Re: [PATCH v2] KVM: SVM: Disable AVIC before setting V_IRQ

[Maxim Levitsky]

On Thu, 2020-05-07 at 10:27 +0200, Paolo Bonzini wrote:
> On 07/05/20 04:35, Suravee Suthikulpanit wrote:
> > The commit 64b5bd270426 ("KVM: nSVM: ignore L1 interrupt window
> > while running L2 with V_INTR_MASKING=1") introduced a WARN_ON,
> > which checks if AVIC is enabled when trying to set V_IRQ
> > in the VMCB for enabling irq window.
> > 
> > The following warning is triggered because the requesting vcpu
> > (to deactivate AVIC) does not get to process APICv update request
> > for itself until the next #vmexit.
> > 
> > WARNING: CPU: 0 PID: 118232 at arch/x86/kvm/svm/svm.c:1372 enable_irq_window+0x6a/0xa0 [kvm_amd]
> >  RIP: 0010:enable_irq_window+0x6a/0xa0 [kvm_amd]
> >  Call Trace:
> >   kvm_arch_vcpu_ioctl_run+0x6e3/0x1b50 [kvm]
> >   ? kvm_vm_ioctl_irq_line+0x27/0x40 [kvm]
> >   ? _copy_to_user+0x26/0x30
> >   ? kvm_vm_ioctl+0xb3e/0xd90 [kvm]
> >   ? set_next_entity+0x78/0xc0
> >   kvm_vcpu_ioctl+0x236/0x610 [kvm]
> >   ksys_ioctl+0x8a/0xc0
> >   __x64_sys_ioctl+0x1a/0x20
> >   do_syscall_64+0x58/0x210
> >   entry_SYSCALL_64_after_hwframe+0x44/0xa9
> > 
> > Fixes by sending APICV update request to all other vcpus, and
> > immediately update APIC for itself.
> > 
> > Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@xxxxxxx>
> > Link: https://lkml.org/lkml/2020/5/2/167
> > Fixes: 64b5bd270426 ("KVM: nSVM: ignore L1 interrupt window while running L2 with V_INTR_MASKING=1")
> > ---
> >  arch/x86/kvm/x86.c | 13 ++++++++++++-
> >  1 file changed, 12 insertions(+), 1 deletion(-)
> > 
> > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
> > index df473f9..69a01ea 100644
> > --- a/arch/x86/kvm/x86.c
> > +++ b/arch/x86/kvm/x86.c
> > @@ -8085,6 +8085,7 @@ void kvm_vcpu_update_apicv(struct kvm_vcpu *vcpu)
> >   */
> >  void kvm_request_apicv_update(struct kvm *kvm, bool activate, ulong bit)
> >  {
> > +	struct kvm_vcpu *except;
> >  	unsigned long old, new, expected;
> >  
> >  	if (!kvm_x86_ops.check_apicv_inhibit_reasons ||
> > @@ -8110,7 +8111,17 @@ void kvm_request_apicv_update(struct kvm *kvm, bool activate, ulong bit)
> >  	trace_kvm_apicv_update_request(activate, bit);
> >  	if (kvm_x86_ops.pre_update_apicv_exec_ctrl)
> >  		kvm_x86_ops.pre_update_apicv_exec_ctrl(kvm, activate);
> > -	kvm_make_all_cpus_request(kvm, KVM_REQ_APICV_UPDATE);
> > +
> > +	/*
> > +	 * Sending request to update APICV for all other vcpus,
> > +	 * while update the calling vcpu immediately instead of
> > +	 * waiting for another #VMEXIT to handle the request.
> > +	 */
> > +	except = kvm_get_running_vcpu();
> > +	kvm_make_all_cpus_request_except(kvm, KVM_REQ_APICV_UPDATE,
> > +					 except);
> > +	if (except)
> > +		kvm_vcpu_update_apicv(except);
> >  }
> >  EXPORT_SYMBOL_GPL(kvm_request_apicv_update);
> >  
> > 
> 
> Queued, thanks.
> 
> Paolo
> 
I tested this patch today on top of kvm/queue,
the patch that add kvm_make_all_cpus_request_except and this patch
(the former patch needs slight adjustment to apply).

Best regards,
	Maxim Levitsky