Close #12400

This handler was a no-op and does not make sense in a fullscreen only
renderer, so drop the handler.
Signed-off-by: Julian Scheel <julian@jusst.de>
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

The fmt in vout_display_t is not allowed to be written to after opening the
plugin without invalidating all the pictures. Instead of doing invalid writes
to vd->fmt, just make use of vd->source and event arguments directly which
shall hold correct values.
As framerate changes are not announced by the core we need to add a local
storage for comparion now as vd->fmt must not be used for it.
Signed-off-by: Julian Scheel <julian@jusst.de>
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Add a new flag, which can be used to let mmal select ahdmi mode which matches
the video size and interlacing type. Using this with a proper screen one can
playback for example 1080i content using the screens deinterlacing filters.
The videocore is configured to scan out picture on odd/even field only
according to the top_field_first mode of the video in case of interlaced
video. For progressive video a progressive mode is selected.
Signed-off-by: Julian Scheel <julian@jusst.de>
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Sort vout_sys_t struct members by size to optimize packing.
Signed-off-by: Julian Scheel <julian@jusst.de>
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This was accidentally left out of b3050f68.
Signed-off-by: Tristan Matthews <tmatth@videolan.org>

fix mediacodec with android-L on Nexus 7v2.
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>

Fix ART warning about java thread attached without a name.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

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

Split es setup so we only call dedicated setup per handler.

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

android-L drops support for pthread_cond_timedwait_monotonic_np and adds
support for pthread_condattr_setclock. So, use the good function depending on
configure detection.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This should avoid the issues of packaging libgcc_s_sjlj-1.dll,
libwinpthread-1.dll and the like.

Return 0 as usual, rather than undefined result.

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

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