Due to non-byte-alignment a read of 32 bits guarantees only
25 usable bits; therefore get_bits_long() (which is used to
potentially read more than 25 bits) performs two reads in case
it needs to read more than 25 bits and combines the result.
Yet this is not necessary: One can just read 64 bits at a time
to get 32 usable bits (57 would be possible). This commit does so.
This reduced the size of .text by 30144B for GCC 11.4 and 5648B
for Clang 14 (both with -O3).
(get_bits_long() is a building block of show_bits_long()
and get_ue_golomb_long(), so this patch affects these, too.)
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
They are currently non-const for reasons unknown, although
avio_write() accepts a const buffer.
Reviewed-by: Anton Khirnov <anton@khirnov.net>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is undefined behaviour even in cases where it works
(it works because both are pointers). Instead change
the functions involved to use the type expected by the AVIO-API
and add inline wrappers for our internal callers.
Reviewed-by: Tomas Härdin <git@haerdin.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
It is only included because two very rarely used functions
use pointers to URLContexts; use struct URLContext instead.
Also move ffio_geturlcontext() so that one can avoid
a forward declaration of struct URLContext (which would be
necessary as soon as FF_API_AVIODIRCONTEXT is no more).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Incompatibility of the flags and the protocol's capabilities
are checked generically (see url_alloc_for_protocol()).
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
A switch is simpler than a lookup over a table with
three entries, only two of which can happen at all.
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes: fc429d785e
cpb_cnt used to be initialized to 1 before
fc429d785e so cpb_cnt_minus1 should be
initialized to 0.
Also add +1 to the decode_sublayer_hrd call to account for the change to
the offset
Broken in 0c6e5f321b.
Also add it to decklink_enc.cpp in order not to rely
on an implicit inclusion via libavfilter/ccfifo.h.
Reviewed-by: あんこ <mailcrystaldiskinfo@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Fixes many warnings when using -Wcast-qual.
Reviewed-by: Tomas Härdin <git@haerdin.se>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
And remove the AVOID_PROBING flag, given it's the last av1 decoder to be tested
either way.
This fixes a regression introduced in 1652f2492f,
where even if forcing the native av1 decoder, if another decoder was present,
like libdav1d or libaom-av1, they'd be used for probing and some fate tests
would have different results.
Signed-off-by: James Almer <jamrial@gmail.com>
Fixes: out of array access
Fixes: 59999/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_MPEG4_fuzzer-5767982157266944
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
There is likely a better way to fix this, this is mainly to show the problem
Fixes: MC within same frame resulting in overlapping memcpy()
Fixes: 60189/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HEVC_fuzzer-4992746590175232
Fixes: 61753/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HEVC_fuzzer-5022150806077440
Fixes: 58062/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_HEVC_fuzzer-4717458841010176
Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
These casts cast const away temporarily; they are safe, because
the pointers that are initialized point to const data. But this
is nevertheless not nice and leads to warnings when using
-Wcast-qual. blend_modes.c generates 546 (2*39*7) such warnings
which is the majority of such warnings for FFmpeg as a whole.
vf_blend.c and vf_blend_init.h also use this pattern;
they have also been changed.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
lzma_stream.next_in is const for more than 15 years now
and has been so in every release of liblzma.
Reviewed-by: Paul B Mahol <onemda@gmail.com>
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Andreas Rheinhardt
Paul B Mahol
llyyr
Michael Niedermayer
James Almer
Lynne