Re: [Xen-ia64-devel] [RFC] Extended I/O port space support

[Alex Williamson <alex.williamson@xxxxxx>]

On Fri, 2007-06-08 at 07:11 +0200, Tristan Gingold wrote:
> On Thu, Jun 07, 2007 at 04:34:01PM -0600, Alex Williamson wrote:
> > Hi,
> > 
> [...]
> >    Next on my TODO list for this is driver domains.  This is where it
> > might get complicated.  Somehow I need to make the driver domain aware
> > that a PCI bus makes use of a different set of I/O port ranges.  This
> > won't work well with the virtual PCI mappings that present the guest
> > with all PCI devices on the same bus.  I think I'm going to need to
> > create a new topology that creates a virtual bus per physical bus, but I
> > haven't dug too deeply there yet.  Suggestions?
> This is certainly the right way to do it, but you also have to specify to
> the driver domain the io space base (although it can be remapped).

   Right.  I think maybe storing the the iospace in xenbus in the
backend and providing a mechanism for the frontend to attach the
resource to the bus is the right way to do this.
 
> >    Please send me comments on the patch below.  It still has some debug
> > printks in it, but it seems to work.  Thanks,
> Seems ok to me.  dom0vp_add_io_space is not thread-safe but who cares :-)

   Thanks for reviewing.  I thought about adding some locking to that
function, but Linux also lacks thread protection there and discovers PCI
root bridges serially.  It works for now, but if someone finds a way to
do parallel bus discovery we'll have to revisit it.  Thanks,

        Alex

-- 
Alex Williamson                             HP Open Source & Linux Org.


_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel