Fixes trac ticket 3226. Also see Andreas' analysis in
https://bugs.debian.org/801745, which was very helpful.
(cherry picked from commit 52f84d82bdf1851ecfcc412c1719e5f6f3396209)
Also fix typo found by Lou Logan:
Sacrifying -> Sacrificing
Reviewed-by: Lou Logan <lou@lrcd.com>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
(cherry picked from commit 28efeb6502df37c99ccf18575d2552381d83a877)
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
This reverts commit 599888a48040f4086c9086aae958b7231b37afcc.
The commit does not silence the warning on ELF-based systems, and will be
fixed in the subsequent commit.
Conflicts:
libavcodec/x86/fft_mmx.asm
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
* rbultje/vp9-bugfixes:
vp9: match another find_ref_mvs() bug in libvpx.
vp9: fix scaled motion vector clipping for sub8x8 blocks.
vp9: improve signbias check.
vp9: don't allow compound references if error_resilience is enabled.
vp9: clamp segmented lflvl before applying ref/mode deltas.
vp9: reset loopfilter mode/ref deltas on keyframe.
vp9: fix crash when playing back 440/440 content with width%64<56.
vp9: extend loopfilter workaround for vp9 h/v mix-up to work for 422.
vp9: clip motion vectors in the same way as libvpx does.
vp9: set skip flag if the block had no coded coefficients.
vp9: apply mv scaling workaround only when subsampling is enabled.
vp9: read all 4x4 blocks in sub8x8 blocks individually with scalability.
vp9: fix segmentation map referencing upon framesize change.
vp9: disable more pmulhrsw optimizations in idct16/32.
vp9: disable all pmulhrsw in 8/16 iadst x86 optimizations.
Merged-by: Michael Niedermayer <michaelni@gmx.at>
See sample vp90-2-14-resize-fp-tiles-16-8-4-2-1.webm from the vp9 test
vector set which reproduces the issue. This probably costs a few cycles,
but I don't think there's an easy way to workaround that.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
When the decoder is deactivated, the x86-optimized versions are
not compiled, resulting in a link error.
The C version is unaffected, as it is part of the idctdsp
subsystem.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
The main difference consists in renaming properly labels, and
letting yasm select the gprs for skipping 1D transforms.
Previous-version-reviewed-by: James Almer <jamrial@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Based on patch by Francisco Blas Izquierdo Riera
Commit message partly taken from carl
fixes a compilation
error in mlpdsp_init.c with -fstack-check and some gcc compilers (I
reproduced the issue with gcc 4.7.3) by simplifying the code.
See also https://bugs.gentoo.org/show_bug.cgi?id=471756
$ make libavcodec/x86/mlpdsp_init.o
libavcodec/x86/mlpdsp_init.c: In function ‘mlp_filter_channel_x86’:
libavcodec/x86/mlpdsp_init.c:142:5: error: can’t find a register in
class ‘GENERAL_REGS’ while reloading ‘asm’
libavcodec/x86/mlpdsp_init.c:142:5: error: ‘asm’ operand has impossible
constraints
4551 -> 4509 dezicycles
Reviewed-by: Ramiro Polla <ramiro.polla@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
The 3*stride value stored in r3src can be loaded much later,
so use r3src instead of a dedicated gpr when possible.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
The second stride is always the internal buffer one, MAX_PB_SIZE (times 2 to
get the value in bytes).
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
