must not be an
incomplete type. Yet they were an incomplete type and while GCC and
Clang did not even warn about this (on default warning levels), it
broke compilation with MSVC. Therefore this commit adds the sizes.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This allows testing parsers with a wider range of input packet sizes.
Which is important and usefull for regression testing, some of our
parsers in fact to not work if the packet size is changed from 1024
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If a file uses unknown-length level 1 elements besides clusters and such
elements are after the first cluster, then these elements will usually
be parsed twice: Once during parsing of the file header and once when
reading the file reaches the position where these elements are located.
The second time the element is parsed leads to a "Duplicate element"
error message. Known-length elements are not affected by this as they
are skipped except during parsing the header.
This commit fixes this by explicitly adding a check for whether the
position of the element to be parsed is the same as the position of the
already known level 1 element.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit converts the MatroskaLevel1Element struct to use file-based
offsets, as opposed to the current practice of using offsets relative to
the beginning of the segment in it. This also includes a change from
uint64_t to int64_t.
This is in preparation to another patch that improves the check for
duplicate level 1 elements.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Up until now, one last kind of unknown-length element hasn't been
properly handled: Unknown-length elements that are supposed to be
skipped, i.e. the level 1 elements that might reside after the
clusters.
This commit changes this. To do this, ebml_parse got a mode that
essentially tries to skip everything except when parsing is needed
(namely for unknown-length elements for which parsing is necessary
as they can't be skipped). This mode is selected by using a NULL
as destination where the parsed data should be written to.
It is used to parse the level 1 elements in matroska_parse_cluster.
The syntax list used for parsing must of course include links to
the syntax of all the master elements that might need to be parsed.
In other words: Instead of matroska_clusters (which contained every
level 1 element except clusters as EBML_NONE elements designated to
be skipped) matroska_segment is needed and used; matroska_clusters has
been removed.
Furthermore, matroska_segment has been reordered so that clusters are at
the front as this is now the most common case for this list.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
matroska_probe did not support the case of an unknown-length EBML header
at all; given that libavformat's Matroska muxer used to produce such
files in the streaming case, support for them has been added.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The current Matroska specifications mandate that only two elements may
use an unknown-length length: Segments and clusters. But this was not
always so: For the greater part of Matroska's existence, all master
elements were allowed to make use of the unknown-length feature.
And there were muxers creating such files: For several years
libavformat's Matroska muxer used unknown-length for all master
elements when the output wasn't seekable. This only stopped in March
2010 with 2529bb30. And even afterwards it was possible (albeit
unlikely) for libavformat to create unknown-length master elements
that are in violation of today's specifications, namely if the master
element was so big that the seek backwards to update the size could
no longer be performed inside the AVIOContext's write buffer. This
has only been fixed in October 2016 (with the patches that introduced
support for writing CRC-32 elements).
Libavformat's Matroska demuxer meanwhile has never really supported
unknown-length elements besides segments and clusters. Support for the
latter was hardcoded. This commit changes this: Now all master elements
for which a syntax to parse them is available are supported. This
includes the files produced by old versions of libavformat's muxer.
More precisely, master elements that have unknown length and are about
to be parsed (not skipped) are supported; only a warning is emitted for
them. For normal files, this means that level 1 elements after the
clusters that are encountered after the clusters have been parsed (i.e.
not because they are referenced by the seekhead at the beginning of the
file) are still unsupported (they would be skipped at this point if
their length were known).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
1. Up until now, the error message for EBML numbers whose length exceeds
the limits imposed upon them because of the element's type did not
distinguish between known-length and unknown-length elements. As a
consequence, the numerical value of the define constant
EBML_UNKNOWN_LENGTH was emitted as part of the error message which is
of course not appropriate. This commit changes this by adding error
messages designed for unknown-length elements.
2. We impose some (arbitrary) sanity checks on the lengths of certain
element types; these checks were conducted before the checks depending
on whether the element exceeds its containing master element. Now the
order has been reversed, because a failure at the (formerly) latter
check implies that the file is truly erroneous and not only fails our
arbitrary length limit. Moreover, this increases the informativeness of
the error messages.
3. Furthermore, the error message in general has been changed by replacing
the type of the element (something internal to this demuxer and
therefore suitable as debug output at best, not as an error message
intended for ordinary users) with the element ID. The element's position
has been added, too.
4. Finally, the length limit for EBML_NONE elements has been changed so
that all unknown-length elements of EBML_NONE-type trigger an error.
This is done because unknown-length elements can't be skipped and need
to be parsed, but there is no syntax to parse available for EBML_NONE
elements. This is done in preparation for a further patch which allows
more unknown-length elements than just clusters and segments.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The Matroska (and WebM) file format achieves forward-compability by
insisting that demuxers ignore and skip elements they don't know about.
Unfortunately, this complicates the detection of errors as errors
resulting from loosing sync can't be reliably distinguished from
unknown elements that are part of a future version of the standard.
Up until now, the strategy to deal with this situation was to skip all
unknown elements that are not obviously erroneous; if an error happened,
it was tried to seek to the last known good position to resync from (and
resync to level 1 elements). This is working fine if the input is
seekable, but if it is not, then the skipped data can usually not be
rechecked lateron. This is particularly acute if unknown-length clusters
are in use, as the check for whether a child element exceeds the
containing master element is ineffective in this situation.
To remedy this, a new heuristic has been introduced: If an unknown
element is encountered in non-seekable mode, an error is presumed to
have happened based upon a combination of the length of the row of the
already encountered unknown elements and of how far away skipping this
element would take us.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Cosmetics include reordering EbmlType so that EBML_SINT is adjacent to
the other numbers (and matches the order in the switch in ebml_parse)
and also reordering the switch for assignment of default values so that
it matches the order in EbmlType.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Up until now, avio_tell was used multiple times in ebml_parse and its
subroutines, although the result of these calls can usually be simply
derived from the result of earlier calls to avio_tell. This has been
changed. Unnecessary calls to avio_tell in ebml_parse are avoided now.
Furthermore, there has been a slight change in the output of some error
messages relating to elements exceeding their containing master element:
The reported position of the element now points to the first byte of the
element ID and no longer to the first byte of the element's payload.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit closes the last hole in the system of checks for a
known-length file ending too early: Now an error message is emitted
in case the file ends directly after an EBML element.
Furthermore, this commit adds a check and a corresponding warning
whether there is data beyond the Matroska segment (only reasonable for
known-length segments). If everything looks alright, then parsing is
stopped as soon as EOF is reached (in contrast, the earlier code would
always call matroska_resync at the end).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
By including SimpleBlocks and BlockGroups twice in the same EbmlSyntax
array (with different semantics), one can reduce the duplication of the
other values.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The new code does not rely on whether the cluster's position is set or
not to infer whether a cluster needs to be closed or not (instead, this
is done in ebml_parse), so there is no need to reset the cluster's
position at all any more. It will be automatically set to the correct
value when a cluster is entered.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Before this commit, the parsing of clusters mixed EBML levels by
allowing elements from different levels in a EbmlSyntax (namely
matroska_cluster_parsing). This has been changed. And the level
is now explicitly used to determine how to parse.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit changes how levels are handled: If the level used for
ebml_parse ends directly after an element that has been consumed, then
ebml_parse ends the level itself (and any known-length levels that end
there as well) and informs the caller via the return value; if the
current level is of unknown-length, then the level is ended as soon as
an element that is not valid on the current level, but on a higher
level is encountered (or if EOF has been encountered).
This is designed for situations where one wants to parse master elements
incrementally, i.e. not in one go via ebml_parse_nest.
The (incremental) parsing of clusters still mixes levels by using a
syntax list that contains elements from different levels and the level
is still ended manually via a call to ebml_level_end if the last cluster
was an unknown-length cluster (known-length clusters are already ended
when their last element is read), but only if the next element is a
cluster, too. A different level 1 element following an unknown-length
cluster will currently simply be presumed to be part of the earlier
cluster. Fixing this will be done in a future patch. The modifications
to matroska_parse_cluster contained in this patch are only intended not
to cause regressions.
Nevertheless, the fact that known-length levels are automatically ended
in ebml_parse when their last element has been read already fixes a bogus
error message introduced in 9326117b that was emitted when a known-length
cluster is followed by another level 1 element other than a cluster in
which case the cluster's level was not ended (which only happened when
a new cluster has been encountered) so that the length check (introduced
in 9326117b) failed for the level 1 element as it is of course not
contained in the previous cluster. Most Matroska files were affected by
this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
By linking to the syntax of the parent (i.e. the containing master
element) one can check whether an element is actually part of a higher
level in the EBML hierarchy. Knowing this is important for
unknown-length levels, because they end when an element that doesn't
belong to this, but to a higher hierarchy level is encountered.
Sometimes there are different syntaxes dealing with the same elements.
In this case it is important to use a parent that contains all the
elements at the parent level; whether this is the syntax actually used
to enter the child's level is irrelevant. This affects the list of level
1 elements (which has been used as parent for matroska_cluster, too) and
it affects recursive elements (currently only the SimpleTag), where the
non-recursive parent has to be choosen.
This is in preparation for a patch that redoes level handling.
Finally, the segment id has been added to ebml_syntax. This will enable
handling of unknown-length EBML headers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Currently, resyncing during reading packets works as follows:
The current position is recorded, then a call to matroska_parse_cluster
is made and if said call fails, the demuxer tries to resync from the
earlier position. If the call doesn't fail, but also doesn't deliver a
packet, then this is looped.
There are two problems with this approach:
1. The Matroska file format aims to be forward-compatible; to achieve
this, a demuxer should simply ignore and skip elements it doesn't
know about. But it is not possible to reliably distinguish unknown
elements from junk. If matroska_parse_cluster encounters an unknown
element, it can therefore not simply error out; instead it returns zero
and the loop is iterated which includes an update of the position that
is intended to be used in case of errors, i.e. the element that is
skipped is not searched for level 1 element ids to resync to at all if
later calls to matroska_parse_cluster return an error.
Notice that in case that sync has been lost there can be a chain of
several unknown/possibly junk elements before an error is detected.
2. Even if a call to matroska_parse_cluster delivers a packet, this does
not mean that everything is fine. E.g. it might be that some of the
block's data is missing and that the data that was presumed to be from
the block just read actually contains the beginning of the next element.
This will only be apparent at the next call of matroska_read_packet,
which uses the (false) end of the earlier block as resync position so
that in the (not unlikely) case that the call to matroska_parse_cluster
fails, the data believed to be part of the earlier block is not searched
for a level 1 element to resync to.
To counter this, a "last known good" position is introduced. When an
element id that is known to be allowed at this position in the hierarchy
(according to the syntax currently in use for parsing) is read and some
further checks (regarding the length of the element and its containing
master element) are passed, then the beginning of the current element is
treated as a "good" position and recorded as such in the
MatroskaDemuxContext. Because of 2., only the start of the element is
treated as a "good" position, not the whole element. If an error occurs
later during parsing of clusters, the resync process starts at the last
known good position.
Given that when the header is damaged the subsequent resync never skips over
data and is therefore unaffected by both issues, the "last known good"
concept is not used there.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Since the changes to the parsing of SimpleBlocks, both ebml_parse_id and
ebml_parse_elem are only called from one place, so that it is possible
to inline these two function calls. This is done, but not completely:
ebml_parse_id still exists in a modified form. This is done in
preparation for a further patch regarding the handling of
unknown-length elements.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The earlier code set the level to zero upon seeking and after a
discontinuity although in both cases parsing (re)starts at a level 1
element.
Also set the segment's length to unkown if an error occured in order not
to drop any valid data that happens to be beyond the designated end of
the segment.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This function will be useful later to reset the status (e.g. current
level and the already parsed id).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When an error happens, the Matroska demuxer tries to resync to level 1
elements from an earlier position onwards. If the seek to said earlier
position fails, the demuxer currently treats this as an unrecoverable
error. And that behaviour is suboptimal as said failure is nothing
unrecoverable or unexpected (when the input isn't seekable).
It is preferable to simply resync from the earliest position available
(i.e. the start of the AVIOContext's buffer) onwards if the seek failed.
Here are some scenarios that might be treated as unrecoverable errors
by the current code if the input isn't seekable. They all have in
common that the current position is so far away from the desired
position that the seek can't be fulfilled from the AVIOContext's buffer:
1. Blocks (both SimpleBlocks as well as a Block in a BlockGroup) for
which reading them as binary EBML elements succeeds, but whose parsing
triggers an error (e.g. an invalid TrackNumber). In this case the
earlier position from which resyncing begins is at the start of the block
(or even earlier).
2. BlockGroups, whose parsing fails in one of the latter elements. Just
as in 1., the start of the BlockGroup (the target of the seek) might be
so far away from the current position that it is no longer in the
buffer.
3. At the beginning of parsing a cluster, the cluster is parsed until a
SimpleBlock or a BlockGroup is encountered. So if the input is damaged
between the beginning of the cluster and the first occurrence of a
SimpleBlock/BlockGroup and if said damage makes the demuxer read/skip so
much data that the beginning of the cluster is no longer in the buffer,
demuxing will currently fail completely.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The dump_extra bitstream filter currently simply adds the extradata to
the packets indicated by the user without checking whether said
extradata already exists in the packets. Besides wasting space
duplicated extradata in the same packet/access unit is also forbidden
for some codecs, e.g. MPEG-2.
This check has been added to be able to use the mpeg2_qsv encoder (which
only adds the sequence headers to the first packet) in broadcast
scenarios where repeating sequence headers are required.
The check used here is not perfect: E.g. dump_extra would add the
extradata to a H.264 access unit consisting of an access unit delimiter,
SPS, PPS and slices.
Fixes#8007.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
It's better to do it before the buffers are actually created. At least in VLC
we currently don't support changing some parameters dynamically easily so we
don't use the information if it comes after the buffer are created.
Co-authored-by: James Almer <jamrial@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
It's better to do it before the buffers are actually created. At least in VLC
we currently don't support changing some parameters dynamically easily so we
don't use the information if it comes after the buffer are created.
Co-authored-by: James Almer <jamrial@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Current version of dvbsub encoder doesn't support HD DVB subtitles. The high
resolution bitmaps are muxed into the stream but without the DDS (display definition
segment) the players asume that the DVB subtitles are in SD (720x576) resolution
which causes them to either render the subtitles too large and misplaced or don't
render them at all. By including the DDS as defined in section 7.7.1 of ETSI EN 300
743 (V1.3.1) this problem is fixed.
7.2.1 Display definition segment The display definition for a subtitle service may
be defined by the display definition segment if present in the stream. Absence of a
DDS implies that the stream is coded in accordance with EN 300 743 (V1.2.1) [5] and
that a display width of 720 pixels and a display height of 576 lines may be assumed.
https://www.etsi.org/deliver/etsi_en/300700_300799/300743/01.03.01_60/en_300743v010301p.pdf
Signed-off-by: Jernej Fijacko <mikrohard@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes: out of array access
Fixes: 15484/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HYMT_fuzzer-5765377054736384
Fixes: 15559/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HYMT_fuzzer-5710295743332352
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
merge_pmt_versions was not usable if multiple programs were present because
when it was searching for candidate streams it did not make sure that the PMT was
of the same program. This caused the streams of all programs to get merged into
a single (garbled) program.
This patch makes sure that the program number (service ID) is also matching
with the old streams when parsing the PMT making the feature useful for multi
program streams.
This change might cause issues for single program streams if the program number
changes, but I think it is acceptable because the goal of the option is to make
the parsing resilient to PID changes, and that is still working as expected.
Signed-off-by: Marton Balint <cus@passwd.hu>
Up until now, it was simply presumed that the first packet had a pts of
zero; otherwise the duration of the first chunk was wrong.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The earlier version of the webm_chunk muxer had several bugs:
1. If the first packet of an audio stream didn't have a PTS of zero,
then no chunk will be started before a packet is delivered to the
underlying Matroska/WebM muxer, i.e. the AVFormatContext used to write
these packets had a NULL as AVIOContext for output. This is behind the
crash in ticket #5752.
2. If an error happens during writing a packet, the underlyimg
Matroska/WebM muxer context is freed. This leads to a use-after-free
coupled with a double-free in webm_chunk_write_trailer (which supposes
that the underlying AVFormatContext is still valid).
3. Even when no error occurs at all, webm_chunk_write_trailer is still
buggy: After the underlying Matroska/WebM muxer has written its trailer,
ending the chunk implicitly flushes it again which is illegal at this
point.
These bugs have been fixed.
Fixes#5752.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
after a successful init if the function fails for another reason close
the codec without requiring FF_CODEC_CAP_INIT_CLEANUP which is meant to
cover init failures themselves. fixes a memory leak in those cases.
BUG=oss-fuzz:15529
Signed-off-by: James Zern <jzern@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
addi/daddi are deprecated by MIPS for years, and MIPS r6 remove
them.
They should be replace with addiu:
ADDIU performs the same arithmetic operation but
does not trap on overflow.
Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Loongson 3A4000 and 2k1000 has supported MSA2.0.
This patch optimized SAD_UB2_UH,UNPCK_R_SH_SW,UNPCK_SB_SH and UNPCK_SH_SW with MSA2.0 instruction.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The truehd_core bitstream filter decreases the sizes of the
major_sync_info structure (if present), of the
substream_directory and of the substreams themselves. As a consequence,
there is enough space available in front of the actual substream data
for the new header, so that one only needs to modify the header in front
of the actual data (which apart from shrinking is left untouched) and
the packet's size and buffer pointer (after having made sure that the
packet is writable).
This and switching to bsf_get_packet_ref also removed the need for
having separate packets for in- and output.
Even if the input is not writable, there are noticable performance
improvements: The average of 10 iterations of processing a file with 262144
runs each (inlcuding about 20 skips per iteration) went down from 5669
to 4362 decicycles. If the input is writable, it goes down to 1363
decicycles.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Words of 16 bit are the unit for TrueHD's size and offset fields;
in particular the sizes of the high-level structures of TrueHD are
always a multiple of a byte; yet truehd_core unnecessarily used
bit offsets at several places. This has been changed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
1. The loop counter of the substream_directory loop is always less than
the number of substreams, yet within the loop it is checked whether it
is less than FFMIN(3, s->hdr.num_substreams), although the check for < 3
would suffice.
2. In case the packet is a major sync packet, the last two bytes of the
major sync structure were initialized to 0xff and then immediately
overwritten afterwards without ever making use of the values just set.
3. When updating the parity_nibble during writing the new
substream_directory, the parity_nibble is updated one byte at a time
with bytes that might be read from the output packet's data. But one can
do both bytes at the same time without resorting to the data just
written by XOR'ing with the variable that contains the value that has
just been written as a big endian number. This changes the intermediate
value of parity_nibble, but in the end it just amounts to a reordering
of the sum modulo two that will eventually be written as parity_nibble.
Due to associativity and commutativity, this value is unchanged.
4. init_get_bits8 already checks that no overflow happens during the
conversion of its argument from bytes to bits. ff_mlp_read_major_sync
makes sure not to overread (the maximum size of a major_sync_info is 60
bytes anyway) and last_offset is < 2^13, so that no overflow in the
calculation of size can happen, i.e. the check for whether size is >= 0
is unnecessary. But then size is completely unnecessary and can be
removed.
5. In case the packet is just passed through, it is unnecessary to read
the packet's dts. This is therefore postponed to when we know that the
packet is not passed through.
6. Given that it seems overkill to use a bitreader just for one
variable, the size of the input access unit is now read directly.
7. A substream's offset (of the end of the substream) is now stored as is
(i.e. in units of words).
These changes amount to a slight performance improvement: It improved
from 5897 decicycles of ten runs with about 262144 runs each (including
an insignificant amount -- about 20-25 usually of skips) to 5747
decicycles under the same conditions.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Several checks (e.g. when the size of the input packet is too small)
simply used "goto fail", but didn't set the return value appropriately
for an error.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
If truehd_core strips Atmos data away, three parts of the output differ
in size compared to the input access unit: a) The major_sync_info block
if the extra_channel_meaning_data is present, as the newly written
output never contains said block; b) the substream_directory (because
entries relating to discarded substreams are discarded, too); and c)
the actual substream data. b) and c) have already been taken into account
when choosing the size of the output packet, but a) has been forgotten.
This is also the reason behind the end of the output buffer having been
uninitialized until 801d78f0. The workaround added in said commit has
been removed, too.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The most serious bit of the substream_info header field (in a mayor sync
packet) indicates whether a 16-channel presentation is present in the
bitstream. If set, the extended_substream_info header field contains
information about the 16-channel presentation. This presentation always
uses substream 3, a substream that is discarded by truehd_core. So
substream_info needs to be changed to no longer indicate the presence
of a 16-channel presentation in order for truehd_core's output to be
consistent. This is implemented in this commit.
This change also makes MediaInfo no longer display the presence of Atmos
in the output of truehd_core.
Also, set the (now irrelevant) extended_substream_info field to zero as
this seems to be the common value for ordinary TrueHD.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Given the recent changes to ff_cbs_delete_unit, it is no longer sensible
to use a return value for ff_cbs_h264_delete_sei_message; instead, use
asserts to ensure that the required conditions are met and remove the
callers' checks for the return value. Also, document said conditions.
An assert that is essentially equivalent to the one used in
ff_cbs_delete_unit has been removed, too.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
ff_cbs_delete_unit never fails if the index of the unit to delete is
valid, as it is with all current callers of the function. So just assert
in ff_cbs_delete_unit that the index is valid and change the return
value to void in order to remove the callers' checks for whether
ff_cbs_delete_unit failed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Fixes: signed integer overflow: 2147475672 + 8192 cannot be represented in type 'int'
Fixes: 15415/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ILBC_fuzzer-5712074128228352
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
libtta 2.3 has a limit of 6 channels, so 16 is substantially above the "official" already
Fixes: OOM
Fixes: 15249/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TTA_fuzzer-5643988125614080
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: 20242 * 121095 cannot be represented in type 'int'
Fixes: 15310/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FOURXM_fuzzer-5737051745419264
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: 2147483646 + 2 cannot be represented in type 'int'
Fixes: infinite loop
Fixes: 15396/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QDM2_fuzzer-5116605501014016
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Instead we ask for a sample, its unclear what to do in this case.
Fixes: index 30 out of bounds for type 'int8_t [30][64]'
Fixes: 15339/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QDM2_fuzzer-5749441484554240
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: 1106434976 + 1041773512 cannot be represented in type 'int'
Fixes: 15421/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WMV2_fuzzer-5669209314426880
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: -1027919784 + -1120041624 cannot be represented in type 'int'
Fixes: 15406/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WMV3_fuzzer-5700646528876544
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This checks the ham value much stricter and avoids hitting cases which cannot be reached
with data from the libavformat demuxer.
Fixes: out of array access
Fixes: 15320/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_IFF_ILBM_fuzzer-5080476840099840
Fixes: 15423/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_IFF_ILBM_fuzzer-5630765833912320
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: 6494225984479297536 - -6043795377581187040 cannot be represented in type 'long'
Fixes: 15285/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FFWAVESYNTH_fuzzer-5632780307791872
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 -9223372036854775808 cannot be represented in type 'int64_t' (aka 'long'); cast to an unsigned type to negate this value to itself
Fixes: 15289/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FFWAVESYNTH_fuzzer-5709034499342336
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: 15360/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FLIC_fuzzer-5653837190266880
Fixes: 15412/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FLIC_fuzzer-5740537648250880
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
lpc_quant of 0 produces undefined behavior, thus disallow this.
If valid samples use this then such a sample would be quite
usefull to confirm the correct&lossles handling of this.
Fixes: libavcodec/alac.c:218:25: runtime error: shift exponent -1 is negative
Fixes: 15273/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALAC_fuzzer-5656388535058432
Fixes: 15276/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALAC_fuzzer-5761238417539072
Fixes: 15315/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALAC_fuzzer-5767260766994432
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 changes vp9_metadata to (a) use ff_bsf_get_packet_ref
instead of ff_bsf_get_packet (thereby avoiding one malloc and free per
filtered packet) and (b) to use only one packet structure at all,
thereby avoiding a call to av_packet_copy_props.
(b) has been made possible by the recent changes to ff_cbs_write_packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
If a new sequence display extension had to be added, this was up until
now done at two places: One where a sequence display extension was
initialized with default values and one where the actual sequence
display extension was inserted into the fragment. This division of
labour is unnecessary and pointless; it has been changed.
Furthermore, if a sequence display extension has to be added, the
earlier code set some fields to their default value twice. This has been
changed, too.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit changes mpeg2_metadata to (a) use ff_bsf_get_packet_ref
instead of ff_bsf_get_packet (thereby avoiding one malloc and free per
filtered packet) and (b) to use only one packet structure at all,
thereby avoiding a call to av_packet_copy_props.
(b) has been made possible by the recent changes to ff_cbs_write_packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit changes h265_metadata to (a) use ff_bsf_get_packet_ref
instead of ff_bsf_get_packet (thereby avoiding one malloc and free per
filtered packet) and (b) to use only one packet structure at all,
thereby avoiding a call to av_packet_copy_props.
(b) has been made possible by the recent changes to ff_cbs_write_packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When looping over an access unit's units in positive direction and
deleting some of them, one needs to make sure that a unit that is at
the position of a unit that just got deleted gets checked, too.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit changes h264_redundant_pps to (a) use ff_bsf_get_packet_ref
instead of ff_bsf_get_packet (thereby avoiding one malloc and free per
filtered packet) and (b) to use only one packet structure at all,
thereby avoiding a call to av_packet_copy_props.
(b) has been made possible by the recent changes to ff_cbs_write_packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
If the fragment is empty after parsing (i.e. it contains no OBUs), then
the check for the type of the fragment's first OBU is nonsensical; so
error out in this case just as h264_metadata and hevc_metadata do.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The recent changes to h264_metadata (enabled by the recent changes to
ff_cbs_write_packet) made it possible to add side_data to the output
packet at any place, not only after the output packet has been written
and the properties of the input packet copied. This means that one can
now localize the code to add display orientation side-data to the packet
to the place dealing with said display-orientation.
Furthermore, the documentation of av_display_rotation_set states that
the matrix will be fully overwritten by it, so there is no need to
allocate it with av_mallocz.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit changes h264_metadata to (a) use ff_bsf_get_packet_ref
instead of ff_bsf_get_packet (thereby avoiding one malloc and free per
filtered packet) and (b) to use only one packet structure at all,
thereby avoiding a call to av_packet_copy_props.
(b) has been made possible by the recent changes to ff_cbs_write_packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When testing whether a particular unit should be kept or discarded, it
is best to start at the very last unit of a fragment and count down,
because that way a unit that will eventually be deleted won't be
memmoved during earlier deletions; and frag/au->nb_units need only be
evaluated once in this case and the counter is automatically correct
when a unit got deleted.
It also works for double loops, i.e. when looping over all SEI messages
in all SEI units of an access unit.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit changes filter_units to (a) use ff_bsf_get_packet_ref
instead of ff_bsf_get_packet (thereby avoiding one malloc and free per
filtered packet) and (b) to use only one packet structure at all,
thereby avoiding a call to av_packet_copy_props (or, in case of
passthrough, to av_packet_move_ref).
(b) has been made possible by the recent changes to ff_cbs_write_packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
According to the BSF API, when a BSF is finished with an input packet,
it should return AVERROR(EAGAIN) to signal that another packet should be
sent to the BSF via av_bsf_send_packet that the actual BSF can receive
via ff_bsf_get_packet[_ref]. filter_units on the other hand simply called
ff_bsf_get_packet again if the first packet received didn't result in
any output. This call of course returned AVERROR(EAGAIN) which was
returned, but it is nevertheless better to not include a fake loop.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
According to the API, the packet structure a bsf receives must not be
touched on failure, yet filter_units nevertheless did it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit changes av1_metadata to (a) use ff_bsf_get_packet_ref
instead of ff_bsf_get_packet (thereby avoiding one malloc and free per
filtered packet) and (b) to use only one packet structure at all,
thereby avoiding a call to av_packet_copy_props.
(b) has been made possible by the recent changes to ff_cbs_write_packet.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Up until now, ff_cbs_write_packet always initialized the packet
structure it received without documenting this behaviour; furthermore,
the packet's buffer would (on success) be overwritten with the new
buffer without unreferencing the old. This meant that the input packet
had to be either clean (otherwise there would be memleaks) in which case
the initialization is redundant or uninitialized. ff_cbs_write_packet
was never used with uninitialized packets, so the initialization was
redundant. Worse yet, it forced callers to use more than one packet and
made it difficult to add side-data to a packet designated for output,
because said side-data could only be attached after the call to
ff_cbs_write_packet.
This has been changed. It is now allowed to use a non-blank packet.
The currently existing buffer will be unreferenced and replaced by
the new one, as will be the accompanying fields (i.e. data and size).
The rest isn't touched at all.
This change will enable us to use only one packet in the bitstream
filters that rely on CBS.
This commit also updates the documentation of ff_cbs_write_extradata
and ff_cbs_write_packet (to better describe existing behaviour and in
the latter case to also describe the new behaviour).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
set AVFrame decode_error_flags in case h->slice_ctx->er.error_occurred is set
after the call to ff_h264_execute_decode_slices. This allows the user to detect
concealed decoding errors in the call to avcodec_receive_frame
Signed-off-by: Amir Pauker <amir@livelyvideo.tv>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fix the quantisation offset - use the whole range, and don't change the
offset size based on bit depth.
Iterate the list in reverse order. The first region in the list is the one
that applies in the case of overlapping regions.
Fixes: out of array access
Fixes: 15277/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALS_fuzzer-5184853437317120
Fixes: 15280/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALS_fuzzer-5741062137577472
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: left shift of negative value -6
Fixes: 15275/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALS_fuzzer-5742361767837696
Fixes: signed integer overflow: 41582592 * 256 cannot be represented in type 'int'
Fixes: 15296/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ALS_fuzzer-5739558227935232
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 fixes an overflow introduced in a569a7b3 that affected EBML
elements that the Matroska demuxer doesn't want to parse like CRC-32
elements. The return value of avio_skip (the new position on success or
an AVERROR on failure) has been assigned to an integer which meant that
new positions in the range of 2GB to 4GB-1 etc. were considered errors.
Fixes ticket #8001.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
From 7.4.3.3.1:
num_tile_columns_minus1 shall be in the range of 0 to PicWidthInCtbsY - 1, inclusive.
num_tile_rows_minus1 shall be in the range of 0 to PicHeightInCtbsY - 1, inclusive.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>
The linearize function (usually refered to EOTF) is the inverse of
delinearize function (usually referred to OETF). Demarcation point of
EOTF should be beta*delta, but the actual value used now in the source
code is beta.
For ITU Rec.709, they are 0.081 (0.018*4.5) and 0.018 respectively
(beta = 0.018 and delta = 4.5), and they correspond to pixel value 5
and 21 for an 8-bit image. Linearized result of pixel within that range
(5-21) will be different, but this commit will make linearize function
of the filter more accurate in the mathematical sense.
Signed-off-by: Yonglin Luo <vincenluo@tencent.com>
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
For example, given TensorFlow model file espcn.pb,
to generate native model file espcn.model, just run:
python convert.py espcn.pb
In current implementation, the native model file is generated for
specific dnn network with hard-code python scripts maintained out of ffmpeg.
For example, srcnn network used by vf_sr is generated with
https://github.com/HighVoltageRocknRoll/sr/blob/master/generate_header_and_model.py#L85
In this patch, the script is designed as a general solution which
converts general TensorFlow model .pb file into .model file. The script
now has some tricky to be compatible with current implemention, will
be refined step by step.
The script is also added into ffmpeg source tree. It is expected there
will be many more patches and community needs the ownership of it.
Another technical direction is to do the conversion in c/c++ code within
ffmpeg source tree. While .pb file is organized with protocol buffers,
it is not easy to do such work with tiny c/c++ code, see more discussion
at http://ffmpeg.org/pipermail/ffmpeg-devel/2019-May/244496.html. So,
choose the python script.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
The subfile protocol treats an end of 0 as meaning "until EOF"; this got
implemented by simply setting the end to INT64_MAX. But seeking relative
to EOF or AVSEEK_SIZE seeking hasn't been adapted; the result is that
e.g. the duration of transport streams isn't correctly determined when
this option is used. This is fixed in this patch.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Throughout hlsenc code, all filename related buffer lengths are set
hardcoded as 1024. This PATCH change it to general value as MAX_URL_SIZE
in internal.h
Reviewed-by: Steven Liu <lq@onvideo.cn>
Signed-off-by: Bela Bodecs <bodecsb@vivanet.hu>
This avoids undefined computations with unchecked values
Fixes: shift exponent -21 is negative
Fixes: 15262/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QDM2_fuzzer-5651261753393152
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
currently master playlist and subtitle playlist creation does not use
temporary files even when temp_file flag is set. Most of the use cases
it is not a problem because master playlist creation happens once on the
beginning of the whole process. But if master playlist is periodically
re-created because of master_pl_refresh_rate is set, non-atomic playlist
creation may cause problems in case of live streaming. This patch
correct this behavior by adding this functionality.
Reviewed-by: Steven Liu <lq@onvideo.cn>
Signed-off-by: Bela Bodecs <bodecsb@vivanet.hu>
This avoids the need for each decoder separately having to handle this case
Fixes: shift exponent -100663046 is negative
Fixes: out of array access
Fixes: 15270/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_IFF_ILBM_fuzzer-5727829913763840
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: addition of unsigned offset to 0x7f56fc26a9b6 overflowed to 0x7f56fc26a8be*
Fixes: clusterfuzz-testcase-minimized-mediasource_MP4_AVC1_pipeline_integration_fuzzer-4917949056679936
Reported-by: Matt Wolenetz <wolenetz@google.com>
Reviewed-by: Matt Wolenetz <wolenetz@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 2314885530818453536 - -8926099139098304480 cannot be represented in type 'long'
Fixes: 15259/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FFWAVESYNTH_fuzzer-5764366093254656
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: 214013 * 2531011 cannot be represented in type 'int'
Fixes: 15254/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QDMC_fuzzer-5698137026461696
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: 3784 * 682038 cannot be represented in type 'int'
Fixes: 15265/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_BINK_fuzzer-5088311799971840
Fixes: 15268/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_BINK_fuzzer-5666502344179712
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Reviewed-by: Peter Ross <pross@xvid.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Commit 31f9032b added the audio_preload feature; its goal is to
interleave audio earlier than the rest. Unfortunately, it has never ever
worked, because the check for whether a packet should be interleaved
before or after another packet was completely wrong: When audio_preload
vanishes, interleave_compare_dts returns 1 if the new packet should be
interleaved earlier than the packet it is compared with and that is what
the rest of the code expects. But the codepath used when audio_preload is
set does the opposite.
Also fixes potential undefined behaviour (namely signed integer
overflow).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -1539565182 + -798086761 cannot be represented in type 'int'
Fixes: 14807/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AAC_FIXED_fuzzer-564925382682214
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: -1727985666 - 538976288 cannot be represented in type 'int'
Fixes: 15031/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TRUEMOTION2_fuzzer-5100228035739648
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
low_power mode will use a fixed HW engine (SFC), thus can offload EU usage.
high quality mode will take EU usage (AVS sampler).
Performance and EU usage (Render usage) comparsion on Intel(R) Xeon(R) CPU E3-1225 v5 @ 3.30GHz:
High quality mode : ffmpeg -hwaccel qsv -c:v h264_qsv -i bbb_sunflower_1080p_30fps_normal_2000frames.h264 \
-vf scale_qsv=w=1280:h=736:mode=hq -f null -
fps=389
RENDER usage: 28.10 (provided by MSDK metrics_monitor)
Low Power mode: ffmpeg -hwaccel qsv -c:v h264_qsv -i ~/bbb_sunflower_1080p_30fps_normal_2000frames.h264 \
-vf scale_qsv=w=1280:h=736:mode=low_power -f null -
fps=343
RENDER usage: 0.00
Low power mode (SFC) may be disabled if not supported by
MSDK/Driver/HW, and replaced by AVS mode interanlly.
Signed-off-by: Zhong Li <zhong.li@intel.com>
The mov flavour of timed text uses the first two bytes of the packet as
a length field. And up until 11bef2fe said length field has been read
correctly in the mov2textsub bsf. But since then the next two bytes are
read as if they were the length field. This is fixed in this commit.
Reviewed-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, when an element was skipped, it was relied upon
ffio_limit to make sure that there is enough data available to skip.
ffio_limit itself relies upon the availability of the file's size. As
this needn't be available, the check has been refined: First one byte
less than intended is skipped, then another byte is read, followed by a
check of the error flags.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit fixes a number of bugs:
1. There was no check that no read error/EOF occured during
ebml_read_uint, ebml_read_sint and ebml_read_float.
2. ebml_read_ascii and ebml_read_binary did sometimes not forward
error codes; instead they simply returned AVERROR(EIO).
3. In particular, AVERROR_EOF hasn't been used and no dedicated error
message for it existed. This has been changed.
In order to reduce code duplication, the new error code NEEDS_CHECKING
has been introduced which makes ebml_parse check the AVIOContext's
status for errors.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
ebml_read_num had a number of flaws:
1. The check for read errors/EOF was totally wrong. E.g. an EBML number
beginning with the invalid 0x00 would be considered a read error,
although it is just invalid data.
2. The check for read errors/EOF was done just once, after reading the
first byte of the EBML number. But errors/EOF can happen inbetween, of
course, and this wasn't checked.
3. There was no way to distinguish when EOF should be an error (because
the data has to be there) for which an error message should be emitted
and when it is not necessarily an error (namely during parsing of EBML
IDs). Such a possibility has been added and used.
All this was fixed; furthermore, the error messages for invalid EBML
numbers were improved and useless initializations were removed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Up until now, webm_dash_manifest_cues used the return values of
ebml_read_num and ebml_read_length without checking for errors,
i.e. return values < 0. This has been changed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
It is only necessary to zero the initial allocated memory used to store
the size of laced frames if the block used Xiph lacing. Otherwise no
unintialized data was ever used, so use av_malloc instead of av_mallocz.
Also use the correct type for the allocations.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Get rid of pr dependency and write the columns strictly
alphabetical without page size considerations (POSIX
specifies 66 lines as default).
Setting the page size via pr's -l option was considered,
but as there is issue #5680 which wants to avoid pr
mainly because it's not in busybox, we chose to replace
pr instead.
Before pr would attempt to write pages, thus if a page
boundary was reached, the output looked confusing as one
couldn't see there was a new page and the alphabetical
order was disrupted when scanning down one of the columns.
This change is based on a shell implementation submitted
before by Yejun.
Possible differences to the current version using pr:
1. pr implementations should truncate items to not overflow columns;
depending on how it's done not truncating shall be better IMHO.
2. pr implementations might balance columns differently;
we use minimum number of lines and might end up not
using all columns or might have lesser entries in the
last column(s)
3. we use spaces only for padding the columns; at least the GNU pr
version on my system also by default stuffs in tabs in addition
to a single space in between columns. I don't see that this
behaviour is demanded by POSIX, though I might be very well
overlooking things. Anyway for our use case I can't see a need
for having the additional tabs, or why it would be better compared
to padding with spaces only.
Fixes output for sizes with width < column width, too.
Fixes remaining part of ticket #5680
Contributor: Guo, Yejun <yejun.guo@intel.com>
When multiple variant streams are specified by var_stream_map option,
%v is expected either in the filename or in the last sub-directory name,
but only in one of them. When both of them contains %v string, current
error message only states half of the truth.
And even %v may appears several times inside the last sub-directory name
or in filename pattern.
This patch clarifies this in the log message and in the doc also.
Signed-off-by: Bela Bodecs <bodecsb@vivanet.hu>
Up until now, the SimpleBlock was treated specially: It basically had
its own EBML category and it was also included in the BlockGroup EBML
syntax (although a SimpleBlock must not exist in a BlockGroup according
to the Matroska specifications). The latter fact also meant that
a MatroskaBlock's buffer was always unreferenced twice.
This has been changed: The type of a SimpleBlock is now an EBML_BIN.
The only way in which SimpleBlocks are still different is that they
share their associated structure with another unit (namely BlockGroup).
This is also used to unref the block: It is always unreferenced via the
BlockGroup syntax.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Before this commit, the Matroska muxer would read a block when required
to do so, parse the block, create and return the necessary AVPackets and
yet keep the blocks (in a dynamically allocated list), although they
aren't used at all any more. This has been changed. There is no list any
more and the block is immediately discarded after parsing.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Every new element of an EbmlList is zeroed initially in
ebml_parse_elem, so that in particular a SimpleBlock's duration is
initialized to zero. Therefore it is unnecessary to initialize this
field again (for SimpleBlocks) in matroska_parse_cluster_incremental.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
By default, the data_offset member of the AVFormatInternal of the
AVFormatContext associated with the MatroskaDemuxContext has not been
initialized explicitly by any Matroska-specific function, so that it was
initialized by default to the offset at the end of matroska_read_header,
i.e. usually to the offset of the length field of the first encountered
cluster. This meant that in case that the Matroska-specific seek-code
fails because there are no index entries for the target track a seek to
data_offset would be performed and ordinary parsing would start from
there which is nonsense: The length field would be treated as EBML ID and
(if the length field is not longer than four bytes (EBML numbers that
long are rejected as invalid EBML IDs)) whatever comes next would be
treated as its EBML size although it simply isn't.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The earlier code relied on the length of clusters always being coded on
eight bytes as was the behaviour of libavformat's Matroska muxer until
recently. But given that our own Matroska muxer now (and mkvmerge from
time immemorial) creates files that don't conform to this assumption,
it is high time to get rid of this assumption.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When the new incremental parser was introduced, the old parser was
kept, because the new parser was unable to handle the way SSA packets
are put into Matroska. But since 2014 (since c7d8dbad) this is no
longer needed, so that the old parser can be completely removed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This commit replaces copying attached pictures by using references to
the already existing buffers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
When multiple variant streams are specified by var_stream_map option,
implementation assumes that each elementary stream is assigned only once
to any variant. But this is not checked. This patch makes this checking.
Signed-off-by: Bela Bodecs <bodecsb@vivanet.hu>
Reviewed-by: Steven Liu<lq@onvideo.cn>
This commit uses smaller types for some static const arrays to reduce
their size in case the entries can be represented in the smaller type.
The biggest savings came from inv_map_table in vp9.c.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
avformat/movenc still relies on AVCodecContext time_base to mux tmcd
tracks and segment muxer did not copy that field to inner streams
leading to SIGFPE in the child muxer instance.
When multiple variant streams are specified by var_stream_map option, %v
placeholder in various names ensures that each variant has its unique
names. Currently %v is substituted by its variant index value (0, 1, 2
etc.). In some use cases it would be handy to specify names for variants
instead of numerical indexes. This patch makes it possible to use names
instead of default indexes. In var_stream_map option each or some of the
variant streams may use an optional name attributum (e.g.
-var_stream_map "v:0,a:0,name:sd v:1,a:1,name:720p") If a name is
specified for a variant, then this name value will be used as
substitution value of %v instead of the default index value.
Signed-off-by: Bela Bodecs <bodecsb@vivanet.hu>
Signed-off-by: Steven Liu <lq@onvideo.cn>
Hi All,
When multiple variant streams are specified by var_stream_map option, %v
placeholder in various names ensures that each variant has its unique
names. Most of %v handlng is done in format_name function. Currently
in this function the result buffer is the same as the
input pattern buffer, so you must allocate it before calling format_name
function. It also means, that it is silently assumed that the result
string will NOT be
longer that the pattern string. It is true most of the time, because %v
may appear only once in the pattern string and number of variant streams
is less than 100 in practical cases. But theoretically it will fail if
specified number of variant streams is greater than 100 (i.e. longer
than 2 digits).
This patch fixes this behaviour by altering format_name function to
allocate the
result buffer and return it to the caller.
Please, review this patch.
best,
Bela
>From 6377ebee8a106a9684d41b270c7d6c8e57cd3e7b Mon Sep 17 00:00:00 2001
From: Bela Bodecs <bodecsb@vivanet.hu>
Date: Mon, 17 Jun 2019 14:31:36 +0200
Subject: [PATCH] av_format/hlsenc: fix %v handling by format_name function
When multiple variant streams are specified by var_stream_map option, %v
placeholder in various names ensures that each variant has its unique
names. Most of %v handlng is done in format_name function. Currently
in this function the result buffer is the same as the input pattern
buffer, so you must allocate it before calling format_name function. It
also means, that it is silently assumed that the result string will NOT
be longer that the pattern string. It is true most of the time, because
%v may appear only once in the pattern string and number of variant
streams is less than 100 in practical cases. But theoretically it will
fail if specified number of variant streams is greater than 100. This
patch fixes this behaviour by altering format_name function to allocate
the result buffer and return it to the caller.
Signed-off-by: Bela Bodecs <bodecsb@vivanet.hu>
Redundant condition: '!A || B' is equivalent to '!A || (A && B)' but
more clearly.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
FF_DECODE_ERROR_CONCEALMENT_ACTIVE is set when the decoded frame has error(s) but the returned value from
avcodec_receive_frame is zero i.e. concealed errors
Signed-off-by: Amir Pauker <amir@livelyvideo.tv>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
1. Currently output format is hard-coded as NV12, thus means
CSC is always done for not NV12 input such as P010.
Follow original input format as default output.
2. Add an option to specify output format.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Avoids returning EAGAIN after signaling EOF to the codec in
ff_mediacodec_dec_send() so we can try to receive a frame before
returning in mediacodec_receive_frame().
This helps avoiding an extra round-trip between avcodec_send_frame() and
avcodec_receive_frame() while draining the remaining frames.
The compound literals assigned to "components"
only exist within the scope of the if/else
block (thanks Mark Thompson for the better
explanation).
Thus, after this if/else block, "components"
ends up pointing to an arbitrary/undefined
array. With some compilers and depending on
optimization settings, these arbitrary values
may end up being the same value (i.e. 0 with
GNU GCC 9.x). Unfortunately, the GNU GCC
compiler, at least, never prints any warnings
about this.
This patch fixes this issue by assigning the
constant arrays to local variables at function
scope and then pointing "components" to those
as necessary.
Fixes#7915
Signed-off-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
The tests previously rounded the timestamps. Its better in a fate test to preserve
the data from the demuxer and decoder.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The horizontal pass get ~2x performance with the patch
under single thread.
Tested overall performance using the command(avx2 enabled):
./ffmpeg -i 1080p.mp4 -vf gblur -f null /dev/null
./ffmpeg -i 1080p.mp4 -vf gblur=threads=1 -f null /dev/null
For single thread, the fps improves from 43 to 60, about 40%.
For multi-thread, the fps improves from 110 to 130, about 20%.
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
READ has already been undefined at this point; it is obviously intended
to undef WRITE.
Furthermore, leb128 (in cbs_av1) was undefined too often and
inconsistently.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Matroska EBML IDs can be only four bytes long maximally, so it is
natural to use uint32_t for them. By doing this and rearranging the
elements of the MatroskaLevel1Element structure, one can reduce the size
of said structure.
Notice that this field is not read via the generic reading process for
EBML_UINT, so one is not forced to use an uint64_t for it.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Pass correct pointer to av_log() and update some error/warning message,
it's will help the debugging
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
Remove the rain in the input image/video by applying the derain
methods based on convolutional neural networks. Training scripts
as well as scripts for model generation are provided in the
repository at https://github.com/XueweiMeng/derain_filter.git.
Signed-off-by: Xuewei Meng <xwmeng96@gmail.com>
The default is to dump extradata to keyframes, not all frames.
Also improve the description of the relevant AVOption.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
This way the clearing can be skipped in case of some errors.
Fixes: Timeout (11sec -> 344ms)
Fixes: 14670/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PAF_VIDEO_fuzzer-5769534503387136
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: 2052526848 + 147237888 cannot be represented in type 'int'
Fixes: 14441/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ARBC_fuzzer-5717632944177152
Fixes: 14453/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ARBC_fuzzer-5739679254577152
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Its unclear if these cases have any relevance in real files
Fixes: shift exponent -2 is negative
Fixes: 14489/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AAC_FIXED_fuzzer-5681941631729664
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This removes the use of the nonstandard combined structures, which
generated some warnings with clang and will cause alignment problems
with some parameter buffer types.
We perfer the coding style like:
/* some stuff */
if (error) {
/* error handling */
return -(errorcode);
}
/* normal actions */
do_something()
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
benchmarking with a simple command:
ffmpeg -i 1080p.mp4 -vf unsharp=la=3:ca=3 -an -f null /dev/null
with the patch, the fps increase from 50 to 120 on my local machine (i7-6770HQ).
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
Used the command for 1080p h264 clip as follow:
a). ffmpeg -i input -vf lutyuv="u=128:v=128" -f null /dev/null
b). ffmpeg -i input -vf lutrgb="g=0:b=0" -f null /dev/null
after enabled the slice threading, the fps change from:
a). 144fps to 258fps (lutyuv)
b). 94fps to 153fps (lutrgb)
in Intel(R) Core(TM) i5-8265U CPU @ 1.60GHz
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
Add slice threading support, use the command like:
./ffmpeg -i input -vf colorlevels -f null /dev/null
with 1080p h264 clip, the fps from 39 fps to 79 fps
in the local(Intel(R) Core(TM) i5-8265U CPU @ 1.60GHz)
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
Tries to find a device backed by the i915 kernel driver and loads the iHD
VAAPI driver to use with it. This reduces confusion on machines with
multiple DRM devices and removes the surprising requirement to set the
LIBVA_DRIVER_NAME environment variable to use libmfx at all.
Opening the device via X11 (DRI2/DRI3) rather than opening a DRM render
node directly is only useful if you intend to use the legacy X11 interop
functions. That's never true for the ffmpeg utility, and a library user
who does want this will likely provide their own display instance rather
than making a new one here.
For example: -init_hw_device vaapi:/dev/dri/renderD128,driver=foo
This may be more convenient that using the environment variable, and allows
loading different drivers for different devices in the same process.
Iterate over available render devices and pick the first one which looks
usable. Adds an option to specify the name of the kernel driver associated
with the desired device, so that it is possible to select a specific type
of device in a multiple-device system without knowing the card numbering.
For example: -init_hw_device vaapi:,kernel_driver=amdgpu will select only
devices using the "amdgpu" driver (as used with recent AMD graphics cards).
Kernel driver selection requires libdrm to work.
The implementation will use some default in this case. The empty string
is not a meaningful device for any existing hardware type, and indeed
OpenCL treats it identically to no device already to work around the lack
of this setting on the command line.
Attempts to pick the set of supported colour properties best matching the
input. Output is then set with the same values, except for the colour
matrix which may change when converting between RGB and YUV.
Fixes two warnings:
libavfilter/avf_showspatial.c:157:26: warning: variable ‘w’ set but not used
libavfilter/avf_showspatial.c:157:23: warning: variable ‘h’ set but not used
Additionally:
- Renamed TIFF_WHITE_LEVEL to DNG_WHITE_LEVEL since it is specified
in the DNG spec.
- Added/changed some comments to be more precise in differentiating
between TIFF, TIFF/EP and DNG values.
Related to ticket: https://trac.ffmpeg.org/ticket/4364
Signed-off-by: Nick Renieris <velocityra@gmail.com>
SubIFDs that were part of more than single-sized "SubIFDs" tags were
being ignored due to existing code ignoring that case.
This patch makes is so the first entry is read, which is not ideal
but enough for some DNG images present in the wild to be decodeable
More specifically, the first SubIFD which we would process with this
patch is the main image and the second one is a second thumbnail,
which is not as important to decode.
In DNG images with the .tiff extension, it solves the issue where
the TIFF thumbnail in IFD 0 was incorrectly parsed (related
confusion: [1]).
Embedded thumbnails for DNG images can still be decoded with the
"-thumbnail" option.
Related to ticket: https://trac.ffmpeg.org/ticket/4364
[1]: https://superuser.com/questions/546879/creating-video-from-dng-images-with-ffmpeg
Signed-off-by: Nick Renieris <velocityra@gmail.com>
Adds the "-thumbnail" option, that works like the current "-subifd"
option, but only for non-full-sized images.
This is particularly useful for DNG images (see next commit) that
have SubIFDs that are not necessarily thumbnails.
Signed-off-by: Nick Renieris <velocityra@gmail.com>
Currently, picref will be freed by calling av_frame_free(&picref) in
submit_frame() in qsvvpp.c when working in system memory mode,and normally it
is freed in filter_frame() in vf_vpp_qsv.c when working in other modes.
Double free happens when working in system memory mode, remove to
fix the memory issue.
Reproduce:
ffmpeg -init_hw_device qsv=foo -filter_hw_device foo -f rawvideo -pix_fmt nv12 -s:v 852x480 \
-i 852x480.nv12 -vf 'vpp_qsv=w=500:h=400' -f rawvideo -pix_fmt nv12 qsv.nv12
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
cbs_mpeg2_free_slice() calls av_buffer_unref() on extra_information_ref,
meaning allocating with av_malloc() was not the intention.
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, things that are merely unsupported by cbs_mpeg2 have been
declared to be invalid input. This has been changed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
The frame_centre_horizontal/vertical_offset values contained in picture
display extensions are actually signed values (i.e. it is possible to
indicate that the display device should add black bars/pillars).
The files sony-ct3.bs and tcela-6.bits (which are both used in fate
tests for mpeg2_metadata) contain picture display extensions; the former
even contains a negative frame_centre_vertical_offset. Fortunately, the
old code did not damage the picture display extensions when one did a
cycle of reading and writing. For the same reason the fate tests needn't
be updated either.
Furthermore these fields now use the trace output for matrices.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
MPEG-2 contains several elements that mustn't be zero according to the
specifications: horizontal/vertical_size_value, aspect_ratio_information,
frame_rate_code, the quantiser matrices, the colour_description
elements, picture_coding_type, the f_code[r][s] values and
quantiser_scale_code. It is now checked that the invalid values don't
occur.
The colour_description elements are treated specially in this regard:
Given that there are files in the wild which use illegal values for the
colour_description elements (some of them created by mpeg2_metadata),
they will be corrected to the value meaning "unknown" (namely 2) during
reading. This has been done in such a way that trace_headers will
nevertheless report the original value, together with a message about
the fixup.
Furthermore, the trace_headers output of user_data has been beautified.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
At present, if the outer stream extradata is empty but first packet
has extradata as a side data element, then only the first segment's
muxer instance may be able to extract this side data and use it.
For all other segments, extradata in packet side data could be missing
and generated segments may be invalid or unplayable in some apps
e.g. for an ADTS AAC stream segmented to MP4, the adtstoasc BSF will
add extradata to the first packet. The MOV muxer for the first segment
will add this to codecpar for the inner stream and write
Decoder Specific Information within the esds box. For other segments,
their esds' will not have this decSpecificInfo and they can't be opened
in Quicktime player or by services like nginx-vod-module.
Fixes: signed integer overflow: -1877966852 + -469491713 cannot be represented in type 'int'
Fixes: 14561/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TRUEMOTION2_fuzzer-5167608359288832
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: Timeout (14sec -> 9sec)
Fixes: 13398/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FFV1_fuzzer-5664106709778432
decode_line() becomes 1% faster for fate/vsynth2-ffv1.avi
for another fate sample there is a 0.5% speedup
the effect should be bigger for files with "flat" colored areas
the new faster branch is used in 97-100% of the cases in fate samples
compared to the older more complex (which i tested)
vsynth3-ffv1-v3-bgr0.avi had the lowest percentual useage of about 97%
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Explain that the default Lanczos filter parameter is 3 and that it can be
changed by the param0 option.
Signed-off-by: Werner Robitza <werner.robitza@gmail.com>
In function ff_dct_unquantize_mpeg2_intra_mmi,
addr0 shoudn't be changed before storage operation.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes infinte loop with -vf loop=loop=1 and also fixes looping when the input
is less frames than the specified loop size.
Possible regressions since ef1aadffc7.
Signed-off-by: Marton Balint <cus@passwd.hu>
Add dilation parameter in dnn native to support dilated convolution.
Signed-off-by: Xuewei Meng <xwmeng96@gmail.com>
Signed-off-by: Steven Liu <lq@onvideo.cn>
frame->enc_ctrl.Payload is malloced in get_free_frame, directly memset
the whole structure of enc_ctrl to zero will cause the memory leak for
enc_ctrl.Payload.
frame->enc_ctrl as a structure will be malloc and init to zero by calling
frame = av_mallocz(sizeof(*frame)), so the memset is redundant and can
be removed.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
ff_filter_get_nb_threads() respect AVFilterContext.nb_threads and
graph->nb_threads both, in most case, we perfer this API than using
ctx->graph->nb_threads directly.
Reviewed-by: Steven Liu <lq@onvideo.cn>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
ff_filter_get_nb_threads() respect AVFilterContext.nb_threads and
graph->nb_threads both, in most case, we perfer this API than using
ctx->graph->nb_threads directly.
Reviewed-by: Steven Liu <lq@onvideo.cn>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
Add another two padding methods "VALID" and "SAME" as tensorflow,
and keep the existing "SAME_CLAMP_TO_EDGE" method suggested by sr filter.
As "SAME_CLAMP_TO_EDGE"can keep the output with the same size as original input,
and gives a slight better result as mentioned by sr filter.
Reviewed-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Xuewei Meng <xwmeng96@gmail.com>
Signed-off-by: Steven Liu <lq@onvideo.cn>
As suggested by Ronald, don't map auto threads to frame threads only, and
instead distribute them between frame and tile more efficiently.
Add a new framethreads override option, similar to the tilethreads one.
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes: "null pointer dereference" (actually misaligned access but the tools seem not to support this)
Fixes: 14551/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_V210_fuzzer-5088609952071680
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This reverts commit dd06f022b0.
Fixes ticket #7871 and reopens ticket #7816.
The introduced regression caused URL's with @ in them to be parsed incorrectly
which is a bigger issue then not being able to specify the slash character as a
password.
I think there are better ways to fix the original issue, like being able to
specify HTTP username and password as a protocol option, or adding a protocol
option to percent-decode the URL first.
Signed-off-by: Marton Balint <cus@passwd.hu>
This fixes the description of the values for src_range and dst_range to
include the possible values and their meanings.
Signed-off-by: Werner Robitza <werner.robitza@gmail.com>
Signed-off-by: Gyan Doshi <ffmpeg@gyani.pro>
missed the category AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT lead to
ffmpeg -devices doesn't show gdigrab as a input device
FIx#7848
Found-by: dangibson
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
The return code is 64bit, so this is more correct, especially in case it
actually would be a file of such large size
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Because the Original repository author loss loss of communication,
add new sr filter script repository.
Thanks to Gyan Doshi for a suggestion.
Signed-Off-By: Steven Liu <lq@chinaffmpeg.org>
I'm not sure why this was written the way it was originally. We
initialise the plane addresses correctly in hwcontext_cuda so
why try and play games to calculate the plane offsets directly
in this code?
When i converted the filter to use texture objects instead of
texture references, I incorrect dropped the `pixel_size` scaling
factor when setting `pitchInBytes`. `src_pitch` is in pixels and
so must be scaled up.
This commit adds a new API to libavutil to allow for arbitrary transformations
on various types of data.
This is a partly new implementation, with the power of two transforms taken
from libavcodec/fft_template, the 5 and 15-point FFT taken from mdct15, while
the 3-point FFT was written from scratch.
The (i)mdct folding code is taken from mdct15 as well, as the mdct_template
code was somewhat old, messy and not easy to separate.
A notable feature of this implementation is that it allows for 3xM and 5xM
based transforms, where M is a power of two, e.g. 384, 640, 768, 1280, etc.
AC-4 uses 3xM transforms while Siren uses 5xM transforms, so the code will
allow for decoding of such streams.
A non-exaustive list of supported sizes:
4, 8, 12, 16, 20, 24, 32, 40, 48, 60, 64, 80, 96, 120, 128, 160, 192, 240,
256, 320, 384, 480, 512, 640, 768, 960, 1024, 1280, 1536, 1920, 2048, 2560...
The API was designed such that it allows for not only 1D transforms but also
2D transforms of certain block sizes. This was partly on accident as the stride
argument is required for Opus MDCTs, but can be used in the context of a 2D
transform as well.
Also, various data types would be implemented eventually as well, such as
"double" and "int32_t".
Some performance comparisons with libfftw3f (SIMD disabled for both):
120:
22353 decicycles in fftwf_execute, 1024 runs, 0 skips
21836 decicycles in compound_fft_15x8, 1024 runs, 0 skips
128:
22003 decicycles in fftwf_execute, 1024 runs, 0 skips
23132 decicycles in monolithic_fft_ptwo, 1024 runs, 0 skips
384:
75939 decicycles in fftwf_execute, 1024 runs, 0 skips
73973 decicycles in compound_fft_3x128, 1024 runs, 0 skips
640:
104354 decicycles in fftwf_execute, 1024 runs, 0 skips
149518 decicycles in compound_fft_5x128, 1024 runs, 0 skips
768:
109323 decicycles in fftwf_execute, 1024 runs, 0 skips
164096 decicycles in compound_fft_3x256, 1024 runs, 0 skips
960:
186210 decicycles in fftwf_execute, 1024 runs, 0 skips
215256 decicycles in compound_fft_15x64, 1024 runs, 0 skips
1024:
163464 decicycles in fftwf_execute, 1024 runs, 0 skips
199686 decicycles in monolithic_fft_ptwo, 1024 runs, 0 skips
With SIMD we should be faster than fftw for 15xM transforms as our fft15 SIMD
is around 2x faster than theirs, even if our ptwo SIMD is slightly slower.
The goal is to remove the libavcodec/mdct15 code and deprecate the
libavcodec/avfft interface once aarch64 and x86 SIMD code has been ported.
New code throughout the project should use this API.
The implementation passes fate when used in Opus, AAC and Vorbis, and the output
is identical with ATRAC9 as well.
b3b7ba62 introduced undefined behaviour: A (non-modifiable) string
literal has been assigned to a modifiable string; said string was indeed
modified later via av_strtok.
This of course caused compiler warnings because of the discarded
qualifier; these are in particular fixed by this commit.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Commit cd48318035 added support for NV24 and NV42, including several
fate tests for these formats, but did not include the reference files
for the tests filter-pixdesc-nv24 and filter-pixdesc-nv42. As a result,
these two tests were broken.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
setting return status following goto will never be executed, so
adjust the location in the code.
Reviewed-by: Steven Liu <lq@onvideo.cn>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
In libswcale/tests/swcale.c, the function fileTest() calls sscanf in
an argument of "%12s" on character srcStr[] and dstStr[], which are
only 12 bytes. So, if the input string is 12 characters, a
terminating null byte can be written past the end of these arrays.
This bug was found by cppcheck.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Hardware-accelerated encoding may not support 10-bit encoding. Use
'-require_sw 1' in this case.
Fixes: #7581
Signed-off-by: Rick Kern <kernrj@gmail.com>
The implementation is pretty straight-forward. Most of the existing
NV12 codepaths work regardless of subsampling and are re-used as is.
Where necessary I wrote the slightly different NV24 versions.
Finally, the one thing that confused me for a long time was the
asm specific x86 path that did an explicit exclusion check for NV12.
I replaced that with a semi-planar check and also updated the
equivalent PPC code, which Lauri kindly checked.
These are the 4:4:4 variants of the semi-planar NV12/NV21 formats.
These formats are not used much, so we've never had a reason to add
them until now. VDPAU recently added support HEVC 4:4:4 content
and when you use the OpenGL interop, the returned surfaces are in
NV24 format, so we need the pixel format for media players, even
if there's no direct use within ffmpeg.
Separately, there are apparently webcams that use NV24, but I've
never seen one.
Fixes: signed integer overflow: -2147483648 + -1 cannot be represented in type 'int'
Fixes: 14444/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_H264_fuzzer-5675880333967360
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
perfer avctx->framerate first than use avctx->time_base when setting
the frame rate to encoder. 1/time_base is not the average frame rate
if the frame rate is not constant, so use avctx->framerate if the
value is not zero.
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
perfer avctx->framerate first than use avctx->time_base when setting
the frame rate to encoder. 1/time_base is not the average frame rate
if the frame rate is not constant. In this case, we need to setting
avctx->framerate and avctx->time_base both, but avctx->framerate not
equal to 1/(avctx->time_base).
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
AVCodecContext->bit_rate is int64_t since 7404f3bd
Unbreaks non-interleaved detection of v210 4k avi files, broken since 0eec40b7.
Reported-by: Xavier Càmara, Centre de Conservació i Restauració, Filmoteca de Catalunya
After the last few commits, the functions for writing master elements
with CRC-32 elements didn't really make use of the ebml_master
structure any more, so remove these parameters from the functions.
The only things that still need to be kept are the positions of the
level 1 elements that are written preliminarily and updated later.
These positions are stored in the MatroskaMuxContext and
replace the corresponding ebml_master structures.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, a block's relative offset has been reported as the offset
in the log messages output when writing blocks; given that it is
impossible to know the real offset from the beginning of the file at
this point due to the fact that it is not yet known how many bytes will
be used for the containing cluster's length field both the relative
offset in the cluster as well as the offset of the containing cluster
will be reported from now on.
Furthermore, the TrackNumber of the written block has been added to the
log output.
Also, the log message for writing vtt blocks has been brought in line
with the message for normal blocks.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, the length field of most level 1 elements has been written
using eight bytes, although it is known in advance how much space the
content of said elements will take up so that it would be possible to
determine the minimal amount of bytes for the length field. This
commit changes this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes intendation, whitespace, a typo and renames a variable
(dyn_bc->cluster_bc) to make its meaning clearer and to bring
it more in line with the naming of similar variables.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Given that in both the seekable as well as the non-seekable mode dynamic
buffers are used to write level 1 elements and that now no seeks are
used in the seekable case any more, the two modes can be combined; as a
consequence, the non-seekable mode automatically inherits the ability to
write CRC-32 elements.
There are no differences in case the output is seekable; when it is not
and writing CRC-32 elements is disabled, there can still be minor
differences because before this commit, the EBML ID and length field
were counted towards the cluster size limit; now they no longer are.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, the writing process for level 1 elements (those elements
for which CRC-32 elements are written by default) was this in case the
output was seekable: Write the EBML ID, write an "unkown length" EBML
number of the desired length, then write the element into a dynamic
buffer, then write the dynamic buffer (after possible calculation and
writing of the CRC-element), then seek back to the size element and
overwrite the unknown-size element with the real size. The seeking and
overwriting part has been eliminated by not writing the size initially.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
A Matroska EBML ID can only be maximally four bytes long, so make the
variables denoting EBML IDs uint32_t instead of unsigned int to
better reflect this.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
All places where end_ebml_master_crc32_preliminary are used already
check for whether the output is seekable, so the check in the function
is redundant.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Since 4e3bdf729a there is no reason any
more to treat the seekable and non-seekable cases separate with regards
to the log message for a new cluster. This effectively reverts
d41aeea8a6.
Also improved the log message: "pts 80dts 0" -> "pts 80, dts 0".
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now, the check for whether to write CRC32 elements was always
mkv->write_crc && mkv->mode != MODE_WEBM. This is equivalent to simply
set write_crc to zero in WebM-mode. And this is what this commit does.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until e7ddafd515 the Matroska muxer
wrote a secondary seek head referencing all the clusters. When this
was changed, a (now completely wrong) comment remained and the unique
remaining seek head was still called main_seekhead. This has been
changed.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Up until now the EBML Header length field has been written with eight
bytes, although the EBML Header is always so small that only one byte
is needed for it. This patch saves seven bytes for every Matroska/Webm
file.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
The upper bounds currently used for determining the size of a CuePoint's
length field can be improved somewhat; as a result, a CuePoint
containing three CueTrackPositions will now only need a size field
with one byte length.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
The earlier code included the size of the BlockGroup's length field and
the EBML ID in the calculation of the size for the payload and ignored
the size of the duration's length field. This meant that Blockgroups
corresponding to packets with size 2^(7n) - 17 - n - i, i = 0,..., n - 1,
n = 1,..., 8 (i.e. 110, 16364, 16365, 2097130..2097132, ...) were written
with length fields that are unnecessarily long.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
At this point, ts already includes the ts_offset so that the relative
time written with the cluster is already given by ts - mkv->cluster_pts.
It is this number that needs to fit into an int16_t.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
currently, only float is supported as model input, actually, there
are other data types, this patch adds uint8.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
some models such as ssd, yolo have more than one output.
the clean up code in this patch is a little complex, it is because
that set_input_output_tf could be called for many times together
with ff_dnn_execute_model_tf, we have to clean resources for the
case that the two interfaces are called interleaved.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
Currently, within interface set_input_output, the dims/memory of the tensorflow
dnn model output is determined by executing the model with zero input,
actually, the output dims might vary with different input data for networks
such as object detection models faster-rcnn, ssd and yolo.
This patch moves the logic from set_input_output to execute_model which
is suitable for all the cases. Since interface changed, and so dnn_backend_native
also changes.
In vf_sr.c, it knows it's srcnn or espcn by executing the model with zero input,
so execute_model has to be called in function config_props
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
remove the requirment that the name of DNN model input/output
should be "x"/"y",
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
remove 'else' since there is always 'return' in 'if' scope,
so the code will be clean for later maintenance
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
otherwise, the following check will return error if layer_add_res
is randomly initialized.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Pedro Arthur <bygrandao@gmail.com>
Adding the support to build FFMPEG with HW accelerated decode and encode on PPC64
little endian architecture.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Cuvid supports clips with a limit on maximum number of macroblocks.
This check was missing after cuvidGetDecoderCaps API call allowing
unsupported clips to proceed.
Added the missing check, same as the one in hwaccel nvdec implementation.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Instead of doing each column one by one, doing several columns
together gives about 30% better performance.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
Currently profile mapping is hard-coded, and not flexible to do extactly
map (E.g: libmfx treats H264 constrained baseline to be baseline profile).
vaapi profile mapping funtion provides a better soultion than current
qsv mapping.
Signed-off-by: Zhong Li <zhong.li@intel.com>
It is helpful to know why some clips decoding failed.
Ticket#7330 is a good example, with this patch it is easily to
know bitstream codec level is out of support range.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Reference: Table 8: Interpretation of valid BITPIX value from FITS standard 4.0
Fixes: runtime error: division by zero
Fixes: 14581/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_FITS_fuzzer-5652382425284608
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
10 bytes (id3v2 header amount of bytes) were being read before any checks
were made on the bitstream. The result was that we were overreading into
the next frame if the current one was 8 or 9 bytes long.
Fixes tickets #7271 and #7869.
Signed-off-by: James Almer <jamrial@gmail.com>
The latest generation video decoder on the Turing chips supports
decoding HEVC 4:4:4. This change adds AV_PIX_FMT_VDPAU as a valid format
for HEVC 4:4:4 8 bit.
Pass SPS, PPS range extensions to VDPAU layer via
VdpPictureInfoHEVC444. Added VdpPictureInfoHEVC444 struct to
VdpPictureInfo union to populate the range extension params. Mapped
FF_PROFILE_HEVC_REXT to VDP_DECODER_PROFILE_HEVC_MAIN_444.
New VdpYCbCr Formats VDP_YCBCR_FORMAT_Y_U_V_444 and,
VDP_YCBCR_FORMAT_Y_UV_444 have been added in VDPAU with libvdpau-1.2
to be used in get/putbits for YUV 4:4:4 surfaces. Earlier mapping of
AV_PIX_FMT_YUV444P to VDP_YCBCR_FORMAT_YV12 is not valid.
Hence this Change maps AV_PIX_FMT_YUV444P to VDP_YCBCR_FORMAT_Y_U_V_444
to access the YUV 4:4:4 surface via read-back API's of VDPAU.
Apparently in the new SDK one cannot query if VANC output is supported, so we
will fall back to non-VANC output if enabling the video output with VANC fails.
Fixes ticket #7867.
Signed-off-by: Marton Balint <cus@passwd.hu>
This commit was merged in a couple years ago as a no-op because we
had already switched from GetProcAddress to dlsym some time before
that. However, not applying the actual cast causes warnings about
FARPROC and when attempting to build FFmpeg in MSVC with AviSynth-GCC
32-bit compatibility, those FARPROC warnings turn into FARPROC errors.
Directed to the AviSynth+ entry on AviSynth Wiki rather than to
the github repository, since the wiki page is both more informative
and has the relevant Git/download links. The github releases page
is little more than a changelog.
These pix_fmts have been added to FFmpeg in the 31 months since
commit 92916e8542 added support for
the larger number of pix_fmts that AviSynth+ can use. They were
present in AviSynth+ even then, just not in libavutil.
As part of the update, it is now possible to test 32-bit GCC builds
of AviSynth+ with FFmpeg by using the AVS_WIN32_GCC32 define. Due to
different calling conventions between MSVC and GCC regarding 32-bit
Windows, this is unfortunately necessary.
The last argument of av_strlcpy is supposed to contain the size of the
destination buffer, but it was filled with the size of the source
string, effectively negating its very purpose.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
* commit 'c4642788e83b0858bca449f9b6e71ddb015dfa5d':
time_internal: Prefix fallback versions of gmtime_r/localtime_r with ff_
Merged-by: James Almer <jamrial@gmail.com>
* commit '9485cce6d55baf547e92ef1f54cad117f2a38287':
time_internal: Do not attempt to override *time_r() macros
This commit is a noop, see a604de4fd8
Merged-by: James Almer <jamrial@gmail.com>
* commit 'c93e92f2b25f4174350ded3f59ad117ec8eb1fe4':
configure: Include time.h when checking for gmtime_r and localtime_r
This commit is a noop, see 1b4dd59e5f
Merged-by: James Almer <jamrial@gmail.com>
* commit 'c0bd865ad60da31282c5d8e1000c98366249c31e':
configure: Add -D_POSIX_C_SOURCE=200112 -D_XOPEN_SOURCE=600 for mingw as well
Merged-by: James Almer <jamrial@gmail.com>
Fixes: signed integer overflow: -2147483648 + -1 cannot be represented in type 'int'
Fixes: 14107/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TRUEMOTION2_fuzzer-5694078680825856
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fix the aligned check in hwupload, input surface should be 16 aligned
too.
Partly fix#7830.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
This follows the spec definition, and removes a field from the relevant
structs.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
Also infer the value time_offset_length as 0 when it's not present.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
KLV length is BER encoded (variable size), but the code assumed the encoding to
always use 4 bytes.
Fixes parsing Random Index Pack in samples/MXF/issue2160/PW0805A0V01.4C5B5636.EFA330.mxf.
Signed-off-by: Marton Balint <cus@passwd.hu>
This affects the following samples:
samples/ffmpeg-bugs/roundup/issue1775/av_seek_frame_failure.mxf
samples/ffmpeg-bugs/trac/ticket1957/16ch.mxf
samples/ffmpeg-bugs/trac/ticket5016/r0.mxf
samples/ffmpeg-bugs/trac/ticket5016/r1.mxf
samples/ffmpeg-bugs/trac/ticket5316/hq.MXF
samples/ffmpeg-bugs/trac/ticket5316/hqx.MXF
Some AVPacket->pos values are changed because for frame wrapped tracks we point
to the KLV offset and not the data.
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes: speed regression with xmap_samsung_gear_2560x1280.pgm
Found-by: Michael Koch
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This reverts nvenc to old behaviour, which in some super rare edge cases
performs better.
The implication of this is that any potential API user who relies on
nvenc cleaning up every frames device resources after it's done using
them will have to change their usage pattern.
That should not be a problem, since pretty much every normal usage
pattern automatically implies that surfaces are reused from a common
pool, since constant re-allocation is also very expensive.
After the commit 9f61abc811, we can use AVFormatContext.strict_std_compliance
instead of HLSContext.strict_std_compliance to avoid the code redundancy.
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
Has a slight speedup.
Can't be carried over to aarch64, since it has no shufps-like instruction.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes: value 1.87633e+10 is outside the range of representable values of type 'int'
Fixes: Undefined behavior
Fixes: 14246/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_JPEG2000_fuzzer-5758393601490944
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The spec defines the valid range of values to be INT32_MIN + 1 to INT32_MAX, inclusive.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
The function in case of n=0 would read more bytes than 0.
The end pointer could be beyond the allocated space, which
is undefined.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: Timeout (158sec -> 36sec)
Fixes: 14214/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DSICINVIDEO_fuzzer-5633569034076160
This is untested with valid cin files as none of the files i found cover the changed
codepath
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
There are 2 types of problems when using adaptive deinterlace with cuvid:
1. Sometimes, in the middle of transcoding, cuvid outputs frames with visible horizontal lines (as though weave deinterlace method was chosen);
2. Occasionally, on scene changes, cuvid outputs a wrong frame, which should have been shown several seconds before (as if the frame was assigned some wrong PTS value).
The reason is that sometimes CUVIDPARSERDISPINFO has property progressive_frame equal to 1 with interlaced videos.
In order to fix the problem we should check if the video is interlaced or progressive in the beginning of a video sequence (cuvid_handle_video_sequence).
And then we just use this information instead of the property progressive_frame in CUVIDPARSERDISPINFO (which is unreliable).
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
I do not know if such vlc trees are allowed in agm, I have no specification
So i do not know if these should be treated as error, or not.
But the code does contain a check for idx < 0 already ...
Fixes: Stack-buffer-overflow in get_tree_codes
Fixes: 14189/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AGM_fuzzer-5745747003179008
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Improves overall speed by about 3%
Testcase: 14124/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PICTOR_fuzzer-5633887734071296
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
It seems the specification does not limit the value to 32bit
Fixes: signed integer overflow: -109611143 * 24 cannot be represented in type 'int'
Fixes: 13477/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5648337460527104
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This patch improves the logs when the message "cur_dts is invalid" appears.
If helps to identify which stream generates the trouble,
and the status of the stream.
A lot of users suffers with the message, and the origin varies.
The improved message can help to discover the cause.
Signed-off-by: Andreas Hakon <andreas.hakon@protonmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The transcode() helper function will already prepend the TARGET_PATH to
the sample path, if its a relative path. This avoids an issue on
Windows, where the relative path check could fail.
Silences several warnings:
libavutil/hwcontext_d3d11va.c:413:49: warning: passing argument 3 of ‘av_image_copy’ from incompatible pointer type
libavutil/hwcontext_d3d11va.c:425:47: warning: passing argument 3 of ‘av_image_copy’ from incompatible pointer type
libavutil/hwcontext_dxva2.c:351:45: warning: passing argument 3 of ‘av_image_copy’ from incompatible pointer type
libavutil/hwcontext_dxva2.c:382:52: warning: passing argument 3 of ‘av_image_copy_uc_from’ from incompatible pointer type
Fixes a bogus compiler warning (max_samples_per_frame is checked):
libavcodec/alac.c: In function ‘allocate_buffers’:
./libavutil/internal.h:142:9: warning: argument 1 value ‘18446744073709551552’ exceeds maximum object size 9223372036854775807
The spec defines it as an array of signed values, inferred to 0 when not
present.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
Use a macro to redirect calling code from the official name to the
ff_ prefixed one.
Detecting these functions in configure can be tricky (on mingw, they
are conditionally available depending on posix feature defines).
If configure didn't detect them, but they still are visible at
compile time (due to an unrelated header defining the posix feature
defines), providing the local fallback versions with a prefixed
name is safer.
Signed-off-by: Martin Storsjö <martin@martin.st>
These functions are available in time.h (conditional on posix thread
safe functions) on mingw.
Previously, these functions weren't detected by configure, and
libavutil/time_internal.h provided replacements, even if time.h
actually contained definitions of them.
These mingw inline functions are currently defined as
"extern __inline __attribute__((__gnu_inline__))". In this case,
redefining a new static inline version of the same function with the
same name is tolerated.
Signed-off-by: Martin Storsjö <martin@martin.st>
Mingw headers have got header inline implementations of localtime_r
and gmtime_r, but only visible if certain posix thread safe functions
have been requested.
This is a preparatory step for improving the detection of those
functions.
Signed-off-by: Martin Storsjö <martin@martin.st>
Cleanup the applehttp as demuxer name, when use the command :
ffmpeg -formats, get the confused information like:
"
E hls Apple HTTP Live Streaming
D hls,applehttp Apple HTTP Live Streaming
"
we don't use applehttp as the demuxer/muxer name usually, so
cleanup the applehttp and update the documents.
After the change, get the information from "ffmpeg -formats":
"
DE hls Apple HTTP Live Streaming
"
Reviewed-by: Steven Liu <lq@onvideo.cn>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
stimeout option is already used in tcp transport, since
http is based on tcp, pass the option to http for tunneling
case.
Reviewed-by: Steven Liu <lq@onvideo.cn>
Signed-off-by: Jun Li <junli1026@gmail.com>
stimeout option is already used in tcp transport, since
http is based on tcp, pass the option to http for tunneling
case.
Reviewed-by: Steven Liu <lq@onvideo.cn>
Signed-off-by: Jun Li <junli1026@gmail.com>
stimeout option is already used in tcp transport, since
http is based on tcp, pass the option to http for tunneling
case.
Reviewed-by: Steven Liu <lq@onvideo.cn>
Signed-off-by: Jun Li <junli1026@gmail.com>
Fixes: signed integer overflow: 255 + 2147483634 cannot be represented in type 'int'
Fixes: 13472/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TRUEMOTION2_fuzzer-5712444142387200
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
./ffmpeg -f lavfi -i yuvtestsrc=duration=1:size=1200x1440 -sws_flags fast_bilinear \
-s 1200x720 -f null -vframes 100 -pix_fmt $i -nostats \
-cpuflags 0 -v error -
32-bit mul, power8 only.
~2x speedup:
rgb24
24431 UNITS in yuv2packed2, 16384 runs, 0 skips
13783 UNITS in yuv2packed2, 16383 runs, 1 skips
bgr24
24396 UNITS in yuv2packed2, 16384 runs, 0 skips
14059 UNITS in yuv2packed2, 16384 runs, 0 skips
rgba
26815 UNITS in yuv2packed2, 16383 runs, 1 skips
12797 UNITS in yuv2packed2, 16383 runs, 1 skips
bgra
27060 UNITS in yuv2packed2, 16384 runs, 0 skips
13138 UNITS in yuv2packed2, 16384 runs, 0 skips
argb
26998 UNITS in yuv2packed2, 16384 runs, 0 skips
12728 UNITS in yuv2packed2, 16381 runs, 3 skips
bgra
26651 UNITS in yuv2packed2, 16384 runs, 0 skips
13124 UNITS in yuv2packed2, 16384 runs, 0 skips
This is a low speedup, but the x86 mmx version also gets only ~2x. The mmx version
is also heavily inaccurate, while the vsx version has high accuracy.
153372 UNITS in postfilter_c, 65536 runs, 0 skips
73164 UNITS in postfilter_neon, 65536 runs, 0 skips -> 2.1x speedup
80591 UNITS in deemphasis_c, 131072 runs, 0 skips
43969 UNITS in deemphasis_neon, 131072 runs, 0 skips -> 1.83x speedup
Total decoder speedup: ~15% on a Raspberry Pi 3 (from 28.1x to 33.5x realtime)
Deemphasis SIMD based on the following unrolling:
const float c1 = CELT_EMPH_COEFF, c2 = c1*c1, c3 = c2*c1, c4 = c3*c1;
float state = coeff;
for (int i = 0; i < len; i += 4) {
y[0] = x[0] + c1*state;
y[1] = x[1] + c2*state + c1*x[0];
y[2] = x[2] + c3*state + c1*x[1] + c2*x[0];
y[3] = x[3] + c4*state + c1*x[2] + c2*x[1] + c3*x[0];
state = y[3];
y += 4;
x += 4;
}
Unlike the x86 version, duplication is used instead of pslldq so
the structure and tables are different.
The `opencl_get_plane_format` function was incorrectly determining the
value used to set the image channel order. This resulted in all RGB
pixel formats being set to the `CL_RGBA` pixel format, regardless of
whether or not they actually *were* RGBA.
This patch fixes the issue by using the `offset` and depth of components
rather than the loop index to determine the value of `order`.
Signed-off-by: Jarek Samic <cldfire3@gmail.com>
Signed-off-by: Mark Thompson <sw@jkqxz.net>
Fix memory leak after write trailer for #7827, only store a audio
packet whose buffer has size greater than zero in cur_audio_pkt.
Audio packets with size zero, but with side-data currently lead to
memleaks, in the Matroska muxer, because they are not properly freed:
They are currently put into an AVPacket in the MatroskaMuxContext to
ensure that the necessary audio is always available for a new cluster,
but are only written and freed when their size is > 0.
As the only use we have for such packets consists in updating the
CodecPrivate it makes no sense to store these packets at all and this
is how this commit solves the memleak.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Jun Zhao <barryjzhao@tencent.com>
Otherwise, AV1 encodes with FFmpeg trigger use-of-uninitialized-value
warnings under MemorySanitizer, and the output buffer potentially
changes from run to run.
Signed-off-by: James Almer <jamrial@gmail.com>
When asetnsamples uses output samples < input samples, remaining samples build up in the fifo over time.
Fix this by marking the filter as ready again if there are enough samples.
Regression since ef3babb2c7
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
./ffmpeg -f lavfi -i yuvtestsrc=duration=1:size=1200x1440 -sws_flags fast_bilinear \
-s 1200x1440 -f null -vframes 100 -pix_fmt $i -nostats \
-cpuflags 0 -v error -
32-bit mul, power8 only.
1.8-2.3x speedup:
rgb24
18192 UNITS in yuv2packed1, 32767 runs, 1 skips
9983 UNITS in yuv2packed1, 32760 runs, 8 skips
bgr24
18665 UNITS in yuv2packed1, 32766 runs, 2 skips
9925 UNITS in yuv2packed1, 32763 runs, 5 skips
rgba
20239 UNITS in yuv2packed1, 32767 runs, 1 skips
8794 UNITS in yuv2packed1, 32759 runs, 9 skips
bgra
20354 UNITS in yuv2packed1, 32768 runs, 0 skips
8770 UNITS in yuv2packed1, 32761 runs, 7 skips
argb
20185 UNITS in yuv2packed1, 32768 runs, 0 skips
8761 UNITS in yuv2packed1, 32761 runs, 7 skips
bgra
20360 UNITS in yuv2packed1, 32766 runs, 2 skips
8759 UNITS in yuv2packed1, 32764 runs, 4 skips
This is a low speedup, but the x86 mmx version also gets only ~2x. The mmx version
is also heavily inaccurate, while the vsx version has high accuracy.
Fixes ticket #4519.
The metadata starting at 0xe00004 is encrypted
with the password "meta" but zlib does not
support decryption, so no kux metadata is read.
There is a calculation error in xcbgrab_reposition() that breaks
vertical repositioning on follow_mouse. It made the bottom
reposition occur when moving the mouse lower than N pixels after
the capture bottom edge, instead of before.
This commit fixes the calculation to match the documentation.
follow_mouse: centered or number of pixels. The documentation says:
When it is specified with "centered", the grabbing region follows
the mouse pointer and keeps the pointer at the center of region;
otherwise, the region follows only when the mouse pointer reaches
within PIXELS (greater than zero) to the edge of region.
Fixes: negation of -2147483648 cannot be represented in type 'int'; cast to an unsigned type to negate this value to itself
Fixes: 13999/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AGM_fuzzer-5644405991538688
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The earlier code had three flaws:
1. The case of an unknown-sized element inside a finite-sized element
(which is against the specifications) was not caught.
2. The error message wasn't helpful: It compared the length of the child
with the offset of the end of the parent and claimed that the first
exceeds the latter, although that is not necessarily true.
3. Unknown-sized elements that are not parsed can't be skipped. Given
that according to the Matroska specifications only the segment and the
clusters can be of unknown-size, this is handled by not allowing any
other units to have infinite size whereas the earlier code would seek
back by 1 byte upon encountering an infinite-size element that ought
to be skipped.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
GPB is the default type, just contains forward references but the
slice_type is B slice with higher encoding efficiency than regular P
slice, but lower performance.
Add an option to allow user to set regular P slice.
Fix ticket#6870
Test data on Intel Kabylake (i7-7567U CPU @ 3.50GHz):
1. ffmpeg -hwaccel qsv -c:v h264_qsv -i bbb_sunflower_1080p_30fps_normal.mp4 -vsync passthrough
-vframes 1000 -c:v hevc_qsv -gpb 0 -bf 0 -q 25 test_gpb_off_bf0_kbl.mp4
transcoding fps: 85
encoded file size of test_gpb_off_bf0_kbl.mp4: 21960100 (bytes)
2. ffmpeg -hwaccel qsv -c:v h264_qsv -i bbb_sunflower_1080p_30fps_normal.mp4 -vsync passthrough
-vframes 1000 -c:v hevc_qsv -gpb 1 -bf 0 -q 25 test_gpb_on_bf0_kbl.mp4
transcoding fps: 79
encoded file size oftest_gpb_on_bf0_kbl.mp4: 21211449 (bytes)
In this case, enable gpb can bring about 7% performance drop but 3.4% encoding efficiency improvment.
Signed-off-by: Zhong Li <zhong.li@intel.com>
write_tmcd allows tmcd track to be created with any mode but in
mov_write_header, index for first tmcd track is only set for modes
MP4 or MOV, causing a crash if tmcd creation is attempted with other
modes.
This brings the channel order in line with that used in 32-bit mode (BGR0).
24-bit decoding is disabled by default (#ifdef ZMBV_ENABLE_24BPP), and no
prior encoders or sample videos are known to exist for this bit depth, so
I consider this change in implementation is unlikely to affect anyone.
The decision has been made in agreement with the DOSBox Development Team
(dosbox.crew@gmail.com), specifically with harekiet, who wrote the original
codec.
Always exposes low_power option for all qsv encoder, and reports a warning
if VDENC is not supported in current version of MSDK.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
This makes sure the data is available when writing the moov atom during the
second pass triggered by the faststart movflag.
Fixes ticket #7780
Signed-off-by: James Almer <jamrial@gmail.com>
Fix ticket #7297
The current setting for send-expect-100 option is either
enabled if applicable or forced enabled, no option to force
disable the header. This change is to expand the option setting
to provide more flexibility, which is useful for rstp case.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Option "-page N" (page index N starts from 1) can now be used to specify which TIFF page/subfile to decode.
Signed-off-by: Nick Renieris <velocityra@gmail.com>
Similarly to the previous changes, we don't need to synchronise
after a memcpy to device memory. On the other hand, we need to
keep synchronising after a copy to host memory, otherwise there's
no guarantee that subsequent host reads will return valid data.
We're also doing a sync here after copying the frame to be passed
on down the pipleine. And it is also unnecessary.
I was able to demonstrate a 33% speedup removing the sync from
an example transcode pipeline.
I put this call in by habit, rather than because there was any
actual need. The filter is simply processing frames one after
the other and has no need to synchronise.
malakudi on the devtalk forums noticed a slowdown when using nvenc
with temporal/spatial aq and that the slowdown went away if the
sync call was removed. I also verified that in the basic encoding
case there's an observable speedup.
I also verified that we aren't doing unnecessary sync calls in any
other filter.
The following are the newly added options:
arnr_max_frames, arnr_strength, aq_mode, denoise_noise_level, denoise_block_size,
rc_undershoot_pct, rc_overshoot_pct, minsection_pct, maxsection_pct, frame_parallel,
enable_cdef, enable_global_motion, and intrabc.
Also added macros for compiling for aom 1.0.0 and fixed the default values.
Signed-off-by: James Almer <jamrial@gmail.com>
cbs trace qsv vps header failed due to some reasons:
1. vps_temporal_id_nesting_flag is not set but spec required it must to
be 1 when vps_max_sub_layers_minus1 is equal to 0.
2. vps_num_hrd_parameters is not set and written.
3. other issues in ff_hevc_encode_nal_vps() (fixed in pervious commit_id: 520226b683).
Reproduce: ffmpeg -hwaccel qsv -v verbose -c:v h264_qsv -i bbb_sunflower_1080p_30fps_normal.mp4 -vframes 1
-c:v hevc_qsv -bsf:v trace_headers -f null -
Signed-off-by: Zhong Li <zhong.li@intel.com>
Right now, the code check for no filter description, but if we use a
filter_complex, the code will use the AVFrame.duration which could be
wrong in case of using fps filter.
How to reproduce the problem:
ffmpeg -f lavfi -i testsrc=duration=1 -vf fps=fps=50 -vsync 1 -f null -
output 50 frames
ffmpeg -f lavfi -i testsrc=duration=1 -filter_complex fps=fps=50 -vsync 1 -f null -
output 51 frames
With this commit, the same command will always output 50 frames.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
There are many problems of current qsv trellis option:
1. Duplicated with AVCodecContext definition
2. MFX_TRELLIS_XXX is introduced by MSDK API 1.17
Currently Without MSDK API checking thus may cause compilation issue.
3. user is inclined to enable trellis when set "-trellis 1", but
actually it is to disable since MFX_TRELLIS_OFF is equal to 1.
4. It is too complex for user to enable trellis for every frame(I/P/B).
Just simply remove the private option, and switch to the AVCodecContext
definition. Compatibility should not a big problem (except can't exact map)
since the option name is same as AVCodecContext.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Reviewed-by: Carl Eugen Hoyos <ceffmpeg@gmail.com>
Reviewed-by: Moritz Barsnick <barsnick@gmx.net>
Currectly just standard header path can be found,
check_type/struct will fail if vaapi is installed somewhere else.
Move them followed "check_pkg_config"
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Reviewed-by: Timo Rothenpieler <timo@rothenpieler.org>
Signed-off-by: Zhong Li <zhong.li@intel.com>
The fields are deprecated in current vaapi,
setting them to 0 in old versions is fine
as FMO is not implemented.
Fixes the following warnings:
libavcodec/vaapi_h264.c:259:10: warning: 'num_slice_groups_minus1' is deprecated [-Wdeprecated-declarations]
.num_slice_groups_minus1 = pps->slice_group_count - 1,
^
libavcodec/vaapi_h264.c:260:10: warning: 'slice_group_map_type' is deprecated [-Wdeprecated-declarations]
.slice_group_map_type = pps->mb_slice_group_map_type,
^
libavcodec/vaapi_h264.c:261:10: warning: 'slice_group_change_rate_minus1' is deprecated [-Wdeprecated-declarations]
.slice_group_change_rate_minus1 = 0, /* FMO is not implemented */
^
Reviewed-by: Mark Thompson
Fixes the following compilation warnings:
libavcodec/vaapi_hevc.c:155:21: warning: initializer overrides prior initialization of this subobject [-Winitializer-overrides]
.pic_fields.bits = {
~^~~~
libavcodec/vaapi_hevc.c:125:57: note: previous initialization is here
.pic_fields.value = 0,
^
libavcodec/vaapi_hevc.c:175:31: warning: initializer overrides prior initialization of this subobject [-Winitializer-overrides]
.slice_parsing_fields.bits = {
~^~~~
libavcodec/vaapi_hevc.c:126:57: note: previous initialization is here
.slice_parsing_fields.value = 0,
Reviewed-by: Mark Thompson
Fixes: NULL pointer dereference and out of array access
Fixes: 13871/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HEVC_fuzzer-5746167087890432
Fixes: 13845/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HEVC_fuzzer-5650370728034304
This also fixes the return code for explode mode
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
./ffmpeg -f lavfi -i yuvtestsrc=duration=1:size=1200x1440 \
-s 1200x1440 -f null -vframes 100 -pix_fmt $i -nostats \
-cpuflags 0 -v error -
This uses 32-bit mul, so POWER8 only.
The following output formats get about 4.5x speedup:
rgb24
39980 UNITS in yuv2packed1, 32768 runs, 0 skips
8774 UNITS in yuv2packed1, 32768 runs, 0 skips
bgr24
40069 UNITS in yuv2packed1, 32768 runs, 0 skips
8772 UNITS in yuv2packed1, 32766 runs, 2 skips
rgba
39759 UNITS in yuv2packed1, 32768 runs, 0 skips
8681 UNITS in yuv2packed1, 32767 runs, 1 skips
bgra
39729 UNITS in yuv2packed1, 32768 runs, 0 skips
8696 UNITS in yuv2packed1, 32766 runs, 2 skips
argb
39766 UNITS in yuv2packed1, 32768 runs, 0 skips
8672 UNITS in yuv2packed1, 32766 runs, 2 skips
bgra
39784 UNITS in yuv2packed1, 32768 runs, 0 skips
8659 UNITS in yuv2packed1, 32767 runs, 1 skips
The lensfun filter wraps the lensfun library which performs
transformations on videos to correct for lens distortion. Often this
results in areas in the input being mapped to areas that fall outside
the boundaries of the output. The library has a parameter called scale
which is a scale factor applied to the output video. By decreasing it it
is possible to regain the areas of the video which would otherwise have
been lost. There is a special value of 0 which indicates that the
library should automatically determine a scale factor that results in
the output frame being filled (i.e. little or no black/unmapped areas).
This patch adds a corresponding scale option to the lensfun filter which
is passed through to the library. The existing behaviour of using the
automatic value of 0 is retained as the default behaviour, while other
values will be passed through to the library.
Signed-off-by: Daniel Playfair Cal <daniel.playfair.cal@gmail.com>
Fixes: signed integer overflow: 1111638592 - -2122219136 cannot be represented in type 'int'
Fixes: 13441/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TRUEMOTION2_fuzzer-5732769815068672
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The first frame contains the sequence header, which is needed to parse every
following frame.
This fixes parsing streams with broken extradata but correct packet data.
Signed-off-by: James Almer <jamrial@gmail.com>
* commit '0676de935b1e81bc5b5698fef3e7d48ff2ea77ff':
arm: Implement a NEON version of 422 h264_h_loop_filter_chroma
Merged-by: James Almer <jamrial@gmail.com>
Some filters may not need to do linearize/delinearize, thus
will even not define them. Add ifdef check, so they could easily
re-use the .cl file.
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
These functions can be reused by other colorspace filters,
so move them to common file. No functional changes.
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
Nobody is going to check the queue anymore, so users must now subscribe to
send messages to ffmpeg-devel. This will prevent orphaned/ignored messages
from rotting in the abandoned queue. This matches the behavior of ffmpeg-user
and libav-user.
Also, this addresses some other nits.
Signed-off-by: Lou Logan <lou@lrcd.com>
The channel loop is now the outer loop for both planar and interleaved. This is
needed by the next patch, and the speed difference is negligable if any.
Signed-off-by: Marton Balint <cus@passwd.hu>
set_metadata with many entries is not very efficient, and with small audio
frames the performance loss is noticable. Also with this very simple
calculations (like peak) can be even further optimized.
Unfoturnately there are some small differences in metadata and av_log info
output, so factorizing calculations and output might not worth the hassle.
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes deadlocks when decoding packets containing more than one of the aforementioned
slices when using frame threads.
Tested-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
This is also what av_base64_encode() expects.
Fixes the following warnings with clang:
libavformat/sdp.c:394:40: warning: implicit conversion from 'int' to 'char' changes value from 254 to -2
libavformat/sdp.c:395:40: warning: implicit conversion from 'int' to 'char' changes value from 205 to -51
libavformat/sdp.c:396:40: warning: implicit conversion from 'int' to 'char' changes value from 186 to -70
Calculate bitrate based on fragment size, only applied when
bitrate is not set, for example rtsp source.
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Signed-off-by: Jun Li <junli1026@gmail.com>
Instead of assuming id 0 is used, use the same logic as used for PPS,
where all available entries in the list are emitted.
Signed-off-by: Aman Gupta <aman@tmm1.net>
This fixes avformat_query_codec incorrectly returning 0 for
mov container and mov_text subtitles.
Signed-off-by: Paweł Wegner <pawel.wegner95@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
* commit '49f9c4272c4029b57ff300d908ba03c6332fc9c4':
aarch64: vp8: Skip saturating in shrn in ff_vp8_idct_add_neon
Merged-by: James Almer <jamrial@gmail.com>
* commit 'cc7ba00c35faf0478f1f56215e926f70ccb31282':
aarch64: vp8: Port missing epel8 functions from arm version
Merged-by: James Almer <jamrial@gmail.com>
* commit '52c9b0a6c0d02cff6caebcf6989e565e05b55200':
aarch64: vp8: Port vp8_luma_dc_wht and vp8_idct_dc_add4uv from arm version
Merged-by: James Almer <jamrial@gmail.com>
* commit 'f1011ea28a4048ddec97794ca3e9901474fe055f':
aarch64: vp8: Reorder the function pointer inits to match the arm original
Merged-by: James Almer <jamrial@gmail.com>
* commit 'b4b27dce95a6d40bfcd78043d3abec7d80dae143':
aarch64: vp8: Move the vp8dsp makefile entries to the right places
aarch64: vp8: Remove superfluous includes
This commit is a noop, see
fecf75a5c4c8bc9d1380
Merged-by: James Almer <jamrial@gmail.com>
* commit '85bfaa4949f4afcde19061def3e8a18988964858':
aarch64: vp8: Use the proper aarch64 form for conditional branches
Merged-by: James Almer <jamrial@gmail.com>
* commit '2eeac79936e83c4495cbe5905064ab797e9b45ff':
aarch64: vp8: Fix assembling with armasm64
aarch64: vp8: Fix assembling with clang
This commit is a noop, see
c950beb68d7ddfa5e908
Merged-by: James Almer <jamrial@gmail.com>
* commit '0801853e640624537db386727b36fa97aa6258e7':
libavcodec: vp8 neon optimizations for aarch64
See 833fed5253
Merged-by: James Almer <jamrial@gmail.com>
* commit 'f8df5e2f31a5ba7b30a0e1caaaf5a03c753b3f9b':
tests: Add a convenience function for video-only lavf tests
Merged-by: James Almer <jamrial@gmail.com>
* commit 'a70eac7a9b193e8434b5bed90bd72aa4cb688363':
tests: Convert image2pipe tests to non-legacy test scripts
Merged-by: James Almer <jamrial@gmail.com>
This consists mostly of the following changes:
- add newly supported pixel formats (RGB555LE, RGB565LE, BGR0)
- select the ZMBV format (c->fmt) and bytes per pixel (c->bypp) based on
avctx->pix_fmt
- multiply widths/x-values by c->bypp, in places where bytes, not pixels, are
expected
- disable palette-writing code for non-palette pix_fmts
- make a note about histogram[]'s datatype (it could need increasing if
ZMBV_BLOCK is increased)
- adjust the c->score_tab length to take up to (and including) 4 times the
number of pixels in a block
- initialise c->score_tab up to c->bypp * the number of pixels
Note: the ZmbvFormat enum allows for additional bit depths:
- 1,2,4-bit (palette)
- 24-bit (RGB)
At time of writing the specifics of these (e.g. channel order, bit alignment)
are not currently defined, and DOSBox only implements support for 8/15/16/32
bpp.
One might expect the 24-bit format - if implemented - to be BGR24, to have the
same channel order as BGR0.
However, the decoder in zmbv.c has been guessed to use RGB24, so I have chosen
to not contradict this, and omitted specific support for this format.
Replaces the libdav1d internal allocator. It uses an AVBufferPool to reduce the
amount of allocated buffers.
About 5% speed up when decoding 720p or higher streams.
Reviewed-by: "Vittorio Giovara <vittorio.giovara@gmail.com>"
Signed-off-by: James Almer <jamrial@gmail.com>
Bump the minimum required version to the first one with the logger API callback.
Reviewed-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Since db77230894 parsing of
mpeg4-extradata lead to a "Failed to parse extradata" warning, because
ff_mpeg4_decode_picture_header returns AVERROR_INVALIDDATA in case that
no VOP was found. This patch adds a parameter to signify whether a
header (where the absence of a VOP does not raise an error) or not is
parsed. The first mode is of course used for parsing headers.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
* commit '8629149816930a43bf5a66b11c6224446cabd044':
tests: Drop duplicate variable declaration
This commit is a noop, as it breaks tests like fate-filter-pixfmts
Merged-by: James Almer <jamrial@gmail.com>
Reduces buffering latency with low bitrate streams, where
8192 bytes can mean several seconds.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Set specific field for repeat in PicStruct if the frame has repeat
flag.
Match the CheckInputPicStruct in MSDK.
Fix#7701.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
libvpx can be compiled with the VP8 decoder and encoder disabled, and
there's no reason to force their presence if the user only wants VP9.
Signed-off-by: James Almer <jamrial@gmail.com>
Dirac internally allocates 5 images per plane and frame currently. One being the actual
image the other 4 being filtered for motion compensation.
Fixes: Out of memory
Fixes: 12870/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5684825871089664
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
autorotate is enabled by default in ffmpeg so the rotation filters
are required and will be attempted for insertion without the user's
knowledge if an input stream has rotation side-data.
Fixes: runtime error: signed integer overflow: 2147483598 + 128 cannot be represented in type 'int'
Fixes: 12926/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_JPEG2000_fuzzer-5705100733972480
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
After this change we always parse the full specifier even if we know the result
in the middle of the parsing. Sligtly slower, but this is needed to
consistently reject incorrect specifiers in both matching and non-matching
cases.
Signed-off-by: Marton Balint <cus@passwd.hu>
This reworks the code to be more strict about accepting stream specifiers. From
now on we strictly enforce the syntax in the documentation up until the
decisive part of the stream specifier. Therefore matching stream specifiers
always need to be correct, non matching specifiers only need to be correct
until the decisive part.
Also recursion is changed to a simple loop.
Signed-off-by: Marton Balint <cus@passwd.hu>
This improves compatibility with some consumer (LG WebOS) TVs which apparently
search a HEVC descriptor (which our mpegts muxer can't generate) or a format
identifier.
Since the HEVC format identifier is not registered (but used in the wild), it is
not written if strict_std_compliance is higher than normal.
This fixes the issue in ticket #7744.
Signed-off-by: Marton Balint <cus@passwd.hu>
With all of our existing users of cuda_sdk switched over to ffnvcodec,
we could remove cuda_sdk completely and say that we should no longer
add code that requires the full sdk, and rather insist that such code
only use ffnvcodec.
As discussed previously, the use of nvcc from the sdk is still
supported with a distinct option.
Signed-off-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
This change switches the vf_thumbnail_cuda filter from using the
full cuda sdk to using the ffnvcodec headers and loader.
Most of the change is a direct mapping, but I also switched from
using texture references to using texture objects. This is supposed
to be the preferred way of using textures, and the texture object API
is the one I added to ffnvcodec.
Signed-off-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
This change switches the vf_scale_cuda filter from using the
full cuda sdk to using the ffnvcodec headers and loader.
Most of the change is a direct mapping, but I also switched from
using texture references to using texture objects. This is supposed
to be the preferred way of using textures, and the texture object API
is the one I added to ffnvcodec.
Signed-off-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
This change switches the vf_thumbnail_cuda filter from using the
full cuda sdk to using the ffnvcodec headers and loader.
Signed-off-by: Philip Langdale <philipl@overt.org>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
The use of nvcc to compile cuda kernels is distinct from the use of
cuda sdk libraries and linking against those libraries. We have
previously not bothered to distinguish these two cases because all
the filters that used cuda kernels also used the sdk. In the following
changes, I'm going to remove the sdk dependency from those filters,
but we need a way to ensure that nvcc is present and functioning, and
also a way to explicitly disable its use so that the filters are not
built.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
Empty edits can occur at any position within the edit list except for at
the end. Empty edits in the middle should not impact the reported stream
start_time or the video PTS adjustment, so only include empty edits at
the start of the list in empty_edits_sum_duration.
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
This avoids making invalid HTTP Range requests for a byte range past the
known end of the file during a seek. Those requests generally return a HTTP
response of 416 Range Not Satisfiable, which causes an error response.
Reference: https://tools.ietf.org/html/rfc7233
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Doesn't change anything, but makes the behaviour better match that of the
other codecs (the CONSTANT_QUALITY_ONLY flag already ensures that CQP is
the only RC mode selectable for MJPEG).
Following b8c45bbcbc they contain allocated
unit arrays which will get leaked. These operations were inconsistently
applied and never actually needed (the old uninit left them in the correct
state), so just drop them entirely.
Currently, a fragment's unit array is constantly reallocated during
splitting of a packet. This commit changes this: One can keep the units
array by distinguishing between the number of allocated and the number
of valid units in the units array.
The more units a packet is split into, the bigger the benefit.
So MPEG-2 benefits the most; for a video coming from an NTSC-DVD
(usually 32 units per frame) the average cost of cbs_insert_unit (for a
single unit) went down from 6717 decicycles to 450 decicycles (based
upon 10 runs with 4194304 runs each); if each packet consists of only
one unit, it went down from 2425 to 448; for a H.264 video where most
packets contain nine units, it went from 4431 to 450.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
This is in preparation for another patch that will stop needless
reallocations of the unit array.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Improves speed of the testcase by about a factor of 10
Fixes: Timeout
Fixes: 13132/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WCMV_fuzzer-5664190616829952
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Unifying the way the EBML unknown length is signaled, rather than using two
incompatible values. UINT64_MAX cannot be read as a valid EBML length with the
current code.
Co-authored-by: Steve Lhomme <robux4@ycbcr.xyz>
Co-authored-by: Dale Curtis <dalecurtis@chromium.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The function typedefs we were using are only present when using the
dynamic loader, which means compilation breaks for code directly
using the cuda SDK.
To fix this, let's just duplicate the function typedefs locally.
These are not going to change.
Improves speed from 5.4 to 4.2 seconds
Fixes: 13149/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PGM_fuzzer-5760833622114304
Fixes: 13166/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PGMYUV_fuzzer-5763216322330624
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Lauri Kasanen <cand@gmx.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The frame is not needed that early so obtaining it later avoids
the costly operation in case other checks fail.
Fixes: Timeout (14sec -> 4sec)
Fixes: 13140/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ZMBV_fuzzer-5738330308739072
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This way re-initializations properly update end_pts, enabling
sub2video_heartbeat to call sub2video_update as expected to re-init
the sub2video AVFrame's contents and to feed a frame into the filter
chain.
This then fixes memory usage ballooning due to framesync waiting
for secondary input in case of no actual subtitle samples being present
for a while in source after a re-init occurs.
Fixes: runtime error: signed integer overflow: 2147421862 - -33624063 cannot be represented in type 'int'
Fixes: 12885/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_H264_fuzzer-5733516975800320
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
When a JACOsub subtitle has two timestamps, they represent its start and
end times (http://unicorn.us.com/jacosub/jscripts.html#l_times); the
duration is the difference between the two, not the sum of the two.
The subtitle end times in the FATE test for this were wrong as a result;
fix them too. (This test is based on JACOsub's demo.txt, and the end
time computed for the last line using @ now matches what the comments
there say it should be.)
Also tested in practice using MPV, a LaserDisc, and some authentic 1993
JACOsub files.
Signed-off-by: Adam Sampson <ats@offog.org>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
* commit '156ea66c91b1986a87916f187216978d686725f6':
h264/x86: sign extend int stride in deblock functions
This commit is a noop, see d5d699ab6e
Merged-by: James Almer <jamrial@gmail.com>
* commit 'eec93e57096aa4804862d62760442380c70d489b':
libopenh264dec: Use a newer decoding entry point function
Merged-by: James Almer <jamrial@gmail.com>
* commit 'ca44fa5d7fda7e954f3ebfeb5b0d6d1be55fcaa3':
avcodec/libdav1d: properly free all output picture references
This commit is a noop, see 10931a0661
Merged-by: James Almer <jamrial@gmail.com>
* commit '70ab2778be9c83dab84340af7e3ba83fa0f98576':
libdav1d: update API usage to the first stable release
libdav1d: fix build after a recent API break
qsvenc: Add VDENC support for H264 and HEVC
avcodec: libdav1d AV1 decoder wrapper.
swscale: Add GRAY10
pixfmt: Add GRAY10
libx264: Pass the reordered_opaque field through the encoder
libavutil: Undeprecate the AVFrame reordered_opaque field
libaom: remove references to yuva444p pixfmt
Revert "decode: copy the output parameters from the last bsf in the chain back to the AVCodecContext"
This commit is a noop, see
87588caf8c4e9cff2824882ae091d43f1b5ca22eb5177c7051beaa350e24e92ce340e6
Merged-by: James Almer <jamrial@gmail.com>
* commit '1ff6cb2ca6652e7d2a929afd33d8ed6268c45568':
lavc/qsvenc_jpeg: set a default quality
lavc/qsvenc_jpeg: add async_depth support
This commit is a noop, see
0e3d7d845d92c25963e8
Merged-by: James Almer <jamrial@gmail.com>
* commit '04e8b8b0530e2aa33010faba3d0b6b6c9c5b704e':
avcodec/libaomenc: export the Sequence Header OBU as extradata
This commit is a noop. aom_codec_get_global_headers() is buggy at the moment.
See https://bugs.chromium.org/p/aomedia/issues/detail?id=2208
Merged-by: James Almer <jamrial@gmail.com>
* commit '97c9a5084479eeb66f4beb100cc7589a2c8bfe81':
avcodec/libaomenc: remove AVOption related to frame partitions
avcodec/extract_extradata: don't uninitialize the H2645Packet on every processed packet
avcodec/extract_extradata: Move the reference in the bsf internal buffer
avcodec/extract_extradata: Do not allocate more space than needed when removing NALUs in h264/hevc
avcodec/extract_extradata: Zero-initialize the padding bytes in all allocated buffers
avcodec/extract_extradata_bsf: Fix leak discovered via fuzzing
avcodec/bsf: Add ff_bsf_get_packet_ref() function
This commit is a noop, see
7ae52f8a6b5a412a5c3cd168e78eff2536bd86329c6dd9d624016d40011ab69ea742ab
Merged-by: James Almer <jamrial@gmail.com>
Even if NEON would be disabled, the init functions should be built
as they are called as long as ARCH_AARCH64 is set.
These functions are part of a generic DSP subsytem, not tied directly
to one decoder. (They should be built if the vp7 decoder is enabled,
even if the vp8 decoder is disabled.)
Signed-off-by: Martin Storsjö <martin@martin.st>
(cherry picked from commit b4b27dce95)
This also partially fixes assembling with MS armasm64 (via
gas-preprocessor).
The movrel macro invocations need to pass the offset via a separate
parameter. Mach-o and COFF relocations don't allow a negative
offset to a symbol, which is handled properly if the offset is passed
via the parameter. If no offset parameter is given, the macro
evaluates to something like "adrp x17, subpel_filters-16+(0)", which
older clang versions also fail to parse (the older clang versions
only support one single offset term, although it can be a parenthesis.
Signed-off-by: Martin Storsjö <martin@martin.st>
(cherry picked from commit 26d7af4c38)
If we fill with black then the generated palette will have one color more
than what the user requested. This also resulted in unwanted black specks in
the output of paletteuse, especially when generating small palettes.
- Clamp ME range to -64..63 (prevents corruption when me_range is too high)
- Allow MV's up to *and including* the positive range limit
- Allow out-of-edge ME by padding the prev buffer with a border of 0's
- Try previous MV before checking the rest (improves speed in some cases)
- More robust logic in code - ensure *mx,*my,*xored are updated together
- Improve block choices by counting 0-bytes in the entropy score
- Make histogram use uint16_t type, to allow byte counts from 16*16
(current block size) up to 255*255 (maximum allowed 8bpp block size)
- Make sure score table is big enough for a full block's worth of bytes
- Calculate *xored without using code in inner loop
The previous version was a pretty exact translation of the arm
version. This version does do some unnecessary arithemetic (it does
more operations on vectors that are only half filled; it does 4
uaddw and 4 sqxtun instead of 2 of each), but it reduces the overhead
of packing data together (which could be done for free in the arm
version).
This gives a decent speedup on Cortex A53, a minor speedup on
A72 and a very minor slowdown on Cortex A73.
Before: Cortex A53 A72 A73
vp8_idct_add_neon: 79.7 67.5 65.0
After:
vp8_idct_add_neon: 67.7 64.8 66.7
Signed-off-by: Martin Storsjö <martin@martin.st>
The original arm version didn't do saturation here. This probably
doesn't make any difference for performance, but reduces the
differences.
Signed-off-by: Martin Storsjö <martin@martin.st>
This makes it similar to put_epel16_v6, and gives a large speedup
on Cortex A53, a minor speedup on A72 and a very minor slowdown on
A73.
Before: Cortex A53 A72 A73
vp8_put_epel16_h6v6_neon: 2211.4 1586.5 1431.7
After:
vp8_put_epel16_h6v6_neon: 1736.9 1522.0 1448.1
Signed-off-by: Martin Storsjö <martin@martin.st>
This makes it similar to put_epel16_v6, and gives a 10-25%
speedup of this function.
Before: Cortex A7 A8 A9 A53 A72
vp8_put_epel16_h6v6_neon: 3058.0 2218.5 2459.8 2183.0 1572.2
After:
vp8_put_epel16_h6v6_neon: 2670.8 1934.2 2244.4 1729.4 1503.9
Signed-off-by: Martin Storsjö <martin@martin.st>
Even if NEON would be disabled, the init functions should be built
as they are called as long as ARCH_AARCH64 is set.
These functions are part of a generic DSP subsytem, not tied directly
to one decoder. (They should be built if the vp7 decoder is enabled,
even if the vp8 decoder is disabled.)
Signed-off-by: Martin Storsjö <martin@martin.st>
The previous form also does seem to assemble on current tools,
but I think it might fail on some older aarch64 tools.
Signed-off-by: Martin Storsjö <martin@martin.st>
This also partially fixes assembling with MS armasm64 (via
gas-preprocessor).
The movrel macro invocations need to pass the offset via a separate
parameter. Mach-o and COFF relocations don't allow a negative
offset to a symbol, which is handled properly if the offset is passed
via the parameter. If no offset parameter is given, the macro
evaluates to something like "adrp x17, subpel_filters-16+(0)", which
older clang versions also fail to parse (the older clang versions
only support one single offset term, although it can be a parenthesis.
Signed-off-by: Martin Storsjö <martin@martin.st>
This was found through the Hacker One program on VLC but is not a security issue in libavformat
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is the equivalent change for cuviddec after the previous change
for nvdec. I made similar changes to the copying routines to handle
pixel formats in a more generic way.
Note that unlike with nvdec, there is no confusion about the ability
of a codec to output 444 formats. This is because the cuvid parser is
used, meaning that 444 JPEG content is still indicated as using a 420
output format.
With the introduction of HEVC 444 support, we technically have two
codecs that can handle 444 - HEVC and MJPEG. In the case of MJPEG,
it can decode, but can only output one of the semi-planar formats.
That means we need additional logic to decide whether to use a
444 output format or not.
The latest generation video decoder on the Turing chips supports
decoding HEVC 4:4:4. Supporting this is relatively straight-forward;
we need to account for the different chroma format and pick the
right output and sw formats at the right times.
There was one bug which was the hard-coded assumption that the
first chroma plane would be half-height; I fixed this to use the
actual shift value on the plane.
We also need to pass the SPS and PPS range extension flags.
We need all the flags to be exposed to be able to pass them on to
HW decoders. I did not attempt to nuance any of the warnings about
flags being unsupported as there's no way, at the point we extract
flags, to say whether an HW decoder is being used.
This removes lots of code duplication and also allows more complex specifiers,
for example you can use p:204:a:m:language:eng to select the English language
audio stream from program 204.
Signed-off-by: Marton Balint <cus@passwd.hu>
ISO-10646 alone means UCS-4 for iconv, the specs refers to the Basic
Multilingual Plane (BMP), therefore we need UCS-2. VLC also using that.
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes some rather embarrassing mistakes that somehow passed my
eyes.
* Now catches if memory allocation has failed during bprint usage
by checking av_bprint_is_complete().
* Now catches if adding an ASS rectangle into an AVSubtitle failed.
* Returns AVERROR_INVALIDDATA if we get an invalid region buffer
length.
Use av_ts2str() for AVFrame.pkt_dts/pts to avoid print the
pkt_dts/pts as negative number like:
"0, 3616613, -9223372036854775808, 1001, 3110400, 0x75e37a65"
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
SRT API Documentation:
This flag is superfluous if both parties are at least version 1.3.0
(this shall be enforced by setting this value to SRTO_MINVERSION if
you expect that it be true) and therefore support HSv5 handshake,
where the SRT extended handshake is done with the overall handshake
process.
This flag is however obligatory if at least one party may be using
SRT below version 1.3.0 and does not support HSv5.
Fix build warning like "warning: ISO C90 forbids mixed declarations
and code" after adjust the location for malloc fail check.
Reviewed-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Need to check malloc fail before using it, so adjust the location
in the code.
Reviewed-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
If we enable a component but a dependant library is disabled, then the enabled
component gets silently disabled. Warning about disabled explicitly enabled components
allows configure to show the missing dependencies and if --fatal-warnings is
used it can also fail if the user wants it so.
For example if libdav1d is not availble ./configure --enable-decoder=libdav1d
succeeds but the libdav1d decoder is not be enabled. After the patch configure
will warn about this:
WARNING: Disabled libdav1d_decoder because not all dependencies are satisfied: libdav1d
Signed-off-by: Marton Balint <cus@passwd.hu>
Binary searching would hang if the fragment items do NOT have timestamp for the
specified stream.
For example, a fmp4 consists of separated 'moof' boxes for each track, and
separated 'sidx' for each segment, but no 'mfra' box. Then every fragment item
only have the timestamp for one of its tracks.
Example:
ffmpeg -f lavfi -i testsrc -f lavfi -i sine -movflags dash+frag_keyframe+skip_trailer+separate_moof -t 1 out.mp4
ffmpeg -ss 0.5 -i out.mp4 -f null none
Also fixes the hang in ticket #7572, but not the reason for having
AV_NOPTS_VALUE timestamps there.
Signed-off-by: Charles Liu <liuchh83@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
when set option fmp4_init_filename to init_%v.mp4
before patch:
the init file will be init_%v_0.mp4, init_%v_1.mp4
after patch:
the init file will be init_0.mp4, init_1.mp4
Reported-By: Gyan Doshi <ffmpeg@gyani.pro>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
This enables us to read the data coding type utilized for
a specific private data stream, of which we currently are
interested in ARIB caption streams.
The component tag limitations are according to ARIB TR-B14,
and the component IDs are defined in ARIB STD-B10.
* Outputs ASS lines with basic coloring and font scaling for each
given region.
* Sets the default style to the resolution of the subtitle plane
(for example, 960x540 / 36pt font for profile A).
* Has options to:
* Disable ruby text (which is coded as regions which have
half-height text in libaribb24).
Enabled by default as without positioning ruby text only
confuses as it is usually coded in the beginning of the decoded
subtitle line.
* Set the working directory, in which libaribb24 will read
configuration as well as into which it may save broadcast extra
symbols as PNG.
Unset by default.
The unconventional library check can be explained by the library's
current master branch being licensed as LGPLv3, but at the time of
writing the latest official release is still licensed under GPLv3.
Thus, one either has to wait for the following release, or enable
GPLv3.
This is robust for some corner case there is incorrect list1 count
in pps header, but it's a P slice and can be decoded well.
Signed-off-by: Decai Lin <decai.lin@intel.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Bad content may contain stsc boxes with a first_chunk index that
exceeds stco.entries (chunk_count). This ammends the existing check to
include cases where chunk_count == 0. It also patches up the case
when stsc refers to unknown chunks, but stts has no samples (so we
can simply ignore stsc).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Detecting missing tfhd avoids re-using tfhd track info from the previous
moof. For files with multiple tracks, this may make a mess of the
avindex and fragindex, which can later trigger av_assert0 in
mov_read_trun().
Reviewed-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes a problem where a sample entry which cannot be written correctly appears to succeed, but produces an invalid file.
For example, this command:
ffmpeg -f lavfi -i sine=frequency=1000:duration=5 -codec:a ac3 -movflags +empty_moov -frag_duration 5000000 /tmp/foo.mp4
produced a file with the ac-3 sample entry, but no AC3SpecificBox (dac3) child, which is invalid according to ETSI TS 102 366.
Reviewed-by: Baptiste Coudurier <baptiste.coudurier@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
./ffmpeg_g -f rawvideo -pix_fmt rgb24 -s hd1080 -i /dev/zero -pix_fmt yuv420p16be \
-s 1920x1728 -f null -vframes 100 -v error -nostats -
9-14 bit funcs get about 6x speedup, 16-bit gets about 15x.
Fate passes, each format tested with an image to video conversion.
Only POWER8 includes 32-bit vector multiplies, so POWER7 is locked out
of the 16-bit function. This includes the vec_mulo/mule functions too,
not just vmuluwm.
With TIMER_REPORT skips disabled:
yuv420p9le
12412 UNITS in planarX, 131072 runs, 0 skips
73136 UNITS in planarX, 131072 runs, 0 skips
yuv420p9be
12481 UNITS in planarX, 131072 runs, 0 skips
73410 UNITS in planarX, 131072 runs, 0 skips
yuv420p10le
12322 UNITS in planarX, 131072 runs, 0 skips
72546 UNITS in planarX, 131072 runs, 0 skips
yuv420p10be
12291 UNITS in planarX, 131072 runs, 0 skips
72935 UNITS in planarX, 131072 runs, 0 skips
yuv420p12le
12316 UNITS in planarX, 131072 runs, 0 skips
72708 UNITS in planarX, 131072 runs, 0 skips
yuv420p12be
12319 UNITS in planarX, 131072 runs, 0 skips
72577 UNITS in planarX, 131072 runs, 0 skips
yuv420p14le
12259 UNITS in planarX, 131072 runs, 0 skips
72516 UNITS in planarX, 131072 runs, 0 skips
yuv420p14be
12440 UNITS in planarX, 131072 runs, 0 skips
72962 UNITS in planarX, 131072 runs, 0 skips
yuv420p16le
10548 UNITS in planarX, 131072 runs, 0 skips
73429 UNITS in planarX, 131072 runs, 0 skips
yuv420p16be
10634 UNITS in planarX, 131072 runs, 0 skips
150959 UNITS in planarX, 131072 runs, 0 skips
Signed-off-by: Lauri Kasanen <cand@gmx.com>
Fixes some random assertion failures with
ffprobe -show_packets async:samples/ffmpeg-bugs/trac/ticket6132/Samsung_HDR_-_Chasing_the_Light.ts > /dev/null
Signed-off-by: Marton Balint <cus@passwd.hu>
instead of an ad-hoc function to search for start codes in order to
remove code duplication and to improve performance.
Improved performance of finding startcodes from 52606 decicycles to
9543 decicycles based upon 262144 runs for a 1 Mb/s MPEG4 video.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Optimize put_hevc_qpel_h_8 with mmi in the case width=4/8/12/16/24/32/48/64.
This optimization improved HEVC decoding performance 2%(2.39x to 2.44x, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Optimize put_hevc_qpel_bi_h_8 with mmi in the case width=4/8/12/16/24/32/48/64.
This optimization improved HEVC decoding performance 2.1%(2.34x to 2.39x, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Optimize put_hevc_epel_bi_hv_8 with mmi in the case width=4/8/12/16/24/32.
This optimization improved HEVC decoding performance 1.7%(2.30x to 2.34x, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Optimize put_hevc_qpel_uni_hv_8 with mmi in the case width=4/8/12/16/24/32/48/64.
This optimization improved HEVC decoding performance 2.7%(2.24x to 2.30x, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes a warning with clang:
libavutil/imgutils.c:314:16: warning: absolute value function 'abs'
given an argument of type 'ptrdiff_t' (aka 'long') but has
parameter of type 'int' which may cause truncation of value
Remove the pdiff_lut_scale in nlmeans and increase weight_lut table size
from 2^9 to 500000, this change will avoid using pdiff_lut_scale in
nlmeans_slice() for weight_lut table search, improving the performance
by about 12%. (in 1080P size picture case).
Use the profiling command like:
perf stat -a -d -r 5 ./ffmpeg -i input -an -vf nlmeans=s=30 -vframes 10 \
-f null /dev/null
without this change:
when s=1.0(default value) 63s
s=30.0 72s
after this change:
s=1.0(default value) 56s
s=30.0 63s
Reviewed-by: Carl Eugen Hoyos <ceffmpeg@gmail.com>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Signed-off-by: Clément Bœsch <u@pkh.me>
No speed difference, or slightly faster (the difference is too small so it may be noise
that this appears faster)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
We assume that if the compressed size is bigger than if each byte is encoded in a single raw packet
that the data is invalid.
Fixes: Out of memory
Fixes: 12208/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_RASC_fuzzer-5648916473708544
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
In Windows if using scaling other than 100% then the grabbed window was not captured fully (cropped)
Signed-off-by: Dilshod Mukhtarov <dilshodm@gmail.com>
Fixes: Timeout
Fixes: 12192/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_RSCC_fuzzer-6279038004363264
Before: clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_RSCC_fuzzer-6279038004363264 in 15423 ms
After: clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_RSCC_fuzzer-6279038004363264 in 190 ms
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: 12447/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HYMT_fuzzer-5201623956062208
Fixes: 12458/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HYMT_fuzzer-5705567736168448
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The addition is moved up into the context where the variable is unsigned avoiding
the undefined behavior
Fixes: runtime error: signed integer overflow: 2147481972 + 4096 cannot be represented in type 'int'
Fixes: 12444/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ILBC_fuzzer-5755706244857856
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
When making a new P-frame when B-frames are present the previous P-frame
is normally in the DPB because it will be referred to by subsequent
B-frames. However, this is not true if there are no B-frames, or in edge
cases where a GOP ends with two P-frames. Fix this by adding the direct
ref pics to the RPS explicitly.
Fixes#7699.
Tested-by: Ullysses A Eoff <ullysses.a.eoff@intel.com>
The timestamp of the changed input frame as well as its relevant
properties can be examined by the user. Only applicable when
reinit_filter is disabled on the input stream.
The specification requires the demuxer to only read the data
atom up to its given size, this is necessary as atoms are
allowed at the end of the file.
This patch duplicates the behaviour of the QuickTime player.
The "new" entry point actually has existed since OpenH264 1.4 in
2015 and is the the recommended decoding entry point.
The name of this function, DecodeFrameNoDelay, is rather backwards
considering that it doesn't return the latest decoded frame immediately,
but actually does proper delaying and reordering of frames.
Signed-off-by: Martin Storsjö <martin@martin.st>
The VP3/4/5/6 reference decoders all use three IDCT versions: one for the
DC-only case, another for blocks with more than 10 coefficients, and an
optimised one for blocks with up to 10 AC coefficents. VP6 relies on the
sparse 10 coefficient version, and without it, IDCT drift occurs.
Fixes: https://trac.ffmpeg.org/ticket/1282
Signed-off-by: Peter Ross <pross@xvid.org>
As .rodata isn't one of the default created sections for COFF, it was
created as a read-write data section. By using the default .rdata
section name for COFF, it automatically becomes a read-only data section.
The existing ".section .rodata" works as intended for ELF though.
This is based on an original patch and diagnose by Tom Tan
<Tom.Tan@microsoft.com>.
Signed-off-by: Martin Storsjö <martin@martin.st>
QVBR mode is to use the variable bitrate control algorithm
with constant quality.
mfxExtCodingOption3 should be supported to enable QVBR mode.
It is neccesary to specify a max_rate for QVBR, else it may be ICQ mode.
Example usage: ffmpeg -hwaccel qsv -c:v h264_qsv -i input.mp4 -c:v
h264_qsv -global_quality 25 -maxrate 2M test_qvbr.mp4 -v verbose
Clip QVBR quality range to be [0, 51] as Mark's commments.
It is similar to qp range of CQP but possibly should be updated when VP8/VP9
encoding can be supported.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: Zhong Li <zhong.li@intel.com>
This attaches the logic of picking the mode of for the next picture to
the output, which simplifies some choices by removing the concept of
the picture for which input is not yet available. At the same time,
we allow more complex reference structures and track more reference
metadata (particularly the contents of the DPB) for use in the
codec-specific code.
It also adds flags to explicitly track the available features of the
different codecs. The new structure also allows open-GOP support, so
that is now available for codecs which can do it.
Now memcpy can be avoided for NAL units containing escapes, too.
Particularly improves performance for files with hardcoded black bars.
For such a file, time spent in cbs_h2645_split_fragment went down from
369410 decicycles to 327677 decicycles. (It were 379114 decicycles when
every NAL unit was copied.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
This is in preparation for a patch for cbs_h2645. Now the packet's
rbsp_buffer can be owned by an AVBuffer.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Dav1dPictures contain more than one buffer reference, so we're forced to use the
API properly to free them all.
Signed-off-by: James Almer <jamrial@gmail.com>
This allows the underlying files to change their duration on subsequent
avformat context opens.
An example use case where this matters:
ffconcat version 1.0
file dummy.mxf
file dummy.mxf
ffmpeg -re -stream_loop -1 -i dummy.ffconcat -f sdl2 none
The user can seamlessly change the input by atomically replacing dummy.mxf.
v2: Set ConcatFile duration in read_header for all segments with known
durations because from now on we always recalculate the start time in
open_file, and an instant seek could have caused unset ConcatFile durations.
Signed-off-by: Marton Balint <cus@passwd.hu>
Optimize put_hevc_qpel_bi_hv_8 with mmi in the case width=4/8/12/16/24/32/48/64.
This optimization improved HEVC decoding performance 11.4%(2.01x to 2.24x, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Optimize put_hevc_qpel_hv_8 with mmi in the case width=4/8/12/16/24/32/48/64.
This optimization improved HEVC decoding performance 11%(1.81x to 2.01x, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The compiled libavcodec/tests/codec_desc was left out of that dir's
.gitignore when the test was added, so it shows up in 'git status'
as an untracked file if it's been built.
Signed-off-by: FeRD (Frank Dana) <ferdnyc@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
if 0 is encountered in the LUT then decompress() will continue to output 0 bytes but never read more data.
Without a specification it is impossible to say if this is invalid or a feature.
None of the valid prosumer files tested cause a 0 to be read, so it is likely
not a intended feature.
Fixes: Timeout
Fixes: 11266/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PROSUMER_fuzzer-5681827423977472
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 strongly based on code by Marton Balint, and depends on the previous commit
Fixes: Timeout
Fixes: 11502/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WCMV_fuzzer-5664893810769920
Before: Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WCMV_fuzzer-5664893810769920 in 11209 ms
After: Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WCMV_fuzzer-5664893810769920 in 4104 ms
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Marton Balint <cus@passwd.hu>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Optimize put_hevc_pel_bi_pixels_8 with mmi in the case width=8/16/24/32/48/64.
This optimization improved HEVC decoding performance 2%(1.77x to 1.81x, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This patch just enables the path from ffmpeg to libx264,
the more encoders can be added later.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
The encoders such as libx264 support different QPs offset for different MBs,
it makes possible for ROI-based encoding. It makes sense to add support
within ffmpeg to generate/accept ROI infos and pass into encoders.
Typical usage: After AVFrame is decoded, a ffmpeg filter or user's code
generates ROI info for that frame, and the encoder finally does the
ROI-based encoding.
The ROI info is maintained as side data of AVFrame.
Signed-off-by: Guo, Yejun <yejun.guo@intel.com>
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Fixes: Timeout
Fixes: 11663/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5636791864918016
Before:Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5636791864918016 in 26006 ms
After: Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5636791864918016 in 106 ms
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: Timeout
Fixes: 11619/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AC3_FIXED_fuzzer-5632398021099520
Fixes: 11620/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AC3_fuzzer-5711996515778560
Fixes: 11658/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_EAC3_fuzzer-5701006524940288
Before: Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AC3_FIXED_fuzzer-5632398021099520 in 20338 ms
After: Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AC3_FIXED_fuzzer-5632398021099520 in 11825 ms
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes a warning using musl:
In file included from libavformat/rtpproto.c:43:0:
/usr/local/musl/include/sys/poll.h:1:2: warning: #warning redirecting incorrect #include <sys/poll.h> to <poll.h>
Fixes: Timeout
Fixes: 11067/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_GDV_fuzzer-5686623711264768
Before change: Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_GDV_fuzzer-5686623711264768 in 34386 ms
After change: Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_GDV_fuzzer-5686623711264768 in 24327 ms
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The B predictor for 4-MV MBs in interlace field pictures is not used
for blocks 0 and 2 when the picture is 1 MB wide.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The position of the second MV predicitor candidate is slightly different
for the old WMV3 format indicated by RES_RTM_FLAG. This patch fixes
decoding of niceday.wmv on the samples server.
Fixes: #6641
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The B predictor for 4-MV macroblocks is only out of bounds when
the A predictor is also out of bounds.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
this was a typo in my original dst decoder. there is no requirement for
64-byte alignment here. the change does not affect decoder performance.
Signed-off-by: Peter Ross <pross@xvid.org>
Report the detailed log with buf_size in parse_nal_units to provide
more information when picture could not be found.
Match the behaviour in h264_parser.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
macros for reading and writing 64-bit aligned little-endian values.
these macros are used by the DST decoder and give a performance boost
on platforms that where the compiler must guard against unaligned
memory access.
Add VDENC(lowpower mode) support for QSV H264
It's an experimental function(like lowpower in vaapi) with
some limitations:
- CBR/VBR require HuC which should be explicitly loaded via i915
module parameter(i915.enable_guc=2 for linux kernel version >= 4.16)
Use option "-low_power 1" to enable VDENC.
Add in dump_video_param() to show the status of VDENC in runtime log.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
Dav1dPictures contain more than one buffer reference, so we're forced to use the
API properly to free them all.
Reviewed-by: BBB
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes: Timeout
Fixes: 11354/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QPEG_fuzzer-5766275943366656
Before: Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QPEG_fuzzer-5766275943366656 in 9470 ms
After : Executed clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QPEG_fuzzer-5766275943366656 in 134 ms
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array access
Fixes: 12381/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HYMT_fuzzer-5705474280783872
Fixes: 12384/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HYMT_fuzzer-5725303345774592
Fixes: 12389/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HYMT_fuzzer-5704033050820608
Fixes: 12391/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HYMT_fuzzer-5707284146028544
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
DXVA2 may be enabled even when every relevant module is disabled,
which would result in the dependency generator not including its
extralibs to avcodec.
Fixes ticket #7642.
Signed-off-by: James Almer <jamrial@gmail.com>
The variable "err" is not initialized, and set only when something went wrong.
When everything is OK, nobody sets it, so using it result in UB.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Before patch:
init nbits = 17, get 10000 samples, average cost: 16175 us
After patch:
init nbits = 17, get 10000 samples, average cost: 14989 us
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
init add three test examples:
1. check no endlist at the end
2. check endlist at the end
3. check hls_list_size 0 full list
Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
Some of the assembly functions e.g. ff_h264_idct_dc_add_neon
has code like:
movrel x14, X(ff_h264_idct_add_neon)
Linker cannot resolve them fully at link time and emits dynamic
relocations.
Use explicit labels instead so that no dynamic relocations are
needed at all.
This avoids lld complains about text relocations.
For background, see https://crbug.com/917919
Signed-off-by: Manoj Gupta <manojgupta@chromium.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
fcmul_add_c: 1228.8
fcmul_add_sse3: 334.3
fcmul_add_avx: 186.3
Tested on a Core i5 4460 @ 3.2GHz
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Fully discarded streams can't be selected for output or mapped or filtered.
Previously, a few packets from such streams, probably buffered for
stream probing, would get smuggled into output files.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This breaks totally valid files that get caught in its heuristic.
This, according to the commit message, is my own doing, having asked
Michael to implement this check and providing a sample that was
"wrong". I am now atoning for my sins, and removing this hack, having
seen the light (aka that this was silly to do in the first place).
Resotores correct behavior on valid files.
This reverts commit 8e5e84c2a2.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
A unsupported codec_id is a internal error and should not happen
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This should never be untrue, if it is, thats a bug
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This improves the speed of decode_run_i()
After: clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SCPR_fuzzer-5656821117747200 in 13516 ms
Before: clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SCPR_fuzzer-5656821117747200 in 14018 ms
Improves: 11270/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SCPR_fuzzer-5656821117747200
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This improves the speed of decode_run_i()
Before: clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SCPR_fuzzer-5656821117747200 in 17420 ms
After: clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SCPR_fuzzer-5656821117747200 in 14018 ms
Improves: 11270/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SCPR_fuzzer-5656821117747200
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Frame properties were not copied, so e.g. PTS was not set for the last frame.
Regression since ef3babb2c7.
Signed-off-by: Marton Balint <cus@passwd.hu>
The problem occurs in slice quant estimation and slice encoding:
If the slice quant is larger than MAX_STORED_Q we don't use pre-calculated
quant matrices, but generate a new one, but both qmat and qmat_chroma both
point to the same table, so the luma table ends up having chroma table
values.
Add custom_chroma_q the same way as custom_q.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Script to download and test ossfuzz testcases
This also includes a list of such testcases.
I intend to subsequently fill this list with the cases we have fixed in the past
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This function wouldn't benefit from VSX instructions, so I put it
under altivec.
./ffmpeg_g -f rawvideo -pix_fmt rgb24 -s hd1080 -i /dev/zero -pix_fmt grayf32le \
-f null -vframes 100 -v error -nostats -
3743 UNITS in planar1, 65495 runs, 41 skips
-cpuflags 0
23511 UNITS in planar1, 65530 runs, 6 skips
grayf32be
4647 UNITS in planar1, 65449 runs, 87 skips
-cpuflags 0
28608 UNITS in planar1, 65530 runs, 6 skips
The native speedup is 6.28133, and the bswapping one 6.15623.
Fate passes, each format tested with an image to video conversion.
Signed-off-by: Lauri Kasanen <cand@gmx.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is a regression introduced by 223d2bde22.
It appears that regression was introduced in 4.1, 4.0.x does not share
this behaviour.
Temp files were not created for MPEG-TS segments options - HLS_TEMP_FILE
flag was never set on AVFormatContext, it is however set on HLSContext object.
In order to fix this issue, proper flags field must be checked. In addition,
renaming code was messed up and apparently was working only for MP4 files.
This commit adds configuration options to libvpxenc.c that can be used to
tune the sharpness parameter for VP8 and VP9.
Signed-off-by: Rene Claus <rclaus@google.com>
Signed-off-by: James Zern <jzern@google.com>
The AV_INPUT_BUFFER_PADDING_SIZE has been increased to 64, but the value is still 32
in function ff_hevc_sao_edge_filter_8_msa. So, use AV_INPUT_BUFFER_PADDING_SIZE directly.
Also, use MAX_PB_SIZE directly instead of 64. Fate tests passed.
Reviewed-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
webrtc uses a int32_t like the existing code in ilbcdec
Fixes: signed integer overflow: 2080245063 + 257939661 cannot be represented in type 'int'
Fixes: 11037/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ILBC_fuzzer-5682976612941824
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This causes windows to fail as the timestamp is outside its supported range
Fixes regression & fate
Reviewed-by: Marton Balint <cus@passwd.hu>
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes issues when a subtitle packet is received before PCR for the
program has been received, leading to wildly jumping timestamps
on the lavf client side as well as in the re-ordering logic.
This usually happens in case of multiplexes where the PCR of a
program is not taken into account with subtitle tracks' DTS/PTS.
The specification allows 2^32-1 to be encoded as any number of zeroes
greater than 31, followed by a one. This previously failed because the
trace code would overflow the array containing the string representation
of the bits if there were more than 63 zeroes. Fix that by splitting the
trace output into batches, and at the same time move it out of the default
path.
(While this seems likely to be a specification error, libaom does support
it so we probably should as well.)
From a test case by keval shah <skeval65@gmail.com>.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
This avoids problems if %4 is the stack pointer
the constraints do not allow %4 to be the stack pointer but gcc 9 may
no longer support specifying such constraints
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Silences a warning with clang on arm:
libavcodec/opus_rc.c:170:17: warning: passing 'const uint8_t *' (aka 'const unsigned char *') to parameter of type 'void *' discards qualifiers
The dynamic metadata contains data for color volume transform -
application 4 of SMPTE 2094-40:2016 standard. The data comes from
HEVC in the SEI_TYPE_USER_DATA_REGISTERED_ITU_T_T35.
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
delta_frame_id_minus1 is not a single value in the bitstream, and can
store values up to 17 bits wide.
Fixes parsing files with frame ids.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
Change the some options location in avcodec_options to make code more
readable. And update the fate test with this change.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Set initialization order of MSA after MMI to make it work on loongson platform(msa is supported by loongson2k、3a4000 etc.).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
1. Old logic meaned: everywhere, except Windows, ffmpeg has to use HW
acceleration, but on Windows ffmpeg has to use (unavailable) software
HEVC by default
2. Software HEVC is available only if you provide corresponding
software MediaSDK library, which isn't provided with ffmpeg. More
information could be found in
https://github.com/Intel-Media-SDK/MediaSDK/blob/master/doc/samples/readme-encode_linux.pdf
3. HW HEVC decoding/encoding are available on Windows in the driver by default
Note: Default case should be the most common case but this change still has potential risk
on windows if HW path is not supported(or doesn't work as expection).
(See the historical disscution: https://lists.libav.org/pipermail/libav-devel/2016-November/080419.html).
In such case, two options suggested:
1. Use the option "-load_plugin hevc_sw" to switch SW path manually.
2. Or report bug to Intel windows driver if your GPU can support HEVC HW codec.
(HEVC decoding is supported since Braswell, and encoding supported since Skylake)
Patch started by Landgraph. Add similar change for hevc decoder and bump a new version.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Reviewed-by: Maxym Dmytrychenko <maxim.d33@gmail.com>
Signed-off-by: Landgraph <me@landgraph.ru>
Signed-off-by: Zhong Li <zhong.li@intel.com>
These are registered identifiers at the MPEG-4 RA, which are
defined as to be utilized for Dolby Vision AVC/HEVC streams that
are not correctly presentable by standards-compliant AVC/HEVC players.
According to the Dolby Vision specification for ISOBMFF, these sample
entry codes are specified to have the standard AVC or HEVC decoder
configuration box in addition to the Dolby custom DOVIConfigurationBox.
This is what enables us to decode the streams without custom parsing.
For correct presentation information from the DOVIConfigurationBox
is required (YCbCr or modified ICtCP, SDR or HDR, base or enhancement
layer).
avio_read may return EOF, leaving the mimetype array unitialized. fail
early when this occurs to avoid using the array in an unitialized state.
Reviewed-by: Tomas Härdin <tjoppen@acc.umu.se>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
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.
Chromium fuzzing produced a whacky file with extra tkhds. This caused
an AVStream that was already in use to be corrupted by assigning it a
new id, which blows up later in mov_read_trun because the
MOVFragmentStreamInfo.index_entry now points OOB.
Reviewed-by: Baptiste Coudurier <baptiste.coudurier@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes#7410.
The value of sub-pixel precision for me/mc can change during an Intra frame. In multi-threaded decoding this change is not propagated to other frame threads causing decoding artifacts. This patch initializes the sub-pixel precision parameter from previous thread, which fixes the issue.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Now "-c copy" works.
Update FATE files.
Demuxer only split file into packets, no data is trimmed.
Encoder & muxer currently expect completely another format
where muxer writes stuff like disposal method which should
be really encoder job.
With this patch muxer only modifies delay between two packets.
Codec copy need to have same behavior between demuxer and
muxer to work correctly.
Fixes#6640.
The color fields were moved to another struct, and a way to propagate
timestamps and other input metadata was introduced, so the packet
fifo can be removed.
Add support for 12bit streams, an option to disable film grain, and
read the profile from the sequence header referenced by the ouput
picture instead of guessing based on output pix_fmt.
Signed-off-by: James Almer <jamrial@gmail.com>
It is unsigned at least on arm, fixes the following warnings:
libavfilter/vf_signalstats.c:833:21: warning: taking the absolute value of unsigned type 'unsigned int' has no effect
dify += abs(yuv - AV_RN16(prev->data[0] + pw + i * 2));
The issue is that the afq still has samples as on init it counts
the overlap used as a delay to adjust the PTS it generates, hence
we can't rely on it right after init.
So just check to see if any frames have been encoded. frame_number
can't be anything but 0 right after init and can only be set by lavc.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
this section has been moved into the CONFIG_NETWORK block, since it only
affects network enabled builds.
sys/socket.h (with WIN32 guard) is needed to check if the SHUT_xxx macro exists.
Instead add the character to the snprintf above as suggested by Mark.
Silences a warning:
libavformat/matroskadec.c: In function 'webm_dash_manifest_cues':
libavformat/matroskadec.c:3947:13: warning: 'strncat' specified bound 1 equals source length [-Wstringop-overflow=]
strncat(buf, ",", 1);
^~~~~~~~~~~~~~~~~~~~
before this change, scale_vaapi hard coding the scaling mode, add a
new option "mode" to setting the scaling mode, it can be use to change
scaling algorithm for performance/quality trade off.
Note: VA_FILTER_SCALING_NL_ANAMORPHIC(nolinear ananorphic scaling
algorithm) doesn't support by i965/iHD driver now, add this just for
completeness.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
This patch fixes the segmentation fault issues due to
unhandled errors from ff_http_do_new_request function.
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Fix the (m)jpeg encoding regression issue as decription in tikect #7593,
due to bs->FrameType is not set in such case in
MSDK (https://github.com/Intel-Media-SDK/MediaSDK/issues/970).
(And assert on a value coming from an external library is not proper.)
Add default type check for bs->FrameType, and return invalid data error in function
ff_qsv_encode to avoid using uninitialized value.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
Silences a warning with clang:
libavcodec/cbs_vp9_syntax_template.c:220:17: warning: implicit conversion from 'int' to 'int8_t' (aka 'signed char')
changes value from 255 to -1
Assertion avctx->codec_id != AV_CODEC_ID_NONE failed at src/libavcodec/parser.c:128
The setting on the internal AVCodecContext used for parsing only is
otherwise irrelevant, so just set it to avoid the assert.
RepeatPPS is enabled by default in mfx. It is not necessary mostly and
wasting encoding bits.
Add an option to control it and disable it by default.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: Zhong Li <zhong.li@intel.com>
The affected functions could also be changed into macros, this is the
smaller change to fix it though. And avoids (probably) less readable macros
The extra code should be optimized out when optimizations are done as all values
are known at build after inlining.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Such low resolution would result in empty output as a minimum of 4x4 is needed
We could also check for multiple of 4 dimensions but that is not needed
Fixes: Timeout
Fixes: 11191/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MSVIDEO1_fuzzer-5739529588178944
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Frames that small are not valid and of limited use for error concealment, while
being very computationally intensive to process.
Fixes: Timeout
Fixes: 11168/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WMV2_fuzzer-5733782032744448
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Frames that small are not valid and of limited use for error concealment, while
being very computationally intensive to process.
Fixes: Timeout
Fixes: 11318/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MSMPEG4V1_fuzzer-5710884555456512
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 ensures that all (potentially, long) filesystem activity is
performed when the user calls av_write_trailer on the DASH libavformat
context, not when freeing the context. Also, this defers media segment
deletion until after the media trailers are written.
Passes fate on LE (with "lavc/jrevdct: Avoid an aliasing violation" applied).
Signed-off-by: Lauri Kasanen <cand@gmx.com>
Tested-by: Michael Kostylev on BE
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This commit ensures that all (potentially, long) filesystem activity is
performed when the user calls av_write_trailer on the DASH libavformat
context, not when freeing the context. Also, this defers media segment
deletion until after the media trailers are written.
When dashenc has to run for long duration(say 24x7 live stream), one can enable this option to ignore the io failure of few segment's upload due to an intermittent network issues.
When the network connection recovers dashenc will continue with the upload of the current segments, leading to the recovery of the stream.
The only native HLS implementation in the wild (Safari browser) doesn't
support WebM. And at least some MSE-based players (e.g. shaka-player)
cannot handle WebM media segments when playing HLS. So just skip non-mp4
streams from HLS manifests. Note that such streams will still be described
by the DASH manifest and therefore consumed by players supporting DASH.
Now the kvazaar warpper didn't setting the threads for kvazaar API,
and kavzaar will auto selecte the thread number.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Previously if only -y was used then nothing happened, if only -x then zero
window height was set which is undefined in SDL and caused a black window.
From now on if only one dimension is set we calculate the other based on
size/aspect of the shown content.
Signed-off-by: Marton Balint <cus@passwd.hu>
Remove invalid operation in the case x and y all equal 0,
this refine made about 2% speedup for H264 decode on loongson platform.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Assert in function ff_qsv_encode to avoid using uninitialized value:
FF_DISABLE_DEPRECATION_WARNINGS
avctx->coded_frame->pict_type = pict_type;
FF_ENABLE_DEPRECATION_WARNINGS
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
This option can be used to repect original input I/IDR frame type.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: Zhong Li <zhong.li@intel.com>
fixed the error in Intel(R) Xeon(R) Gold 6152 CPU like:
[libaom-av1 @ 0x469f340] Failed to initialize encoder: Invalid parameter
[libaom-av1 @ 0x469f340] Additional information: g_threads out of range [..MAX_NUM_THREADS]
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
This is in coherence with dashenc, which can now generate segments with
webm file name extension by default. Dashdec should be able to handle
such streams by default as well.
The alloc_size attribute is valid only on functions that return a
pointer. GCC 9 (not yet released) warns about invalid usage:
./libavutil/mem.h:342:1: warning: 'alloc_size' attribute ignored on a function returning int' [-Wattributes]
342 | av_alloc_size(2, 3) int av_reallocp_array(void *ptr, size_t nmemb, size_t size);
| ^~~~~~~~~~~~~
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
./ffmpeg_g -f rawvideo -pix_fmt rgb24 -s hd1080 -i /dev/zero -pix_fmt yuv420p \
-f null -vframes 100 -v error -nostats -
1158 UNITS in planar1, 65528 runs, 8 skips
-cpuflags 0
19082 UNITS in planar1, 65533 runs, 3 skips
16.48 speedup ratio. On x86, SSE2 is ~7. Curiously, the Power C version
takes as many cycles as the x86 SSE2 version, yikes it's fast.
Note that this function uses VSX instructions, but is not marked so.
This is because several existing functions also make that mistake.
I'll submit a patch moving them once this is reviewed.
Signed-off-by: Lauri Kasanen <cand@gmx.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The color fields were moved to another struct, and a way to propagate
timestamps and other input metadata was introduced, so the packet fifo
can be removed.
Signed-off-by: James Almer <jamrial@gmail.com>
When stream time bases are very fine grained (e.g. nanoseconds), 32-bit
segment duration may overflow for even for rather small segment duration
(about 4 seconds long). Therefore we use 64-bit values for segment duration.
Now memcpy is avoided for NAL units that don't contain 0x03 escape
characters.
Improves performance of cbs_h2645_fragment_add_nals from 36940
decicycles to 6364 decicycles based on 8 runs with a 5.1 Mb/s H.264
sample (262144 runs each).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Mark Thompson <sw@jkqxz.net>
The file name template options now support a new "$ext$" placeholder,
which is replaced with a filename extension specific for the selected
file format. This is useful for the new "auto" format mode, when
different streams may use different file formats, and it is not
possible to specify the correct file name extension exactly.
Resolves warnings in the log about webm segments not having webm extensions.
This commit restores the ability to create DASH streams with codecs
that require different containers that was lost after commit
2efdbf7367. It adds a new "auto" value for
the dash_segment_type option and makes it the default. When in this mode,
the segment format will be chosen based on the codec used in the stream:
webm for Vorbis, Opus, VP8 or VP9, mp4 otherwise.
This commit adds configuration options to libvpxenc.c that can be used to
enable VP8 temporal scalability. It also adds a way to programmatically set the
per-frame encoding flags which can be used to control usage and updates of
reference frames while encoding with temporal scalability enabled.
Signed-off-by: James Zern <jzern@google.com>
The current wording regarding size and min_size is completely wrong and
ignores that min_size is indeed only a desired minimal size, not the
actually allocated size.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The SAR of the input could have a numerator or denominator greater than
2^16 which would then be truncated to a 16-bit integer when written to
the VUI parameters, giving a random result. Instead, reduce the SAR to
the nearest representable fraction.
Fixes#7502.
Now the fragment is uninitialized and the input packet freed in case of
errors.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Mark Thompson <sw@jkqxz.net>
The ISOBMFF and Matroska specs allow the last OBU in a Sample/Block to have
obu_has_size_field equal to 0.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
These may be used by hwaccel decoders when the standard tables are not
otherwise available. At the same time, clean up that code into an array
so it's a little less repetitive.
Supports both prefix and suffix SEI, decoding all of the common SEI
types and some more obscure ones. Most of this is tested by the
existing tests in fate.
Fixes the following warning:
libavutil/avsscanf.c: In function 'decfloat':
libavutil/avsscanf.c:354:9: warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement]
int bitlim = bits-3*(int)(rp-9);
^~~
The header guards were unnecessarily non-standard and the c file
inclusion trick means the files dont't have standard licence
headers.
Based on a patch by: Martin Vignali <martin.vignali@gmail.com>
Fixes the following warnings:
libavformat/hlsenc.c: In function 'hls_write_trailer':
libavformat/hlsenc.c:2364:17: warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement]
uint8_t *buffer = NULL;
^~~~~~~
libavformat/hlsenc.c:2372:17: warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement]
int byterange_mode = (hls->flags & HLS_SINGLE_FILE) || (hls->max_seg_size > 0);
^~~
libavformat/hlsenc.c:2379:13: warning: ISO C90 forbids mixed declarations and code [-Wdeclaration-after-statement]
int range_length = 0;
^~~
method 0 (inflate/deflate) is the only specified in the specification and the only supported
Fixes: Timeout
Fixes: 10976/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PNG_fuzzer-5729372588736512
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The bwdif filter can use common yadif frame management if we track
when a field is the first or last field in a sequence. While this
information is not used by yadif, the added benefit of removing the
duplicated frame management logic makes it worth tracking this state
in the common code.
We have a pattern of wrapping CUDA calls to print errors and
normalise return values that is used in a couple of places. To
avoid duplication and increase consistency, let's put the wrapper
implementation in a shared place and use it everywhere.
Affects:
* avcodec/cuviddec
* avcodec/nvdec
* avcodec/nvenc
* avfilter/vf_scale_cuda
* avfilter/vf_scale_npp
* avfilter/vf_thumbnail_cuda
* avfilter/vf_transpose_npp
* avfilter/vf_yadif_cuda
Internally in ISOBMFF the FLAC-in-ISOBMFF draft uses "fLaC"
as the identifier for FLACSampleEntry, and there seems to be no
MPEG-DASH specification for the in-manifest identifier for FLAC.
After testing the browsers' implementations, it seems like all of
the major browser vendors have decided to utilize the MIME type for
FLAC ("audio/flac") as the identifier. This change set leads to
that string being utilized for FLAC streams instead of the sample
entry identifier ("fLaC"), which is the default behavior.
Verified by auri_ on IRC to play with the major browsers.
Fixes: signed integer overflow: -1625810908 - 582229060 cannot be represented in type 'int'
Fixes: 10977/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SHORTEN_fuzzer-5732602018267136
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Using the value of buffering frames inferred from the old level is
not a sensible approach when one wants to guess the level.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Mark Thompson <sw@jkqxz.net>
Add VDENC(lowpower mode) support for QSV h264 and HEVC
It's an experimental function(like lowpower in vaapi) with
some limitations:
- CBR/VBR require HuC which should be explicitly loaded via i915
module parameter(i915.enable_guc=2 for linux kerner version >= 4.16)
- HEVC VDENC was supported >= ICE LAKE
use option "-low_power 1" to enable VDENC.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
videotoolbox returns an already cropped stream which led to double
cropping. this issue was introduced with the refactor of the cropping
mechanism in commit 07596e4 for h264 and 000fb61 for HEVC.
to fix this we set the cropping of the frame and the output frame to 0.
Tested-by: ponpon
Fixes ticket #7544.
fix ticket: 7527
check dirname before use it
refine webvtt code in the hls_delete_old_segments
Reported-by: caspy
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
Instead of using a combination of bitreader and -writer for copying data,
one can byte-align the (obsolete and removed) bitreader to improve performance.
With the right alignment one can even use memcpy. The right alignment
normally exists for CABAC and hence for H.265 in general.
For aligned data this reduced the time to copy the slicedata from
776520 decicycles to 33889 with 262144 runs and a 6.5mb/s H.264 video.
For unaligned data the number went down from 279196 to 97739 decicycles.
Signed-off-by: Mark Thompson <sw@jkqxz.net>
Also add SIMD which works on lines because it is faster then calculating it on
8x8 blocks using pixelutils.
Signed-off-by: Marton Balint <cus@passwd.hu>
Instead of using a combination of bitreader and -writer for copying data,
one can byte-align the (obsolete and removed) bitreader to improve performance.
One can even use memcpy in the normal case.
This improved the time needed for writing the slicedata from 33618 to
2370 decicycles when tested on a video originating from a DVD (4194394
runs).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Signed-off-by: Mark Thompson <sw@jkqxz.net>
After inspecting the source code of x265, mpv and ffmpeg, I've found that
ffmpeg mistakenly regards EVC_NAL_BLA_N_LP and HEVC_NAL_IDR_N_LP as non-
reference frames, which are acutally reference frames according to the
specification in x265, and drops them.
This patch should address the problem. I have tested it with mpv.
Signed-off-by: Mark Wu <wfwf1997@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
only 16b alpha is supported (not 8 bits)
following official encoder, alpha data doesn't impact
yuv plane quality.
So the alpha data encoding is done after the yuv part.
It's also avoid to loose quality in yuv part when
alpha is not uniform.
the alpha encoding funcs is mainly take from prores_ks
encoder, except for the alpha data reorganization
The earlier code used the most recent non-auxiliary slice to determine
whether an auxiliary slice has the syntax of an IDR slice, even when
the most recent slice was from a slice of a redundant frame. Now only
slices of the primary coded picture are used, as the specifications
mandate.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@googlemail.com>
Flush the buffered data in libmfx before video param reinit
in case the frames drop.
Cache the first frame causing the reinit and decode zero-size
pkt to flush the buffered pkt before reinit. After all the
buffered pkts being flushed, resume to reinit and decode.
Fix the issue in ticket #7399.
[V2]: Move the definition of zero_pkt to where it is exactly
used.
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Zhong Li <zhong.li@intel.com>
libx264 does have a field for opaque data to pass along with frames
through the encoder, but it is a pointer, while the libavcodec
reordered_opaque field is an int64_t. Therefore, allocate an array
within the libx264 wrapper, where reordered_opaque values in flight
are stored, and pass a pointer to this array to libx264.
Update the public libavcodec documentation for the AVCodecContext
field to explain this usage, and add a codec capability that allows
detecting whether an encoder handles this field.
Signed-off-by: Martin Storsjö <martin@martin.st>
This was marked as deprecated (but only in the doxygen, not with an
actual deprecation attribute) in 81c623fae0 in 2011, but was
undeprecated in ad1ee5fa7.
Signed-off-by: Martin Storsjö <martin@martin.st>
libx264 does have a field for opaque data to pass along with frames
through the encoder, but it is a pointer, while the libavcodec
reordered_opaque field is an int64_t. Therefore, allocate an array
within the libx264 wrapper, where reordered_opaque values in flight
are stored, and pass a pointer to this array to libx264.
Update the public libavcodec documentation for the AVCodecContext
field to explain this usage, and add a codec capability that allows
detecting whether an encoder handles this field.
Signed-off-by: Martin Storsjö <martin@martin.st>
This was marked as deprecated (but only in the doxygen, not with an
actual deprecation attribute) in 81c623fae0 in 2011, but was
undeprecated in ad1ee5fa7.
Signed-off-by: Martin Storsjö <martin@martin.st>
buffer_delay_length_minus_1 is five bits long, meaning decode_buffer_delay and
encoder_buffer_delay can have values up to 32 bits long.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
partitioned_frame is also set/cleared in decode_vop_header()
Fixes: out of array read
Fixes: 9789/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-5638681627983872
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 a cuda implementation of yadif, which gives us a way to
do deinterlacing when using the nvdec hwaccel. In that scenario
we don't have access to the nvidia deinterlacer.
I'm writing a cuda implementation of yadif, and while this
obviously has a very different implementation of the actual
filtering, all the frame management is unchanged. To avoid
duplicating that logic, let's make it shareable.
From the perspective of the existing filter, the only real change
is introducing a function pointer for the filter() function so it
can be specified for the specific filter.
For HEVC streams, only the FourCC tag is written without profile, level etc.,
This is breaking playout support in native Safari.
Native Safari playout expects the full info in CODECS tag or None at all.
The driver bugs that caused decoded HEVC content to have an incorrect
memory layout have been fully fixed in the 410.xx driver release so
we can start exposing support.
This fixes the grammar of two HLS option descriptions and makes them less
ambiguous.
Signed-off-by: Werner Robitza <werner.robitza@gmail.com>
Signed-off-by: Lou Logan <lou@lrcd.com>
a thread count of 0 is treated the same as 1, use av_cpu_count() to get
the correct thread count when auto threads is requested.
this matches the fix in libvpxenc:
27df34bf1f avcodec/libvpxenc: fix setting amount of threads used for encoding
Reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: James Zern <jzern@google.com>
-1 will be map to error number "EPERM", and will be map to the error
message like "Error while decoding stream #0:0: Operation not permitted",
it's a strange error message when debug update_frame_pool fail,
now only return the error code from av_image_fill_pointers in case
of av_image_fill_pointers failure.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
In fmp4 & sub-range mode, the output's duration always smaller than expected,
because the size of the last #EXT-X-BYTERANGE is too small.
Signed-off-by: Charles Liu <liuchh83@gmail.com>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
The size of init.mp4 is zero in fmp4 mode,
when the input duraton smaller than the expected segment time.
fix ticket: 7166
Signed-off-by: Charles Liu <liuchh83@gmail.com>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
In fmp4 mode, the duration of the second m4s segment is
unusually smaller than the expected segment time.
Signed-off-by: Charles Liu <liuchh83@gmail.com>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
Adds an option to specify the number of tile rows and columns, then uses
a uniform tiling if possible and otherwise a fixed tiling with equal-sized
tiles to fill the frame.
Also adds -tile-columns and -tile-rows options to make tilings with
power-of-two numbers of tiles, matching the behaviour of the libvpx/VP9
encoder.
This is needed because of 32bit float formats (which are difficult to
store in 16bits)
This also fixes undefined behavior found by fate
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
found_ref is not a single value in the bitstream. Fixes parsing files with
frame size changes.
Based on code from cbs_vp9.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
This adds common code to query driver support and set appropriate
address/size information for each slice. It only supports rectangular
slices for now, since that is the most common use-case.
This reverts commit 662558f985.
The avcodec_parameters_to_context() call was freeing and reallocating
AVCodecContext->extradata, essentially taking ownership of it, which according
to the doxy is user owned. This is an API break and has produces crashes in
some library users like Firefox.
Revert until a better solution is found to internally propagate the filtered
extradata back into the decoder context.
Signed-off-by: James Almer <jamrial@gmail.com>
This error isn't particularly helpful as checking for mixed IDR/non-IDR
NALUs would need to be done at a higher level to actually be accurate.
Removing the error allows an API user to send individual slice NALUs
(i.e. incomplete frames) so they can take advantage of slice
threading. The ticket which this error was added for (#4408) no
longer segfaults after removing this error (as the bug was likely
fixed more properly elsewhere).
Libx264 uses strtok which is not thread safe. Strtok is used in
x264_param_default_preset in param_apply_tune in x264/common/base.c.
Therefore the flag must be removed.
x264 fixed the issue, once the fix is pushed to stable, an #if can be added
to re-enable the flag based on X264_BUILD number.
Fixes ticket #7446.
Signed-off-by: Marton Balint <cus@passwd.hu>
This reverts commit f631c328e6.
The avcodec_parameters_to_context() call was freeing and reallocating
AVCodecContext->extradata, essentially taking ownership of it, which according
to the doxy is user owned. This is an API break and has produced crashes in
some library users like Firefox[1].
Revert until a better solution is found to internally propagate the filtered
extradata back into the decoder context, or a decision is made to change the
API.
[1] https://bugzilla.mozilla.org/show_bug.cgi?id=1486080
Signed-off-by: James Almer <jamrial@gmail.com>
Several SRT options are missing. Since pkg_config requires libsrt v1.3.0 and above, it should be able to support options added in libsrt v1.3.0 and below.
This commit adds 8 SRT options.
sndbuf, rcvbuf, lossmaxttl, minversion, streamid, smoother, messageapi and transtype
The keys of option are equivalent to stransmit.
https://github.com/Haivision/srt/blob/v1.3.0/apps/socketoptions.hpp#L196-L223
Signed-off-by: Marton Balint <cus@passwd.hu>
Allows arrangement of multiple windows such as:
ffmpeg -re -f lavfi -i mandelbrot -f sdl -window_x 1 -window_y 1 mandelbrot -vf waveform,format=yuv420p -f sdl -window_x 641 -window_y 1 waveform -vf vectorscope,format=yuv420p -f sdl -window_x 1 -window_y 481 vectorscop
Some changes by Marton Balint:
- allow negative position (partially or fully out-of-screen positions seem to
be sanitized automatically by SDL (or my WM?), so no special handling is
needed)
- only show window after the position is set
- do not use resizable and borderless flags at the same time, that caused
issues in ffplay
- add docs
Signed-off-by: Marton Balint <cus@passwd.hu>
Create SMPTE ST 12-1 timecodes based on H.264 SEI picture timing
info.
For framerates > 30 FPS, the field flag is used in conjunction with
pairs of frames which contain the same frame timestamp in S12M.
Ensure the field is properly set per the spec.
Currently qsv (m)jpeg encoding is broken.
Regression introducing by the commit(id: c1bcd3): fix async support,
which requires the minimum async_depth to be 1, instead previous zero.
But the default async_depth of qsv (m)jpeg encoding is still initialized
(mostly) as zero.
This patch also abviously improves qsv (m)jpeg encoding performance
due to the default async_depth is changed to 4.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
64c50c0e97 declared support for decomposing
them but omitted to implement it; this adds an implementation.
Also do the same for end-of-stream NAL units, since they are equivalent.
case 1:
use the hexdump -C SMM0005.rcv get:
size skip (size - 4)
| |
V V
00000000 18 00 00 c5 05 00 00 00 4d f1 0a 11 00 e0 01 00
00000010 00 d0 02 00 00 0c 00 00 00 88 13 00 00 c0 65 52
^
|
size + 16
case 2:
same the command for SMM0015.rcv get:
size
|
V
00000000 19 00 00 c5 04 00 00 00 41 f3 80 01 40 02 00 00
00000010 d0 02 00 00 0c 00 00 00 00 00 00 10 00 00 00 00
^
|
size + 16
There are different the RCV file format for VC-1, vc1test
just handle the case 2 now, this fix will support the case 1.
(Both of test clips come from: SMPTE Recommended Practice -
VC-1 Decoder and Bitstream Conformance). And I think I got
a older VC-1 test clip in the case 1.
Reviewed-by: Carl Eugen Hoyos <ceffmpeg@gmail.com>
Reviewed-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Jun Zhao <jun.zhao@intel.com>
Signed-off-by: Yan, FengX <fengx.yan@intel.com>
Fixes bug with HTTP DELETE when HTTP Persistent is ON.
Right now, HTTP Persistent connections is supported only for POSTs and PUTs.
HTTP DELETE will still open a new connection every time.
Make the function static, or else Clang complains with:
error: no previous prototype for function 'decklink_get_attr_string' [-Werror,-Wmissing-prototypes]
Signed-off-by: Devin Heitmueller <dheitmueller@ltnglobal.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes: 10651/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ILBC_fuzzer-5202341540659200
Fixes: signed integer overflow: -1707705920 - 1703592888 cannot be represented in type 'int'
This tries to follow the webrtc code. For example using cliping and 64 bit as in WebRtcSpl_DotProductWithScale()
and not doing so in other places.
I could not find anything in rfc3951 and the reference code inside which would
explain what to do in these corner cases.
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
These limits are based on limiting done in WebRtcIlbcfix_CreateAugmentedVec()
Fixes: out of array accesses
Fixes: 10652/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ILBC_fuzzer-5638941487661056
Fixes: 10655/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ILBC_fuzzer-5699970020147200
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: -2144033225 + -5208934 cannot be represented in type 'int'
Fixes: 10633/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_RA_144_fuzzer-5679133791617024
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: 7738135736989908991 - -7954308516317364223 cannot be represented in type 'long'
Fixes: find_stream_info_usan
Reported-by: Thomas Guilbert <tguilbert@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes trac issue #7215
Output for files created by xWMAEncode and various videogames is correct now.
1ch 32000hz files are still broken, would need fixes in WMA decoder.
Signed-off-by: bnnm <bananaman255@gmail.com>
Fixes crash noticed in the cbs_userdata patchset.
====ERROR: AddressSanitizer: heap-buffer-overflow on address 0x609000026c89 at pc 0x00010725d37b bp 0x7ffeea04e750 sp 0x7ffeea04e748
READ of size 4 at 0x609000026c89 thread T0
#0 0x10725d37a in ff_cbs_read_unsigned get_bits.h:274
#1 0x1072d2767 in ff_cbs_read_a53_user_data cbs_misc_syntax_template.c:119
#2 0x1078251a7 in h264_metadata_filter h264_metadata_bsf.c:595
#3 0x105c1321d in output_packet ffmpeg.c:853
0x609000026c89 is located 1 bytes to the right of 8-byte region [0x609000026c80,0x609000026c88)
allocated by thread T0 here:
#0 0x10aef08d7 in wrap_posix_memalign (libclang_rt.asan_osx_dynamic.dylib:x86_64h+0x578d7)
#1 0x10aca95e6 in av_malloc mem.c:87
#2 0x10ac545fe in av_buffer_allocz buffer.c:72
#3 0x107263b27 in cbs_h264_read_nal_unit cbs_h264_syntax_template.c:722
#4 0x10725b688 in cbs_read_fragment_content cbs.c:155
Signed-off-by: Aman Gupta <aman@tmm1.net>
Currently qsv (m)jpeg encoding is broken.
Regression introducing by the commit(id: c1bcd3): fix async support,
which requires the minimum async_depth to be 1, instead previous zero.
But the default async_depth of qsv (m)jpeg encoding is still initialized
(mostly) as zero.
This patch also abviously improves qsv (m)jpeg encoding performance
due to the default async_depth is changed to 4.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Prior to Xcode 9.3, the clang built-in assembler didn't support
altmacro, and gas-preprocessor was used for assembling for arm/darwin.
For thumb functions, gas-preprocessor took care of adding the .thumb_func
directives, but when now being able to assemble without gas-preprocessor,
we need to add these directives ourselves.
Signed-off-by: Martin Storsjö <martin@martin.st>
The existing av_mediacodec_release_buffer allows the user to render
or discard the Surface-backed frame. This new method allows the user
to control exactly when the frame will be rendered to its SurfaceView.
Available since Android API 21.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Such streams are invalid according to
4.5.2.1 Top level payloads for the audio object types AAC main, AAC SSR, AAC LC and AAC LTP
4.5.2.1.1 Definitions
...cIn the raw_data_block(), several instances of the
same syntactic element may occur, but must have a different 4 bit
element_instance_tag, except for data_stream_element()'s and
fill_element()'s.
Fixes: Ticket7477
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Tool mediastreamvalidator reports error "Variant media_[N].m3u8 is
missing audio group" for audio streams in HLS master playlist. As audio
streams are already listed in audio group, skip them as variant media
streams in master playlist.
Libmfx requires 16 bytes aligned input/output for uploading.
Currently only output is 16 byte aligned and assigning same width/height to
input with smaller buffer size actually, thus definitely will cause segment fault.
Can reproduce with any 1080p nv12 rawvideo input:
ffmpeg -init_hw_device qsv=qsv:hw -hwaccel qsv -filter_hw_device qsv -f rawvideo -pix_fmt nv12 -s:v 1920x1080
-i 1080p_nv12.yuv -vf 'format=nv12,hwupload=extra_hw_frames=16,hwdownload,format=nv12' -an -y out_nv12.yuv
It can fix#7418
Signed-off-by: Zhong Li <zhong.li@intel.com>
RGB32(AV_PIX_FMT_BGRA on intel platforms) format may be used as overlay with alpha blending.
So add AV_PIX_FMT_BGRA format support.
One example of alpha blending overlay: ffmpeg -hwaccel qsv -c:v h264_qsv -i BA1_Sony_D.jsv
-filter_complex 'movie=lena-rgba.png,hwupload=extra_hw_frames=16[a];[0:v][a]overlay_qsv=x=10:y=10'
-c:v h264_qsv -y out.mp4
Rename RGB32 to be BGRA to make it clearer as Mark Thompson's suggestion.
V2: Add P010 format support else will introduce HEVC 10bit encoding regression.
Thanks for LinJie's discovery.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Verified-by: Fu, Linjie <linjie.fu@intel.com>
This allows switching between absolute (LUFS) and relativ (LU) display
in the status line.
Signed-off-by: Daniel Molkentin <daniel@molkentin.de>
Signed-off-by: Conrad Zelck <c.zelck@imail.de>
This eases meeting the target level during live mixing.
Signed-off-by: Daniel Molkentin <daniel@molkentin.de>
Signed-off-by: Conrad Zelck <c.zelck@imail.de>
Allow to show short-term instead of momentary in gauge. Useful for monitoring
whilst live mixing.
Signed-off-by: Daniel Molkentin <daniel@molkentin.de>
Signed-off-by: Conrad Zelck <c.zelck@imail.de>
If we don't copy this value first, it is seen as 0 by h264_slice_header_init,
due to zero-allocation of the new context, triggering an old hack that
multiplied the denominator by 2 for files produced by old x264 versions, but
only if more than one thread was used.
Fixes#7475 and #7083.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
This avoids surprising developers. Its bad to surprise developers with
such unexpected things.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
We check for the documented explanation of the "Ignore code" in extract_extradata_check() already
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This was the only case in the source that uses a hexadecimal shift value.
The change removed a special case in respect to greping
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Based on hevc_parser code. This prevents repeated unnecessary allocations
and frees on every packet processed by the bsf.
Reviewed-by: Jun Zhao <mypopydev@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Fixes ticket #7441.
for block contrast calculate, the block is like this:
|<---------------- stride-----------------------|
+-----------------------------------------------> X
|
| w = 16
| (cx,cy)+------+
| | |
|h=blocksize| |
| | |
| +------+
V
Y
so we calc the block contrast use:
(cy + y) * stride + (cx + x)
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
yae_set_tempo was overlooked when max tempo limit was raised to 100.
tested with:
./ffmpeg_g -i Delerium/SemanticSpaces/Gateway.mp3 \
-af asendcmd=f=asendcmd.cfg,atempo=1.0 -y /tmp/asendcmd-atempo.wav
where asendcmd.cfg was:
15.0-45.0 [enter] atempo tempo 2.0,
[leave] atempo tempo 0.5;
60.0-300.0 [enter] atempo tempo 4.0,
[leave] atempo tempo 1.0;
Simple parser to set keyframes, frame type, structure, width, height, and pixel
format, plus stream profile and level.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
That alone supports specifying the interface based on its address. Getting the
interface index from the local address seems quite a bit of work in a platform
independent way...
Obviously for IPv6 we still always use MCAST_JOIN_SOURCE_GROUP.
As a side effect this also fixes ticket #7459.
Signed-off-by: Marton Balint <cus@passwd.hu>
We already use localaddr for the multicast joins without source filters, so we
should use them for source filters as well. This patch only fixes the
IP_ADD_SOURCE_MEMBERSHIP and the IP_BLOCK_SOURCE case.
Unless we do this, the kernel automatically selects an interface based on the
source address, and that interface might be different from the one set in
localaddr. For blocked sources this even casues EINVAL because we joined the
multicast group on a different interface.
Signed-off-by: Marton Balint <cus@passwd.hu>
This allows getting data only from a specific source IP. This is useful not
only for unicast but for multicast as well because multicast source
subscriptions do not act as source filters for the incoming packets.
Signed-off-by: Marton Balint <cus@passwd.hu>
Variable 'ret' hasn't been initialized，thus introducing a random
hwupload failure regression due to qsv session uninitialized.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
This option is useful for maintaining input synchronization across N
different hardware devices deployed for 'N-way' redundancy.
The system time of different hardware devices should be synchronized
with protocols such as NTP or PTP, before using this option.
Signed-off-by: Marton Balint <cus@passwd.hu>
hevc parser mistakenly reports the following message if a dummy buffer
is padded for EOF
[hevc @ 0x559b63848610] missing picture in access unit
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Reviewed-by: "Li, Zhong" <zhong.li@intel.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Consider a component to be damaged if more than 50% of its subbands are damaged
Fixes: Timeout (part 1 of 2)
Fixes: 9774/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5748957085958144
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Divisions tend to be slower than shifts unless the compiler optimizes them out.
And some of these are in inner loops.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Because it will be used by avformat/segment.c or other module which
need to automatically create sub-directories operation.
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
The value in AVCodecParameters->codec_tag may not be correct for IVF,
as it's the case when remuxing AV1 streams from mp4, so ignore it and
write the correct value based on codec ID instead.
Reviewed-by: Jan Ekström <jeebjp@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
It seems what this function does is a vertical prediction filter, thus
the new name should improve understanding.
rename the related table_b too
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Also bump the API version requirement to 10.9.5, because on olders versions
there were some reports of crashes using the undocumented, yet available
BMDDeckLinkDeviceHandle.
Signed-off-by: Marton Balint <cus@passwd.hu>
If vaEndPicture() failed in ff_vaapi_decode_issue(), free
the pic->slice_buffers.
Fixes the memory leak issue in ticket #7385
Signed-off-by: Linjie Fu <linjie.fu@intel.com>
Signed-off-by: Mark Thompson <sw@jkqxz.net>
Sets the level based on the stream properties if it is not explicitly
set by the user. Also add a tier option to set general_tier_flag, since
that affects the level choice.
Set profile compatibility/constraint flags properly (including the
constraint flags used for RExt profiles, as all streams we can currently
generate are RExt-compatible), and use that to add support for the "Main
Intra" and "Main 10 Intra" RExt subprofiles (for which we can re-use the
existing Main and Main10 VAAPI profiles).
A recent version of the standard changed the max and default to 15, from
16 in older versions. This updates the default to 15 to match, but the
max stays as 16 so that we don't reject older streams.
Give the entries in the VAAPI format map table an explicit type and add
functions to do the necessary lookups. Add another field to this table
indicating whether the chroma planes are swapped (as in YV12), and use
that rather than explicit comparisons where swapping is needed.
Clarify that the list is the naughty list, and therefore being on it is
not desirable. The i965 driver does not need to be on the list after
version 2.0 (when the standard parameter buffer rendering behaviour was
changed).
constraint_set1_flag should be set for constrained baseline and main
profiles, because the stream conforms to main profile.
constraint_set3_flag should be set for high profile when the stream
is intra-only.
constraint_set4_flag should always be set for main and high profiles
because interlaced encoding is not supported.
constraint_set5_flag should be set for main and high profiles when
B-frames are not used.
Also fix the setting of max_num_ref_frames - use the gop_size value
to check for intra-only rather than the constraint flag (which is not
necessarily set).
Add a larger warning more clearly explaining the consequences of missing
packed header support in the driver. Also only write the extradata if the
user actually requests it via the GLOBAL_HEADER flag.
Choose what types of reference frames will be used based on what types
are available, and make the intra-only mode explicit (GOP size one,
which must be used for MJPEG).
This was added in libva 2.1.0 (VAAPI 1.1.0). Use AVCodecContext.qmax,
matching the existing behaviour for qmin, and clean up the defaults so
that we only pass min/max when explicitly set.
Query which modes are supported and select between VBR and CBR based
on that - this removes all of the codec-specific rate control mode
selection code.
The codec sequence headers may contain fields which can overwrite the
fine parameters given in the specific settings (e.g. a crude bitrate
value vs. the max-rate / target-percentage / etc. values in
VAEncMiscParameterRateControl). Always reapply all global parameters
after a sequence header to avoid this causing problems.
Previously there was one fixed choice for each codec (e.g. H.265 -> Main
profile), and using anything else then required an explicit option from
the user. This changes to selecting the profile based on the input format
and the set of profiles actually supported by the driver (e.g. P010 input
will choose Main 10 profile for H.265 if the driver supports it).
The entrypoint and render target format are also chosen dynamically in the
same way, removing those explicit selections from the per-codec code.
Set the minimum version to 0.35.0 (libva 1.3.0) and remove redundant
configure tests. This also allows the proprietary libmfx fork of libva,
which always shows the version number 0.99.0 (independent of the actual
version).
Nothing prevents it to work except this check. AV1 is already supported
by Matroska muxer and aomenc produces WebM/AV1 files as well.
Signed-off-by: Kagami Hiiragi <kagami@genshiken.org>
Signed-off-by: James Almer <jamrial@gmail.com>
SIDX atom being inserted for every MOOF atom increases the muxing overhead.
This behaviour can be disabled for chunked CMAF format by enabling Global SIDX option of mov muxer.
Fixes: out of array read
Fixes: 10064/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-5766801384800256
Fixes: 10225/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-5071833448054784
Fixes: 10261/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-5115048024866816
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This optimization improved h264 decoding performance about 4%(from 74fps to 77fps, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This checks the value exactly for intra frames and checks it against a
minimum for inter frames as they can be variable.
Fixes: Timeout
Fixes: 10182/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_ZMBV_fuzzer-6245951174344704
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
decomp_len is used in raw frames, so it should not be left at the value from
whatever was decoded previously (which may be any other frame)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Combined 1st and 2nd loop into one inline asm in function ff_vc1_inv_trans_8x8_mmi to
reduce memory operation, and made some small optimization in ff_vc1_inv_trans_4x8_mmi.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes high memory usage and prevents over allocation of the frames via
proper unref.
Can be checked as:
-hwaccel qsv -c:v h264_qsv -i ../h264-conformance/CANL2_Sony_E.jsv -c:v
h264_qsv -b:v 2000k -y qsv.mp4
For example bitdepth should be printed as 10 instead of 0A. Thanks to Hendrik Leppkes for pointing this out
Signed-off-by: James Almer <jamrial@gmail.com>
Failed case: svq3-watermark
When minimum loop count of following functions are greater than parameter h passed to them, svq3-watermark failed.
1. ff_put_pixels4_8_mmi
2. ff_avg_pixels4_8_mmi
3. ff_put_pixels4_l2_8_mmi
4. ff_avg_pixels4_l2_8_mmi
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes ticket #5654.
The linesize can be greater than the minimum required. This copies the
frame taking linesize into account.
Signed-off-by: Rick Kern <kernrj@gmail.com>
MSVC expands the preprocessor directives differently, making the
version check fail in the previous form.
Clang can warn about this with -Wexpansion-to-defined (not currently
enabled by default):
warning: macro expansion producing 'defined' has undefined behavior [-Wexpansion-to-defined]
Signed-off-by: Martin Storsjö <martin@martin.st>
The libaom doxy says that a value of 0 for the threads fields is
equivalent to a value of 1, whereas for avctx->thread_count it means
the maximum amount of threads possible for the host system.
Use av_cpu_count() to get the correct thread count when auto threads
is requested.
Reviewed-by: Jan Ekström <jeebjp@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
* commit '2edaafe5b93832715781851dfe2663da228a05ad':
libfdk-aacdec: Allow setting the new dynamic range control effect setting
Merged-by: James Almer <jamrial@gmail.com>
* commit 'ffb9b7a6bab6c6bfd3dd9a7c32e3724209824999':
libfdk-aac: Consistently use a proper version check macro for detecting features
Merged-by: James Almer <jamrial@gmail.com>
* commit '83678dbbae64ad8c501e0c732c1117e642c25dae':
libopenh264dec: Export the decoded profile and level in AVCodecContext
Merged-by: James Almer <jamrial@gmail.com>
* commit '8c76bfacf663ff71cee5264a74d0f9c86addd325':
tcp: Use ff_connect_parallel for RFC 8305 style connecting
Merged-by: James Almer <jamrial@gmail.com>
* commit '69caad8959982580504643d36aef22528e4aa6ce':
qsvdec: Release packet on decoding failure for mpeg2/vp8/vc1
Merged-by: James Almer <jamrial@gmail.com>
* commit 'e05e5920a4e1f1f15cc8a7c843159d519f6ec18e':
qsv: Error out if getting session handle failed in avfilter
Merged-by: James Almer <jamrial@gmail.com>
* commit '662558f985f50834eebe82d6b6854c66f33ab320':
decode: copy the output parameters from the last bsf in the chain back to the AVCodecContext
decode: flush the internal bsfs instead of constantly reinitalizing them
h264_redundant_pps_bsf: implement a AVBSFContext.flush() callback
vp9_superframe_bsf: implement a AVBSFContext.flush() callback
vp9_superframe_split_bsf: implement a AVBSFContext.flush() callback
h264_mp4toannexb_bsf: implement a AVBSFContext.flush() callback
bsf: add a flushing mechanism to AVBSFContext
This commit is a noop, see
b33f5299a5390f15645163e0846c66e9980c451e2954e5139394fe138de0f631c328e6
Merged-by: James Almer <jamrial@gmail.com>
* commit '6a9c00c09d2bc50c0ea64ba092b2f4afc46aa978':
tls_openssl: Fix checks for SSL_ERROR_WANT_WRITE in nonblocking operation
Merged-by: James Almer <jamrial@gmail.com>
* commit 'c194b9ad6dbe65f5abd68158c4811ed84e2a2b95':
network: Use ff_neterrno instead of AVERROR(errno) for poll errors
This commit is a noop, see 54b6bef6e1
Merged-by: James Almer <jamrial@gmail.com>
* commit '5d01bd181bb77e6740462095d7be4e0733a59420':
http: pass return code from http_open_cnx_internal() on its failure
This commit is a noop, see 70c9d40008
Merged-by: James Almer <jamrial@gmail.com>
* commit 'f89ec87afaf0d1abb6d450253b0b348fd554533b':
frame: Simplify the video allocation
Merged-by: James Almer <jamrial@gmail.com>
Padding-Remixed-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes bug id #7386
Muxer overhead calculations was intented for HLS playlist as Apple's mediastreamvalidator tests were failing.
But applying the same fix for DASH manifest proved counterproductive, as Bandwidth can be used for segment name templates.
Add missing dnxhr mxf container essence ULs to the mxf encoder.
This fixes dnxhr mxf files being quarantined by Avid Media Composer.
Signed-off-by: Jason Stevens <jay@wizardofthenet.com>
Reviewed-by: Baptiste Coudurier
refactor ff_dnxhd_get_hr_frame_size to avpriv_dnxhd_get_hr_frame_size,
to allow cross library usage in libavformat/mxfenc this change makes
this function no longer be always inlined.
Signed-off-by: Jason Stevens <jay@wizardofthenet.com>
When encoding to V210, make sure the AFD side data makes it through
in the resulting AVPacket. This is needed so the decklink output
module can put out AFD when in 10-bit mode.
Signed-off-by: Devin Heitmueller <dheitmueller@ltnglobal.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Create a new AVPacket side data type for Active Format Description,
which mirrors the side data type found in AVFrame. The primary
use case for this is ensuring AFD gets preserved in the V210
encoder, so that the decklink libavdevice can output AFD.
Signed-off-by: Devin Heitmueller <dheitmueller@ltnglobal.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Hook in libklvanc and use it for output of EIA-708 captions over
SDI. The bulk of this patch is just general support for ancillary
data for the Decklink SDI module - the real work for construction
of the EIA-708 CDP and VANC line construction is done by libklvanc.
Libklvanc can be found at: https://github.com/stoth68000/libklvanc
Updated to reflect feedback from Marton Balint <cus@passwd.hu>,
Carl Eugen Hoyos <ceffmpeg@gmail.com>, Aaron Levinson
<alevinsn_dev@levland.net>, and Moritz Barsnick <barsnick@gmx.net>
Signed-off-by: Devin Heitmueller <dheitmueller@ltnglobal.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
When encoding to V210, make sure the CC side data makes it through
in the resulting AVPacket. This is needed so the decklink output
module can put out captions when in 10-bit mode.
Signed-off-by: Devin Heitmueller <dheitmueller@ltnglobal.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
ISOBMFF does not allow AudioSampleEntryV1 in stsd version 0, so
assume the descriptor format is QTFF SoundDescriptionV1. ISOBMFF does
not define a version 2.
This fixes audio decoding for some MP4 files generated with Apple
tools. The additional fields present in SoundDescriptionV1/V2 need to
be read in order to correctly read additional boxes that contain
information required for decoding the stream.
Fixes#7376.
Also see: https://github.com/HandBrake/HandBrake/issues/1555
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Constraint "g" means compiler can store variable in memory or register.
When we use constraint "g" for a variable and this variable was operated by
instruction which only support register operands may lead "invalid operands" error.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
It refers to the uncompressed quantization, therefore is not correct for AAC.
Also change mxf_set_pts to work based on current edit unit if
bits_per_coded_sample is not available.
Fixes error messages in the sample of ticket #7366.
Signed-off-by: Marton Balint <cus@passwd.hu>
Simplify the usage of intermediate variable addr and remove unused variable all64
in following functions:
1. ff_put_pixels_clamped_mmi
2. ff_put_signed_pixels_clamped_mmi
3. ff_add_pixels_clamped_mmi
This optimization speed up mpeg4 decode about 2% on loongson platform(tested with 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The previous version checks checked explicitly for the version
where the version define was added to the installed headers,
making an "#ifdef AACDECODER_LIB_VL0" enough. Now that we have
a need for more diverse version checks than this, convert all checks
to such checks.
Signed-off-by: Martin Storsjö <martin@martin.st>
Simplify the usage of intermediate variable addr in following functions:
1. ff_put_pixels4_8_mmi
2. ff_put_pixels8_8_mmi
3. ff_put_pixels16_8_mmi
4. ff_avg_pixels16_8_mmi.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Removing unused VPP sessions by initializing only when used in order to help
reduce CPU utilization.
Thanks to Maxym for the guidance.
Signed-off-by: Joe Olivas <joseph.k.olivas@intel.com>
Signed-off-by: Maxym Dmytrychenko <maxim.d33@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Failed case: mss2-wmv
In following functions, pmullh was used to multiply two 16-bit data, this will cause data overflow.
1. ff_vc1_inv_trans_8x8_dc_mmi
2. ff_vc1_inv_trans_8x8_mmi
3. ff_vc1_inv_trans_8x4_mmi
4. ff_vc1_inv_trans_4x8_mmi
5. ff_vc1_inv_trans_4x4_mmi
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
When flushing the encoder, we now need to provide non-null buffer
parameters for everything, even if they are unused.
The encoderDelay parameter has been replaced by two, nDelay and
nDelayCore.
Signed-off-by: Martin Storsjö <martin@martin.st>
* commit 'c011beda2611acfeb6f67d4fdf30d1eceed9e62f':
avconv: make sure packets put into the muxing FIFO are refcounted
This commit is a noop, see 33580a8625
Merged-by: James Almer <jamrial@gmail.com>
* commit 'b93026777aada7742583d8c5ab079e9f4dfe9a5d':
libfdk-aac: Use enum names instead of literal numbers for the output format
Merged-by: James Almer <jamrial@gmail.com>
* commit '52fd2afce8436c59c05765f3a6e95f9adb6f9f2f':
configure: fix inline asm checks
This commit is a noop, see ad94f1c8ab
Merged-by: James Almer <jamrial@gmail.com>
* commit 'f8060865f3e1a16c62e0d337ef0979b6ee4ba457':
qsvenc: use the compression_level to replace private option
Merged-by: James Almer <jamrial@gmail.com>
This requires us to pre-parse the skip data, as we want to
detect this before allocating all the arrays
Fixes: Timeout
Fixes: 9708/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WMV2_fuzzer-5729709861109760
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
* commit 'ad5bbc408637cffd4cc2ba990abef529cf5fa6a3':
configure: Rename require_header() --> require_headers()
This commit is a noop, see ce47f1589e
Merged-by: James Almer <jamrial@gmail.com>
* commit '4130e05ff496667565ff7c386a514bd46434eddf':
libavformat: add mbedTLS based TLS
This commit is a noop, see c24d247e2c
Merged-by: James Almer <jamrial@gmail.com>
* commit '39f3b6f3fc2b46b405b680cce3599f1b370e342d':
configure: Move add_fooflags() helper functions into canonical order
Merged-by: James Almer <jamrial@gmail.com>
* commit '5691c746cf62e69806aae1baf0a6e8252d519444':
configure: Group toolchain parameter mangling functions together
Merged-by: James Almer <jamrial@gmail.com>
* commit '25c2a27c9ec0150210d75ee5ac8ed1bfa14c1a56':
configure: Make require_cc() and require_cpp_condition() functions consistent
Merged-by: James Almer <jamrial@gmail.com>
* commit '78149d6657302b58d5e46e8bc0a521ed009f86f7':
amfenc: Retain a reference to D3D frames used as input during the encoding process
This commit is a noop, see 05f1a3face
Merged-by: James Almer <jamrial@gmail.com>
* commit 'abf806f7f1601c7e54de7f863bbb816af144a88c':
random_seed: use bcrypt instead of the old wincrypt API
This commit is a noop, see aedbf1640c
Merged-by: James Almer <jamrial@gmail.com>
* commit '347aa8f72356124ec6b95bf8ebd1faf72db03f8d':
x86: Don't declare a non-static function as inline
This commit is a noop
Merged-by: James Almer <jamrial@gmail.com>
Optimized memset with mmi in following functions:
1. ff_h264_add_pixels4_8_mmi.
2. ff_h264_idct_add_8_mmi.
3. ff_h264_idct8_add_8_mmi.
This optimization improved h264 decoding performance about 1.3%(tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Reoptimize function ff_put_h264_chroma_mc8_mmi and ff_avg_h264_chroma_mc8_mmi.
Performance of h264 decoding improved about 5%(from 69fps to 73fps, tested on loongson 3A3000).
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Performance of mpeg4 decoding improved about 23%(from 128fps to 158fps, tested on loongson 3A3000).
Reoptimized following functions with mmi.
1. ff_simple_idct_put_8_mmi
2. ff_simple_idct_add_8_mmi
3. ff_simple_idct_8_mmi
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Also make sure we set the URL context max packet size accordingly.
Based on a patch by Tudor Suciu <tudor.suciu@gmail.com>
Signed-off-by: Marton Balint <cus@passwd.hu>
Entries are always at least 8 bytes per the parsing code, so if we
see an impossible entry count avoid massive allocations. This is
similar to an existing check in mov_read_stsc().
Since ff_mov_read_stsd_entries() does eof checks, an alternative
approach could be to clamp the entry count to atom.size / 8.
Signed-off-by: Dale Curtis <dalecurtis@chromium.org>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
AV_CODEC_FLAG_GLOBAL_HEADER should be set before calling avcodec_open2() to have any effect.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
For cases with dual stack (IPv4 + IPv6) connectivity, but where one
stack potentially is less reliable, strive to trying to connect over
both protocols in parallel, using whichever address connected first.
In cases with a hostname resolving to multiple IPv4 and IPv6
addresses, the current connection mechanism would try all addresses
in the order returned by getaddrinfo (with all IPv6 addresses ordered
before the IPv4 addresses normally). If connection attempts to the
IPv6 addresses return quickly with an error, this was no problem, but
if they were unsuccessful leading up to timeouts, the connection process
would have to wait for timeouts on all IPv6 target addresses before
attempting any IPv4 address.
Similar to what RFC 8305 suggests, reorder the list of addresses to
try connecting to, interleaving address families. After starting one
connection attempt, start another one in parallel after a small delay
(200 ms as suggested by the RFC).
For cases with unreliable IPv6 but reliable IPv4, this should make
connection attempts work as reliably as with plain IPv4, with only an
extra 200 ms of connection delay.
Signed-off-by: Martin Storsjö <martin@martin.st>
- Allow to add deps in any order rather than "in linking order".
- Expand deps chains as required rather than just once.
- Validate that there are no cycles.
- Validate that [after expansion] deps are limited to other fflibs.
- Remove expectation for a specific output order of unique().
Previously when adding items to <fflib>_deps, developers were
required to add them in linking order. This can be awkward and
bug-prone, especially when a list is not empty, e.g. when adding
conditional deps.
It also implicitly expected unique() to keep the last instance of
recurring items such that these lists maintain their linking order
after removing duplicate items.
This patch mainly allows to add deps in any order by keeping just
one master list in linking order, and then reordering all the
<fflib>_deps lists to align with the master list order.
This master list is LIBRARY_LIST itself, where otherwise its order
doesn't matter.
The patch also removes a limit where these deps lists were expanded
only once. This could have resulted in incomplete expanded lists,
or forcing devs to add already-deducable deps to avoid this issue.
Note: it is possible to deduce the master list order automatically
from the deps lists, but in this case it's probably not worth the
added complexity, even if minor. Maintaining one list should be OK.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes the following warnings:
libavcodec/v4l2_m2m_enc.c:51:12: warning: missing braces around initializer
libavcodec/v4l2_m2m_enc.c:71:12: warning: missing braces around initializer
PTS is in microseconds, so correct field name is out_time_us.
Old field out_time_ms kept for now - will be removed after a suitable transition
period.
Fixes#7345
Just remove some dead variable assignments, unneeded variables and
change the FFMAX order to something more readable. Still identical.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Much simpler than regular decoding, does allow for 5.1 and 7.1
streams to be decoded without desync.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Unlike the range, the gradient start value does not have to be lower
than the end value.
Does allow more files to be correctly decoded without errors.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
x4 - x25 faster.
check_deps() recursively enables/disables components, and its loop is
iterated nearly 6000 times. It's particularly slow in bash - currently
consuming more than 50% of configure runtime, and about 20% with other
shells.
This commit applies few local optimizations, most effective first:
- Use $1 $2 ... instead of pushvar/popvar, and same at enable_deep*
- Abort early in one notable case - empty deps, to avoid costly no-op.
- Smaller changes which do add up:
- Handle ${cfg}_checking locally instead of via enable[d]/disable
- ${cfg}_checking: test done before inprogress - x2 faster in 50%+
- one eval instead of several at the empty-deps early abort path.
- The "actual work" part is unmodified - just its surroundings.
Biggest speedups (relative and absolute) are observed with bash.
Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Tested-by: Helmut K. C. Tessarek <tessarek@evermeet.cx>
Tested-by: Dave Yeo <daveryeo@telus.net>
Tested-by: Reino Wijnsma <rwijnsma@xs4all.nl>
Signed-off-by: James Almer <jamrial@gmail.com>
x4 - x10 faster.
Inside print_enabled components, the filter_list case invokes sed
about 350 times to parse the same source file and extract different
info for each arg. This is never instant, and on systems where fork is
slow (notably MSYS2/Cygwin on windows) it takes many seconds.
Change it to use sed once on the source file and set env vars with the
parse results, then use these results inside the loop.
Additionally, the cases of indev_list and outdev_list are very
infrequent, but nevertheless they're faster, and arguably cleaner, with
shell parameter substitutions than with command substitutions.
Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Tested-by: Helmut K. C. Tessarek <tessarek@evermeet.cx>
Tested-by: Dave Yeo <daveryeo@telus.net>
Tested-by: Reino Wijnsma <rwijnsma@xs4all.nl>
Signed-off-by: James Almer <jamrial@gmail.com>
x50 - x200 faster.
Currently configure spends 50-70% of its runtime inside a single
function: flatten_extralibs[_wrapper] - which does string processing.
During its run, nearly 20K command substitutions (subshells) are used,
including its callees unique() and resolve(), which is the reason
for its lengthy run.
This commit avoids all subshells during its execution, speeding it up
by about two orders of magnitude, and reducing the overall configure
runtime by 50-70% .
resolve() is rewritten to avoid subshells, and in unique() and
flatten_extralibs() we "inline" the filter[_out] functionality.
Note that logically, "unique" functionality has more than one possible
output (depending on which of the recurring items is kept). As it
turns out, other parts expect the last recurring item to be kept
(which was the original behavior of uniqie()). This patch preservs
its output order.
Tested-by: Michael Niedermayer <michael@niedermayer.cc>
Tested-by: Helmut K. C. Tessarek <tessarek@evermeet.cx>
Tested-by: Dave Yeo <daveryeo@telus.net>
Tested-by: Reino Wijnsma <rwijnsma@xs4all.nl>
Signed-off-by: James Almer <jamrial@gmail.com>
Encoder frame_number may be double-counted if some frames are cached and then flushed.
Take qsv encoder (some frames are cached firsty for asynchronism) as example,
./ffmpeg -loglevel verbose -hwaccel qsv -c:v h264_qsv -i in.mp4 -vframes 100 -c:v h264_qsv out.mp4
frame_number passed to encoder is double-counted and larger than the accurate value.
Libx264 encoding with B frames can also reproduce it.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Fixes: signed integer overflow: -19818 + -2147483648 cannot be represented in type 'int'
Fixes: 9545/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SNOW_fuzzer-4928769537081344
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
ISMV lacks any sort of edit list support, as well as tfxd is
effectively the PTS of the fragment for most intents and purposes.
Thus, if b-frames are requested without negative CTS offsets you
end up with N frames' worth of delay (tfxd PTS plus the CTS offset
of the first sample). Negative CTS offsets enable the first sample
to have CTS=DTS, and thus a/v desync due to b-frame reorder delay
is avoided.
Since libopus 1.2, packets of sizes 80ms, 100ms and 120ms are allowed.
Fixes assertion failures when trying to mux such streams.
Signed-off-by: James Almer <jamrial@gmail.com>
Packets of sizes 80ms, 100ms and 120ms are allowed since libopus 1.2
Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
This reverts commit 7e0df5910e.
"complete frames" containers, even if they don't need to assemble
packets, still depended on this code for proper packet duration and
timestamp generation.
remove redundant av_init_packet after av_packet_unref.
av_packet_unref have call av_init_packet and reset the packet size.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
H264/265 have been fixed such an issue with commit
559370f2c4.
Similar fixing is needed for other codecs.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Solve some issues found by an automated code scansion.
Suppress the complain "variables 'handle' is used but maybe
uninitialized".
Signed-off-by: Zhong Li <zhong.li@intel.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
If there is a saio/saiz in clear content, we shouldn't create the
encryption index if we don't already have one. Otherwise it will
confuse the cenc_filter.
The changed method is also used for senc atoms, but they should not
appear in clear content.
Found by Chromium's ClusterFuzz: https://crbug.com/873432
Signed-off-by: Jacob Trimble <modmaker@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
fix the waring: libavcodec/libkvazaar.c:210:27: warning: passing argument 3 of ‘av_image_copy’ from incompatible pointer type [-Wincompatible-pointer-types]
frame->data, frame->linesize,
^~~~~
In file included from libavcodec/libkvazaar.c:31:0:
./libavutil/imgutils.h:119:6: note: expected ‘const uint8_t ** {aka const unsigned char **}’ but argument is of type ‘uint8_t * const* {aka unsigned char * const*}’
void av_image_copy(uint8_t *dst_data[4], int dst_linesizes[4],
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
fix the build warning for "ISO C90 forbids mixed declarations and code"
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Currently float are converted to 16b uint in input part
using src depth (32 bits) in hScale16To19 and hScale16to15,
make an invalid shift for the data
So shift the value when using float input
like 16 bpc uint.
Add fix a memory leak issue as James's comments.
V2: use a local pict_type since coded_frame is deprecated.
Signed-off-by: Zhong Li <zhong.li@intel.com>
The specification states "NSV files may contain a single file header. "
Fixes: out of array access
Fixes: nsv-asan-002f473f726a0dcbd3bd53e422c4fc40b3cf3421
Found-by: Paul Ch <paulcher@icloud.com>
Tested-by: Paul Ch <paulcher@icloud.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is a temporary workaround for transcoding scenarious using libaom-av1
encoder, which currently can't propagate extradata during initialization.
Signed-off-by: James Almer <jamrial@gmail.com>
This fixes the creation of the hls manifest in hlsenc.c by writing the
entire manifest at the end for VOD playlists. Live & Event Playlists are unaffected.
This also fixes the behavior with HLS_TEMP_FILE to work correctly when
-hlsflags temp_file is specified, instead of always relying on use_rename, which caused these problems.
Files that would previously take over a week to fragment now take
1 minute on the same hardware. This was a 153 hour audio file (2.2GB of audio).
Signed-off-by: Ronak Patel <ronak2121@yahoo.com>
Fixes: long running loop
Fixes: ivr-timeout-42468cb797f52f025fb329394702f5d4d64322d6
Found-by: Paul Ch <paulcher@icloud.com>
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This will get ISOBMFF and Matroska up to date with the revised AV1 Codec
Configuration Box spec.
For now keep propagating raw OBUs as extradata until all libavcodec modules
are adapted to handle AV1CodecConfigurationRecord formatted extradata.
Tested-by: Thomas Daede <bztdlinux@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Also only initialize it in ff_av1_packet_split() and not ff_av1_extract_obu(),
same as h2645_parse, so GetBitContext specific failures may not affect the
latter.
Signed-off-by: James Almer <jamrial@gmail.com>
Certain AVCodecParameters, like the contents of the extradata, may be changed
by the init() function of any of the bitstream filters in the chain.
Signed-off-by: James Almer <jamrial@gmail.com>
Initialize the bsfs once when opening the codec and uninitialize them once when
closing it, instead of at every codec flush/seek.
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes: runtime error: left shift of 1 by 31 places cannot be represented in type 'int'
Fixes: 9480/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SHORTEN_fuzzer-6647324284551168 -rss_limit_mb=2000
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: -540538872 + -2012739576 cannot be represented in type 'int'
Fixes: 9255/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SHORTEN_fuzzer-5758630052757504
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Previously, AVERROR(EIO) was returned on failure of
http_open_cnx_internal(). Now the value is passed to upper level, thus
it is possible to distinguish ECONNREFUSED, ETIMEDOUT, ENETUNREACH etc.
Signed-off-by: Martin Storsjö <martin@martin.st>
Fixes vorbis mp4 audio files, with edit list specified. Since
st->skip_samples is not set in case of vorbis , ffmpeg computes the
start_time as negative.
Signed-off-by: Sasi Inguva <isasi@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Neutrals are supposed to be anything not black (0,0,0) and not white
(N,N,N).
Previous neutral filtering code was too strict by excluding colors with
any of its RGB component maxed instead of just the white color.
Reported-by: Royi Avital <royiavital@yahoo.com>
Existing link is broken.
This patch updates the existing url with a working one.
Signed-off-by: Mina <minasamy_@hotmail.com>
Signed-off-by: Gyan Doshi <ffmpeg@gyani.pro>
Currently extra_hw_frames can't be applied to qsv since it
doesn't call function avcodec_get_hw_frames_parameters().
Give an option to fix ticket #7261 though it is not a perfect soultion
(allocate the minimum pool size internally and automatically).
Signed-off-by: Zhong Li <zhong.li@intel.com>
we need to make sure that memory allocation for Y/UV planes is continuous and re-used from a
pool
Signed-off-by: Maxym Dmytrychenko <maxim.d33@gmail.com>
When there is no metadata attached to a frame, take into account both
the PQ and HLG transfers, and change the HLG default value to 10:
the value of 12 is the maximum range in scene referred light, but
the reference OOTF maps this from 0 to 1000 cd/m² on the ideal HLG
monitor.
This matches what vf_tonemap_opencl does.
Alternatively the parser could be re implemented / redesigned so as to better
and more efficiently find frame boundaries
Fixes: Timeout
Fixes: 9210/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PBM_fuzzer-4770771833454592
Fixes: 9214/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PPM_fuzzer-5741633353023488
Fixes: 9219/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PGM_fuzzer-6249230237696000
Fixes: 9550/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_PAM_fuzzer-5312669836902400
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: simple_idct_template.c:184:30: runtime error: signed integer overflow: -1065517056 - 1392182838 cannot be represented in type 'int'
Fixes: simple_idct_template.c:269:21: runtime error: signed integer overflow: 16384 * 259254 cannot be represented in type 'int'
Fixes: simple_idct_template.c:164:17: runtime error: signed integer overflow: 21407 * 210162 cannot be represented in type 'int'
Fixes: simple_idct_template.c:167:17: runtime error: signed integer overflow: 21407 * 210162 cannot be represented in type 'int'
Fixes: simple_idct_template.c:169:19: runtime error: signed integer overflow: 22725 * 259190 cannot be represented in type 'int'
Fixes: simple_idct_template.c:171:19: runtime error: signed integer overflow: 19265 * 259190 cannot be represented in type 'int'
Fixes: simple_idct_template.c:173:19: runtime error: signed integer overflow: 12873 * 259190 cannot be represented in type 'int'
Fixes: simple_idct_template.c:183:28: runtime error: signed integer overflow: 1860878336 + 585177665 cannot be represented in type 'int'
Fixes: simple_idct_template.c:159:17: runtime error: signed integer overflow: 16384 * 189520 cannot be represented in type 'int'
Fixes: simple_idct_template.c:170:22: runtime error: signed integer overflow: 19265 * 130147 cannot be represented in type 'int'
Fixes: simple_idct_template.c:174:23: runtime error: signed integer overflow: -22725 * 130147 cannot be represented in type 'int'
Fixes: simple_idct_template.c:183:20: runtime error: signed integer overflow: 16384 * -175206 cannot be represented in type 'int'
Fixes: simple_idct_template.c:184:22: runtime error: signed integer overflow: -16384 * -175206 cannot be represented in type 'int'
Fixes: simple_idct_template.c:185:22: runtime error: signed integer overflow: -16384 * -175206 cannot be represented in type 'int'
Fixes: simple_idct_template.c:186:20: runtime error: signed integer overflow: 16384 * -175206 cannot be represented in type 'int'
Fixes: simple_idct_template.c:195:26: runtime error: signed integer overflow: 19265 * 150747 cannot be represented in type 'int'
Fixes: simple_idct_template.c:198:27: runtime error: signed integer overflow: -22725 * 150747 cannot be represented in type 'int'
Fixes: simple_idct_template.c:184:37: runtime error: signed integer overflow: 21407 * -171941 cannot be represented in type 'int'
Fixes: simple_idct_template.c:185:37: runtime error: signed integer overflow: 21407 * -171941 cannot be represented in type 'int'
Fixes: simple_idct_template.c:192:27: runtime error: signed integer overflow: -12873 * 206341 cannot be represented in type 'int'
Fixes: simple_idct_template.c:185:30: runtime error: signed integer overflow: 1196441600 + 1703756981 cannot be represented in type 'int'
Fixes: simple_idct_template.c:176:23: runtime error: signed integer overflow: -12873 * 168461 cannot be represented in type 'int'
Fixes: simple_idct_template.c:191:27: runtime error: signed integer overflow: -22725 * -140062 cannot be represented in type 'int'
Fixes: simple_idct_template.c:197:26: runtime error: signed integer overflow: 19265 * -140062 cannot be represented in type 'int'
Fixes: simple_idct_template.c:183:34: runtime error: signed integer overflow: 8867 * -243046 cannot be represented in type 'int'
Fixes: simple_idct_template.c:186:34: runtime error: signed integer overflow: 8867 * -243046 cannot be represented in type 'int'
Fixes: simple_idct_template.c:186:28: runtime error: signed integer overflow: -816234496 - 2139878414 cannot be represented in type 'int'
Fixes: simple_idct_template.c:188:26: runtime error: signed integer overflow: 12873 * -239872 cannot be represented in type 'int'
Fixes: simple_idct_template.c:165:16: runtime error: signed integer overflow: 8867 * -260084 cannot be represented in type 'int'
Fixes: simple_idct_template.c:166:16: runtime error: signed integer overflow: 8867 * -260084 cannot be represented in type 'int'
Fixes: 9135/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-6324422955761664
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: -1813244069 + -1407981383 cannot be represented in type 'int'
Fixes: 8823/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AAC_FIXED_fuzzer-5643295618236416
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This code came originally from gstreamer, where it was added in [1]
as a work-around for the Tegra 3. (The alignment was changed in [2]
as a response to [3], from 32-bit to 16-bit).
gstreamer only used this workaround in the case where the decoder
didn't return a slice-height property, but when the code was copied
into avcodec the conditional got lost. This commit restores the guard
and prefers the slice-height from the decoder when it is available.
This fixes segfaults decoding 1920x1080 h264 and mpeg2 videos on the
NVidia SHIELD after upgrading to Android Oreo.
[1] a870e6a5c3
[2] 21ff3ae0b0
[3] https://bugzilla.gnome.org/show_bug.cgi?id=748867
Signed-off-by: Aman Gupta <aman@tmm1.net>
Fixes some SVQ3 encoded files which fail to decode correctly after 6d6faa2a2d.
These files exhibit lots of artifacts and logs show "Media key encryption is not implemented".
However they decode without artifacts before 6d6faa2a2d.
The attatched patch allows these files to successfully decode, but also reject media key files.
Tested on the files in #6094 and http://samples.mplayerhq.hu/V-codecs/SVQ3/Vertical400kbit.sorenson3.mov
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Make sure to not write forbidden OBUs to CodecPrivate, and do the same with
unnecessary OBUs for packets.
Signed-off-by: James Almer <jamrial@gmail.com>
Some containers, like Matroska, may propagate key frames with no Sequence
Header OBU since it's provided in extradata instead.
With this change, the Sequence Header will be appended to the packet data
before calling aom_codec_decode().
Signed-off-by: James Almer <jamrial@gmail.com>
Some callers (like do_subtitle_out(), or do_streamcopy()) call this
with an AVPacket that is not refcounted. This can cause undefined
behavior.
Calling av_packet_move_ref() does not make a packet refcounted if it
isn't yet. (And it can't be made to, because it always succeeds,
and can't return ENOMEM.)
Call av_packet_ref() instead to make sure it's refcounted.
Cc: libav-stable@libav.org
Signed-off-by: Martin Storsjö <martin@martin.st>
Add tests for upmixing and downmixing with audio channel counts that
have a corresponding default layout and also tests where there is no
default layout.
Update the existing "stereo4" test so it actually outputs stereo like
the other stereo tests. Rename the previous "stereo4" test into
"upmix1".
Reviewed-by: Nicolas George <george@nsup.org>
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
Also remove the superfluous aandcttables dependency from all the modules
that only need it because of mpegvideoenc
Fixes ticket #7333
Signed-off-by: James Almer <jamrial@gmail.com>
This can change the the MSS value announced to the other end in
the initial TCP packet, it's can be used when failed Path MTU
discovery.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Current implementations of qsv components incorrectly work with async level, they
actually try to work in async+1 level stepping into MFX_WRN_DEVICE_BUSY and polling
loop. This change address this misbehaviour.
Signed-off-by: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
Cc: Maxym Dmytrychenko <maxim.d33@gmail.com>
Cc: Zhong Li <zhong.li@intel.com>
Signed-off-by: Maxym Dmytrychenko <maxim.d33@gmail.com>
Rematrixing supports up to 64 channels. However, there is only a limited number of channel layouts defined. Since the in/out channel count is currently obtained from the channel layout, for undefined layouts (e.g. for 9, 10, 11 channels etc.) the rematrixing fails.
This patch changes rematrix init methods to use in (used) and out channel count directly instead of computing it from channel layout.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
libavutil/hwcontext_d3d11va.c: In function 'd3d11va_device_create':
libavutil/hwcontext_d3d11va.c:554:46: warning: passing argument 2 of 'pAdapter->lpVtbl->GetDesc' from incompatible pointer type [-Wincompatible-pointer-types]
hr = IDXGIAdapter2_GetDesc(pAdapter, &desc);
^
libavutil/hwcontext_d3d11va.c:554:46: note: expected 'DXGI_ADAPTER_DESC * {aka struct DXGI_ADAPTER_DESC *}' but argument is of type 'DXGI_ADAPTER_DESC2 * {aka struct DXGI_ADAPTER_DESC2 *}'
Reviewed-by: Jean-Baptiste Kempf <jb@videolan.org>
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes a compilation warning if size_t != uint64_t:
libavformat/mov.c: In function ‘mov_read_saio’:
libavformat/mov.c:6207:45: warning: assignment from incompatible pointer type [-Wincompatible-pointer-types]
encryption_index->auxiliary_offsets = auxiliary_offsets;
^
This way if an index table segment is present multiple times, we can always use
the proper one instead of the invalid one.
Fixes seeking in the sample of ticket #5671.
Signed-off-by: Marton Balint <cus@passwd.hu>
And add it to the CONFIGURABLE_COMPONENTS list in Makefile. This way, changes
to the new file will be tracked and the usual warning to suggest re-running
configure will be shown.
Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes the following warnings:
In file included from libavcodec/fft_fixed.c:21:0:
libavcodec/fft_template.c:528:6: warning: ‘pass_big’ defined but not used [-Wunused-function]
PASS(pass_big)
^
libavcodec/fft_template.c:505:13: note: in definition of macro ‘PASS’
static void name(FFTComplex *z, const FFTSample *wre, unsigned int n)\
^~~~
CC libavcodec/ffv1.o
In file included from libavcodec/fft_float.c:21:0:
libavcodec/fft_template.c:528:6: warning: ‘pass_big’ defined but not used [-Wunused-function]
PASS(pass_big)
^
libavcodec/fft_template.c:505:13: note: in definition of macro ‘PASS’
static void name(FFTComplex *z, const FFTSample *wre, unsigned int n)\
^~~~
GCC requires the argument to vec_splat_u32 to be a literal. The easiest
way to accomplish this is to change 'shift' to be const in scale (as it
is in the transform routine above), and convert both routines to be
inline. This way, GCC can coerce the values to literals.
Tested on a 970 (Apple G5) and POWER9 (Talos II); passed fate and played
a clip of Big Buck Bunny correctly.
Fixes ticket #7048
Signed-off-by: A. Wilcox <AWilcox@Wilcox-Tech.com>
aom_codec_get_global_headers() is not implemented as of libaom 1.0.0 for AV1, so
we're forced to extract the relevant header OBUs from the first packet and propagate
them as packet side data.
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes two warnings:
libavfilter/af_afir.c:194:45: warning: assuming signed overflow does not occur when assuming that (X - c) > X is always false [-Wstrict-overflow]
int dx = FFABS(x1-x0), sx = x0 < x1 ? 1 : -1;
~~~~~~~~~~~~^~~~
libavfilter/af_aiir.c:689:45: warning: assuming signed overflow does not occur when assuming that (X - c) > X is always false [-Wstrict-overflow]
int dx = FFABS(x1-x0), sx = x0 < x1 ? 1 : -1;
~~~~~~~~~~~~^~~~
On macOS, a zero rc_max_rate cause an error from
VTSessionSetProperty(kVTCompressionPropertyKey_DataRateLimits).
on iOS (depending on device/version), a zero rc_max_rate cause invalid
arguments from the vtenc_output_callback after few frames and then a crash
within the VideoToolbox library.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Fixes a bug that would prevent using multiple comma-separated filters,
and allows options to be passed to each filter.
Based on similar loop in ffmpeg_opt.c's new_output_stream().
Signed-off-by: Aman Gupta <aman@tmm1.net>
Lensfun is a library that applies lens correction to an image using a
database of cameras/lenses (you provide the camera and lens models, and
it uses the corresponding database entry's parameters to apply lens
correction). It is licensed under LGPL3.
The lensfun filter utilizes the lensfun library to apply lens
correction to videos as well as images.
This filter was created out of necessity since I wanted to apply lens
correction to a video and the lenscorrection filter did not work for me.
While this filter requires little info from the user to apply lens
correction, the flaw is that lensfun is intended to be used on indvidual
images. When used on a video, the parameters such as focal length is
constant, so lens correction may fail on videos where the camera's focal
length changes (zooming in or out via zoom lens). To use this filter
correctly on videos where such parameters change, timeline editing may
be used since this filter supports it.
Note that valgrind shows a small memory leak which is not from this
filter but from the lensfun library (memory is allocated when loading
the lensfun database but it somehow isn't deallocated even during
cleanup; it is briefly created in the init function of the filter, and
destroyed before the init function returns). This may have been fixed by
the latest commit in the lensfun repository; the current latest release
of lensfun is almost 3 years ago.
Bi-Linear interpolation is used by default as lanczos interpolation
shows more artifacts in the corrected image in my tests.
The lanczos interpolation is derived from lenstool's implementation of
lanczos interpolation. Lenstool is an app within the lensfun repository
which is licensed under GPL3.
v2 of this patch fixes license notice in libavfilter/vf_lensfun.c
v3 of this patch fixes code style and dependency to gplv3 (thanks to
Paul B Mahol for pointing out the mentioned issues).
v4 of this patch fixes more code style issues that were missed in
v3.
v5 of this patch adds line breaks to some of the documentation in
doc/filters.texi (thanks to Gyan Doshi for pointing out the issue).
v6 of this patch fixes more problems (thanks to Moritz Barsnick for
pointing them out).
v7 of this patch fixes use of sqrt() (changed to sqrtf(); thanks to
Moritz Barsnick for pointing this out). Also should be rebased off of
latest master branch commits at this point.
Signed-off-by: Stephen Seo <seo.disparate@gmail.com>
MP3 frames may not be aligned to aa chunk boundaries. When seeking,
calculate the expected frame offset in the target chunk. Adjust the
timestamp and truncate the next packet accordingly.
This solution works for the majority of tested audio material. For
some rare encodings with mp3 padding or embedded id3 tags, it will
mispredict the correct offset, and at worst skip an extra frame.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
In commit 975a1a8,function ff_vc1_h_s_overlap_mmi was refactored,
but the declaration in libavcodec/mips/vc1dsp_mips.h was unchanged.
Change-Id: I90beae683511622a0cc1130ab1660ac8669ec3ef
Signed-off-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Reviewed-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Without this check some crafted files might crash because a packet might be
demuxed which have no corresponding mxf track.
Signed-off-by: Marton Balint <cus@passwd.hu>
This was reduced from 128 in libav commit
192f1984b1, but since we support unknown channel
layouts, we can increase this limit.
Fixes ticket #6332.
Signed-off-by: Marton Balint <cus@passwd.hu>
SDL from version 2.0.8 has support for full range YUV and specifying
BT601/BT709 color space for YUV->RGB conversion.
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes the following warnings:
libavcodec/aarch64/h264dsp_init_aarch64.c: In function ‘ff_h264dsp_init_aarch64’:
libavcodec/aarch64/h264dsp_init_aarch64.c:84:38: warning: assignment from incompatible pointer type [enabled by default]
c->weight_h264_pixels_tab[0] = ff_weight_h264_pixels_16_neon;
^
libavcodec/aarch64/h264dsp_init_aarch64.c:85:38: warning: assignment from incompatible pointer type [enabled by default]
c->weight_h264_pixels_tab[1] = ff_weight_h264_pixels_8_neon;
^
libavcodec/aarch64/h264dsp_init_aarch64.c:86:38: warning: assignment from incompatible pointer type [enabled by default]
c->weight_h264_pixels_tab[2] = ff_weight_h264_pixels_4_neon;
^
libavcodec/aarch64/h264dsp_init_aarch64.c:88:40: warning: assignment from incompatible pointer type [enabled by default]
c->biweight_h264_pixels_tab[0] = ff_biweight_h264_pixels_16_neon;
^
libavcodec/aarch64/h264dsp_init_aarch64.c:89:40: warning: assignment from incompatible pointer type [enabled by default]
c->biweight_h264_pixels_tab[1] = ff_biweight_h264_pixels_8_neon;
^
libavcodec/aarch64/h264dsp_init_aarch64.c:90:40: warning: assignment from incompatible pointer type [enabled by default]
c->biweight_h264_pixels_tab[2] = ff_biweight_h264_pixels_4_neon;
^
At present, box size is clipped to frame size before being drawn,
which can lead to the box not fully covering animated text which is
longer than one or both frame dimensions.
Since ff_blend_rectangle correctly takes care of clipping, it is skipped
here which results in correct box sizing
This fixes the check for the reserved MPEG audio version ID,
used to detect an invalid frame header.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
fix ticket: 7305
vs->sequence - hls->start_sequence - vs->nb_entries is the
after_init_list_dur fragment numbers
fix the wrong compute way vs->sequence - vs->nb_entries
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
Set make variable KEEP to non-zero value to preserve temp files
when a test has passed.
Helpful in diagnosing failed tests when test outfile is some type of
single hash and does not reveal differences in processed output.
The version 1 needs the channel count and would divide by 0
Fixes: division by 0
Fixes: fpe_movenc.c_1108_1.ogg
Fixes: fpe_movenc.c_1108_2.ogg
Fixes: fpe_movenc.c_1108_3.wav
Found-by: #CHEN HONGXU# <HCHEN017@e.ntu.edu.sg>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
read_packet reads content in chunks. Thus seek must be clamped to valid
chunk positions in the file, which in turn are relative to chapter start
positions.
So in read_header, scan for chapter headers once by skipping through the
content. Set stream time_base based on bitrate in bytes/s, for easy
timestamp to position conversion.
Then in read_seek, find the chapter containing the seek position, calculate
the nearest chunk position, and reinit the read_seek state accordingly.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Check the MPEG version ID for the reserved bit pattern 01, and abort the
header check in that case. This reduces the chance of misinterpreting
arbitrary data as a valid header, and prevents resulting audio artifacts.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Remember the end position of audio content in the file and check it during
read_packet. There always seems to be other data beyond it, which could be
misinterpreted as more audio. Also add some extra avio_read error checks,
to bail early in case of a broken/truncated file.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is identical to what the VP9 parser does
Fixes: 9215/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_LIBVPX_VP8_fuzzer-5768227253649408
Fixes: out of memory access
This may also fix oss fuzz issue 9212
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: use after free()
Fixes: rmdec-crash-ffe85b4cab1597d1cfea6955705e53f1f5c8a362
Found-by: Paul Ch <paulcher@icloud.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: integer overflow and out of array access
Fixes: asfo-crash-46080c4341572a7137a162331af77f6ded45cbd7
Found-by: Paul Ch <paulcher@icloud.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array access
Fixes: mxf-crash-1c2e59bf07a34675bfb3ada5e1ec22fa9f38f923
Found-by: Paul Ch <paulcher@icloud.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array read
Fixes: asff-crash-0e53d0dc491dfdd507530b66562812fbd4c36678
Found-by: Paul Ch <paulcher@icloud.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
For clip wrapped essences this should work. Also, since index_edit_rate can now
be different from track edit rate, remove overriding track edit rate.
Signed-off-by: Marton Balint <cus@passwd.hu>
The profile field is changed by code inside and outside the decoder,
its not a reliable indicator of the internal codec state.
Maintaining it consistency with studio_profile is messy.
Its easier to just avoid it and use only studio_profile
Fixes: assertion failure
Fixes: ffmpeg_crash_9.avi
Found-by: Thuan Pham, Marcel Böhme, Andrew Santosa and Alexandru Razvan Caciulescu with AFLSmart
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 1139785606 + 1454196085 cannot be represented in type 'int'
Fixes: 8937/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SHORTEN_fuzzer-6202943597445120
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This exposes encryption info from the container to the app. This
includes key ID, IV, and subsample byte ranges. The info is passed
using the new side-data AV_PKT_DATA_ENCRYPTION_DATA and
AV_PKT_DATA_ENCRYPTION_INIT_DATA.
Signed-off-by: Jacob Trimble <modmaker@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This commit implements a full ATRAC9 decoder, a simple low-delay codec
developed by Sony and used in most PSVita games, some PS3 games and some
PS4 games. Its similar to AAC in that it uses Huffman coded scalefactors
but instead of vector quantization it just Huffman codes the spectral
coefficients (in a way similar to how Opus splits band energy coding
into coarse and fine precision). It opts to write rather large Huffman
codes by packing several small coefficients into one Huffman coded
symbol, though I don't believe this increases efficiency at all.
Band extension implements SBC in a simple way, first it mirrors the
lower spectrum onto the higher frequencies and then it uses one of 5
filters to shape it. Noise substitution is implemented via 2 of them.
Unlike previous ATRAC codecs, there's no QMF, this is a standard MDCT
codec.
Based off of the reverse engineering work of Alex Barney.
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Testcase with large transparent rectangles changes from 67 sec to 3 sec decode time
Fixes: Timeout
Fixes: 8728/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DVDSUB_fuzzer-5190088756559872
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
And let the generic code figure out the proper PTS. This is needed because apng
does not provide seek functions, but after a generic seek (e.g. to file start)
timestamps are not reset which causes broken timestamps when looping apngs,
like in ticket #6121.
Signed-off-by: Marton Balint <cus@passwd.hu>
The input thread needs to be properly cleaned up and re-initalized before we
can start reading again in threaded mode. (Threaded input reading is used when
there is mode than one input file).
Fixes ticket #6121 and #7043.
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes: out of array access
Fixes: ffmpeg_bof_4.avi
Fixes: ffmpeg_bof_5.avi
Fixes: ffmpeg_bof_6.avi
Found-by: Thuan Pham, Marcel Böhme, Andrew Santosa and Alexandru Razvan Caciulescu with AFLSmart
Reviewed-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Set pixel format and color_range for YUVJ pixel formats. Also set
color_range based on AVFormatContext.
Signed-off-by: Wang Cao <wangcao@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The overlap filter is not correct for vertical edges in frame interlaced
I and P pictures. When filtering macroblocks with different FIELDTX values,
we have to match the lines at both sides of the vertical border. In addition,
we have to use the correct rounding values, depending on the line we are
filtering.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The overlap filter needs to cover a full macroblock vertical edge when the
FIELDTX value for two neighbouring macroblocks is not equal. By changing
the internal ordering of the blocks from row major to column major, we do
not need to reinterlace a FIELDTX coded macroblock before running the overlap
filter.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
When not using libavformat for demuxing, AVCodecContext.has_b_frames
gets set too late causing the recovery frame heuristic in h264_refs to
incorrectly flag an early frame as recovered.
This patch sets has_b_frames earlier to prevent improperly flagging the
frame as recovered.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: null pointer dereference
Fixes: ffmpeg_crash_7.avi
Found-by: Thuan Pham, Marcel Böhme, Andrew Santosa and Alexandru Razvan Caciulescu with AFLSmart
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array read
Fixes: ffmpeg_crash_8.avi
Found-by: Thuan Pham, Marcel Böhme, Andrew Santosa and Alexandru Razvan Caciulescu with AFLSmart
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: null pointer dereference
Fixes: ffmpeg_crash_6.avi
Found-by: Thuan Pham, Marcel Böhme, Andrew Santosa and Alexandru Razvan Caciulescu with AFLSmart
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: out of array access
Fixes: ffmpeg_bof_1.avi
Found-by: Thuan Pham, Marcel Böhme, Andrew Santosa and Alexandru Razvan Caciulescu with AFLSmart
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: -4096 * -524288 cannot be represented in type 'int'
Fixes: 8650/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_RA_144_fuzzer-5734816036159488
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This fixes a bug/regression with very small packets
Fixes: output_file
Regression since: 0782fb6bcb
Reported-by: Thierry Foucu <tfoucu@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Also make sure we set a valid track index sid and a valid track edit rate in
order for the index to be useful.
Signed-off-by: Marton Balint <cus@passwd.hu>
It is possible for there to be multiple encryption init info structure.
For example, to support multiple key systems or in key rotation. This
changes the AVEncryptionInitInfo struct to be a linked list so there
can be multiple structs without breaking ABI.
Signed-off-by: Jacob Trimble <modmaker@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This filter does HDR(HDR10/HLG) to SDR conversion with tone-mapping.
An example command to use this filter with vaapi codecs:
FFMPEG -init_hw_device vaapi=va:/dev/dri/renderD128 -init_hw_device \
opencl=ocl@va -hwaccel vaapi -hwaccel_device va -hwaccel_output_format \
vaapi -i INPUT -filter_hw_device ocl -filter_complex \
'[0:v]hwmap,tonemap_opencl=t=bt2020:tonemap=linear:format=p010[x1]; \
[x1]hwmap=derive_device=vaapi:reverse=1' -c:v hevc_vaapi -profile 2 OUTPUT
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
Current ffplay code assumes that the read thread is in its main loop before any
key events are captured, but apparently on IOS even keypresses without a window
are forwared.
Fixes ticket #7252.
Signed-off-by: Marton Balint <cus@passwd.hu>
da9cc22d5b allowed the MOV muxer to relay a custom stream handler name,
whether populated from the input stream or user-set. However, the entry
key didn't match the key set by the MOV demuxer, so it wasn't
effective. Fixed.
Due to the change, four FATE refs have to be updated. Verified that the
target payload of the tests hasn't changed in terms of CRC.
CLI options -maxrate, -bufsize and -rc_init_occupancy can now be picked
up by the x265 wrapper. Min. rc init has to be 1001 to avoid x265
setting it to vbv-bufsize.
In 9152c1e495, the mpegts parser was taught how to parse
PMT sections which contained multiple tables. That commit
fixed parsing of PMT packets from some cable providers,
which included a special SCTE table (0xc0) before the
standard program map table (0x2).
Sometimes, however, the combined 0xc0 and 0x2 tables are
larger than a single TS packet (188 bytes). The mpegts parser
already attempts to parse sections which span multiple packets,
but still assumed that the split section only contained one
table.
This patch fixes parsing of such a sample[1].
Before:
Input #0, mpegts, from 'combined-pmt-tids-split.ts':
Duration: 00:00:01.26, start: 39188.931756, bitrate: 597 kb/s
Program 1
No Program
Stream #0:0[0xeff]: Audio: ac3, 48000 Hz, mono, fltp, 64 kb/s
Stream #0:1[0xefd]: Audio: mp3, 0 channels, fltp
Stream #0:2[0xefe]: Unknown: none
After:
Input #0, mpegts, from 'combined-pmt-tids-split.ts':
Duration: 00:00:01.27, start: 39188.931756, bitrate: 589 kb/s
Program 1
Stream #0:0[0xefd]: Video: h264 ([27][0][0][0] / 0x001B), none, 59.94 fps, 59.94 tbr, 90k tbn, 180k tbc
Stream #0:1[0xefe](eng): Audio: ac3 ([129][0][0][0] / 0x0081), 48000 Hz, stereo, fltp, 384 kb/s
Stream #0:2[0xeff](spa): Audio: ac3 ([129][0][0][0] / 0x0081), 48000 Hz, mono, fltp, 64 kb/s
Stream #0:3[0xf00]: Data: scte_35
Stream #0:4[0xf01]: Unknown: none (ETV1 / 0x31565445)
Stream #0:5[0xf02]: Unknown: none (ETV1 / 0x31565445)
Stream #0:6[0xf03]: Unknown: none ([192][0][0][0] / 0x00C0)
With the patch, the PMT is parsed correctly so the streams are
created in the correct order, are associated with "Program 1",
and their codecs are set correctly.
[1] https://s3.amazonaws.com/tmm1/combined-pmt-tids-split.ts
Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
@xyz reported a regression on his Sony Xperia Z3 Tablet Compact where
playback would intermittently fail to start, essentially deadlocking in
the decoder. Bisecting narrowed down the issue to this commit, which was
meant as an optimization but is not necessary.
This reverts commit a75bb5496a.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Split vtenc_init() into vtenc_init() (VTEncContext initialization) and
vtenc_configure_encoder() (creates the vt session).
This commit will allow to restart the vt session while encoding.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Simple and Main Profile also need unsigned put_pixels_clamped. Add an argument
to choose between signed and unsigned put_pixels and change function name to
vc1_put_blocks_clamped.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
When using hardware accelerated decoding for multi-slice field interlaced pictures,
only the first slice was decoded. This patch adds the neccesary looping over the
remaining slices that may exist in field interlaced pictures. Additionally, we align
the calculation of mby_start for the second field with the method given in VC-1 spec.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Contrary to VC-1 spec, VAAPI expects the row address of the first
macroblock row in the first slice to start from zero for the second
field in a field interlaced picture.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Simple and Main profile also need unsigned put_pixels_clamped. Add an argument
to choose between signed and unsigned put_pixels and change function name to
vc1_put_blocks_clamped.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Overlap filtering I and BI frames for Simple and Main profile is only
dependent on PQUANT. Restrict testing for CONDOVER and OVERFLAGS to
advanced profile. Change from mb_width to end_mb_x in ff_vc1_i_loop_filter
to avoid breaking the Microsoft Screen 2 decoder.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The v_overlap_filter needs to run on the colocated block of the previous
macroblock. For the luma plane, the colocated block is located two blocks
on the left instead of one. In addition, the overlap filter needs to run
on the non-edge blocks of the first macroblock row and column.
Fixes ticket #7171.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Fixes: signed integer overflow: 1195517 * 2048 cannot be represented in type 'int'
Fixes: 8636/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AAC_FIXED_fuzzer-4695836326887424
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: 8521/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DIRAC_fuzzer-5639024952737792
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
STRIDE_ALIGN is not known in libavutil so av_image_check_size* cannot consider it
Fixes: OOM
Fixes: 8291/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SNOW_fuzzer-5176528009691136
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Regression since: af1761f7
Fixes: Division by 0
Fixes: ffmpeg_crash_1
Found-by: Thuan Pham, Marcel Böhme, Andrew Santosa and Alexandru Razvan Caciulescu with AFLSmart
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
verify that the stco atom is upgraded to co64 when the addition of moov
size to the offsets results in an overflow
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
when the last offsets in the stco atom are close to 4GB, the addition of
the moov atom size can overflow, causing corruption near the end of the
mp4 file.
this patch upgrades all stco atoms to co64 when such an edge case is
detected. in order to accomplish this, the implementation was changed to
walk the atom tree, instead of searching for the strings 'stco'/'co64'.
this was required since when an stco atom is changed to co64, its size
changes, and the sizes of all containing atoms (moov, trak, etc.) have
to be updated as well.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: assertion failure
Fixes: ffmpeg_crash_5.avi
Found-by: Thuan Pham <thuanpv@comp.nus.edu.sg>, Marcel Böhme, Andrew Santosa and Alexandru Razvan Caciulescu with AFLSmart
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
For chapter images, the mov demux produces streams with disposition set
to attached_pic+timed_thumbnails. This patch fixes to properly recognize
streams that should be encoded as cover image (ones with only and only
attached_pic disposition set).
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Find codec tag for attached images using appropriate list of
supported image formats.
This fixes writing the cover image to m4v/m4a and other container
formats that do not allow these codecs as a track.
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
A generic lavf flag for AAC LATM packetization for the RTP muxer was
added in ef409645f0 and then made inert 20 days later in 0832122880
when a private muxer option was added and the generic flag no longer
read.
If the user provides a valid timecode_format look for timecode of that
format in the capture and if found store it on the video avstream's
metadata.
Slightly modified by Marton Balint to capture per-frame timecode as well.
Signed-off-by: Marton Balint <cus@passwd.hu>
The default memory allocator is limited in the max number of frames available,
and therefore caused frame drops if the frames were not freed fast enough.
Signed-off-by: Marton Balint <cus@passwd.hu>
Some of these enums have gaps in between their values, since they correspond
to the values in various specs, instead of being an incrementing list.
Fixes segfaults when, for example, using the valid API call:
av_color_primaries_from_name("jecdec-p22");
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
1. vcm mode is only available for H264.
2. vcm is not supported on Linux, but it is shown when run "./avconv -h
encoder=h264_qsv |grep vcm". This shouldn't happen.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Signed-off-by: Maxym Dmytrychenko <maxim.d33@gmail.com>
Use a common way to control target_usage, keeping consistent with vaapi
encoders. The private option preset is kept only for compatibility.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Signed-off-by: Maxym Dmytrychenko <maxim.d33@gmail.com>
Fixes: signed integer overflow: 2146907204 + 26846088 cannot be represented in type 'int'
Fixes: 8105/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WAVPACK_fuzzer-6233036682166272
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes the following warnings:
In file included from libswscale/rgb2rgb.c:128:0:
libswscale/rgb2rgb_template.c:346:13: warning: 'shuffle_bytes_3210_c' defined but not used
libswscale/rgb2rgb_template.c:346:13: warning: 'shuffle_bytes_3012_c' defined but not used
libswscale/rgb2rgb_template.c:346:13: warning: 'shuffle_bytes_1230_c' defined but not used
move the the function aacsbr_tableinit definition from header file
to .c file to fix make checkheaders warning.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
move the the function init_tables() definitions from header file
to .c file to fix make checkheaders warning.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
This validates that the common encryption saio/saiz atoms only appear
when the data is actually encrypted. This also ignores those atoms
in clear content.
Found by Chrome's ClusterFuzz: http://crbug.com/850389
Signed-off-by: Jacob Trimble <modmaker@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: signed integer overflow: 3 * 1006632960 cannot be represented in type 'int'
Fixes: 8278/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SHORTEN_fuzzer-5692857166856192
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: OOM
Fixes: 8195/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SHORTEN_fuzzer-5179785826271232
The reference software appears to use longs for 32bits and it uses int for nmeans
hinting that the intended maximum size was not 32bit.
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This fix a build error on Windows:
C2440: connot convert from 'void (__cdecl *) (...)' to 'void (__stdcall *)(...)'.
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
Fixes: signed integer overflow: 2147483647 + 1 cannot be represented in type 'int'
Fixes: 8024/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_SHORTEN_fuzzer-5109204648984576
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: 32768 + 2147450880 cannot be represented in type 'int'
Fixes: 7885/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_THP_fuzzer-5298834394578944
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: 1077952576 + 1077952576 cannot be represented in type 'int'
Fixes: 7712/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TRUEMOTION2_fuzzer-5056281753681920
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If start_time is not set, ffmpeg takes the duration from the global
movie instead of the per stream duration.
Signed-off-by: Sasi Inguva <isasi@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
We already do this for audio, but it should be done for video too.
If we don't, seeking back to the start of the file, for example, can
become quite broken, since the first N packets will have repeating
and nonmonotonic PTS, yet they need to be decoded even if they are
to be discarded.
Signed-off-by: Sasi Inguva <isasi@isasi.mtv.corp.google.com>
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Fixes: runtime error: signed integer overflow: -1440457022 - 785819492 cannot be represented in type 'int'
Fixes: 7700/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_OPUS_fuzzer-6595838684954624
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Forced key frames generation functionality was assuming the first PTS
value as zero, but, when 'copyts' is enabled, the first PTS can be any
big number. This was eventually forcing all the frames as key frames.
To resolve this issue, update has been made to use first input pts as
reference pts.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Out-of-bounds reference pixel replication should take into account the frame
coding mode of the reference frame(s), not the frame coding mode of the
current frame.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
HMS is formatted as HH:MM:SS.mmm, but, HH part is not limited to
24 hours. For example, the the drawn text may look like this:
243029:20:30.342. To present the timestamp in more readable and
user friendly format, this patch provides an additional option
to limit the hour part in the range 0-23.
Note: Actually the above required format can be obtained with
format options 'localtime' and 'gmtime', but, milliseconds part
is not supported in those formats.
These files depend on libavformat, and the vf_srcnn filter
currently is the only thing utilizing these dnn_* files and
already happens to have a dependency on libavformat.
This fixes compilation in cases where libavformat is not a
dependency for libavfilter.
Reported by Kam_ on IRC.
These 2 fields are not always the same, it is simpler to always use the same field
for detecting studio profile
Fixes: null pointer dereference
Fixes: ffmpeg_crash_3.avi
Found-by: Thuan Pham <thuanpv@comp.nus.edu.sg>, Marcel Böhme, Andrew Santosa and Alexandru RazvanCaciulescu with AFLSmart
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Generic C implementation of vf_blend performs reads and writes of 16-bit
elements, which requires the buffers to be aligned to at least 2-byte
boundary.
Also, the change fixes source buffer overrun caused by src_offset being
added to to test handling of misaligned buffers.
Fixes: #7226
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
1. validate the moov size before checking for cmov atom
2. avoid performing arithmetic operations on unvalidated numbers
3. verify the stco/co64 offset count does not overflow the stco/co64
atom (not only the moov atom)
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This improves performance and makes qtrle behave more similar to other decoders.
Libavcodec does generally not output known duplicated frames, instead the calling Application
can insert them as it needs.
Fixes: Timeout
Fixes: 6383/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_QTRLE_fuzzer-6199846902956032
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The producer reference time box supplies relative wall-clock times
at which movie fragments, or files containing movie fragments
(such as segments) were produced.
The box is mainly useful in live streaming use cases. A media player
can parse the box and utilize the time fields to measure and improve
the latency during real time playout.
This utility function creates 64-bit NTP time format as per the RFC
5905.
A simple explaination of 64-bit NTP time format is here
http://www.beaglesoft.com/Manual/page53.htm
Direct prediction for interlace frame B pictures references the mv in the
second block in an MB in the backward reference frame for the twomv case.
When the backward reference frame is an I frame, this value may be unset.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
For interlace field pictures s->mb_height indicates the height of the full
picture in MBs, i.e. the two fields combined. A single field is half this
size. When calculating mquant for interlace field pictures, the bottom edge
is the last MB row of the field.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The last workaround is not sufficient to make oss fuzz work with the iterate API
as it did not provide a FFmpeg that external libs can be linked to.
This patch does not fully restore the pre iterate functionality. My attempts to
do this have so far failed.
The problem with this solution is that it renders the fuzzers virtual system
ffmpeg (libs) non functional. Which differs from a real system compared to the
virtual system tested by the fuzzer.
It should theoretically not matter as the system ffmpeg wouldnt be used.
But with more cases being fuzzed we likely will hit a case where a external
lib is involved and it does matter ...
Working around this may be possible with weak symbols but so far my attempts
failed
Alternatively multiple ffmpeg could be built, this becomes messy though
quickly as they need to be all linked together. That is we need a FFmpeg
that has the iterate API modified so it can work with the resources
available to ossfuzz. And at the same time we need a ffmpeg that has
its full functionality for any external libs which use ffmpeg and are
used by ffmpeg.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Yet another case of forgotten 0 =! EOF translation.
While the documentation for this specific synchronous read
function does not mention it, the documentation for
`sftp_async_read` documents it, as well as looking at the
implementation of this function leads one to find
`if (handle->eof) { return 0; }`.
Reported by stnutt on IRC.
Applicable only to webm output format.
By default all the segment filenames end with .m4s extension.
When someone chooses webm output format, we recommend they also override the relevant segment name options to end with .webm extension. This patch will issue a warning for he same
Right now segment file format is chosen to be either mp4 or webm based on the codec format.
This patch makes that choice configurable by the user, instead of being decided by the muxer.
Also with this change per-stream choice segment file format(based on codec type) is not possible.
All the output audio and video streams should be in the same file format.
Fixes: left shift of 1 by 63 places cannot be represented in type 'long long'
Fixes: out of array access
Fixes: 7284/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AC3_fuzzer-5767914968842240
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: -2141499320 + -14469590 cannot be represented in type 'int'
Fixes: 7351/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AAC_FIXED_fuzzer-6351214791884800
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: 1073741842 + 1784008138 cannot be represented in type 'int'
Fixes: 6792/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AAC_FIXED_fuzzer-5677589835284480
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes truncation
Fixes Assertion n <= 31 && value < (1U << n) failed at libavcodec/put_bits.h:169
Fixes: ffmpeg_crash_2.avi
Found-by: Thuan Pham <thuanpv@comp.nus.edu.sg>, Marcel Böhme, Andrew Santosa and Alexandru RazvanCaciulescu with AFLSmart
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: inconsistency
Fixes:runtime error: index 8 out of bounds for type 'int32_t [8]'
Fixes: 6686/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_TRUEHD_fuzzer-5191383498358784
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 inconsistent value combinations.
Alternatively it would be possible to add more checks and careful use of
temporary variables, but my try of this quickly seemed to become
a rather large change.
The disadvantage of this, is that the struct is copied back and forth.
Fixes: index 6 out of bounds for type 'const uint16_t [5][16]'
Fixes: 6557/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_INDEO4_fuzzer-4787296550256640
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Provide proper aliases to enable/disable MFE.
The numeric values are ambiguous and misleading (e.g: user may misunderstand
setting mfmode to 1 is to enable MFE but actually it is to disable MFE, and
set it to be 5 or above is meaningless).
MFX_MF_MANUAL hasn't been exposed since it is to be implemented.
Signed-off-by: Zhong Li <zhong.li@intel.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
The libvpx doxy says that a value of 0 for the g_threads field is
equivalent to a value of 1, whereas for avctx->thread_count it means
the maximum amount of threads possible for the host system.
Use av_cpu_count() to get the correct thread count when auto threads
is requested.
Reviewed-by: James Zern <jzern@google.com>
Signed-off-by: James Almer <jamrial@gmail.com>
This uses any devices it can find on the host system - on a system with no
hardware device support or in builds with no support included it will do
nothing and pass.
Both stream_id and stream_identifier are used in this file,
and have different meanings. The latter comes from the
stream_identifier_descriptor.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Without this some operations might overflow (undefined behavior)
even though the index adding loop would never execute
No testcase known
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
If v->fieldtx_is_raw is not reset to zero, it may spill over from a previous
interlaced frame I/BI picture.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This massively reduces the detection of random data as low score mp3
It may improve security by making it harder to read non multimedia data
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Only for the last slice of the first field is the last line of the slice
equal to the height of the field.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
DIRECTBIT was decoded before the intra/inter MB branching when decoding
interlace frame B pictures. Resulting in mistakenly also decoding it for intra
MBs where this syntax element is not present.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This new optional flag makes it easier to deal with mpegts
samples where the PMT is updated and elementary streams move
to different PIDs in the middle of playback.
Previously, new AVStreams were created per PID, and it was up
to the user to figure out which streams had migrated to a new PID
(by iterating over the list of AVProgram and making guesses), and
switch seamlessly to the new AVStream during playback.
Transcoding or remuxing these streams with ffmpeg on the CLI was
also quite painful, and the user would need to extract each set
of PIDs into a separate file and then stitch them back together.
With this new option, the mpegts demuxer will automatically detect
PMT changes and feed data from the new PID to the original AVStream
that was created for the orignal PID. For mpegts samples with
stream_identifier_descriptor available, the unique ID is used to
merge PIDs together. If the stream id is not available, the demuxer
attempts to map PIDs based on their position within the PMT.
With this change, I am able to playback and transcode/remux these
two samples which previously caused issues:
https://tmm1.s3.amazonaws.com/pmt-version-change.tshttps://kuroko.fushizen.eu/videos/pid_switch_sample.ts
I also have another longer sample in which the PMT changes
repeatedly and ES streams move to different pids three times
during playback:
https://tmm1.s3.amazonaws.com/multiple-pmt-change.ts
Demuxing this sample with the new option shows several new log
messages as the PMT changes are handled:
[mpegts] detected PMT change (program=1, version=3/6, pcr_pid=0xf98/0xfb7)
[mpegts] re-using existing video stream 0 (pid=0xf98) for new pid=0xfb7
[mpegts] re-using existing audio stream 1 (pid=0xf99) for new pid=0xfb8
[mpegts] re-using existing audio stream 2 (pid=0xf9a) for new pid=0xfb9
[mpegts] detected PMT change (program=1, version=6/3, pcr_pid=0xfb7/0xf98)
[mpegts] detected PMT change (program=1, version=3/4, pcr_pid=0xf98/0xf9b)
[mpegts] re-using existing video stream 0 (pid=0xf98) for new pid=0xf9b
[mpegts] re-using existing audio stream 1 (pid=0xf99) for new pid=0xf9c
[mpegts] re-using existing audio stream 2 (pid=0xf9a) for new pid=0xf9d
[mpegts] detected PMT change (program=1, version=4/5, pcr_pid=0xf9b/0xfa9)
[mpegts] re-using existing video stream 0 (pid=0xf98) for new pid=0xfa9
[mpegts] re-using existing audio stream 1 (pid=0xf99) for new pid=0xfaa
[mpegts] re-using existing audio stream 2 (pid=0xf9a) for new pid=0xfab
[mpegts] detected PMT change (program=1, version=5/6, pcr_pid=0xfa9/0xfb7)
Signed-off-by: Aman Gupta <aman@tmm1.net>
With these fields, the user has enough information to
detect PMT changes and switch to new streams when the PMT
is updated with new ES pids.
To do so, the user would monitor the AVProgram they're interested
in for changes to pmt_version. If the version changes, they would
iterate over the program's streams to find new streams added with
the updated version number.
If new versions of streams are found, then the user would first try
to replace existing streams where stream_identifier matched.
If stream_identifier is not available, then the user would compare
pmt_stream_idx instead to replace the stream that was previously
at the same position within the PMT.
Signed-off-by: Aman Gupta <aman@tmm1.net>
These fields will allow the mpegts demuxer to expose details about
the PMT/program which created the AVProgram and its AVStreams.
In mpegts, a PMT which advertises streams has a version number
which can be incremented at any time. When the version changes,
the pids which correspond to each of it's streams can also change.
Since ffmpeg creates a new AVStream per pid by default, an API user
needs the ability to (a) detect when the PMT changed, and (b) tell
which AVStream were added to replace earlier streams.
This has been a long-standing issue with ffmpeg's handling of mpegts
streams with PMT changes, and I found two related patches in the wild
that attempt to solve the same problem:
The first is in MythTV's ffmpeg fork, where they added a
void (*streams_changed)(void*); to AVFormatContext and call it from
their fork of the mpegts demuxer whenever the PMT changes.
The second was proposed by XBMC in
https://ffmpeg.org/pipermail/ffmpeg-devel/2012-December/135036.html,
where they created a new AVMEDIA_TYPE_DATA stream with id=0 and
attempted to send packets to it whenever the PMT changed.
Signed-off-by: Aman Gupta <aman@tmm1.net>
In a normal hwaccel, the AVHWFramesContext sets AVFrame.hw_frames_ctx
when it initializes a new AVFrame in av_hwframe_get_buffer().
But the VT hwaccel doesn't know what hw_frames_ctx to assign when
the AVFrame is first created, because it depends on the format of
the pixbuf that the decoder eventually decides to return. Thus
newly created AVFrames always have a NULL hw_frames_ctx, and the
hwaccel would only assign the ctx once a frame was done decoding.
This worked fine with the H264 decoder, but with the HEVC decoder
the frame's data may be moved to another empty AVFrame. Since the
empty AVFrame never had hw_frames_ctx set, a frame with a NULL
ctx could be returned to the API user.
This patch works around the issue by moving the derived
hw_frames_ctx from the AVFrame to a new VTHWFrame which now holds
both the CVPixelBufferRef and the AVBuffer. The hw_frames_ctx
is only copied to the AVFrame right before it is about to be
returned to the user in videotoolbox_postproc_frame() (since
in the case of VT, the hw_frames_ctx is only there for the API
user anyway).
Fixes playback on macOS and iOS of some hevc videos like
https://s3.amazonaws.com/tmm1/videotoolbox/germany-hevc-zdf.ts
Signed-off-by: Aman Gupta <aman@tmm1.net>
Some filtered mpegts streams may erroneously include PMTs for
programs that are not advertised in the PAT. This confuses ffmpeg
and most players because multiple audio/video streams are created
and it is unclear which ones actually contain data.
See for example https://tmm1.s3.amazonaws.com/unknown-pmts.ts
In this sample, the PAT advertises exactly one program. But the
pid it points to for the program's PMT contains PMTs for other
programs as well. This is because the broadcaster decided to
re-use the same pid for multiple program PMTs.
The hardware that filtered the original multi-program stream
into a single-program stream did so by rewriting the PAT to
contain only the program that was requested. But since it just
passed through the PMT pid referenced in the PAT, multiple PMTs
are still present for the other programs.
Before:
Input #0, mpegts, from 'unknown-pmts.ts':
Duration: 00:00:10.11, start: 80741.189700, bitrate: 9655 kb/s
Program 4
Stream #0:2[0x41]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv, bt709, progressive), 1280x720 [SAR 1:1 DAR 16:9], Closed Captions, 11063 kb/s, 59.94 fps, 59.94 tbr, 90k tbn, 119.88 tbc
Stream #0:3[0x44](eng): Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, 5.1(side), fltp, 384 kb/s
Stream #0:4[0x45](spa): Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, stereo, fltp, 128 kb/s
No Program
Stream #0:0[0x31]: Video: mpeg2video ([2][0][0][0] / 0x0002), none(tv), 90k tbr, 90k tbn, 90k tbc
Stream #0:1[0x34](eng): Audio: ac3 (AC-3 / 0x332D4341), 0 channels, fltp
Stream #0:5[0x51]: Video: mpeg2video ([2][0][0][0] / 0x0002), none, 90k tbr, 90k tbn
Stream #0:6[0x54](eng): Audio: ac3 (AC-3 / 0x332D4341), 0 channels
With skip_unknown_pmt=1:
Input #0, mpegts, from 'unknown-pmts.ts':
Duration: 00:00:10.11, start: 80741.189700, bitrate: 9655 kb/s
Program 4
Stream #0:0[0x41]: Video: mpeg2video (Main) ([2][0][0][0] / 0x0002), yuv420p(tv, bt709, progressive), 1280x720 [SAR 1:1 DAR 16:9], Closed Captions, 11063 kb/s, 59.94 fps, 59.94 tbr, 90k tbn, 119.88 tbc
Stream #0:1[0x44](eng): Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, 5.1(side), fltp, 384 kb/s
Stream #0:2[0x45](spa): Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, stereo, fltp, 128 kb/s
Signed-off-by: Aman Gupta <aman@tmm1.net>
Generates color bar test patterns based on EBU PAL recommendations.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
The FATE tests for MSVC versions older than 2013 are untested in FATE
and apparently are no longer supported.
This commit makes the configure process error out in case an older version
is used, and suggests to use a supported version of MSVC to compile.
This also changes the documentation to reflect this.
As discussed on IRC:
2018-05-12 19:45:16 jamrial then again, most of those were for old msvc, and i think we're not supporting versions older than 2013 (first one c99 compliant) anymore
2018-05-12 19:45:43 +JEEB yea, I think 2013 update 2 is needed
22:53 <@atomnuker> nevcairiel: which commit broke/unsupported support for msvc 2013?
23:23 <@atomnuker> okay, it was JEEB
23:25 <+JEEB> which was for 2012 and older
23:25 <+JEEB> and IIRC we no longer test those in FATE so that was my assumption
23:26 <+JEEB> 2013 is when MS got trolled enough to actually update their C part
23:26 <+JEEB> aand actually advertised FFmpeg support
23:26 <+JEEB> (although it was semi-failing until VS2013 update 1 or 2)
Signed-off-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Parses the video_stream_descriptor (H.222 2.6.2) to look
for the still_picture_flag. This is exposed to the user
via a new AV_DISPOSITION_STILL_IMAGE.
See for example https://tmm1.s3.amazonaws.com/music-choice.ts,
whose video stream only updates every ~6 seconds.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
For seekable mpegts streams, duration is calculated from
pts by seeking to the end of the file for a pts and subtracting
the initial pts to compute a duration.
This can be expensive in terms of added latency during
probe, especially when streaming over a network. This new
option lets you skip the duration calculation, which is useful
when you don't care about the value and want to save some overhead.
This patch is particularly useful when dealing with live mpegts
streams. Normally such streams are not seekable, so durations
are not calculated. However in my case I am dealing with a seekable
live mpegts stream (networked access to a .ts file which is still
being appended to).
Signed-off-by: Aman Gupta <aman@tmm1.net>
Unbreaks files with unknown extradata, the Canopus decoder accepts both files
with and without this extradata (24 byte "INFO", 16 byte "RDRT", rest "FIEL").
Reported-by: Peter Bubestinger
Tested-by: Piotr Bandurski
Fixes: runtime error: shift exponent -1 is negative
Fixes: 7486/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-4977380939530240
Fixes: runtime error: index 36 out of bounds for type 'const uint8_t [32]'
Fixes: 7566/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-6536620682510336
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: runtime error: left shift of 1876744317 by 16 places cannot be represented in type 'int'
Fixes: 6799/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_G2M_fuzzer-5115274731716608
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
A few days ago ossfuzz stoped testing new FFmpeg as it run out of diskspacee
https://oss-fuzz-build-logs.storage.googleapis.com/index.html
An alternative would be to revert the API.
This changes for example
-rwxr-x--- 1 michael michael 144803654 May 14 12:54 tools/target_dec_ac3_fixed_fuzzer*
to
-rwxr-x--- 1 michael michael 30333852 May 14 12:51 tools/target_dec_ac3_fixed_fuzzer*
Which should massively decrease space requirements
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
My conversation from AVFormatContext->filename to AVFormatContext->url was
wrong in this case because get_chunk_filename uses filename as an output
buffer, and not as an input buffer.
Fixes ticket #7188.
Signed-off-by: Marton Balint <cus@passwd.hu>
This does require libmysofa with today's latest commit (08f243d1ec).
They already had a pkg-config file, but the dependencies weren't setup right. Until now.
This should be included as `<lilv/lilv.h>`, same as is done in af_lv2.c.
Forcing the extra lilv-0 breaks platforms where the include dir is
`/usr/include/lilv/lilv.h` rather than
`/usr/include/lilv-0/lilv/lilv.h`.
The new include path works for both, because the `pkg-config --cflags`
includes `-I/usr/include/lilv-0`.
Fixes PMT parsing in some mpegts streams which contain
multiple tables within the PMT pid. Previously, the parser
assumed only one table was present in each packet, and discarded
the rest of the section data after attempting to parse the first
table.
A similar issue was documented in the BeyondTV software[1], which
helped me diagnose the same bug in the ffmpeg mpegts demuxer. I also
tried DVBInspector, libdvbpsi's dvbinfo, and tstools' tsinfo to
help debug. The former two properly read PMTs with multiple tables,
whereas the last has the same bug as ffmpeg.
I've created a minimal sample[2] which contains the combined PMT.
Here's what ffmpeg probe shows before and after this patch:
Before:
Input #0, mpegts, from 'combined-pmt-tids.ts':
Duration: 00:00:01.08, start: 4932.966167, bitrate: 741 kb/s
Program 1
No Program
Stream #0:0[0xf9d]: Audio: ac3, 48000 Hz, mono, fltp, 96 kb/s
Stream #0:1[0xf9b]: Audio: mp3, 0 channels, fltp
Stream #0:2[0xf9c]: Unknown: none
After:
Input #0, mpegts, from 'combined-pmt-tids.ts':
Duration: 00:00:01.11, start: 4932.966167, bitrate: 718 kb/s
Program 1
Stream #0:0[0xf9b]: Video: mpeg2video ([2][0][0][0] / 0x0002), none(tv, top first), 29.97 fps, 29.97 tbr, 90k tbn, 90k tbc
Stream #0:1[0xf9c](eng): Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, 5.1(side), fltp, 384 kb/s
Stream #0:2[0xf9d](spa): Audio: ac3 (AC-3 / 0x332D4341), 48000 Hz, mono, fltp, 96 kb/s
With the patch, the PMT is parsed correctly so the streams are
created in the correct order, are associated with "Program 1",
and their codecs are set correctly.
[1] http://forums.snapstream.com/vb/showpost.php?p=343816&postcount=201
[2] https://s3.amazonaws.com/tmm1/combined-pmt-tids.ts
Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This allows remuxing streams from one mpegts container to another,
without requiring avformat_find_stream_info() (or using `ffmpeg
-probesize 32` on the cli).
Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
SMPTE 386M (D-10) lists 4 as value to be used
SMPTE 377-1-2009 says
"The definitions of 00h (coSiting) and 04h (Rec 601) are equivalent. The value of 04h is deprecated. New
MXF encoders shall use the value of 00h instead."
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Adds tests for the hue angle and brightness filter parameters.
Renames the existing saturation parameter test for consistency.
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
The artificial sample file sei-1.h264 contains five frames (IDR P B I B)
and the following SEI message types:
* Buffering period
* Picture timing
* Pan-scan rectangle (display as 4:3)
* User data registered, containing A/53 closed captions (captions match
frame content, including reordering)
* Recovery point (at the I frame)
* Display orientation (identity transformation)
* Mastering display (with arbitrary contents)
* Undefined SEI type 1234 (containing ascending bytes)
Formats ARGB32, XRGB32, ABGR32, and XBGR32 were added to V4L2 instead
of ill-defined deprecated RGB32/BGR32 pixel formats.
When pixel format is not specified explicitly FFmpeg tries formats in
order in which they are stored in the table. Therefore formats are
sorted as follows: BGR is preferred over RGB and XBGR is preferred
over ARGB, because it could give better performance by ignoring alpha
component.
'-sei xxx' is added to control SEI insertion, so far only mastering
display colour volume is available for testing.
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
Similar to H264, cbs_h265_{read, write}_nal_unit() can handle HEVC
prefix SEI NAL units. Currently mastering display colour volume SEI
message is added only, we may add more SEI message if needed later
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
Every fourcc in vaapi_drm_format_map should be in
vaapi_format_map, so add an assert to ensure we have the right maps.
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
The flag of input_available must be set when pic_start is not NULL, so
add an assert to ensure it is true. In addition, the assert on last_pic
is unnecessary now, so remove this assert.
Signed-off-by: Haihao Xiang <haihao.xiang@intel.com>
The main issue here was the use of [i] instead of [i * 3] for the 32x32
matrix. As part of fixing this, I changed the code to match that used
in vdpau_hevc, which I spent a lot of time verifying.
I also changed to calculating NumPocTotalCurr using the existing helper,
which is what vdpau does.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
fix the warning: "function declaration isn’t a prototype", in C
int foo() and int foo(void) are different functions. int foo()
accepts an arbitrary number of arguments, while int foo(void) accepts 0
arguments.
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
Replaces the data pointers with the mapped cuvid ones.
Adds buffer_refs to the frame to ensure the needed contexts stay alive
and the cuvid idx stays allocated.
Adds another buffer_ref to unmap the frame when it's unreferenced itself.
This reverts commit 7d4e1f7cfb.
Accidentially pushed this with a batch of other patches, and it didn't
seem to break anything, so I went with it.
Except it does, so reverting it it is.
This mimics the logic flow in all the other callbacks
(pat_cb, sdt_cb, m4sl_cb), and avoids calling skip_identical()
for non PMT_TID packets.
Since skip_identical modifies internal state like
MpegTSSectionFilter.last_ver, this change prevents unnecessary
reprocessing on some streams which contain multiple tables in
the PMT pid. This can be observed with streams from certain US
cable providers, which include both tid=0x2 and another unspecified
tid=0xc0.
Signed-off-by: Aman Gupta <aman@tmm1.net>
No longer required since 63d875772d. The equivalent hack
for h264 was removed in that commit, but this one was missed.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Removes unnecessary data copies, and partially fixes potential issues
with dangling references held in said lists.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
This helps figuring out where the filter is slow:
70.53% ffmpeg_g ffmpeg_g [.] nlmeans_slice
25.73% ffmpeg_g ffmpeg_g [.] compute_safe_ssd_integral_image_c
1.74% ffmpeg_g ffmpeg_g [.] compute_unsafe_ssd_integral_image
0.82% ffmpeg_g ffmpeg_g [.] ff_mjpeg_decode_sos
0.51% ffmpeg_g [unknown] [k] 0xffffffff91800a80
0.24% ffmpeg_g ffmpeg_g [.] weight_averages
(Tested with a large image that takes several seconds to process)
Since this function is irrelevant speed wise, the file's TODO is
updated.
before: ssd_integral_image_c: 49204.6
after: ssd_integral_image_c: 44272.8
Unrolling by 4 made the biggest difference on odroid-c2 (aarch64);
unrolling by 2 or 8 both raised 46k cycles vs 44k for 4.
Additionally, this is a much better reference when writing SIMD (SIMD
vectorization will just target 16 instead of 4).
SIMD code will not have to deal with padding itself. Overwriting in that
function may have been possible but involve large overreading of the
sources. Instead, we simply make sure the width to process is always a
multiple of 16. Additionally, there must be some actual area to process
so the SIMD code can have its boundary checks after processing the first
pixels.
Most decoders (pgssubdec, ccaption_dec) are using -1 or UINT32_MAX for a
subtitle event which should be cleared at the next event.
Signed-off-by: Marton Balint <cus@passwd.hu>
This is large enough for all jpeg2000 files i tested. If some need more then this
should be changed to dynamic allocation. Dynamic allocation would need to be done
carefully as these are many relatively small arrays so repeatly reallocating them
would not be good.
The decrease is a clean and simple solution assuming it works for all files.
Fixes: OOM
Fixes: 6534/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_JPEG2000_fuzzer-4821490731057152
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
It exists, so why not use it? Helps one get rid of additional
search path related flags in addition to PKG_CONFIG_{PATH,LIBDIR}
when utilizing a cross-prefix separate from the sysroot.
Older iOS devices don't have a hardware HEVC decoder, but the
software decoder offered by VideoToolbox is well-optimized and
performs much better than the ffmpeg decoder.
Signed-off-by: Aman Gupta <aman@tmm1.net>
The patch enables dynamic bitrate through ReconfigureEncoder method
from nvenc API.
This is useful for live streaming in case of network congestion.
Signed-off-by: pkviet <pkv.stream@gmail.com>
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
If there is input like DVB-T streams it can change aspect ratio
on-the-fly, so nvenc should respect this change and change aspect ratio
in encoder.
Signed-off-by: Timo Rothenpieler <timo@rothenpieler.org>
The thread id was invalid because it was not initialised
during the calls to init_complex_filtergraph.
This adds a flag to check for initialisation before trying to
peform the join.
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: Kevin Wheatley <kevin.j.wheatley@gmail.com>
I tested the previous mediacodec changes on seven different Android
TV devices, with both mpeg2 and h264 content. All except one worked
as expected. The exception was the MiBox3 running Android 6.0.1,
where playback would freeze on a frame every few seconds. I tested
two other AMLogic devices with newer Android versions that did not
show the same problem. H264 decoding on the MiBox3 was also not affected,
so this workaround applies only to OMX.amlogic.mpeg2.decoder.awesome
on Android API22.
There is a rumor that Xiaomi is planning to release Android Oreo for
the MiBox3, so I will revisit in a few months to confirm whether this
is specific to os/driver version or the chipset used in that device.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>
The output_buffer_count==0 special case is no longer required, and
can cause spurious EAGAIN to surface to the user when input buffers
are filled up. Since the caller now knows if the decoder is accepting
new input (via current_input_buffer>=0), let the wait parameter
control whether we block or not.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>
The new logic follows a recommendation by @rcombs to use
dequeueInputBuffer with a timeout of 0 as a way to detect
whether the codec wants more data. The dequeued buffer index is
kept in MediaCodecDecContext until it can be used next.
A similar technique is also used by the Google's official media
player Exoplayer: see MediaCodecRenderer.feedInputBuffer().
Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>
This can "demux" .vpy files. Autodetection of .vpy scripts is
intentionally not done, because it would be a major security issue. You
need to force the format, for example with "-f vapoursynth" for the
FFmpeg CLI tools.
Some minor code copied from other LGPL parts of FFmpeg.
I did not find a good way to test a few of the more obscure VS features,
like VFR nodes, compat pixel formats, or nodes with dynamic size/format
changes. These can be easily implemented on demand.
This code will print a warning if any user agent is set - even if the
API user used the proper non-deprecated "user_agent" option.
This change should not even break anything, because even if the user
sets the deprecated "user-agent" option, http.c copies it to the
"user_agent" option anyway.
If the API user doesn't set avg_frame_rate, matroskaenc will write the
current timebase as "default duration" for the video track. This makes
no sense, because the "default duration" implies the framerate of the
video. Since the timebase is forced to 1/1000, this will make the
resulting file claim 1000fps.
Drop it and don't write the element. It's optional, so it's better not
to write it if the framerate is unknown.
Strangely does not require FATE changes.
Enables one to test possibly nonstandard formats such as Opus or
FLAC in ISOBMFF, among other things.
This becomes much more useful if output segment format becomes an
option, or if the WebM segment feature gets removed.
It has not ever been working and has not been validated,
Additionally, mention that the segment file names should be changed
to end with webm instead of m4s, which is utilized for ISOBMFF
fragments.
Fixes stream field order written by avformat_write_header when "top"
option is specified on the command-line.
Signed-off-by: Tobias Rapp <t.rapp@noa-archive.com>
Uses the same mechanism as other codecs - conformance test files are
passed through the metadata filter (which, with no options, reads the
input and writes it back) and the output verified to match the input.
Create a buffer from the data instead and use the buffer destructor to free the
DeckLink frame. This avoids a memcpy of the frame data.
Signed-off-by: Marton Balint <cus@passwd.hu>
Temporarily keep the old method for ffmpeg_filters.c choose_pix_fmt and
avfiltergraph.c pick_format() until a paletted pixel format without alpha is
introduced.
Signed-off-by: Marton Balint <cus@passwd.hu>
The specs says that the the first color component in the color array is
not alpha, but simply 0.
Fixes 0 alpha of fate-suite/cvid/catfight-cvid-pal8-partial.mov
Signed-off-by: Marton Balint <cus@passwd.hu>
Fixes: runtime error: signed integer overflow: 2147483637 + 128 cannot be represented in type 'int'
Fixes: 6701/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WAVPACK_fuzzer-5358324934508544
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes: runtime error: signed integer overflow: 2147483531 + 16384 cannot be represented in type 'int'
Fixes: 6615/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_WAVPACK_fuzzer-5165715515506688
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
In vc1_decode_i_blocks_adv mquant needs to be reset to its default value for
each macroblock, instead of once at the beginning of the slice.
DQPROFILE specifies which macroblocks can have an alternative quantizer step
size. When DQPROFILE specifies edges, the selection is applicable to the edges
of the picture. Slice edges are not selected by DQPROFILE.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Useful when transcoding videos at 29.97 fps because delivers a more accurate result for monitoring.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The -benchmark and -benchmark_all options now show user, system, and real time,
instead of just user time.
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
1. an audio component with an ISO_639_language_descriptor in the PMT with the
audio_type field set to 0x03
2. a supplementary_audio_descriptor with the editorial_classification field set
to 0x01
3. an ac-3_descriptor or an enhanced_ac-3_descriptor with a component_type field
with the service_type flags set to Visually Impaired
Tested-by: Łukasz Krzciuk <lkrzciuk@vewd.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This way, every CodedBitstreamType->split_fragment() function can
safely assume the fragment passed to them will be reference counted,
potentially simplifying code.
Reviewed-by: Mark Thompson <sw@jkqxz.net>
Signed-off-by: James Almer <jamrial@gmail.com>
Overlap filtering of the first row and first column must be guarded
for out of bounds access of v->over_flags_plane.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The track's media duration from the mdhd atom takes precedence
over both the stts and elst atom for calculating and setting
the track's total duraion.
Technically, we shouldn't be using the stts atom at all for
calculating stream durations.
This fixes incorrect stream and final packet durations on files
with edit lists that are longer than the media duration.
The FATE changes are expected, and output is more correct (the
AAC frame is not 1028 samples).
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
The av_rc4_crypt() documentation allows src == dst.
Silences the following warning:
libavformat/rtmpcrypt.c:304:36: warning: passing argument 2 of 'av_rc4_crypt' discards 'const' qualifier from pointer target type
Reported-by: Reino Wijnsma
There is a separate muxer(webmdashenc.c) for supporting VP9+webm output in DASH.
Hence in this muxer we will focus on supporting VP9 in MP4
Have verified playout support of VP9+MP4 in Chrome and Firefox.
Fixes: runtime error: signed integer overflow: 197710 * 10923 cannot be represented in type 'int'
Fixes: 7010/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-5667127596941312
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
HALFQP should only be added to the inverse quantizer when the block is
coded with PQUANT. When PQUANT is equal to ALTPQUANT, the original test
for the addition of HALFQP fails. A negative value for mquant indicates
that the value was derived from VOPDQUANT.
Fixes#4372
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
HALFQP should only be added to the inverse quantizer when the block is
coded with PQUANT. See 8.1.3.8 in VC-1 spec.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
mspel indicates the use of bicubic interpolation. The check wrongly included
MVMODE MV_PMODE_1MV_HPEL as using bilinear interpolation.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Add previously omitted overlap smooting and loop filtering for
frame/field-interlace pictures. For progressive pictures switch to the
re-implemented versions of overlap smooting and loop filtering.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The existing implementation only used vc1_put_signed_blocks_clamped for I and
BI frames. This rewritten version is also applicable to P frame both
progressive and frame/field-interlace.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The existing implementation did out-of-bounds reference pixel replication for
progressive reference frames. In interlaced reference frames both the even and
odd line on the horizontal edges need to be replicated.
Fixes#3262.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
According to VC-1 spec table 74, the last value in ff_vc1_dqscale should be
0x1041 instead of 0x1000.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
According to VC-1 spec 10.7.3.4, FIELDTX shall be set to the same type as the
motion vector for zero-coded blocks.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The loop filter for P interlace field pictures needs the reference field type.
For luma, the reference field type was already available. Store the reference
field type for color-difference as well.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The existing implementation did loop filtering for progressive
frames only. This rewritten version implements loop filtering for
all applicable frame types for both progessive and
frame/field-interlace.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The new overlap smooting filter smoothes image pixels stored in v->block.
Switch to v->block instead of s->block for storing decoded image pixels for P
frames. Additionally, we must take incrementing *_blk_idx out of the
vc1_put_signed_blocks_clamped function.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
The existing implementation did overlap smoothing for progressive
frames only. This rewritten version implements overlap smoothing
for all applicable frame types for both progessive and
frame/field-interlace.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
before:
419022 decicycles in assemble_fragment, 2047 runs, 1 skips
after:
104621 decicycles in assemble_fragment, 2045 runs, 3 skips
Benched with a 2 minutes long 720x480 DVD mpeg2 sample.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: James Almer <jamrial@gmail.com>
Also fixes a bug where EOS buffer was sent with incorrect
pts when not using surface generation.
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>
Signed-off-by: Aman Gupta <aman@tmm1.net>
As of 2a0eb8685, ff_mediacodec_dec_is_flushing() only returns
true in delay_flush mode. Make this more obvious by adding
delay_flush to the if statement.
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>
Signed-off-by: Aman Gupta <aman@tmm1.net>
also fixes: runtime error: index 1456 out of bounds for type 'int16_t [16]'
Found-by: durandal_1707
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The crc flag is only stored since version 3 thus before this crcs do not
work. We increase the version as needed same as we do with pix_fmts
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Interlaced frame pictures do not contain the MVMODE or MVMODE2 bitstream
element. Trying to parse this element and passing a nonzero value to the
hardware decoder results in small inaccuracies in the decoded picture.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Should be useful for muxers that require values as defined in the
vpcc atom but don't need to write the atom itself.
Signed-off-by: James Almer <jamrial@gmail.com>
When use new decode APIs(avcodec_send_packet/avcodec_receive_frame),
don't need to setting the deprecated field refcounted_frames.
Reviewed-by: wm4 <nfxjfg@googlemail.com>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
When use new decode APIs(avcodec_send_packet/avcodec_receive_frame),
don't need to setting the deprecated field refcounted_frames.
Reviewed-by: wm4 <nfxjfg@googlemail.com>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
When use new decode APIs(avcodec_send_packet/avcodec_receive_frame),
don't need to setting the deprecated field refcounted_frames.
Reviewed-by: wm4 <nfxjfg@googlemail.com>
Signed-off-by: Jun Zhao <mypopydev@gmail.com>
scaleforsame_y references ref_field_type. Therefore, it needs to be set
before scaleforsame is called.
Fixes#2557.
Signed-off-by: Jerome Borsboom <jerome.borsboom@carpalis.nl>
Fixes a warning:
libavformat/dashdec.c:1900:65: warning: argument to 'sizeof' in 'memcpy' call is the same pointer type 'struct fragment *' as the destination; expected 'struct fragment' or an explicit length
For filters based on framesync, the input frame was managed
by framesync, so we should not directly keep and destroy it,
instead we make a clone of it here, or else double-free will occur.
But for other filters not based on framesync, we still need to
free the input frame inside filter_frame.
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
The existing version which was cherry-picked from Libav does not work
with FFmpeg framework, because ff_request_frame() was totally
different between Libav (recursive) and FFmpeg (non-recursive).
The existing overlay_qsv implementation depends on the recursive version
of ff_request_frame to trigger immediate call to request_frame() on input pad.
But this has been removed in FFmpeg since "lavfi: make request_frame() non-recursive."
Now that we have handy framesync support in FFmpeg, so I make it work
based on framesync. Some other fixing which is also needed to make
overlay_qsv work are put in a separate patch.
Signed-off-by: Ruiling Song <ruiling.song@intel.com>
This doesn't support saio atoms with more than one offset.
Signed-off-by: Jacob Trimble <modmaker@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Similar to 4c9c4fe8b2, but for durations. This fixes#7151, where
the report duration and bitrate on a mpegts stream is wildly off
due to the dvb_teletext stream's timings.
Signed-off-by: Aman Gupta <aman@tmm1.net>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
In write only mode, the TCP receive buffer's data keeps growing with
http response messages and the buffer eventually becomes full.
This results in zero tcp window size, which in turn causes unwanted
issues, like, terminated tcp connection. The issue is apparent when
http persistent connection is enabled in hls/dash live streaming use
cases. To overcome this issue, the logic here reads the buffer data
when a file transfer is completed, so that any accumulated data in
the recieve buffer gets flushed out.
reference hls support fmp4 file from draft-pantos-http-live-streaming-20
the spec describes version 7 of hls protocol
Suggested-by: Ronak <ronak2121@yahoo.com>
Signed-off-by: Steven Liu <lq@chinaffmpeg.org>
The headers from where the dimensions are read in actual files
are limited to 16bit per component.
Fixes: Timeout
Fixes: 6305/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_DFA_fuzzer-4824270749302784
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This refactors get_cookies to simplify some code paths, specifically for
skipping logic in the while loop or exiting it. It also simplifies the logic
for appending additional values to *cookies by replacing strlen/malloc/snprintf
with one call av_asnprintf.
This refactor fixes a bug where the cookie_params AVDictionary would get leaked
if we failed to allocate a new buffer for writing to *cookies.
Branch to global symbol results in reference to PLT, and when compiling
for THUMB-2 - in a R_ARM_THM_JUMP19 relocation. Some linkers don't
support this relocation (ld.gold), while others can end up truncating
the relocation to fit (ld.bfd).
Convert this branch through PLT into a direct branch that the assembler
can resolve locally.
See https://github.com/android-ndk/ndk/issues/337 for background.
The current workaround is to disable neon during gstreamer build,
which is not optimal and can be reverted after this patch:
41556c4157
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This fixes frame corruption issue when decoder started reusing frames
while they are still in use of encoding process
Issue with frame corruption was reproduced using:
avconv.exe -y -hwaccel d3d11va -hwaccel_output_format d3d11 -i input.h264 -an -c:v h264_amf output.mkv
It is recommended to use -extra_hw_frames 16 option in case if hw frames
number in pool is not enough
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
- enable the parsing code
- use the new buffer instead of replacing the context one
- do not push/pop configuration, just discard the exiting one
- propagate errors correctly
Without properly grouping the checks, the second test would execute for
MSVC cl.exe, which results in configure getting stuck since cl.exe -? is
an interactive paginated help screen, waiting for input.
Remove the wincrypt API calls since we don't support XP anymore and
bcrypt is available since Vista, even on Windows Store builds.
Signed-off-by: Martin Storsjö <martin@martin.st>
- Parse schm atom to get different encryption schemes.
- Allow senc atom to appear in track fragments.
- Allow 16-byte IVs.
- Allow constant IVs (specified in tenc).
- Allow only tenc to specify encryption (i.e. no senc/saiz/saio).
- Use sample descriptor to detect clear fragments.
This doesn't support:
- Different sample descriptor holding different encryption info.
- Only first sample descriptor can be encrypted.
- Encrypted sample groups (i.e. seig).
- Non-'cenc' encryption scheme when using -decryption_key.
Signed-off-by: Jacob Trimble <modmaker@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Show a more useful error message which specifies the required driver version
for the build, and use the correct context in the error message for WIN32.
Signed-off-by: Marton Balint <cus@passwd.hu>
Silences several warnings:
libavcodec/dxva2_internal.h:107:98: warning: pointer type mismatch in conditional expression
libavcodec/dxva2_internal.h:109:94: warning: pointer type mismatch in conditional expression
Reported-by: Reino Wijnsma
Fixes the following warnings:
libavdevice/vfwcap.c:331:35: warning: passing argument 1 of 'av_parse_video_size' from incompatible pointer type
libavdevice/vfwcap.c:331:59: warning: passing argument 2 of 'av_parse_video_size' from incompatible pointer type
Reported-by: Reino Wijnsma
SDL_QueryTexture and SDL_DestroyTexture require that the input texture
pointer be non-null. Debug builds of SDL will correctly check for this
and break program execution. This patch fixes this by checking the
status of the texture pointer.
Signed-off-by: Matt Oliver <protogonoi@gmail.com>
Fixes https://trac.ffmpeg.org/ticket/2798
This makes movenc handle AV_DISPOSITION_ATTACHED_PIC and write
the associated pictures in iTunes cover atom. This corresponds
to how 'mov' demuxer parses and exposes the cover images when
reading.
Most of the existing track handling loops properly ignore
these 'virtual streams' as MOVTrack->entry is never incremented
for them. However, additional tests are added as needed to ignore
them.
Tested to produce valid output with:
ffmpeg -i movie.mp4 -i thumb.jpg -disposition:v:1 attached_pic \
-map 0 -map 1 -c copy movie-with-cover.mp4
The cover image is also copied correctly with:
ffmpeg -i movie-with-cover.mp4 -map 0 -c copy out.mp4
AtomicParseley says that the attached_pic stream is properly
not visible in the main tracks of the file.
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
av_guess_sample_aspect_ratio() will return undefined or missing
value as {0,1}. This fixes show_stream() to check numerator to
display 'N/A' when appropriate. show_frame() does this already
correctly.
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
The logic is applicable only when use_template is enabled and use_timeline
is disabled. The logic monitors the flow of segment indexes. If a streams's
segment index value is not at the expected real time position, then
the logic corrects that index value.
Typically this logic is needed in live streaming use cases. The network
bandwidth fluctuations are common during long run streaming. Each
fluctuation can cause the segment indexes fall behind the expected real
time position. Without this logic, players will not be able to consume
the content, even after encoder's network condition comes back to
normal state.
availability time of Nth segment = availabilityStartTime + (N*segment duration) - availabilityTimeOffset.
This field helps to reduce the latency by about a segment duration in streaming mode.
@availabilityStartTime specifies the anchor for the computation of the earliest
availability time (in UTC) for any Segment in the Media Presentation.
As per this requirement, the @AvailabilityStartTime should be set to the
wallclock time at which the first frame of the first segment begins encoding.
But, it was getting set only when the first segment was completely ready. Making
the required correction in this patch. This correction is mainly needed to reduce
the latency in live streaming use cases.
Calling 'write_manifest' from 'write_header' was causing creation of
first MPD with invalid values. Ex: zero @duration param value. Also,
the manifest files (MPD or M3U8s) should be created when at-least
one media frame is ready for consumption.
When use_template is enabled and use_timeline is disabled, typically
it is required to generate the segments at the configured segment duration
rate on an average. This commit is particularly needed to handle the
segmentation when video frame rates are fractional like 29.97 or 59.94 fps.
There are use cases where average segment duration needs to be configured
and muxer is expected to maintain the average segment duration. So, using
the name 'min_seg_duration' will be misleading. So, changing the parameter
name to 'seg_duration', where it can be minimum segment duration or average
segment duration based on the use-case. The additional updates needed for
this functinality are made the sub-sequent patches of this patch series.
The HLSContext struct contains fields which duplicate the data stored in the
avio_opts field. This change removes those fields in favor of avio_opts, and
updates the code accordingly.
The original patch caused the buffer pointed to by new_cookies in open_url to be
leaked. The only thing that buffer is used for is to store the value until it
can be passed to av_dict_set. To fix the leak, v2 of the patch simply calls
av_dict_set with the AV_DICT_DONT_STRDUP_VAL flag, so that the dictionary takes
ownership of the memory instead of copying it again.
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Signed-off-by: Richard Shaffer <rshaffer@tunein.com>
The rw_timeout option is currently not applied when opening media playlist,
segment, or encryption key URLs. This can cause the HLS demuxer to block
indefinitely, even when the rw_timeout option has been specified. This change
simply enables carrying over the rw_timeout option when the demuxer opens these
URLs.
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Signed-off-by: Richard Shaffer <rshaffer@tunein.com>
The rw_timeout option is currently not applied when opening media playlist,
segment, or encryption key URLs. This can cause the HLS demuxer to block
indefinitely, even when the rw_timeout option has been specified. This change
simply enables carrying over the rw_timeout option when the demuxer opens these
URLs.
Reviewed-by: Steven Liu <lq@chinaffmpeg.org>
Signed-off-by: Richard Shaffer <rshaffer@tunein.com>
This prevents creating potentially broken files, as both the AV1 and
the AV1 in ISOMBFF specs are unfinished.
Signed-off-by: James Almer <jamrial@gmail.com>
Some old mingw-w64 builds seem to provide an incomplete implementation
of the API. Add an extra check to make sure it's disabled for those.
Signed-off-by: James Almer <jamrial@gmail.com>
@@ -178,6 +178,9 @@ Capture the mouse pointer. Default is 0.
@item -capture_mouse_clicks
Capture the screen mouse clicks. Default is 0.
@item -capture_raw_data
Capture the raw device data. Default is 0.
Using this option may result in receiving the underlying data delivered to the AVFoundation framework. E.g. for muxed devices that sends raw DV data to the framework (like tape-based camcorders), setting this option to false results in extracted video frames captured in the designated pixel format only. Setting this option to true results in receiving the raw DV stream untouched.
@end table
@subsection Examples
@@ -208,6 +211,13 @@ Record video from the system default video device using the pixel format bgr0 an
- new bitstream reader (see http://ffmpeg.org/pipermail/ffmpeg-devel/2017-April/209609.html)
- use av_cpu_max_align() instead of hardcoding alignment requirements (see https://ffmpeg.org/pipermail/ffmpeg-devel/2017-September/215834.html)
- f44ec22e0 lavc: use av_cpu_max_align() instead of hardcoding alignment requirements
- 4de220d2e frame: allow align=0 (meaning automatic) for av_frame_get_buffer()
- Support recovery from an already present HLS playlist (see 16cb06bb30)
- Remove all output devices (see 8e7e042d41, 8d3db95f20, 6ce13070bd, d46cd24986 and https://ffmpeg.org/pipermail/ffmpeg-devel/2017-September/216904.html)
- avcodec/libaomenc: export the Sequence Header OBU as extradata (See a024c3ce9a)
Collateral damage that needs work locally:
------------------------------------------
- Merge proresdec2.c and proresdec_lgpl.c
- Merge proresenc_anatoliy.c and proresenc_kostya.c
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
@var{use_timeline} is disabled and as minimum segment duration for all the other
use cases.
@item window_size @var{size}
Set the maximum number of segments kept in the manifest.
@item -extra_window_size @var{size}
@item extra_window_size @var{size}
Set the maximum number of segments kept outside of the manifest before removing from disk.
@item -remove_at_exit @var{remove}
@item remove_at_exit @var{remove}
Enable (1) or disable (0) removal of all segments when finished.
@item -use_template @var{template}
@item use_template @var{template}
Enable (1) or disable (0) use of SegmentTemplate instead of SegmentList.
@item -use_timeline @var{timeline}
@item use_timeline @var{timeline}
Enable (1) or disable (0) use of SegmentTimeline in SegmentTemplate.
@item -single_file @var{single_file}
@item single_file @var{single_file}
Enable (1) or disable (0) storing all segments in one file, accessed using byte ranges.
@item -single_file_name @var{file_name}
DASH-templated name to be used for baseURL. Implies @var{single_file} set to "1".
@item -init_seg_name @var{init_name}
DASH-templated name to used for the initialization segment. Default is "init-stream$RepresentationID$.m4s"
@item -media_seg_name @var{segment_name}
DASH-templated name to used for the media segments. Default is "chunk-stream$RepresentationID$-$Number%05d$.m4s"
@item -utc_timing_url @var{utc_url}
@item single_file_name @var{file_name}
DASH-templated name to be used for baseURL. Implies @var{single_file} set to "1". In the template, "$ext$" is replaced with the file name extension specific for the segment format.
@item init_seg_name @var{init_name}
DASH-templated name to used for the initialization segment. Default is "init-stream$RepresentationID$.$ext$". "$ext$" is replaced with the file name extension specific for the segment format.
@item media_seg_name @var{segment_name}
DASH-templated name to used for the media segments. Default is "chunk-stream$RepresentationID$-$Number%05d$.$ext$". "$ext$" is replaced with the file name extension specific for the segment format.
@item utc_timing_url @var{utc_url}
URL of the page that will return the UTC timestamp in ISO format. Example: "https://time.akamai.com/?iso"
@item -http_user_agent @var{user_agent}
@item method @var{method}
Use the given HTTP method to create output files. Generally set to PUT or POST.
@item http_user_agent @var{user_agent}
Override User-Agent field in HTTP header. Applicable only for HTTP output.
@item -http_persistent @var{http_persistent}
@item http_persistent @var{http_persistent}
Use persistent HTTP connections. Applicable only for HTTP output.
@item -hls_playlist @var{hls_playlist}
@item hls_playlist @var{hls_playlist}
Generate HLS playlist files as well. The master playlist is generated with the filename master.m3u8.
One media playlist file is generated for each stream with filenames media_0.m3u8, media_1.m3u8, etc.
@item -streaming @var{streaming}
@item streaming @var{streaming}
Enable (1) or disable (0) chunk streaming mode of output. In chunk streaming
mode, each frame will be a moof fragment which forms a chunk.
@item -adaptation_sets @var{adaptation_sets}
@item adaptation_sets @var{adaptation_sets}
Assign streams to AdaptationSets. Syntax is "id=x,streams=a,b,c id=y,streams=d,e" with x and y being the IDs
of the adaptation sets and a,b,c,d and e are the indices of the mapped streams.
To map all video (or audio) streams to an AdaptationSet, "v" (or "a") can be used as stream identifier instead of IDs.
When no assignment is defined, this defaults to an AdaptationSet for each stream.
@item -timeout @var{timeout}
@item timeout @var{timeout}
Set timeout for socket I/O operations. Applicable only for HTTP output.
@item index_correction @var{index_correction}
Enable (1) or Disable (0) segment index correction logic. Applicable only when
@var{use_template} is enabled and @var{use_timeline} is disabled.
When enabled, the logic monitors the flow of segment indexes. If a streams's
segment index value is not at the expected real time position, then the logic
corrects that index value.
Typically this logic is needed in live streaming use cases. The network bandwidth
fluctuations are common during long run streaming. Each fluctuation can cause
the segment indexes fall behind the expected real time position.
@item format_options @var{options_list}
Set container format (mp4/webm) options using a @code{:} separated list of
key=value parameters. Values containing @code{:} special characters must be
escaped.
@item global_sidx @var{global_sidx}
Write global SIDX atom. Applicable only for single file, mp4 output, non-streaming mode.
@item dash_segment_type @var{dash_segment_type}
Possible values:
@table @option
@item auto
If this flag is set, the dash segment files format will be selected based on the stream codec. This is the default mode.
@item mp4
If this flag is set, the dash segment files will be in in ISOBMFF format.
@item webm
If this flag is set, the dash segment files will be in in WebM format.
@end table
@item ignore_io_errors @var{ignore_io_errors}
Ignore IO errors during open and write. Useful for long-duration runs with network output.
@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.
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.
