Further performance improvements and security fixes by
Vittorio Giovara, Luca Barbato and Diego Biurrun.
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Additional fixes and enhancements by Vittorio Giovara, Gonzalo Garramuno,
Nicolas George, Paul B Mahol and Michael Niedermayer.
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
Build 13 has a crucial fix which makes x265_encode_headers work
as its always been documented.
Previous to this, the check was incorrect.
Also allows for use of the repeat headers flag.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
The DCBZL instruction is not available for the e500v1 and e500v2
architectures, but may still be recognized by the toolchain, so we
need to explicitly disable it for these architectures.
References: PowerPC™ e500 Core Family Reference Manual (Freescale)
Found-by: Ståle Kristoffersen <staalebk@ifi.uio.no>
Signed-off-by: Vittorio Giovara <vittorio.giovara@gmail.com>
The values of {FLT,DBL}_{MAX,MIN} macros on some systems (older musl
libc, some BSD flavours) are not exactly representable, i.e.
(double)DBL_MAX == DBL_MAX is false
This violates (at least some interpretations of) the C99 standard and
breaks code (e.g. in vf_fps) like
double f = DBL_MAX;
[...]
if (f == DBL_MAX) { // f has not been changed yet
[....]
}
Not guaranteed to be in nanosecond resolution. On iOS 7 the duration
of one tick is 125/3 ns which is still more than an order of magnitude
better then microseconds.
Replace decicycles with the neutral UNITS. Decicycles is strange but
tenths of a nanosecond and unspecific "deci"-ticks for mach_absolute_time
is just silly.
The SVQ3 decoder reuses large parts of the H.264 decoder so it
makes no sense to enable the former but not the latter.
Also drop unnecessary h263.o object from SVQ3 decoder object list.
The s390 architecture requires shared libraries to be built in PIC mode.
Otherwise applications will get wrong relocations at run-time, leading
to confusing segmentation faults.
CC: libav-stable@libav.org
A threading type might be detected originally, but later disabled
if one of its dependencies is unavailable.
This makes sure that the threading support item in the configure
output is right for setups where w32threads are available but
native atomics aren't.
Signed-off-by: Martin Storsjö <martin@martin.st>
These codecs compile all of the MJPEG code anyway, so there is little
point in not enabling the MJPEG decoder directly. This also simplifies
the dependency declarations for the MJPEG codec family.
This codec compiles all of the SP5X code anyway, so there is little
point in not enabling the decoder directly. This also simplifies the
dependency declaration for the AMV decoder.
The vector dequantization has a test in a loop preventing effective SIMD
implementation. By moving it out of the loop, this loop can be DSPized.
Therefore, modify the current DSP implementation. In particular, the
DSP implementation no longer has to handle null loop sizes.
The decode_hf implementations have following timings:
For x86 Arrandale:
C SSE SSE2 SSE4
win32: 260 162 119 104
win64: 242 N/A 89 72
The arm NEON optimizations follow in a later patch as external asm. The
now unused check for the y modifier in arm inline asm is removed from
configure.
Framerate is now a sane rational instead of an integer, and
inputDepth is changed to what it actually is.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Moving cpunop from the HAVE_LIST to the ARCH_EXT_LIST_X86 has the side
effect of enabling it. The semantics of the check have to be changed
from enable if successful to disable if unsuccessful. This was missing
in 2b0bb69997 causing build errors with
nasm.
Fixes dependency file generation with gas-preprocessor.pl and clang.
Flags copied from GCC and tested with Apple's clang from Xcode 5 and
5.1 and clang 3.2, 3.3, 3.4 on Linux.
NEON and VFP are currently mandatory for all ARMv8 profiles. Both are
handled as extensions as far as cpuflags are concerned. This is
consistent with handling x86_64 which always has SSE2, but still
handles it as an extension.
Stack is always 16 byte aligned and clz, 64bit operations and unaligned
memory access are fast in aarch64 mode on ARMv8.
Signed-off-by: Janne Grunau <janne-libav@jannau.net>
Previously, if neither of the checks for the closesocket function
succeeded, we still kept winsock2.h and networking in general
enabled.
When targeting the WinRT API subset, the winsock2.h header is
available (making the check for it succeed, giving the impression
that winsock is available), but tests that actually try to use
such a function will fail. In this case, disable the winsock2.h
feature and networking in general, as if the winsock2.h header
test would have failed in the first place.
Signed-off-by: Martin Storsjö <martin@martin.st>
New versions of FreeType have moved the location of their API
header(s) and hide the location behind a macro.
Since the location changes between versions and no other way
to know the location exists, this workaround becomes necessary.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Some pthreads symbols might be present in libc (as shown on various *BSD)
but not all of them, leading to false positives.
Check for the most common compiler flags before the plain symbol check
to avoid known pitfalls.
Directly loads AviSynth through LoadLibrary instead of relying on
Video for Windows, and supports using AvxSynth (via dlopen) to
open scripts on Linux and OS X.
Error messages from AviSynth/AvxSynth are now reported through
av_log and exit, rather than the traditional behavior of generating
an error video that the user would need to watch to diagnose.
The main rewrite was authored by d s <avxsynth.testing@gmail.com>
from the AvxSynth team, with additional contributions by
Oka Motofumi <chikuzen.mo@gmail.com>
Stephen Hutchinson <qyot27@gmail.com>
Diego Biurrun <diego@biurrun.de>
Anton Khirnov <anton@khirnov.net>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
Certain instrumentation addons leads to a false positive in configure
and link failures at the end of the build phase.
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
This will check if -wN '..@*' is available and fall back on -x if not;
when none are available, do not run strip at all to prevent removing
functions that might be actually needed.
Msys is unable to convert unix style absolute paths to windows style
paths when combined with certain multichar MSVC options such as
-Fo<file>. We used to work around this issue by passing them as two
separate parameters separated by a space to c99wrap, which then mapped
them back to the actual parameter format that MSVC uses.
The only paths that actually are an issue are absolute unix style
paths, and the only place such absolute paths are used with the output
arguments (-Fo, -Fe, -Fi, -out:) are for the temp files within configure.
By setting TMPDIR to . for msvc/icl builds, we never need to use
absolute unix style paths for the file output, and we can use the
actual proper form of the file output parameters. This avoids requiring
the c99wrap wrapper for remapping the parameters for cases where the
c99 converter isn't invoked at all (MSVC2013 and ICL).
Signed-off-by: Martin Storsjö <martin@martin.st>
ICL doesn't return an error on unknown parameters, and will
always pass the symver_gnu_asm test, and since Windows
never has symbol versioning, just always disable it.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
On some platforms (such as msys), symlinks are (poorly) emulated
by simply creating a copy of the file.
This means that when building out of tree, the build tree gets
a copy of the original makefile, which can lead to unintuitive
build errors when the original makefile gets updated later.
Instead simply create a stub makefile which includes the real
one.
Signed-off-by: Martin Storsjö <martin@martin.st>
Originally written by Ronald S. Bultje <rsbultje@gmail.com> and
Clément Bœsch <u@pkh.me>
Further contributions by:
Anton Khirnov <anton@khirnov.net>
Diego Biurrun <diego@biurrun.de>
Luca Barbato <lu_zero@gentoo.org>
Martin Storsjö <martin@martin.st>
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
MSVC does support enough of C99 to work without the converter since
the 2013 version. Try to detect which version of the compiler in
the path needs to run the C99 converter or not. When the converter
is omitted, compilation time is reduced quite drastically.
Prior to this, users could still use --cc="c99conv -noconv cl"
when running MSVC 2013 to achieve the same.
This checks the version number instead of doing a normal compile
test, since this check needs to be done earlier in configure, before
the normal compile test helpers are usable.
Signed-off-by: Martin Storsjö <martin@martin.st>
As another example of bizarre compiler behavior clang groks the
-Wmaybe-uninitialized option, but not -Wno-maybe-uninitialized
and spews a warning for every file that gets compiled.
For some weird reason gcc does not check if the -Wno disabling variants
of warning flags match existing warning flags. Instead it swallows them
silently. That is, unless other warning or error messages are generated,
because then - for some even more bizarre reason - a complaint about the
unknown disable warning flag is issued along with the error or warning
message.
Thus to check for the availability of a warning disabling option, one
needs to check for the enabling variant instead and then add the
disabling variant to CFLAGS.
Initially written by Guillaume Martres <smarter@ubuntu.com> as a GSoC
project. Further contributions by the OpenHEVC project and other
developers, namely:
Mickaël Raulet <mraulet@insa-rennes.fr>
Seppo Tomperi <seppo.tomperi@vtt.fi>
Gildas Cocherel <gildas.cocherel@laposte.net>
Khaled Jerbi <khaled_jerbi@yahoo.fr>
Wassim Hamidouche <wassim.hamidouche@insa-rennes.fr>
Vittorio Giovara <vittorio.giovara@gmail.com>
Jan Ekström <jeebjp@gmail.com>
Anton Khirnov <anton@khirnov.net>
Martin Storsjö <martin@martin.st>
Luca Barbato <lu_zero@gentoo.org>
Yusuke Nakamura <muken.the.vfrmaniac@gmail.com>
Reimar Döffinger <Reimar.Doeffinger@gmx.de>
Diego Biurrun <diego@biurrun.de>
Signed-off-by: Anton Khirnov <anton@khirnov.net>
F4V is Adobe's mp4/iso media variant, with the most significant
addition/change being supporting other flash codecs than just
aac/h264.
Signed-off-by: Martin Storsjö <martin@martin.st>
This is in preparation for removing a .rodata kludge
which was only required for older YASM versions.
The movbe instruction was introduced in 0.8.0, which already
had AVX, which was introduced in 0.7.0, and NASM introduced
movbe in 2.0.3, which is the same version which introduced
AVX support.
Also, make the failure message more accurate.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
The implementation of 25cb0c1a involves lots of spurious labels.
The effect of keeping those labels around is making debugging harder.
Those labels are meaningless, and complicate the disassembly. Also,
gdb can't tell the difference between them and function entry points.
This new strip command is irrelevant to any usage of Libav that would
have used the old fully stripped version, because the old one was for
non-debug use.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
Qansi-alias worked on 12.x, then caused miscompilation on 13.x, but now
works again passing all FATE tests for icl version 14.
Signed-off-by: Diego Biurrun <diego@biurrun.de>
Prior to this on msvc/icl there was no handling of deprecated functions
and the deprecated warning was disabled.
After enabling there are a number of warnings relating to the CRT and
the use of the non-secure versions of several functions. Defining
_CRT_SECURE_NO_WARNINGS silences these warnings.
Signed-off-by: Martin Storsjö <martin@martin.st>
The i686 feature really is a CPU feature and should be handled as such.
The cpunop dependency on i686 should be expressed with a standard _deps
declaration instead of a manual test.
The hls muxer itself doesn't have any direct (object file level)
dependencies on mpegtsenc.o, and including that object file
directly doesn't ensure that it is registered so that the muxer
actually is accessible.
Signed-off-by: Martin Storsjö <martin@martin.st>
There is no record of this ever being used at all, anywhere,
since the feature was added in 2effd27446.
This gets rid of extra linker tricks just to support a feature
that isn't used, simplifying portability to other platforms.
Signed-off-by: Martin Storsjö <martin@martin.st>
Use this for enabling the ppc timer.h implementation only on
assemblers that support labels in the inline assembly.
Signed-off-by: Martin Storsjö <martin@martin.st>
Because O1 or O2 are required to build libav with msvc/icl, this must be
explicitly set instead of just omitting Oy.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
It is implied by O1 or O2, both of which are required to build libav
with msvc/icl. Silences warnings when targeting x64 with icl.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
This avoids warnings about this option not having any effect on
this platform.
We still want to enable the pic configure item for these platforms
(if detected via the compiler builtin define __PIC__) to get proper
inline assembly workarounds.
Signed-off-by: Martin Storsjö <martin@martin.st>
This reverts e08c946c6 and 05165c2f7a. The actual intention of
e08c946c6 was to fix shared library builds for arm/win32, which
can also be accomplished in other ways.
Disabling pic on those platforms broke inline assembly on cygwin/64
(since some inline assembly requires knowing whether we are building
as PIC or not), and might also break inline assembly on other
compilers on windows.
As a side-effect, this unfortunately brings back all the warnings
about PIC not having any effect on that platform.
Signed-off-by: Martin Storsjö <martin@martin.st>
The object file format doesn't support PIC loads in ARM assembly,
there are no relocation types in PE/COFF that correspond to
BFD_RELOC_32_PCREL (R_ARM_REL32 in ELF).
Signed-off-by: Martin Storsjö <martin@martin.st>
Allows for easier handling of flags that may be specific to icl or msvc.
Furthermore, simplify the handling of warnings and remarks thanks to
icl's support of -Wall on Windows.
Signed-off-by: Derek Buitenhuis <derek.buitenhuis@gmail.com>
The modern MSVC for ARM always builds for thumb, and it can't be
disabled.
Also just use the default arch instead of trying to map the -march
parameter to MSVC's -arch parameter (which only takes the values
ARMv7VE and VFPv4).
Signed-off-by: Martin Storsjö <martin@martin.st>
Initial support for the ICL compiler on windows. Requires a new
c99wrap with ICL support (1.0.2+).
Currently not much different speed wise compared to msvc. In the
future with a few changes it can be made to support the inline asm.
This would be the primary reason for using it.
Passed all fate tests, versions tested:
13.1.1.171 (2013 Update 3) x86 and x64
12.1.5.344 (2011 Update 11) x86 and x64
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
Change the check_exec_crash test to use a function pointer instead of
simply calling the function. The EBP availability test will crash when
compiled with ICL likely due to compiler optimization shenanigans.
Originally the check_exec_crash code was moved out of main to fix a
problem with gcc's treatment of non-leaf main on x86_32. Libav already
moved the code out of main but the addition of the function pointer will
prevent any inlining which fixes the remaining problem.
A function pointer is used since it is compiler agnostic (as opposed to
say __attribute__ ((noinline)) which would only work with gcc compatible
compilers).
Signed-off-by: Luca Barbato <lu_zero@gentoo.org>
MSVC 2010 (or more precisely, Windows SDK 7.0 which comes with MSVC
2010) sets _WIN32_WINNT to the constant for Windows 7 if nothing is
set. This could lead to the libav configure script detecting and
using functions only present in Windows 7 or newer, which in most
cases isn't desired. If the caller explicitly wants this, the caller
can add the _WIN32_WINNT define via --extra-cflags, setting the desired
version.
Signed-off-by: Martin Storsjö <martin@martin.st>
With the parameter --toolchain valgrind-massif, the configure
script sets reasonable defaults that can be overridden as explained
in the documentation.
This makes it consistent with the msvc builds which automatically set
the DEP and ASLR flags by default. There really is no good reason why
they shouldn't be set.
The fact that binutils does not set them on by default boggles the mind.
Signed-off-by: Martin Storsjö <martin@martin.st>
If this is explicitly disabled for win32/mingw, it should also
be disabled for cygwin, for consistency and for the same reasons
as for win32/mingw.
Signed-off-by: Martin Storsjö <martin@martin.st>
These platforms do not have any notion of PIC. On some compilers,
enabling pic produces a number of warnings.
This avoids trying to produce PIC loads in the ARM assembly - there
are no relocation types in PE/COFF that correspond to
BFD_RELOC_32_PCREL (R_ARM_REL32 in ELF).
As a side-effect, this avoids enabling PIC on mingw64, getting rid
of the warnings about PIC not having any effect on that platform.
Signed-off-by: Martin Storsjö <martin@martin.st>
This allows it to be overridden, either by the user on the command
line, or by other sections of the configure script.
Signed-off-by: Martin Storsjö <martin@martin.st>
When targeting the "windows store application" (metro) API subset
(or the windows phone API subset), the getenv function isn't
available. If it is unavailable, just define getenv to NULL.
The check uses check_func_headers, since the function actually
might exist in the libraries, but is hidden in the headers.
The fallback is in config.h since msvc can't do -D defines with
parameters on the command line, and it's used both within the
libraries and the frontend applications (so a libavutil internal
header wouldn't be enough).
Signed-off-by: Martin Storsjö <martin@martin.st>
With the parameter --valgrind-memcheck, the configure script sets
reasonable defaults that can be overridden as explained in the
documentation.
The idea of using set_defaults is from Luca Barbato.
If building libav with -MD in the cflags (for making the MSVC compiler
generate code for using a dynamically linked libc), the system headers
that declare strtod, snprintf and vsnprintf declare the functions as
imported from a DLL. To hook up wrappers of our own for these functions,
the function names are defined to avpriv_*, so that the calling code
within libav calls the wrappers instead. Since these functions
are declared to be imported from DLLs, the calling code expects to
load them from DLL import function pointers (creating references to
_imp__avpriv_strtod instead of directly to avpriv_strtod). If the
libav libraries are not built as DLLs, no such function pointers (as
the calling code expects) are created.
The linker can fix this up automatically in some cases (producing
warnings LNK4217 and LNK4049), if the object files are already
included. By telling the linker to try to include those symbols
(without the _imp prefix as the calling code ends up using),
we get the object files included, so that the linker can do the
automatic fixup. This is done via config.h, so that all (or at least
most) of the object files in our libraries force including the compat
files, to make sure they are included regardless of what files from our
static libraries actually are included.
Signed-off-by: Martin Storsjö <martin@martin.st>
This avoids cases where configure tries to weakly enable an item
which actually is disabled, ending up still enabling dependencies
of the item which itself is only enabled weakly.
More concretely, the h264 decoder suggests error resilience, which
is then enabled weakly (unless manually disabled). Previously,
dsputil, which is a dependency of error resilience, was enabled
even if error resilience wasn't enabled in the end.
Signed-off-by: Martin Storsjö <martin@martin.st>
The variable name 'var' is commonly used to iterate through arguments
in other functions. When the pushvar function internally uses the
variable 'var', it makes pushing/popping the variable 'var' not
work as intended.
Signed-off-by: Martin Storsjö <martin@martin.st>
