[Bug 779145] dmabuf export from vaapi encoders fails

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=779145

--- Comment #1 from Scott D Phillips <scott.d.phillips at intel.com> ---
Created attachment 346591
  --> https://bugzilla.gnome.org/attachment.cgi?id=346591&action=edit
[PATCH] vaapivideomemory: dmabuf: destroy derived imagein al in all cases

The derived image will cause vaBeginPicture to fail for both
upstream and downstream cases, so change the dmabuf creation to
destroy the derived image in both cases.

https://bugzilla.gnome.org/show_bug.cgi?id=779145

---
Note that this logic conflicts with the defined usage of buffer
handles in the VA-API, which states:

> Besides, no additional operation is allowed on any of the buffer
> parent object until vaReleaseBufferHandle() is called. e.g.
> decoding into a VA surface backed with the supplied VA buffer
> object buf_id would fail with a VA_STATUS_ERROR_SURFACE_BUSY
> error.

Obviously the intel-vaapi-driver is not currently behaving this
way, but the dmabuf allocator here should be adapted later to work
once the intel-vaapi-driver begins conforming to this defined
restriction.

 gst/vaapi/gstvaapivideomemory.c | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.