On Fri, 2009-12-04 at 01:10 +0000, Jeremy Fitzhardinge wrote:
> On 12/03/09 16:24, Brendan Cully wrote:
> > On Thursday, 03 December 2009 at 15:22, Jeremy Fitzhardinge wrote:
> >> On 12/03/09 11:35, Brendan Cully wrote:
> >>> Not a patch, but I've just tried out xm save -c again with the latest
> >>> xen changes, and while I no longer see the grant table version panic,
> >>> the guest's devices (aside from the console) appear to be wedged on
> >>> resume. Is anyone else seeing this?
> >>> After a while on the console I see messages like this:
> >>> INFO: task syslogd:2219 blocked for more than 120 seconds.
> >>> which I assume is trouble with the block device.
> >> Is there any backtrace or other info associated with this?
> > I take it you couldn't reproduce? Let me know what you want
> > captured. In the meantime, here's the kernel log:
> Nope, works for me :) Ian has been doing testing with hundreds or
> thousands of suspend/resumes, so it *can* be reliable.
I've been doing regular suspend/resumes not checkpoint ones as Brendan
is doing, I did try a couple of checkpointed ones yesterday and they
failed, IIRC with a similar softlockup to this one.
> That said, I am seeing some odd hangs after suspend/resume with the
> bugfix branch merged into Linux mainline...
Me too, my many successful iterations were with 2.6.31 and 2.6.30 (plus
the fixes). I've also seen some hangs with 2.6.32-rc8 but they don't
look much like this one.
Xen-devel mailing list