commit 6c9bc13f

See also a6b259b7.

DecoderSignalWait() had no observable effects if has_data was false. It
could wake up the condition variable, but no thread would wait on that
event.

In some cases, the thread(s) consuming a FIFO needs to wake up in other
circumstances than the FIFO being non-empty. For that purpose, this new
set of functions is vastly more flexible than block_FifoWake().

There are ARM servers and or HD/UHD-capable STBs. The distinction no
longer makes sense.

es->p_dec is obviously non-NULL when the loop starts, and the loop
cannot change the value.

Calling input_DecoderStopWait() from the decoder thread makes no sense
as it signals a condition variable that only the decoder thread itself
would ever wait on.

More importantly, the call breaks assumptions made by the input thread.
For instance, input_DecoderDecode() assumes p_owner->b_waiting is only
changed by the decoder owner (i.e. input) thread.

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

The renderer may be updated in the DMO thread which uses COINIT_MULTITHREADED
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Fixes WMA 9.1 Lossless playback through avcodec
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Fixes #14210
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Fixes #14205
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Just like in AVI
See modules/demux/avi/avi.c l461
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Fixes #14202

All the other CoInitializeEx() calls and now using multithreading, except for
this code which is called in UpdateRects()
See modules/video_output/msw/common.c l286
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

commit 8fd834512cf778a89d0fbf2540b0055dec002f42.

should go to pre es queue otherwise first frames/gop are dropped

Fixes #13855

The small menu buttons on the left doesn't move downwards anymore when
the menu is expanded
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

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

similar to what avformat does

I forgot to re-submit that one after the other commits went through.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Since the FIFO is being destroyed, there is no need to lock and to
update the FIFO internal state.

p_dec->fmt_out is owned by the decoder plugin, and can only safely be
accessed by the core from within decoder callbacks, notably the format
update callbacks.

Outside that context, p_owner->fmt has to be used. It contains a copy
of p_dec->fmt_out at the last format update.