Re: [Xen-devel] [PATCH v2 3/4] xen/arm: introduce GNTTABOP_cache_flush

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 03.10.14 at 18:33, <stefano.stabellini@xxxxxxxxxxxxx> wrote:
> On Fri, 3 Oct 2014, Andrew Cooper wrote:
>> On 03/10/14 15:50, Stefano Stabellini wrote:
>> > +        }
>> > +
>> > +        spin_lock(&owner->grant_table->lock);
>> 
>> Is this safe without taking the paging lock and doing a dying check on
>> 'owner' ?
> 
> In gnttab_release_mappings we do take the grant_table lock before
> releasing any entries so it seems to me that it should be sufficient.

Now did you check at all when gnttab_release_mappings() gets
called? Only ever on a dying domain (there's even a BUG_ON() to
that effect at the start of the function). Yet here you want to
avoid dying domains, i.e. your point of reference would rather be
gnttab_transfer().

That said, since you take a reference to the page, I don't think
you need to pay attention to whether the page owner is dying.

Jan


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