RE: [Xen-users] AMD-8131

["Petersson, Mats" <Mats.Petersson@xxxxxxx>]

> -----Original Message-----
> From: xen-users-bounces@xxxxxxxxxxxxxxxxxxx 
> [mailto:xen-users-bounces@xxxxxxxxxxxxxxxxxxx] On Behalf Of 
> Mark Williamson
> Sent: 30 January 2006 14:24
> To: xen-users@xxxxxxxxxxxxxxxxxxx
> Cc: Ernst Bachmann; Sebastian Böhm
> Subject: Re: [Xen-users] AMD-8131
> 
> > > Did you check the dom0 kernel config if it contains the 
> neccessary 
> > > functions? If you create a non-xen kernel with the same 
> config (copy 
> > > the .config into a vanilla kernel dir, run make oldconfig), does 
> > > that one work with PCI-X?
> >
> > This was the first thing I tried. And yes: without xen it 
> does work (i 
> > use the same kernel-config for xen2, and there pcix works also).
> 
> Unfortunately you can't just copy the config from the Xen2 
> kernel to the Xen3
> kernel: the PCI functionality that you saw in Xen 2 hasn't 
> been *removed*, it's just been *moved* into Linux itself and 
> out of Xen.
> 
> The Linux/Xen2 kernel config won't have the PCI quirk 
> handling code because PCI quirks were handled by Xen itself.  
> In Xen 3, Linux needs to have this enabled because Xen will 
> not help it anymore.
> 
> Can you verify that any config options for your board that 
> are enabled in your native Linux build are also enabled under 
> the Xen 3-patched Linux kernel?
> 
> Cheers,
> Mark
> 
The fix in quirks.c isn't something that will make the devices behind the PCI-X 
bridge disappear or appear - it sets a signle bit in a configuration register, 
which I think deals with Message Signaled Interrupts [based on the printout 
comments]. 

If the devices behind 8131 don't appear, then there's something wrong with the 
PCI enumeration, which not directly related to the quirks.c code. 

--
Mats


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