This commit repairs the libvlc equalizer API.
Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

macosx: move video filter control code to the global libvlccore control object since it does not belong in a window controller

While vaDeriveImage is slightly faster than vaGetImage, there's an added cost of having to convert NV12 into YV12 later, which in practice negates any benefits.
Prefer YV12 and YUV420 formats over NV12.
This decrease time to process a VAAPI frame by over 50%: from 2.1ms/frame to 0.9ms/frame on an i7-4650U
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Adopt suggestions as per https://software.intel.com/en-us/articles/copying-accelerated-video-decode-frame-buffers:
- 64kB aligned memory buffer
- Call to mfence only around copy to cache routine.

Use a single instruction to copy non-aligned memory at startup.
Remove compilation warning on mac

This result in a small 1.3% speed increase on an i7-4650U
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This solves compilation issues from the future.

- do not misuse a block_t as a plain array
- use correct size of buffers
- do not overwrite buffer pointers of the second block_t, instead
copy the data

close #7886

The FTP server sometimes gives the following message:

230-Your bandwidth usage is restricted
230-
230-GMP -- GNU Multiple Precision Arithmetic Library
230-
230->>>  This is an intensionally throttled ftp server.                 <<<
230->>>  For faster and less tamper-prone file download, please go to:  <<<
230->>>                                                                 <<<
230->>>                  https://gmplib.org/download/                   <<<
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

SI NON CONFECTVS, NON REFICIAT.

And activate BD-J

A subtitle track chosen by hotkey "v" is remembered in variable "spu-choice".
The hotkey "Shift+v" toggles subtitle visibility.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

When building current HEAD of master for x86_64-w64-mingw32 against current
contribs, compilation fails at libmod_plugin due to undefined symbols of
the form __imp_ModPlug_*. This patch modifies libmodplug in contribs by
adding -DMODPLUG_STATIC to CFlags in the pkg-config file. A bug is also
fixed in the main build system to make sure the CFLAGS are picked up from
this pkg-config file.

This problem occurs because of the following code in
contrib/x86_64-w64-mingw32/include/libmodplug/modplug.h. This code is
present in libmodplug obtained from running "make fetch & make" in the
contribs directory, but not present in the prebuilt contribs obtained from
running "make prebuilt".

 #if defined(_WIN32) || defined(__CYGWIN__)
 # if defined(MODPLUG_BUILD) && defined(DLL_EXPORT)
 #   define MODPLUG_EXPORT __declspec(dllexport)
 # elif defined(MODPLUG_BUILD) || defined(MODPLUG_STATIC)
 #   define MODPLUG_EXPORT
 # else
 #   define MODPLUG_EXPORT __declspec(dllimport)
 # endif

When building for Windows, if MODPLUG_STATIC is not defined, MODPLUG_EXPORT
is defined as "__declspec(dllimport)". On Windows, dynamic symbols are
prefixed with "__imp_", so this causes GCC to adjust the symbol references
to the libmodplug functions accordingly.

This patch therefore adds a patch to contribs for libmodplug's pkg-config
file. It adds -DMODPLUG_STATIC to CFlags, so that when libmod_plugin is
building it will have the correct symbol names.

However, there is also a bug in the build system that needs to be fixed for
this to work. The configure script sets a CXXFLAGS_mod containing the
output of pkg-config for libmodplug, but not CFLAGS_mod. Additionally, the
Makefile.ac for libmod_plugin sets CXXFLAGS when in fact the relevant files
(mod.c) are plain C. Autotools therefore ignores these CXXFLAGS when
generating a makefile.

The solution is to add a macro to configure.ac to set CFLAGS_mod. This then
needs to be used in modules/demux/Makefile.am to populate
libmod_plugin_la_CFLAGS instead of libmod_plugin_la_CXXFLAGS (which gets
ignored).
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

libva will never return I420 fourcc code, instead it uses IYUV. The first image format returned by vaQueryImageFormat with Intel VA backend is YV12, so using I420 wasn't an issue.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This commit prevents the leak of the semaphore.

This reverts commit 703cfc64 as it does not solve any issue.

Allow restricting auto matching to some parent nodes.
There was some size checks on stsd/wave vs wave to prevent
bad matchs, but this could happen with any other codec name.