The function is used on unaligned buffers (such as those provided
by AVPacket), accessing them as uint16_t causes SIGBUS crashes on
architectures like SPARC.
This fixes ubsan runtime error: load of misaligned address for type
'const uint16_t', which requires 2 byte alignment
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Fixes out of array accesses
Fixes: 0cf176e6d3ab9fe924f39738e513f547/asan_generic_4a54aa_3431_aaa28be1cb32e307a9890cad06f84fba.avi
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Fixes out of array access
Fixes: 0cf176e6d3ab9fe924f39738e513f547/asan_generic_4a54aa_3431_aaa28be1cb32e307a9890cad06f84fba.avi
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
Sometimes rollup captions can move around the screen. This fixes "ghost"
captions from below the current rollup area from continuing to be
captured when a rollup moves higher up on the screen.
This new mode is useful for realtime decoding of closed captions so they
can be display along with mpeg2 frames.
Closed caption streams contain two major types of captions:
- POPON captions, which are buffered off-screen and displayed
only after EOC (end of caption, aka display buffer)
- PAINTON/ROLLUP captions, which are written to the display as soon as
they arrive.
In a typical real-time eia608 decoder, commands like EOC (end of
caption; display buffer), EDM (erase display memory) and EBM (erase
buffered memory) perform their expected functions as soon as the
commands are processed. This is implemented in the real_time branches
added in this commit.
Before this commit, and in the !real_time branches after this commit,
the decoder cleverly implements its own version of the decoder which is
specifically geared towards buffered decoding. It does so by actively
ignoring commands like EBM (erase buffered memory), and then re-using
the non-display buffer to hold the previous caption while the new one is
received. This is the opposite of the real-time decoder, which uses the
non-display buffer to hold the new caption while the display buffer is
still showing the current caption.
In addition to ignoring EBM, the buffered decoder also has custom
implementations for EDM and EOC. An EDM (erase display memory) command
flushes the existing contents before clearing the screen, and EOC
similarly always flushes the active buffer (the previous subtitle)
before flipping buffers.
This is quite an accurate approximation; testing shows ~ 2ulp error in
the floating point result. Tested with FATE.
Alternatively, if one wants "full accuracy", one can use powf, or sqrt
instead of sqrtf. With powf, one gets 1 ulp error (theoretically should be 0, as
0.75 is exactly representable) on GNU libm, with sqrt, 0 ulp error.
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Abstracted into pos_pow34 utility function
Signed-off-by: Claudio Freire <klaussfreire@gmail.com>
src and dst are only 8-bit-aligned, so accessing them as uint16_t causes
SIGBUS crashes on architectures like sparc.
This fixes ubsan runtime error: load of misaligned address for type
'const uint16_t', which requires 2 byte alignment
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
Fixes out of array read
Fixes: 04442da73d935b776d2236282588d4f9/signal_sigsegv_2625a69_8790_ae85ffc889070663319b3417ede777b0.mov
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This avoids closing and opening the bit reader
Reviewed-by: Rostislav Pehlivanov <atomnuker@gmail.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
All MDCT outputs must be checked in case of 128point MDCTs
Fixes: out of array read
Fixes: 04442da73d935b776d2236282588d4f9/signal_sigsegv_2625a69_351_52ca6226eb83547a2d26e322ce84ed84.mov
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This is used to check if the input buffer is larger enough, so if this
overflows it can cause a false negative leading to a segmentation fault
in bytestream2_get_bufferu.
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
More don't fit into the integer output.
Also use get_bits_long, since get_bits only supports reading up to 25
bits, while get_bits_long supports the full integer range.
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Use the ability to invert phase with ms_mask instead of changing
the codebook when possible, to avoid having to switch codebooks
if some bands are INTENSTY_BT and others are INTENSITY_BT2, since
usually a set ms_mask uses less bits that a codebook change. While
it may not always be a win (ie: if it causes an ms_mask bitmap
to be sent when it wouldn't have been otherwise), it's unlikely
since the ms_mask bitmap will almost always be there already for
M/S itself.
The relative error between two encoding strategies is the simple
difference of rate-distortion values, and not the absolute
difference. An absolute measure would allow worsening of the
quantization error as well as improving.
1. Fix sf_idx and band_type addressing to address only the first
subwindow in the group (others could hold garbage values)
2. Don't step on ms_mask when is_mask is set. I/S selection
already sets the ms_mask properly and shouldn't be overridden.
3. Use mid/sid cb/sf when computing coding error, as should be
since those are the cb/sfs that will eventually be set.
4. Fix distortion computation on multi-subwindow groups (was
subtracting the bits terms multiple times)
5. Clear ms_mask when one side uses PNS and the other doesn't.
When using PNS, ms_mask signals correlated noise, which can be
detected just like regular M/S detection, so we don't skip
noise bands, but when only one side uses PNS setting the flag
can confuse some encoders, so avoid that.
Correct addressing of sf_idx and band_type arrays in I/S code.
Both arrays don't guarantee valid values for subwindows at all
times, depending on which coder is being used, which could result
in reading of garbage values. Instead, only the first subwindow
in the window group has to be used, ie: use w*16+g instead of
(w+w2)*16+g
Change the condition for application of the M/S transform to match
that of the decoder. Namely, that no special coding books must be
in use in either channel. While the condition ought to be
equivalent to the current one when the invariant of is_mask is
kept, matching the decoder's condition is safer and easier to
maintain.
Fixes an assertion error reported in #2686 that happens when
using prediction (either explicitly or implicitly by setting
the AAC main profile), since prediction code would allow
creating new zeroes or removing existing ones, without
properly checking for SF delta violations.
This patch forbids creating/removing zeroes, perhaps an
overly conservative approach, but a safe one. More permissive
and sophisticated approaches may be attempted in the future.
since:
cbcc88c libvpx: Support setting color range for vp9.
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Zern <jzern@google.com>
When there is a non-zero decoding delay due to reordering, the first dts
should be lower than the first pts (since the first packet fed to the
decoder does not produce any output).
Use the same scheme used in mpegvideo_enc (which comes from x264
originally) -- wait for first two timestamps and extrapolate linearly to
the past to produce the first dts value.
When B-frames are enabled and the encoder returns success, all currently
pending buffers immediately become valid and can be returned to the
caller. We can only return one packet at a time, so all the other
pending buffers should be transferred to a new 'ready' fifo, from where
they can be returned in subsequent calls (in which the encoder does not
produce any new output). This bug was hidden by the incorrect testing of
the encoder return value (the return value was overwritten before it was
tested).
Fixes assertion failure
Fixes out of memory access
Fixes: test_casex.ivf
Found-by: Tyson Smith <twsmith@mozilla.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
The transformation to bytes must happen after alignment to get the same
resulting pointers as before.
This fixes segmentation faults in the assembler code.
The regression was introduced in commit 9553689.
Reviewed-by: Kieran Kunhya <kierank@obe.tv>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
This exploits an approach based on the sieve of Eratosthenes, a popular
method for generating prime numbers.
Tables are identical to previous ones.
Tested with FATE with/without --enable-hardcoded-tables.
Sample benchmark (Haswell, GNU/Linux+gcc):
prev:
7860100 decicycles in cbrt_tableinit, 1 runs, 0 skips
7777490 decicycles in cbrt_tableinit, 2 runs, 0 skips
[...]
7582339 decicycles in cbrt_tableinit, 256 runs, 0 skips
7563556 decicycles in cbrt_tableinit, 512 runs, 0 skips
new:
2099480 decicycles in cbrt_tableinit, 1 runs, 0 skips
2044470 decicycles in cbrt_tableinit, 2 runs, 0 skips
[...]
1796544 decicycles in cbrt_tableinit, 256 runs, 0 skips
1791631 decicycles in cbrt_tableinit, 512 runs, 0 skips
Both small and large run count given as this is called once so small run
count may give a better picture, small numbers are fairly consistent,
and there is a consistent downward trend from small to large runs,
at which point it stabilizes to a new value.
Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Ganesh Ajjanagadde <gajjanagadde@gmail.com>
Fixes index out of bounds error
Fixes: aac_index_out_of_bounds.wmv
Found-by: Piotr Bandurski <ami_stuff@o2.pl>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
