On Mon, May 11, 2015 at 06:51:03PM +0200, Daniel Vetter wrote: > On Mon, Apr 27, 2015 at 01:41:16PM +0100, Chris Wilson wrote: > > As we perform the mmio-flip without any locking and then try to acquire > > the struct_mutex prior to dereferencing the request, it is possible for > > userspace to queue a new pageflip before the worker can finish clearing > > the old state - and then it will clear the new flip request. The result > > is that the new flip could be completed before the GPU has finished > > rendering. > > > > The bugs stems from removing the seqno checking in > > commit 536f5b5e86b225dab94c7ff8061ae482b6077387 > > Author: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@xxxxxxxxx> > > Date: Thu Nov 6 11:03:40 2014 +0200 > > > > drm/i915: Make mmio flip wait for seqno in the work function > > > > Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx> > > Cc: Ander Conselvan de Oliveira <ander.conselvan.de.oliveira@xxxxxxxxx> > > I think I grumbled about this before, but the rq vs. req distinction > elludes me. rq = runqueue in my reading ... What do we need to use "req" > for that we're forced to have such an ambigious name for requests? Because I use rq everywhere and runqueues aren't very common in the kernel? Besides which why did you change some of my _request to _req? -Chris -- Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
