Re: k10+ pstate fix.

2012-01-06 Thread Jonathan Gray
On Tue, Jan 03, 2012 at 01:11:05PM -0500, Brynet wrote: > Check for valid ACPI _PSS object before doing the MSR read, fixes KVM > brokeness > as reported/tested by Walter Haidinger. > > It could also potentially avoid a general protection fault on theoretically > real (..yet so far not witnesse

k10+ pstate fix.

2012-01-03 Thread Brynet
Check for valid ACPI _PSS object before doing the MSR read, fixes KVM brokeness as reported/tested by Walter Haidinger. It could also potentially avoid a general protection fault on theoretically real (..yet so far not witnessed) systems. I'm still open to doing the CPUID check, but as I explai