[Bug 699382] v4l2: dmabuf handling is broken

[GStreamer (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=699382
  GStreamer | gst-plugins-good | git

Nicolas Dufresne <nicolas.dufresne> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |nicolas.dufresne at collabora.
                   |                            |co.uk

--- Comment #7 from Nicolas Dufresne <nicolas.dufresne at collabora.co.uk> 2014-03-29 21:02:29 UTC ---
Adding some thought, note that even if you negotiated everything correctly, the
dmabuf you get from upstream might still not be usable for your device driver.
V4L2 driver may use different memory model, e.g. UVC could (if it was
implement, I think Hans has a branch), import any dmabuf with supported
alignment and stride, because it's using the vmalloc allocator. Though, most
driver use the contiguous allocator, for which a vmalloc dmabuf (e.g. exported
by UVC) would not work. So, just like we are doing in the USERPTR experimental
branch, we need all appropriate fallback.

Final note, v4l2 really need to track it's memory, since it can't detach from
it. I'm working on an allocator for that, but I have the impression that the
dmabuf allocator will be unusable in that context since we still have to track
our buffers, something the dmabuf allocator don't do (it also don't allocate
anything, as allocating dmabuf is a per-driver action). Anything I'm missing ?
Was this considered in this attempt to provide an allocator ?

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.