[Bug 761059] kmssink: add new plugin and element

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=761059

--- Comment #16 from Víctor Manuel Jáquez Leal <vjaquez at igalia.com> ---
(In reply to Nicolas Dufresne (stormer) from comment #15)
> Review of attachment 320439 [details] [review]:
> 
> As weird as it sound, I'm not exactly a fan of the io-mode properties, even
> though, I agree they can be handy. Though, how hard would it be to check if
> the incoming frame is containing dmabuf and choose the right method
> automatically ? I don't see any setup like in v4l2 that makes this obviously
> non-trivial. (not a review, just a question ;-P)

I experimented with different approaches but none of them worked. Though, I
ought to say, that my test environment for v4l2videc with dmabuf is constrained
to GStreamer 1.4.5.

I have pending to try with master and try to negotiate the feature
memory:DMABuf and play with the bufferpool's options... Is that already landed
in v4l2viddec?

My general idea is:

If the feature memory:DMABuf is negotiated, we know that it could be either to
import or export buffers.

If upstream request a buffer pool option, say GST_BUFFER_POOL_OPTION_DMABUF, so
the sink will offer a buffer pool with exported KMS buffers to PRIME dmabufs
(that's missing in the current implementation).

Otherwise (no GST_BUFFER_POOL_OPTION_DMABUF of the buffer pools is not used in
upstream) the buffers will be imported following the current mechanism.

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