Re: [Xen-devel] [PATCH] [RFC PATCH] xen: vmx: Use an INT 2 call to process real NMI's instead of self_nmi() in VMEXIT handler

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 13.11.12 at 13:05, Tim Deegan <tim@xxxxxxx> wrote:
> In that case, I agree that we should just run the NMI handler, but first
> I would really like to know what _unblocks_ NMIs in this case.  Any of
> the things I can think of (the next vmenter, the next iret, ??) will
> need some handling to make sure they actually happen before, say, we
> take this CPU into the idle loop...

Wouldn't we have that exact same problem on the "normal" NMI
path? We do switch away from the NMI stack when it arrived
while in guest context, but I don't think we do anything to end the
NMI masking window before going into the scheduler (or handling
other softirqs).

If so, rather than doing this potentially dangerous "int $2" (in
terms of IST stack usage), we might be better off dealing with
the ending of the mask window in a uniform way.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel