On Mon, May 30, 2011 at 02:16:41PM +0200, Markus Schade wrote:
> On Thu, 26 May 2011, Joerg Roedel wrote:
>
> >On Thu, May 26, 2011 at 05:20:32PM +0200, Markus Schade wrote:
> >>On 05/26/2011 01:28 PM, Markus Schade wrote:
> >>>On 05/26/2011 08:44 AM, Avi Kivity wrote:
> >>>>On 05/25/2011 09:49 AM, Markus Schade wrote:
> >>>>>
> >>>>>Git bisect tells me that this is the first bad commit:
> >>>>>-----------------%<-----------------
> >>>>>aff48baa34c033318ad322ecbf2e4bcd891b29ca is the first bad commit
> >>>>
> >>>>Does your machine have ept?
> >>>>
> >>>>(cat /sys/module/kvm_intel/parameters/ept)
> >>>>
> >>>
> >>>Sure. (Assuming that the Y means yes). I am no C developer, but I was
> >>>wondering, if the issue is related to the difference in
> >>>ept_update_paging_mode_cr0 between original patch in the kvm git and the
> >>>linux-2.6. git tree.
> >>
> >>I have re-added the missing 4 lines to ept_update_paging_mode_cr0 in
> >>vmx.c, which resolves this issue for in Kernel 2.6,37 and 2.6.39.
> >>I haven't tested all guests, but neither Squeeze nor 2008 R2 reboot anymore.
> >
> >Hmm, not 100% sure, but the issue might be that the arch.cr3 value does
> >not make it into the vmcs after cr3 was changed from user-space? This
> >would also be fixed with the change above.
>
> So, basically, Linus just has to re-pull, since the
> "change" I made is already in the tree. Or is there another way to solve
> this? I find it hard to believe that I am the only one using
> save/restore on Nehalem CPUs or that this bug only affects me.
> Although I can verify this on dozens of machines.
Markus,
Please check whether the following patch fixes the problem for you.
-----
Only decache guest CR3 value if vcpu->arch.cr3 is stale.
Signed-off-by: Marcelo Tosatti <mtosatti@xxxxxxxxxx>
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 750b0ff..772c4d2 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -2049,7 +2049,9 @@ static void ept_update_paging_mode_cr0(unsigned long *hw_cr0,
unsigned long cr0,
struct kvm_vcpu *vcpu)
{
- vmx_decache_cr3(vcpu);
+
+ if (!test_bit(VCPU_EXREG_CR3, (ulong *)&vcpu->arch.regs_avail))
+ vmx_decache_cr3(vcpu);
if (!(cr0 & X86_CR0_PG)) {
/* From paging/starting to nonpaging */
vmcs_write32(CPU_BASED_VM_EXEC_CONTROL,
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
