Commit Graph

982 Commits

Author SHA1 Message Date
Ronald S. Bultje 8283f90a52 swscale: handle unaligned buffers in yuv2plane1
The issue had been introduced in
c435653627

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2011-11-13 08:27:20 +01:00
Sean McGovern 124e56454d swscale: add padding to conversion buffer.
Altivec does unaligned reads from this buffer in
hscale_altivec_real(), and can thus read up to 16 bytes beyond
the end of the buffer. Therefore, add an extra 16 bytes of
padding at the end of the conversion buffer.

This fixes fate-lavfi-pixfmts_scale on AltiVec-enabled builds
under valgrind.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2011-11-11 07:44:35 -08:00
Ronald S. Bultje c435653627 swscale: write yuv2plane1 MMX/SSE2/SSE4/AVX functions. 2011-11-05 20:48:14 -07:00
Ronald S. Bultje 1deb08fcb6 swscale: align vertical filtersize by 2 on x86.
The vertical scaler handles 2 rows at a time and thus requires
alignment by 2, or else it'll read invalid memory and result in
corrupt output.
2011-11-05 07:06:38 -07:00
Ronald S. Bultje 9e66b892e8 swscale: add missing colons to x86 assembly yuv2planeX.
This fixes assembling using "nasm".
2011-10-23 09:44:03 -07:00
Ronald S. Bultje f48b12e0a6 swscale: update altivec yuv2planeX asm to new per-plane API. 2011-10-22 10:35:14 -07:00
Ronald S. Bultje 6cacecdca3 swscale: make yuv2yuvX_10_sse2/avx 8/9/16-bits aware.
Also implement MMX/MMX2 versions and SSE4 versions.
2011-10-22 10:35:14 -07:00
Kieran Kunhya 7fbbf95293 yuv2planeX10 SIMD
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2011-10-22 10:35:14 -07:00
Ronald S. Bultje 109f62e8f8 swscale: decide whether to use yuv2plane1/X on a per-plane basis. 2011-10-22 10:35:14 -07:00
Ronald S. Bultje f99654d470 swscale: reintroduce full precision in 16-bit output. 2011-10-22 10:35:14 -07:00
Kieran Kunhya ff7913aef1 Split up yuv2yuvX functions
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2011-10-22 10:35:13 -07:00
Kieran Kunhya 34e8d147b3 Split out yuv2yuv1 luma and chroma in order to make them generic DSP functions
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2011-10-22 10:35:13 -07:00
Mans Rullgard 41ac093f7e swscale: fix signed shift overflows in ff_yuv2rgb_c_init_tables()
Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-10-21 20:56:59 +01:00
Ronald S. Bultje dc49bf1270 sws/pixfmt/pixdesc: add support for yuv420p9le/be. 2011-10-21 00:58:01 -07:00
Ronald S. Bultje 8305041e13 swscale: prevent overflow in coefficient calculation. 2011-10-21 00:14:11 -07:00
Ronald Bultje d1d421cbc0 swscale: prevent overflow during initialization
Signed-off-by: Janne Grunau <janne-libav@jannau.net>
2011-10-18 10:29:49 +02:00
Anton Khirnov 145f741e11 AVOptions: rename FF_OPT_TYPE_* => AV_OPT_TYPE_* 2011-10-12 16:51:16 +02:00
Anton Khirnov 04de1569cd sws: support yuv444p9/10 output. 2011-10-12 08:27:30 +02:00
Ronald S. Bultje 6aa3cac6bf swscale: use aligned move for storage into temporary buffer.
The intermediate buffer is always aligned.
2011-10-11 07:50:48 -07:00
Mans Rullgard d853e571ad ppc: fix some pointer to integer casts
Use uintptr_t instead of plain int.  Without this change, the
comparisons will come out wrong for pointers in certain ranges.
Fixes random failures on ppc64.  Also fixes some compiler warnings.

Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-09-25 18:33:38 +01:00
Kieran Kunhya 4d4d0e8176 Fix unnecessary shift with 9/10bit vertical scaling
Signed-off-by: Diego Biurrun <diego@biurrun.de>
2011-09-23 02:13:30 +02:00
Ronald S. Bultje ea540401d6 swscale: fix byte overreads in SSE-optimized hscale().
SSE-optimized hScale() scales up to 4 pixels at once, so we need to
allocate up to 3 padding pixels to prevent overreads. This fixes
valgrind errors in various swscale-tests on fate.
2011-09-15 07:30:46 -07:00
Ronald S. Bultje e0c3e07387 sws: implement MMX/SSE2/SSSE3/SSE4 versions for horizontal scaling.
Speed: from 3.9x to 9.6x speed improvement over C, and some small
(up to 15%) speed improvements over existing MMX code (particularly
for bigger filters).
2011-09-13 09:53:42 -07:00
Anton Khirnov fb4ca26bdb lavf,lavc,sws: add {avcodec,avformat,sws}_get_class() functions. 2011-09-03 20:53:35 +02:00
Ronald S. Bultje 3f04ab4fcd swscale: split hScale() function pointer into h[cy]Scale().
This allows using more specific implementations for chroma/luma, e.g.
we can make assumptions on filterSize being constant, thus avoiding
that test at runtime.
2011-08-17 20:56:06 -07:00
Luca Barbato 3304a1e69a swscale: add dithering to yuv2yuvX_altivec_real
It just does that part in scalar form, I doubt using a vector store
over 2 array would speed it up particularly.

The function should be written to not use a scratch buffer.
2011-08-13 00:06:04 +02:00
Ronald S. Bultje 28c1115a91 swscale: use 15-bit intermediates for 9/10-bit scaling. 2011-08-12 11:54:25 -07:00
Stefano Sabatini 6a92551234 lsws: remove optimization debug logs in sws_init_context()
The logged information is possibly false, and it tends to be outdated
after each change since the logging code needs to be manually updated.

Simplify and prevent confusing wrong debug messages.

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2011-08-05 00:45:06 +02:00
Stefano Sabatini 347167ecf5 lsws: use array for storing the supported in/out information
Also remove the unnecessary isSupportedIn/Out macros.

Make the code more compact/readable, and simplify the access to
lsws-specific pixel format information.

Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
2011-08-05 00:44:11 +02:00
Ronald S. Bultje 62ee0e6a97 Revert "swscale: use 15-bit intermediates for 9/10-bit scaling."
This reverts commit ac0fb59348. It
causes valgrind errors which I'll want to investigate before
resubmitting this.
2011-08-02 12:27:43 -07:00
Ronald S. Bultje ac0fb59348 swscale: use 15-bit intermediates for 9/10-bit scaling. 2011-08-02 10:34:02 -07:00
Kostya Shishkov 42ba12888b Do not convert RGB buffer at once when stride does not fit exact samples.
When converting RGB format to RGB format with the same bits per sample,
unscaled path performs conversion on the whole buffer at once. For
non-multiple-of-16 BGR24 to RGB24 conversion it means that padding at the
end of line will be converted too. Since it may be of arbitrary length
(e.g. 8 bytes), operating on the whole buffer produces obviously wrong
results.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2011-07-30 09:51:23 -07:00
Joseph Artsimovich eedb1f2034 swscale: mark YUV422P10(LE,BE) as supported for output
Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-07-21 02:10:31 +01:00
Ronald S. Bultje baba2eedac swscale: extend mmx padding.
Fixes a crash when forcing libc to strictly adhere to malloc sizes.
2011-07-14 13:27:06 -07:00
Ronald S. Bultje 55eda370cb swscale: clip unscaled colorspace conversion path.
Prevents overflows on very bright scenes when adding dither,
which may lead to black dots.
2011-07-14 13:27:06 -07:00
Diego Biurrun 96c1e6d40d doxygen: Make sure parameter names match between .c and .h files. 2011-07-14 04:09:49 +02:00
Anton Mitrofanov 142e76f105 swscale: fix crash with dithering due incorrect offset calculation.
ptrdiff_t can be 4 bytes, which leads to the next element being 4-byte
aligned and thus at a different offset than intended. Forcing 8-byte
alignment forces equal offset of dither16/32 on x86-32 and x86-64.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2011-07-10 14:52:21 -07:00
Ronald S. Bultje 5c391a161a swscale: rename uv_off/uv_off2 to uv_off_px/byte. 2011-07-08 15:15:31 -07:00
Ronald S. Bultje 1ce724ee39 swscale: implement error dithering in planarCopyWrapper.
Based on a somewhat similar idea in FFmpeg's swscale.
2011-07-08 15:15:31 -07:00
Ronald S. Bultje 4e3e333a79 swscale: error dithering for 16/9/10-bit to 8-bit.
Based on a somewhat similar idea in FFmpeg's swscale copy.
2011-07-08 15:15:30 -07:00
Ronald S. Bultje 7d7bacf0f1 swscale: fix overflow in 16-bit vertical scaling.
We operated on 31-bits, but with e.g. lanczos scaling, values can
add up to beyond 0x80000000, thus leading to output of zeroes. Drop
one bit of precision fixes this.
2011-07-08 15:15:30 -07:00
Ronald S. Bultje bf2cba4532 swscale: fix crash in 8-bpc bilinear output without alpha.
We accessed the alpha array even it wasn't used and didn't
exist, hence leading to a NULL pointer segfault.
2011-07-08 15:15:30 -07:00
Ronald S. Bultje f44d50a94c swscale: fix 16-bit horizontal scaling underflow.
When using e.g. lanczos scaling, values can drop below 0, so they
should never be unsigned.
2011-07-08 15:15:30 -07:00
Ronald S. Bultje 42d622fab3 swscale: fix 16-bit scaling when output is 8-bits.
We would use the second half of the U plane buffer, rather than the
V plane buffer, to output the V plane pixels.
2011-07-08 15:15:30 -07:00
Ronald S. Bultje f28aaae1c5 swscale: amend documentation to mention use of native depth for scaling.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
2011-07-06 03:33:36 +02:00
Diego Biurrun e589e4b82d Remove unused static tables and static inline functions. 2011-07-04 14:59:35 +02:00
Mans Rullgard abc78a5a7c Do not include log.h in avutil.h
Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-07-03 21:42:07 +01:00
Mans Rullgard d49ea4afb4 Do not include pixfmt.h in avutil.h
Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-07-03 21:42:07 +01:00
Diego Biurrun 01c17c88ed doxygen: Remove spurious documentation for non-existing function parameters. 2011-07-03 18:30:02 +02:00
Ronald S. Bultje 8f440223f6 swscale: disable full_chroma_int when converting to non-24/32bpp RGB.
This functionality is only implemented for RGB24/32 and causes crashes
otherwise.
2011-07-02 08:33:52 -07:00