now goes through pf_
8728715d

Zhao Zhili authored 7 years ago and François Cartegnie committed 7 years ago

Signed-off-by: Francois Cartegnie <fcvlcdev@free.fr>

Use the VADisplay from the src pic and not upload one (that is NULL in case of
download).

Steve Lhomme authored 7 years ago and Jean-Baptiste Kempf committed 7 years ago

Apply the filter setting *after* initializing the context.

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

Favor DRM and use X11/WAYLAND if it's not available.

Fixes #18445

This fixes undefined behaviors when the native display is destroyed before the
VADisplay is terminated. This caused a crash or invalid surfaces with the DRM
native display when changing between 2 vouts. For X11, no crashes were
observed, probably because the native display was already hold by the window.

These values are always positive.

Shaleen Jain authored 7 years ago and Thomas Guillem committed 7 years ago

According to the spec, calloc can return a NULL or a unique pointer value if
either of the arguments are 0 depending on the implementation. Add a guard to
prevent allocation by returning an error in the above case.

Signed-off-by: Thomas Guillem <thomas@gllm.fr>

See XXX comments.

We still need a *not global* vaapi instance: struct vlc_vaapi_instance. This
struct holds a VADisplay and a refcount. Every pictures of the vout pool hold
a reference to this struct in order to avoid that the VADisplay is terminated
while being used by pictures.

Filters now get the vaapi instance via filter_NewPicture(). Static filters like
the deinterlace one may not have access to a picture allocated by the vout.
This is fixed by next patches.

Refs #18445

This is most likely coming from a Qt thread, which is already
initialized as STA. If it is not coming from a Qt thread, then we don't
really care since the API also work when called from a MTA, and we
release the threading apartment once we're done (which was the real
culprit for #16935)

Fix #18482

We need to ignore the unguarded-availability diagnostic, we are already
using partial-availability here anyway.

Otherwise creation of new dmg file will fail. Also, zip would
update an existing zip file.

Matrix check boxes in simple prefs need an explicit height
constraint. Otherwise they will not shown on 10.7.

This will crash at runtime on 10.7.