[Xen-users] question about hvmloader and "No PCI config space access function found"

[Kai Schaetzl <maillists@xxxxxxxxxxxxx>]

Is the error message "No PCI config space access function found" connected 
to hvmloader enabled hypervisor kernels?

I started to get this message in VMs yesterday. Yesterday I compiled 3.2.1 
according to the instructions at 
http://grantmcwilliams.com/index.php?
option=com_content&view=article&id=204:recompile-xen-321-for-centos-
5&catid=97:xen-howtos&Itemid=379

I started to get this message yesterday, but *not* after installing the 
new rpms, but already before it. I only noticed it today. It's possible 
that there is some time glitch that makes it appear as having already 
happened before. The sure thing is it started yesterday.

So, I wonder if this has anything to do with hvm. Per the instructions at 
the grantmcwilliams site I uncommented the "# /usr/lib/xen/boot/hvmloader" 
line from that spec file. I first compiled with it commented and then 
uncommented and ran both hypervisors. Is it correct to assume that I only 
need that line if I want to run fully virtualized guests? (I do not want 
to.) Or does it have other functionality as well? I just uncommented and 
compiled with it, because I was troubleshooting the freqency scaling 
problem and wanted to see if it makes a difference. As I installed this 
after the first test without hvmloader I'm now running with it.

So, basic question: is the error message harmless or even expected if I 
run paravirtualized guests? Is there any advantage or disadvantage 
compiling with hvmloader if I do not run fully virtualized guests?


Kai

-- 
Kai Schätzl, Berlin, Germany
Get your web at Conactive Internet Services: http://www.conactive.com




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