Re: [Xen-devel] [PATCH v4 7/8] ioreq-server: make buffered ioreq handling optional

[Paul Durrant <Paul.Durrant@xxxxxxxxxx>]

> -----Original Message-----
> From: Ian Campbell
> Sent: 07 April 2014 17:07
> To: Paul Durrant
> Cc: xen-devel@xxxxxxxxxxxxx; Ian Jackson; Stefano Stabellini; Jan Beulich
> Subject: Re: [PATCH v4 7/8] ioreq-server: make buffered ioreq handling
> optional
> 
> On Wed, 2014-04-02 at 16:11 +0100, Paul Durrant wrote:
> > Some emulators will only register regions that require non-buffered
> > access. (In practice the only region that a guest uses buffered access
> > for today is the VGA aperture from 0xa0000-0xbffff). This patch therefore
> > makes allocation of the buffered ioreq page and event channel optional for
> > secondary ioreq servers.
> >
> > If a guest attempts buffered access to an ioreq server that does not
> > support it, the access will be handled via the normal synchronous path.
> 
> In terms of the guest PFN space do the magic PFNs get packed together or
> are there holes? (Mostly just curious...)
> 

There can be holes. The allocator is just a dead-simple bitmap 
test-and-set/clear so which actual PFNs get assigned to which ioreq server 
depends on the sequence of creation/destruction.

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