The use of ff_inverse speeds up slightly arches + compilers that
do not provide a division faster than the whole machinery, such
as ppc32 + gcc4.7, but has operational limits.
Drop the always-enable assert and provide a fallback.
Some fixes provided by Paul B Mahol <onemda@gmail.com>
and Michael Niedermayer <michaelni@gmx.at> and me.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Signed-off-by: Kostya Shishkov <kostya.shishkov@gmail.com>
Currently, if VIS is enabled by configure, it will also be enabled at
run-time regardless of its support in the hardware. Thus, masking VIS
usage as it is done in vis.h by constructing binary instructions is
pointless. Using normal VIS mnemonics in inline assembly allows to take
advantage of automatic register allocation, gets rid of register
variables, which are unsupported by suncc for SPARC, and improves code
readability.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
MPV_decode_mb_internal needs 3 * 16 * linesize bytes of scratch buffer
For interlaced content, linesize is multiplied by two after the allocation
of the scratch buffer, and the dest_cr pointer ends past the buffer.
This patch makes ff_mpv_frame_size_alloc allocate a total of
(aligned line_size) * 2 * 16 * 3 bytes, which suffices even for the
interlaced case.
CC:libav-stable@libav.org
Signed-off-by: Jindrich Makovicka <makovick@gmail.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Those were useful when avcodec_thread_init() was a public functions. It
was deprecated and removed some time ago, so those checks are not needed
anymore.
When JPEG-LS support is disabled the decoder would feed the
data to the JPEG Lossless decode_*_scan function resulting in
faulty decoding.
CC: libav-stable@libav.org
This is required if we return other error codes than explicitly
-1, which so far has been the only other possible return value
besides 0.
Signed-off-by: Martin Storsjö <martin@martin.st>
This replaces a large number of checks for the second field by
fixing the pointers when they are setup.
This should also fix I/BI field pictures.
Changes checksums for vc1_sa10143, the file becomes slightly closer
to what the reference decoder outputs.
Based on "vc1dec: the second field is written wrong to the picture"
by Sebastian Sandberg <sebastiand.sandberg@gmail.com>.
Signed-off-by: Martin Storsjö <martin@martin.st>
Also eliminate an unnecessary cast. Fixes the following warning:
jpeg2000dec.c:1200:20: warning: format ‘%lX’ expects argument of type ‘long unsigned int’, but argument 5 has type ‘long long unsigned int’
This was due to a miscomputation of s->cur_channel, which led to
psy-based encoders using the psy coefficients for the wrong channel.
Signed-off-by: Martin Storsjö <martin@martin.st>
This fixes a case where multichannel bitrate isn't accurately
targetted by psy model alone, never achieving the target bitrate.
Signed-off-by: Martin Storsjö <martin@martin.st>
