On Wed, Jan 23, 2013 at 5:38 PM, Takashi Iwai <tiwai at suse.de> wrote: > At Wed, 23 Jan 2013 17:25:08 +0100, > Daniel Vetter wrote: >> >> From: Alan Cox <alan at lxorguk.ukuu.org.uk> >> >> Adjust the console layer to allow a take over call where the caller already >> holds the locks. Make the fb layer lock in order. >> >> This s partly a band aid, the fb layer is terminally confused about the >> locking rules it uses for its notifiers it seems. >> >> Signed-off-by: Alan Cox <alan at linux.intel.com> >> [danvet: Tiny whitespace cleanup.] >> Reported-and-tested-by: Hugh Dickins <hughd at google.com> >> Reported-and-tested-by: Sasha Levin <levinsasha928 at gmail.com> >> References: https://lkml.org/lkml/2012/10/25/516 >> Signed-off-by: Daniel Vetter <daniel.vetter at ffwll.ch> > > FYI, the latest patch of this is found in mm tree: > http://ozlabs.org/~akpm/mmots/broken-out/fb-rework-locking-to-fix-lock-ordering-on-takeover.patch > > Also I hit the same problem in another code paths (for unbind and > unregister): > http://marc.info/?t=135309396400003&r=1&w=2 > > My additional patch is found in mm tree, too: > http://ozlabs.org/~akpm/mmots/broken-out/fb-yet-another-band-aid-for-fixing-lockdep-mess.patch Indeed, there's more stuff :( And I've missed the export_symbol fix for the first patch. Still, is there any chance we can at least merge the first one (I don't think that many people unload framebuffers) into 3.8 with cc: stable? I'd like to get at least rid of the known deadlocks at kms takeover time, since we have quite a few unexplained such bug reports floating around ... Andrew, Dave? -Daniel -- Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch
