On Wed, 2010-11-24 at 15:01 +0100, Philipp Matthias Hahn wrote:
> reopen 599507
> found 599507 2.6.32-27
> thanks
>
> Package: linux-2.6.32
> Severity: normal
>
> The bug is not fixed, since
> features/all/KVM-x86-Add-KVM_GET-SET_VCPU_EVENTS.patch from
> debian/pacthes/series/12 introduces the new function svm_set_nmi_mask(),
> which also has the wrong "1UL" instead of the correct "1ULL".
>
> --- a/arch/x86/kvm/svm.c 2010-11-24 14:29:05.000000000 +0100
> +++ b/arch/x86/kvm/svm.c 2010-11-24 13:24:35.000000000 +0100
> @@ -2566,10 +2479,10 @@ static void svm_set_nmi_mask(struct kvm_
>
> if (masked) {
> svm->vcpu.arch.hflags |= HF_NMI_MASK;
> - svm->vmcb->control.intercept |= (1UL << INTERCEPT_IRET);
> + svm->vmcb->control.intercept |= (1ULL << INTERCEPT_IRET);
> } else {
> svm->vcpu.arch.hflags &= ~HF_NMI_MASK;
> - svm->vmcb->control.intercept &= ~(1UL << INTERCEPT_IRET);
> + svm->vmcb->control.intercept &= ~(1ULL << INTERCEPT_IRET);
> }
> }
>
> Stable (2.6.32.26) applied the patch, but since it doesn't habe
> svm_set_nmi_mask(), its okay.
> Debian on the other hand has the back-ported svm_set_nmi_mask() and
> needs the above patch.
Sorry, I hadn't thought of that. I'll apply this patch, thanks.
Ben.
--
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.
Attachment:
signature.asc
Description: This is a digitally signed message part