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

Re: [Xen-devel] Regression introduced by changeset "evtchn: Freepirq_to_evtchn/pirq_mask arrays on domain destruction."



On 29/05/2009 08:19, "Jan Beulich" <JBeulich@xxxxxxxxxx> wrote:

> Yeah, evtchn_destroy() runs much earlier than arch_domain_destroy().
> 
> Keir, is there any reason evtchn_destroy() cannot be deferred accordingly?
> If there is, then un-tying the freeing of pirq_mask and/or pirq_to_evtchn
> from evtchn_destroy() would be needed. I have to admit that it seems not
> logical even in the original code to have IRQ (and hence indirectly evtchn)
> related activities going on for a domain past evtchn_destroy().

There are subtleties which make it a bad idea to defer evtchn_destroy() to
complete_domain_destroy(). Changeset 15465 actually deliberately moved
evtchn_destroy() earlier, and that cleaned up some issues.

Moving the xfree()s to complete_domain_destryoy was my plan B in this case.

 -- Keir



_______________________________________________
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®.