[Bug 25884] xine-ui's on-screen display triggers KMS errors with r100 and rv280

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Sun Mar 21 09:06:48 PDT 2010


http://bugs.freedesktop.org/show_bug.cgi?id=25884





--- Comment #28 from Alex Deucher <agd5f at yahoo.com>  2010-03-21 09:06:47 PST ---
(In reply to comment #27)
> (In reply to comment #26)
> > The problem is you need to re-emit all the previous 3D state (default state and
> > state setup in the textured video functions) if you cross over into a new
> > command buffer.
> 
> So does this mean that R300+ was already splitting command buffers, like the
> patch does for R100 and R200?

R3xx->R5xx only emits one set of verts per draw command since it doesn't have a
native rect primitive but rather uses clipped triangles, so it has to update
the clipping info between draws.

> 
> Note that my RV790 has no problems with xine-ui's on-screen display. Is this
> because it already correctly splits across command buffers, or because its
> default buffer size is big enough not to need splitting in the first place?
> 

R6xx+ is less of an issue since it uses vertex buffers rather than embedding
verts in the command stream (r6xx+ doesn't support inline verts like older
asics).

None of the ddx accel code (EXA, Xv) tracks 3D state which would be needed to
solve this properly.


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.


More information about the xorg-driver-ati mailing list