This completes the conversion of h264dsp to yasm; note that h264 also
uses some dsputil functions, most notably qpel. Performance-wise, the
yasm-version is ~10 cycles faster (182->172) on x86-64, and ~8 cycles
faster (201->193) on x86-32.
* qatar/master: (35 commits)
h264_idct_10bit: port x86 assembly to cpuflags.
x86inc: clip num_args to 7 on x86-32.
x86inc: sync to latest version from x264.
fft: rename "z" to "zc" to prevent name collision.
wv: return meaningful error codes.
wv: return AVERROR_EOF on EOF, not EIO.
mp3dec: forward errors for av_get_packet().
mp3dec: remove a pointless local variable.
mp3dec: remove commented out cruft.
lavfi: bump minor to mark stabilizing the ABI.
FATE: add tests for yadif.
FATE: add a test for delogo video filter.
FATE: add a test for amix audio filter.
audiogen: allow specifying random seed as a commandline parameter.
vc1dec: Override invalid macroblock quantizer
vc1: avoid reading beyond the last line in vc1_draw_sprites()
vc1dec: check that coded slice positions and interlacing match.
vc1dec: Do not ignore ff_vc1_parse_frame_header_adv return value
configure: Move parts that should not be user-selectable to CONFIG_EXTRA
lavf: remove commented out cruft in avformat_find_stream_info()
...
Conflicts:
Makefile
configure
libavcodec/vc1dec.c
libavcodec/x86/h264_deblock.asm
libavcodec/x86/h264_deblock_10bit.asm
libavcodec/x86/h264dsp_mmx.c
libavfilter/version.h
libavformat/mp3dec.c
libavformat/utils.c
libavformat/wv.c
libavutil/x86/x86inc.asm
Merged-by: Michael Niedermayer <michaelni@gmx.at>
Without this, cglobal will expand "z" to "zh" to access the high byte
in a register's word, which causes a name collision with the ZH(x) macro
further up in this file.
* qatar/master:
proresdsp: port x86 assembly to cpuflags.
lavr: x86: improve non-SSE4 version of S16_TO_S32_SX macro
lavfi: better channel layout negotiation
alac: check for truncated packets
alac: reverse lpc coeff order, simplify filter
lavr: add x86-optimized mixing functions
x86: add support for fmaddps fma4 instruction with abstraction to avx/sse
tscc2: fix typo in array index
build: use COMPILE template for HOSTOBJS
build: do full flag handling for all compiler-type tools
eval: fix printing of NaN in eval fate test.
build: Rename aandct component to more descriptive aandcttables
mpegaudio: bury inline asm under HAVE_INLINE_ASM.
x86inc: automatically insert vzeroupper for YMM functions.
rtmp: Check the buffer length of ping packets
rtmp: Allow having more unknown data at the end of a chunk size packet without failing
rtmp: Prevent reading outside of an allocate buffer when receiving server bandwidth packets
Conflicts:
Makefile
configure
libavcodec/x86/proresdsp.asm
libavutil/eval.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master:
libopenjpeg: support YUV and deep RGB pixel formats
Fix typo in v410 decoder.
vf_yadif: unset cur_buf on the input link.
vf_overlay: ensure the overlay frame does not get leaked.
vf_overlay: prevent premature freeing of cur_buf
Support urlencoded http authentication credentials
rtmp: Return an error when the client bandwidth is incorrect
rtmp: Return proper error code in handle_server_bw
rtmp: Return proper error code in handle_client_bw
rtmp: Return proper error codes in handle_chunk_size
lavr: x86: add missing vzeroupper in ff_mix_1_to_2_fltp_flt()
vp8: Replace x*155/100 by x*101581>>16.
vp3: don't use calls to inline asm in yasm code.
x86/dsputil: put inline asm under HAVE_INLINE_ASM.
dsputil_mmx: fix incorrect assembly code
rtmp: Factorize the code by adding handle_invoke
rtmp: Factorize the code by adding handle_chunk_size
rtmp: Factorize the code by adding handle_ping
rtmp: Factorize the code by adding handle_client_bw
rtmp: Factorize the code by adding handle_server_bw
Conflicts:
libavcodec/libopenjpegdec.c
libavcodec/x86/dsputil_mmx.c
libavfilter/vf_overlay.c
libavformat/Makefile
libavformat/version.h
Merged-by: Michael Niedermayer <michaelni@gmx.at>
Mixing yasm and inline asm is a bad idea, since if either yasm or inline
asm is not supported by your toolchain, all of the asm stops working.
Thus, better to use either one or the other alone.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
In ff_put_pixels_clamped_mmx(), there are two assembly code blocks.
In the first block (in the unrolled loop), the instructions
"movq 8%3, %%mm1 \n\t", and so forth, have problems.
From above instruction, it is clear what the programmer wants: a load from
p + 8. But this assembly code doesn’t guarantee that. It only works if the
compiler puts p in a register to produce an instruction like this:
"movq 8(%edi), %mm1". During compiler optimization, it is possible that the
compiler will be able to constant propagate into p. Suppose p = &x[10000].
Then operand 3 can become 10000(%edi), where %edi holds &x. And the instruction
becomes "movq 810000(%edx)". That is, it will stride by 810000 instead of 8.
This will cause a segmentation fault.
This error was fixed in the second block of the assembly code, but not in
the unrolled loop.
How to reproduce:
This error is exposed when we build using Intel C++ Compiler, with
IPO+PGO optimization enabled. Crashed when decoding an MJPEG video.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
In file libavcodec/x86/dsputil_mmx.c, function ff_put_pixels_clamped_mmx(), there are two assembly code blocks. In the first block (in the unrolled loop), the instructions "movq 8%3, %%mm1 \n\t" etc have problem.
For above instruction, it is clear what the programmer wants: a load from p + 8. But this assembly code doesn’t guarantee that. It only works if the compiler puts p in a register to produce an instruction like this: “movq 8(%edi), %mm1”. During compiler optimization, it is possible that the compiler will be able to constant propagate into p. Suppose p = &x[10000]. Then operand 3 can become 10000(%edi), where %edi holds &x. And the instruction becomes “movq 810000(%edx)”. That is, it will stride by 810000 instead of 8.
This will cause the segmentation fault.
This error was fixed in the second block of the assembly code, but not in the unrolled loop.
How to reproduce:
This error is exposed when we build the ffmpeg using Intel C++ Compiler, IPO+PGO optimization. The ffmpeg was crashed when decoding a mjpeg video.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master:
v410dec: Implement explode mode support
zerocodec: fix direct rendering.
wav: init st to NULL to avoid a false-positive warning.
wavpack: set bits_per_raw_sample for S32 samples to properly identify 24-bit
h264: refactor NAL decode loop
RTMPTE protocol support
RTMPE protocol support
rtmp: Add ff_rtmp_calc_digest_pos()
rtmp: Rename rtmp_calc_digest to ff_rtmp_calc_digest and make it global
swscale: add missing HAVE_INLINE_ASM check.
lavfi: place x86 inline assembly under HAVE_INLINE_ASM.
vc1: Add a test for interlaced field pictures
swscale: Mark all init functions as av_cold
swscale: x86: Drop pointless _mmx suffix from filenames
lavf: use conditional notation for default codec in muxer declarations.
swscale: place inline assembly bilinear scaler under HAVE_INLINE_ASM.
dsputil: ppc: cosmetics: pretty-print
dsputil: x86: add SHUFFLE_MASK_W macro
configure: respect CC_O setting in check_cc
Conflicts:
Changelog
configure
libavcodec/v410dec.c
libavcodec/zerocodec.c
libavformat/asfenc.c
libavformat/version.h
libswscale/utils.c
libswscale/x86/swscale.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master:
Print full compiler identification, not only version number
flacdec: reverse lpc coeff order, simplify filter
x86: dsputil: drop some unused CPU flag debug code
Conflicts:
cmdutils.c
configure
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master:
ppc: fix build with altivec disabled
vp3: move idct and loop filter pointers to new vp3dsp context
build: add CONFIG_VP3DSP, reduce repetition in OBJS lists
tscc2: do not add/subtract 128 bias during DCT
tscc2: fix typo in DCT
configure: clarify external library section of help output
configure: mark libfdk-aac as nonfree
configure: cosmetics: drop some unnecessary backslashes
os_support: K&R formatting cosmetics
Conflicts:
configure
libavcodec/vp3.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
This moves all VP3-specific function pointers from dsputil to a
new vp3dsp context. There is no reason to ever use the VP3 IDCT
where an MPEG2 IDCT is expected or vice versa.
Signed-off-by: Mans Rullgard <mans@mansr.com>
* qatar/master:
mxfdec: replace x>>av_log2(sizeof(..)) by x/sizeof(..).
x86: h264_intrapred: Don't add the 'd' suffix to the SPLATB_REG macro
Conflicts:
libavformat/mxfdec.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
The SPLATB_REG macro already adds the 'd' suffix internally.
This fixes building on Win64, which has been broken since 878e66902.
This worked for unix, where r2 happened to be rdx in this case, which
with the first suffix rdxd was mapped to eax, and eaxd is defined back
to eax. On win64 however, r2 happened to be R8 in this case, and
R8d mapps to R8D just fine, but there's no mapping for R8Dd to anything.
Signed-off-by: Martin Storsjö <martin@martin.st>
* qatar/master:
qdm2: remove broken and disabled dump_context() debug function
x86: h264_intrapred: use newly introduced SPLAT* and PSHUFLW macros
x86inc: add SPLATB_LOAD, SPLATB_REG, PSHUFLW macros
x86inc: modify ALIGN to not generate long nops on i586
x86: h264_intrapred: port to cpuflag macros
avplay: update input filter pointer when the filtergraph is reset.
avconv: fix parsing of -force_key_frames option.
h264: use templates to avoid excessive inlining
xtea: Make the count parameter match the documentation
blowfish: Make the count parameter match the documentation
mpegvideo: Don't use ff_mspel_motion() for vc1
xtea: invert branch and loop precedence
blowfish: invert branch and loop precedence
flvdec: optionally trust the metadata
avconv: Set audio filter time base to the sample rate
vp8: Add ifdef guards around the sse2 loopfilter in the sse2slow branch too
Conflicts:
ffmpeg.c
ffplay.c
libavcodec/h264.c
libavcodec/mpegvideo_common.h
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master: (29 commits)
lavfi: reclassify showfiltfmts as a TESTPROG
graph2dot: fix printf format specifier
swscale: yuv2planeX 8bit >=sse2 functions need aligned stack on x86-32.
vp8: loopfilter >=sse2 functions need aligned stack on x86-32.
amr: remove shift out of the AMR_BIT() macro.
dsputilenc: group yasm and inline asm function pointer assignment.
mov: use forward declaration of a function instead of a table.
Clarify Doxygen comment for FF_API_* #defines.
configure: simplify get_version()
Create version.h headers for libraries that lack them
gitignore: Use full path instead of relative path to specify patterns
mpegvideo: remove VLAs
Add XTEA encryption support in libavutil
Add Blowfish encryption support in libavutil
eval: Add the isinf() function and tests for it
flacdec: move lpc filter to flacdsp
flacdec: split off channel decorrelation as flacdsp
avplay: Add an option for not limiting the input buffer size
FATE: add a test for WMA cover art.
FATE: add a test for apetag cover art
...
Conflicts:
.gitignore
configure
ffplay.c
libavcodec/Makefile
libavcodec/error_resilience.c
libavcodec/mpegvideo.c
libavcodec/ratecontrol.c
libavdevice/avdevice.h
libavfilter/Makefile
libavfilter/filtfmts.c
libavfilter/version.h
libavformat/mov.c
libavformat/version.h
libavutil/Makefile
libavutil/avutil.h
libavutil/version.h
libswscale/swscale.h
libswscale/x86/swscale_mmx.c
tests/fate/libavutil.mak
tests/lavfi-regression.sh
tools/graph2dot.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
* qatar/master:
dsputilenc_mmx: split assignment of ff_sse16_sse2 to SSE2 section.
dnxhdenc: add space between function argument type and comment.
x86: fmtconvert: add special asm for float_to_int16_interleave_misc_*
attributes: Add a definition of av_always_inline for MSVC
cmdutils: Pass the actual chosen encoder to filter_codec_opts
os_support: Add fallback definitions for stat flags
os_support: Rename the poll fallback function to ff_poll
network: Check for struct pollfd
os_support: Don't compare a negative number against socket descriptors
os_support: Include all the necessary headers for the win32 open function
x86: vc1: fix and enable optimised loop filter
Conflicts:
cmdutils.c
cmdutils.h
ffmpeg.c
ffplay.c
libavformat/os_support.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
The problem is that the ssse3 psign instruction does the wrong
thing here. Commit ea60dfe incorrectly removed a macro emulating
this instruction for pre-ssse3 code. However, the emulation is
incorrect, and the code relies on the behaviour of the macro.
Specifically, the psign sets destination elements to zero where
the corresponding source element is zero, whereas the emulation
only negates destination elements where the source is negative.
Furthermore, the PSIGNW_MMX macro in x86util.asm is totally bogus,
which is why the original VC-1 code had an additional right shift
when using it. Since the psign instruction cannot be used here,
skip all the macro hell and use the working instruction sequence
directly.
None of this was noticed due a stray return statement in
ff_vc1dsp_init_mmx() which meant that only the mmx version of the
loop filter was ever used (before being removed in ea60dfe).
Signed-off-by: Mans Rullgard <mans@mansr.com>
* qatar/master:
mss1: validate number of changeable palette entries
mss1: report palette changed when some additional colours were decoded
x86: fft: replace call to memcpy by a loop
udp: Support IGMPv3 source specific multicast and source blocking
dxva2: include dxva.h if found
libm: Provide fallback definitions for isnan() and isinf()
tcp: Pass NULL as hostname to getaddrinfo if the string is empty
tcp: Set AI_PASSIVE when the socket will be used for listening
Conflicts:
configure
libavcodec/mss1.c
libavformat/udp.c
Merged-by: Michael Niedermayer <michaelni@gmx.at>
The function call was a mess to handle, and memcpy cannot make
the assumptions we do in the new code.
Tested on an IMC sample: 430c -> 370c.
Signed-off-by: Mans Rullgard <mans@mansr.com>
