Bug#605673: [lenny] KVM host crash at kvm:gfn_to_rmap+0x17/0x49
El mié, 25-01-2012 a las 16:39 -0600, Jonathan Nieder escribió:
> > System was rebooted 3 days ago. There are about 8 virtual machines.
> > One of them was doing heavy I/O during the crash.
> > After a cold-reboot everything worked ok and the heavy I/O task has
> > been re-run and completed successfully.
> > BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
> > IP: [<ffffffffa025e963>] :kvm:gfn_to_rmap+0x17/0x49
> > [498738.031444] PGD 338c91067 PUD 338cc9067 PMD 0
> > [498738.031444] Oops: 0000  SMP
> > [498738.031444] CPU 3
> > [498738.031444] Modules linked in: tun kvm_intel kvm ipv6 bridge loop snd_pcm snd_timer snd sou
> > i2c_core parport_pc parport shpchp pcspkr rng_core i5000_edac container button pci_hotplug edac_
> > mirror dm_log dm_snapshot dm_mod ide_cd_mod cdrom piix ide_pci_generic ide_core ses enclosure sd
> > ci_hcd uhci_hcd tg3 aacraid scsi_mod thermal processor fan thermal_sys [last unloaded: scsi_wait
> > [498738.031444] Pid: 3275, comm: kvm Not tainted 2.6.26-2-amd64 #1
> Sorry we missed this before. Was it reproducible?
I think we maybe saw another crash like that after a few months, but
nothing very recurring in that 24h running server.
We changed to proxmox distro for easing VM administration, that is based
on lenny but uses different kernels backported from squeeze, ubuntu and
redhat; we first used the 2.6.32 kernel based on squeeze and now are
using a 2.6.35 backported from ubuntu (for KSM support); we haven't seen
this kind of problem again, so I guess it was fixed upstream.
Thanks a lot
Zuzendari Teknikoa / Director Técnico
Binovo IT Human Project, S.L.
Astigarraga bidea 2, planta 6 dcha., ofi. 3-2; 20180 Oiartzun (Gipuzkoa)