On 25/1/08 10:19, "Andi Kleen" <ak@suse.de> wrote:

quoted text
>> Whether this a problem in light of Xen spurious faults depends on whether
>> NMI handlers touch dynamically-allocated data.
> 
> How do you define dynamically-allocated data?

Anything that could have been a read-only pte or ldt page in a previous life
with no intervening TLB flush. So get_free_page(), kmalloc(), vmalloc(), ...

Actually I think we are fine, now I think about it some more, because we
only clear the software NMI-in-flight flag if the guest executes IRET via
the hypervisor. Most Xen Linux guests only do IRET via the hypervisor when
the current context is an NMI handler (additionally x86_64 also does so when
returning to ring 3). Most importantly for this case, we will *not* IRET via
the hypervisor when returning from a #PF context nested in an NMI context.
Hence the NMI-in-flight flag will not be cleared, and guest virtual NMIs
will not nest. So that's a relief!

 -- Keir


--
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/