Subject : [Regression 2.6.39][nouveau][bisected?] leaving fullscreen XV "crashes" with KDE desktop effects enabled Submitter : Jan Seiffert <kaffeemonster@googlemail.com> Date : 2011-05-25 2:32 Message-ID : BANLkTimd6nvXTbn2hM+cKDNKFpxzsQLHog@mail.gmail.com References : http://marc.info/?l=linux-kernel&m=130629073407685&w=2 This entry is being used for tracking a regression from 2.6.38. Please don't close it until the problem is fixed in the mainline.
This is very interesting beast. First of all, I also notice similar problem with compiz, not that often though. Once again, open totem, maximize, minimize, boom. It doesn't happen often, but when I start watching videos often it bites. Also XV sometimes hangs the system without any user intervention. Now that commit that Jan Seiffert bisected looks very very suspicions. I can say I almost sure he found something very interesting. Kudos to Jan Seiffert
On Mon, 2011-06-20 at 15:04 +0000, bugzilla-daemon@bugzilla.kernel.org wrote: > https://bugzilla.kernel.org/show_bug.cgi?id=36962 > > > Maxim Levitsky <maximlevitsky@gmail.com> changed: > > What |Removed |Added > ---------------------------------------------------------------------------- > CC| |maximlevitsky@gmail.com > > > > > --- Comment #1 from Maxim Levitsky <maximlevitsky@gmail.com> 2011-06-20 > 15:04:12 --- > This is very interesting beast. > > First of all, I also notice similar problem with compiz, not that often > though. > Once again, open totem, maximize, minimize, boom. > It doesn't happen often, but when I start watching videos often it bites. > Also XV sometimes hangs the system without any user intervention. > > Now that commit that Jan Seiffert bisected looks very very suspicions. > I can say I almost sure he found something very interesting. > > Kudos to Jan Seiffert > I have seen that on my NV86, and that could be the reason for hangs I have quite often (at least one type of them...) Best regards, Maxim Levitsky
Hmm, few days ago on irc someone else reported similar bug and Ben provided this patch: http://members.iinet.net.au/~skeggsb/0001-drm-nouveau-fix-assumption-that-semaphore-dmaobj-is-.patch which seemed to fix it. I'm not sure why he didn't apply it yet.
The patch from comment #3 is in nouveau/linux-2.6, but on the second page ;) http://cgit.freedesktop.org/nouveau/linux-2.6/commit/?id=b16a5a18ff994532120c1d18e678bbc5fb477b62 It is taged as drm-nouveau-fixes, so it should go into Linus Tree. Hopefully fast! I had a Linux-3.0.0-rc3 git tree here, applied the 3 patched till drm-nouveau-fixes by hand (left out the endian fix, no BE here), and yes, the problem is gone. I repeat: This seems to resolve the problem! Where can i put my "Tested-by: Jan Seiffert"? Greetings Jan
Merged already in v3.0-rc4: commit b16a5a18ff994532120c1d18e678bbc5fb477b62 Author: Ben Skeggs <bskeggs@redhat.com> Date: Fri Jun 17 23:41:54 2011 +1000 drm/nouveau: fix assumption that semaphore dmaobj is valid in x-chan sync