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

Re: [Xen-devel] RFC: [0/2] Remove netloop by lazy copying in netback

On Tue, Mar 27, 2007 at 11:40:00AM +0100, Keir Fraser wrote:
> Yeah, but: how does Xen efficiently work out whether a not-present fault is
> due to a lazy grant mapping and, if so, what it is that needs to be demand
> mapped? PV guests can store what they like in not-present pagetable entries.
> Perhaps we could make use of reserved bits in ptes instead (not that there
> are any in non-pae x86, but I'd very much like to obsolete non-pae x86 Xen
> after 3.0.5 anyway -- I don't think anyone actually ships a product with
> non-pae bits).

That's a good point.  Actually, on x86 we have the accessed bit so
we can do this in a different way.

Instead of setting up PTE/P2M entries that cause a page fault on the
first access, we setup the real thing every time and simply check the
accessed bit when we unmap the grant table entry.  That will then allow
us to flush the TLB iff the accessed bit is set.

Now I haven't looked at the other architectures, but I would think
the accessed bit would likely to be present there as well, do you
know whether this is the case for ia64/ppc?

Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@xxxxxxxxxxxxxxxxxxx>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.