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

Re: [Xen-devel] RE: transferring ownership of domheap pages


  • To: Dan Magenheimer <dan.magenheimer@xxxxxxxxxx>, "Xen-Devel (E-mail)" <xen-devel@xxxxxxxxxxxxxxxxxxx>
  • From: Keir Fraser <keir.fraser@xxxxxxxxxxxxx>
  • Date: Fri, 13 Feb 2009 15:23:01 +0000
  • Cc:
  • Delivery-date: Fri, 13 Feb 2009 07:23:34 -0800
  • List-id: Xen developer discussion <xen-devel.lists.xensource.com>
  • Thread-index: AcmN7fYgl+sJmTXiQ3OE+JjPO0ZT7QAAP7zV
  • Thread-topic: [Xen-devel] RE: transferring ownership of domheap pages

On 13/02/2009 15:14, "Dan Magenheimer" <dan.magenheimer@xxxxxxxxxx> wrote:

>> Yeah, that's pretty... optimistic. ;-)
> 
> What else do you see that is missing in my first cut?
> Do you see any other code there that's just wrong?

You can't just assert that count_info == PGC_allocated|1, for example. You
need to be careful that mappings aren't created or destroyed while you are
changing the ownership. Steal_page() does that. Once a page is anonymous,
transferring it to a domain is easier, since mappings can't be
created/destroyed while it is anonymous. So steal_page() just does the hard
transition for you (non-anon -> anon). A benefit of using steal_page() is
that we will maintain that for other callers -- a private implementation in
tmem risks getting subtly unsafely broken every now and then.

 -- 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®.