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

Signed-off-by: Rémi Duraffort <ivoire@videolan.org>

This will be useful for RTSP.

For use within already variadic functions

pointed out by courmish

  - remove the awkward global lock and serializer mechanism
  - adapt to latest module API (fix compil errors)
  - be more specific about hwnd or xid depending on port
  - set display.x11 on Linux port

We can now pass any number and type of parameter to the callbacks.
This way, we do not need to clutter the object structure with
properties that only make sense during activation. In fact, we do
not need to use VLC objects at all anymore.

This patch resolves two heap corruption vulnerabilities in the CDG
decoder for VLC media player.  In both cases, a failure to properly
validate indexes into statically-sized arrays on the heap could allow a
maliciously crafted CDG video to corrupt the heap in a controlled
manner, potentially leading to code execution.

The patch is against v1.1.5 from vlc git, but this decoder hasn't been
touched in awhile, so I'd expect it to cleanly apply to older versions.
I've tested it and confirmed it resolves the heap corruption issues and
does not break functionality.

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

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

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

Without this, clang will complain about the asm block.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

It is a bit cleaner and properly check for slow write pictures. It also
prepare for letting the vout display to do the blending, and when not
available to blend after the resize (not activated).

You have to set the list of chromas the module support in
vout_display_t::info.subpicture_chromas and then handle the subpicture_t
parameters in vout_display_t::prepare/display.

Reorder the function declarations to avoid forward declarations.

I have merged spu_RenderSubpictures and spu_SortSubpictures and removed the
blending from spu_RenderSubpictures. No functionnal changes except maybe
a bit for the transcode plugin.

It blends a subpicture onto a picture.

This reverts commit 8acffd0c.

It has nothing to do with the patch in itself (which is perfectly fine)
but it conflicts too much with the following patches (which will also fix
the issue). I hope it's ok.