| Message ID | 1599620043-12908-1-git-send-email-wanpengli@tencent.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [1/3] KVM: SVM: Get rid of handle_fastpath_set_msr_irqoff() | expand |
Wanpeng Li <kernellwp@gmail.com> writes: > From: Wanpeng Li <wanpengli@tencent.com> > > Analysis from Sean: > > | svm->next_rip is reset in svm_vcpu_run() only after calling > | svm_exit_handlers_fastpath(), which will cause SVM's > | skip_emulated_instruction() to write a stale RIP. > This should only happen when svm->vmcb->control.next_rip is not set by hardware as skip_emulated_instruction() itself sets 'svm->next_rip' otherwise, right? > Let's get rid of handle_fastpath_set_msr_irqoff() in svm_exit_handlers_fastpath() > to have a quick fix. which in the light of the whole seeries seems to be appropriate, so: Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com> > > Reported-by: Paul K. <kronenpj@kronenpj.dyndns.org> > Suggested-by: Sean Christopherson <sean.j.christopherson@intel.com> > Cc: Paul K. <kronenpj@kronenpj.dyndns.org> > Cc: <stable@vger.kernel.org> # v5.8-rc1+ > Fixes: 404d5d7bff0d (KVM: X86: Introduce more exit_fastpath_completion enum values) > Signed-off-by: Wanpeng Li <wanpengli@tencent.com> > --- > arch/x86/kvm/svm/svm.c | 5 ----- > 1 file changed, 5 deletions(-) > > diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c > index 19e622a..c61bc3b 100644 > --- a/arch/x86/kvm/svm/svm.c > +++ b/arch/x86/kvm/svm/svm.c > @@ -3349,11 +3349,6 @@ static void svm_cancel_injection(struct kvm_vcpu *vcpu) > > static fastpath_t svm_exit_handlers_fastpath(struct kvm_vcpu *vcpu) > { > - if (!is_guest_mode(vcpu) && > - to_svm(vcpu)->vmcb->control.exit_code == SVM_EXIT_MSR && > - to_svm(vcpu)->vmcb->control.exit_info_1) > - return handle_fastpath_set_msr_irqoff(vcpu); > - > return EXIT_FASTPATH_NONE; > }
On Wed, 9 Sep 2020 at 16:23, Vitaly Kuznetsov <vkuznets@redhat.com> wrote: > > Wanpeng Li <kernellwp@gmail.com> writes: > > > From: Wanpeng Li <wanpengli@tencent.com> > > > > Analysis from Sean: > > > > | svm->next_rip is reset in svm_vcpu_run() only after calling > > | svm_exit_handlers_fastpath(), which will cause SVM's > > | skip_emulated_instruction() to write a stale RIP. > > > > This should only happen when svm->vmcb->control.next_rip is not set by > hardware as skip_emulated_instruction() itself sets 'svm->next_rip' > otherwise, right? The bug is reported here https://bugzilla.kernel.org/show_bug.cgi?id=209155 , the old machine which the reporter uses doesn't have NRIP save on #VMEXIT support. :) Wanpeng
diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c index 19e622a..c61bc3b 100644 --- a/arch/x86/kvm/svm/svm.c +++ b/arch/x86/kvm/svm/svm.c @@ -3349,11 +3349,6 @@ static void svm_cancel_injection(struct kvm_vcpu *vcpu) static fastpath_t svm_exit_handlers_fastpath(struct kvm_vcpu *vcpu) { - if (!is_guest_mode(vcpu) && - to_svm(vcpu)->vmcb->control.exit_code == SVM_EXIT_MSR && - to_svm(vcpu)->vmcb->control.exit_info_1) - return handle_fastpath_set_msr_irqoff(vcpu); - return EXIT_FASTPATH_NONE; }