Gerry Reno wrote:
  

Gerry Reno wrote:
    

Gerry Reno wrote:
      

Daniel P. Berrange wrote:
        

On Sat, May 23, 2009 at 03:43:12PM -0400, Gerry Reno wrote:
  
          

I upgraded the host from F10 to F11 (x86_64) with no issues.  Now when I 
start a F10 (i386) guest it runs very very slow.  I also see messages on 
the guest boot console about "clocksource tsc unstable" and some kernel 
oops.  Once it got far enough to start network I logged in and checked 
the clocksource and it currently is 'acpi_pm' even though the kernel 
line says clocksource=pit.  The available clocksources are acpi_pm, 
jiffies, and tsc.  I do not see 'pit' in the list.  How do I fix this issue?
    
            

If the guest runs 'extrememly' slowly then the most like thing is that 
it has fallen back to using QEMU emulation, instead of KVM hardware
acceleration. Check the /var/log/libvirt/qemu/$GUEST.log to see if there
is any mesage about not being able to open /dev/kvm.  Also make sure that
KVM modules are loaded, and that 'virsh capabilities' lists KVM as a valid
domain.

Daniel
  
          

Ok, I checked the guest log and it says:
/dev/kvm: no such file or directory.

So how do I make this node?  Shouldn't libvirt have made it for us?

        

Ok, once I got both kernel modules loaded, it created the /dev/kvm
device and now everything runs fine.

      

Well, not quite so fine.  If I reboot the machine then the kvm modules
are no longer loaded.  How do I keep these modules loaded?
    

I'm assuming that you haven't installed qemu from the F-11 packages.  If you
install the qemu-system-x86 on F-11, it comes with a file
/etc/sysconfig/modules/kvm.modules.  On bootup, any scripts in that directory
are executed, and that command automatically loads the appropriate modules for
you.  If you don't want to install the F-11 qemu-system-x86 package for some
reason, you'll have to arrange to do the same with a custom script in that
directory, or just in /etc/rc.d/rc.local.