The pixel format is not known until the frame header is parsed.
Guessing it here only causes trouble for the caller if the guess
turns out to be wrong (and actually causes very wrong output by
avconv/avplay).
Signed-off-by: Mans Rullgard <mans@mansr.com>
The low 32 bits of a multiplication are the same for signed
and unsigned operands. Casting to unsigned before multiplying
is thus equivalent while avoiding signed overflow, which is
undefined by the C99 standard.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Making 'samples' a pointer to uint8_t simplifies the DECODE
macro and reduces the amount of type casting overall.
This also fixes some signed overflows on left shift.
Signed-off-by: Mans Rullgard <mans@mansr.com>
This function intentionally overflows the signed range on
the left shift. Using this type-punning avoids errors from
the overflow checker without disabling this test globally.
Signed-off-by: Mans Rullgard <mans@mansr.com>
This code relies on the result wrapping as for unsigned
values, and the sign is not used. Thus an unsigned type
is proper here.
Signed-off-by: Mans Rullgard <mans@mansr.com>
The bit_rate_value_minus1 and cpb_size_value_minus1 elements
allow a wider range than get_ue_golomb() supports. This
adds a get_ue_golomb_long() function supporting up to 31
leading zeros, which is the maximum for these syntax
elements, and uses it in decode_hrd_parameters().
Signed-off-by: Mans Rullgard <mans@mansr.com>
The level_code expression includes a shift which is invalid in
those cases where the value is not used. Moving the calculation
to the branch where the result is used avoids these.
Signed-off-by: Mans Rullgard <mans@mansr.com>
These additions might overflow the signed range for large
input values. Converting to unsigned before the addition
rather than after avoids such undefined behaviour. The
result under normal two's complement wraparound remains
unchanged.
Signed-off-by: Mans Rullgard <mans@mansr.com>
The PPS may contain a few trailing elements whose presence is
only signalled by data remaining after the the mandatory part
has been parsed. The current code fails to take into account
the rbsp_trailing_bits() when deciding whether to parse these
optional elements. Assuming no unnecessary padding bytes are
passed to this function, the optional elements are present if
either more than 8 extra bits remain or the remaining bits do
not form a valid rbsp_trailing_bits() after the mandatory PPS
elements have been parsed.
Signed-off-by: Mans Rullgard <mans@mansr.com>
The way this value is used, it should be an unsigned type.
While the numerical value has no meaning, unsigned wraparound
is relied upon.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Keeping byte values read from the file as unsigned is consistent
with how they are subsequently used and avoids an undefined left
shift by 24 when bit 7 is set.
Signed-off-by: Mans Rullgard <mans@mansr.com>
This fixes a signed overflow from i << 24 when i == 255 by
making i unsigned. The result of the shift is already
assigned to an variable of unsigned type.
Signed-off-by: Mans Rullgard <mans@mansr.com>
On the first iteration through this code, last_dts is always
INT64_MIN (AV_NOPTS_VALUE) and the subtraction overflows in
an invalid manner. Although the result is only used if the
input values are valid, performing the subtraction is still
not allowed in a strict environment.
Signed-off-by: Mans Rullgard <mans@mansr.com>
In addition to avoiding undefined behaviour, an unsigned type
makes more sense for packing multiple 8-bit values.
Signed-off-by: Mans Rullgard <mans@mansr.com>
If flush_put_bits() is called when the 32-bit buffer is empty,
e.g. after writing a multiple of 32 bits, and invalid shift by
32 is performed. Since flush_put_bits() is called infrequently,
this additional check should have negligible performance impact.
Signed-off-by: Mans Rullgard <mans@mansr.com>
The inline asm added in bf5d46d uses the 'y' modifier which
is only supported from gcc 4.5. This check allows building
with older compilers.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Writing zeros to the high entries in the array need only be
done once as the cutoff position is constant throughout the
loop.
Signed-off-by: Mans Rullgard <mans@mansr.com>
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>
On 32-bit ppc, the GOT pointer must be loaded manually.
This adds a "get_got" assembler macro to compute the
GOT address. The "movrel" macro is updated to take an
additional parameter containing the GOT address since
no register is reserved for this purpose on ppc32.
These changes have no effect on ppc64 builds.
Signed-off-by: Mans Rullgard <mans@mansr.com>
If the demuxer did not set a codec_tag, there is none and
inventing one makes no sense. This change stops the rawvideo
"decoder" over-writing user-supplied pixfmt with one derived
from the codec_tag. The pixfmt-codec_tag-pixfmt round-trip
is lossy since several pixfmts map to the same codec_tag.
This fixes fate-lavf-pixfmt with avfilter disabled.
Signed-off-by: Mans Rullgard <mans@mansr.com>
This fixes some nonsensical code by moving some should-be-shared
code to separate functions and using these.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Inspection of compiled code shows gcc handles these fine on its own.
Benchmarking also shows no measurable speed difference.
Removing the remaining cases in get_cabac_bypass_sign_x86() does
cause more substantial changes to the compiled code with uncertain
impact.
Signed-off-by: Mans Rullgard <mans@mansr.com>
The functions and tables initialised in this function rely on an
initialised DSPContext. Make sure they always have one.
Signed-off-by: Mans Rullgard <mans@mansr.com>
The upcoming gcc 4.7 has more advanced constant propagation
resulting some inline asm operands becoming constants and thus
emitted as literals, sometimes in contexts where this results
in invalid instructions.
This patch changes the constraints of the relevant operands
to "rm" thus forcing a valid type. While obviously suboptimal,
this is what older gcc versions already did, and there is no
change to the code generated with these.
Signed-off-by: Mans Rullgard <mans@mansr.com>
High bitdepth H.264 needs 32-bit transform coefficients, whereas
dnxhd does not. This creates a conflict with the templated
functions operating on DCTELEM data. This patch adds a field
allowing the caller to choose the element size in dsputil_init()
and adds the required functions.
Signed-off-by: Mans Rullgard <mans@mansr.com>
This ensures the linker picks the just built libraries even
if LDFLAGS for some reason contains -L flags pointing at
other directories containing libav libraries.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Use of these has been broken ever since the h264 idct was changed
to always use transposed inputs. Furthermore, they were only
ever used if some *other* non-default idct was requested.
Signed-off-by: Mans Rullgard <mans@mansr.com>
The change to LOCAL_ALIGNED means the declared object must be an
array and the subsequent test should not use the & operator.
Noticed by Uoti Urpala.
Signed-off-by: Mans Rullgard <mans@mansr.com>
This variable is set to the same value for all directories.
Adding the -L flags directly to LDFLAGS is simpler and achieves
the same thing.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Declaring tools associated with each library in their respective
makefiles allows these tools to easily depend on the correct
prerequisites and link against the libs they need.
Signed-off-by: Mans Rullgard <mans@mansr.com>
The output from -v with gcc 4.6 has changed such that the search
pattern matches too soon without making it more strict.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Mans Rullgard
