Bug#595187: linux-2.6: lots of OOPses on resume (NULL pointer dereference in file_ra_state_init)
On Mon, Sep 20, 2010 at 02:04:23AM +0100, Ben Hutchings wrote:
> On Mon, 2010-09-06 at 17:05 -0700, Vagrant Cascadian wrote:
> > On Tue, Sep 07, 2010 at 12:10:09AM +0100, Ben Hutchings wrote:
> > > On Mon, 2010-09-06 at 12:27 -0700, Vagrant Cascadian wrote:
> > i also didn't mention, largely because i'm pretty unsure if it's really the
> > case, but the issue *seems* to occur more reliably when hibernating (to disk)
> > while on battery, but i'm really unsure of that.
that didn't appear to really make any difference... i tried all combinations of
hibernate with AC, with battery, and resume with AC, with battery, and was
unable to reproduce it reliably. in fact, i hadn't seen it since my last
comment on the bug report nearly two weeks ago... until today. gah.
so it's not exactly easy to reproduce. :(
> > i'll see if i can more reliably trigger the same problem on a clean squeeze
> > install on the same hardware in another partition... where i'm not as worried
> > about crashing.
i was also unable to reproduce it on a clean squeeze install either, so it's
probably something particular with my existing configuration or useage
the test install was just on another partition, and my typical environment is
using lvm on an encrypted volume, if that seems likely to make any difference.
my typical environment is also upgraded from a lenny install.
> Please add 'memory_corruption_check=1' to the kernel parameters and
> report whether that fixes the problem and whether it results in any new
> log messages.
added this now, will reboot and see if that fixes it... presuming it just
doesn't hide itself for another several weeks.