[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Xen panic on guest shutdown with PCI Passthrough



On Mon, 7 Feb 2011, Anthony PERARD wrote:

> On Mon, Feb 7, 2011 at 09:27, Jan Beulich <JBeulich@xxxxxxxxxx> wrote:
> >>>> On 04.02.11 at 17:34, Anthony PERARD <anthony.perard@xxxxxxxxxx> wrote:
> >> Hi,
> >>
> >> Each time I shutdown a guest with a passthrough pci device, Xen panic
> >> on a NMI - MEMORY ERROR.
> >
> > And this happening in msi_set_mask_bit() is consistent? Always at
> > the same RIP, or (slightly) varying? If always at the same address,
> > could you disassemble the respective instruction so we know what
> > address is being read/written?
>
> Unfortunately, this error happened in many place, even in default_idle.
>
> Maybe this can help:
> (XEN) irq.c:1585: dom1: forcing unbind of pirq 16
> (XEN) irq.c:1585: dom1: forcing unbind of pirq 54
> the error happened just after these debug print.
>
> > And quite certainly only on one (type of) machine?
>
> Yes, only one machine, I didn't try on other machine.

I tried to use 'xl pci-attach' and 'xl pci-detach' with the module
acpiphp in the guest, this work fine. But xen say:
(XEN) domctl.c:992:d0 memory_map:remove: gfn=20000 mfn=da000 nr_mfns=2000
(XEN) p2m.c:2723:d0 clear_mmio_p2m_entry: gfn_to_mfn failed! gfn=0002000c

The error append only when the pci device is still attach to the guest
when I type 'halt' in the guest.

-- 
Anthony PERARD

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.