Re: [Xen-devel] Xen pciback not working: "address space collision: [mem ...] conflicts with System RAM [mem ...]"

[Andrea Brugiolo <andrea.brugiolo@xxxxxxxx>]

On Fri, Aug 30, 2013 at 12:08:05PM -0400, Konrad Rzeszutek Wilk wrote:
> On Wed, Aug 28, 2013 at 06:44:50PM +0200, Andrea Brugiolo wrote:
> > On Wed, Aug 28, 2013 at 11:45:01AM -0400, Konrad Rzeszutek Wilk wrote:
> > > On Tue, Aug 27, 2013 at 11:03:02PM +0200, Andrea Brugiolo wrote:
> > > > On Fri, Aug 23, 2013 at 02:54:30PM -0400, Konrad Rzeszutek Wilk wrote:
> > > > > > > Awaiting your response with the usage of 'xl'.
> > > > > > 
> > > > > > Now I am using `xl' and the
> > > > > > 
> > > > > >     e820_host  =   1
> > > > > > 
> > > > > > option in the domU configuration but I get the same result as before
> > > > > > and no device attached:
> > > > > 
> > > > > And this is with Xen 4.2 I presume?
> > > > 
> > > > No, Xen 4.1 (Debian 4.1.4-3+deb7u1 Xen hypervisor and libs)
> > > 
> > > OK, Xen 4.1 did not have the code to handle the e820_host so hence the
> > > reason for this not working :-(.
> > > 
> > > You will need to upgrade to Xen 4.2 at least. I don't know how that
> > > is done under Debian.
> > 
> > Argh... the 4.2 hypervisor is still in the Debian unstable release so
> > I am not sure about how I could put that into production without
> > relevant breaks.
> > 
> > As a last resort, would you suggest tecniques other than pciback to
> > delegate a device to a guest in Xen 4.1?
> 
> The other option is to limit the amount of memory the guest will use
> from 16GB to say 2GB. That would work as well.

I confirm, I have limited the guest memory to 2 GB and now I can
attach the device as I used to do.

> Sure thing. Sadly the patch did not make Xen 4.1 hence the need to use
> a more modern version of Xen stack.

I am looking forward to have Xen 4.2 or higher in Debian stable.. :)

Thank you again

Andrea

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