On Fri, Mar 23, 2012 at 08:03:23PM +0100, Daniel Vetter wrote: > On Wed, Mar 21, 2012 at 10:48:18AM +0000, Chris Wilson wrote: > > The BLT commands on gen2/3 utilize the fence registers and so we cannot > > modify any fences for the object whilst those commands are in flight. > > Currently we marked tiled commands as occupying a fence, but forgot to > > restrict the untiled commands from preventing a fence being assigned > > before they were completed. > > > > One side-effect is that we ten have to double check that a fence was > > allocated for a fenced buffer during move-to-active. > > > > Reported-by: Jiri Slaby <jirislaby at gmail.com> > > Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=43427 > > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk> > > Cc: stable at kernel.org > > I've had a few comments on irc about a preliminary version of this patch, > and Chris addressed them all. So > > Reviewed-by: Daniel Vetter <daniel.vetter at ffwll.ch> > > Futhermore I've stitched together a little testcase in i-g-t, and it > indeed blows up without this patch. So > > Testcase: i-g-t/tests/gem_tiled_after_untiled_blt > Tested-by: Daniel Vetter <daniel.vetter at ffwll.ch> ... and now also the report from our QA: Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=47990 -- Daniel Vetter Mail: daniel at ffwll.ch Mobile: +41 (0)79 365 57 48
