Re: [Xen-devel] Re: [PATCH] EPT: Flush running cpus, add mask to flush when scheduled in

["Jan Beulich" <JBeulich@xxxxxxxxxx>]

>>> Keir Fraser <keir.fraser@xxxxxxxxxxxxx> 22.09.09 10:20 >>>
>On 22/09/2009 09:07, "Keir Fraser" <keir.fraser@xxxxxxxxxxxxx> wrote:
>
>> On 22/09/2009 08:07, "Jan Beulich" <JBeulich@xxxxxxxxxx> wrote:
>> 
>>> Passing a pointer to the global cpu mask looks racy here: What if a CPU
>>> disappears from domain_dirty_cpumask under your feet?
>> 
>> I'm fixing this race before I apply the patch.
>
>George, Jan,
>
>Please see what you think of xen-unstable:20244.

With no assertion in ept_sync_domain() on any locks held, is it guaranteed
that the function cannot be entered twice at the same time for a given
guest? If not, passing a pointer to the new ept_synced member isn't any
better than passing the one to domain_dirty_cpumask.

Jan


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