| Message ID | 20190426065414.GC12768@blackberry |
|---|---|
| State | Accepted |
| Headers | show |
| Series | KVM: PPC: Book3S HV: XIVE: Clear escalation interrupt pointers on device close | expand |
On 4/26/19 8:54 AM, Paul Mackerras wrote: > This adds code to ensure that after a XIVE or XICS-on-XIVE KVM device > is closed, KVM will not try to enable or disable any of the escalation > interrupts for the VCPUs. Yes. This is a required cleanup. Reviewed-by: Cédric Le Goater <clg@kaod.org> Thanks, C. > We don't have to worry about races between > clearing the pointers and use of the pointers by the XIVE context > push/pull code, because the callers hold the vcpu->mutex, which is > also taken by the KVM_RUN code. Therefore the vcpu cannot be entering > or exiting the guest concurrently. > > Signed-off-by: Paul Mackerras <paulus@ozlabs.org> > --- > arch/powerpc/kvm/book3s_xive.c | 15 +++++++++++++++ > 1 file changed, 15 insertions(+) > > diff --git a/arch/powerpc/kvm/book3s_xive.c b/arch/powerpc/kvm/book3s_xive.c > index 4280cd8..4953957 100644 > --- a/arch/powerpc/kvm/book3s_xive.c > +++ b/arch/powerpc/kvm/book3s_xive.c > @@ -1096,6 +1096,21 @@ void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu) > arch_spin_unlock(&sb->lock); > } > } > + > + /* Disable vcpu's escalation interrupt */ > + if (vcpu->arch.xive_esc_on) { > + __raw_readq((void __iomem *)(vcpu->arch.xive_esc_vaddr + > + XIVE_ESB_SET_PQ_01)); > + vcpu->arch.xive_esc_on = false; > + } > + > + /* > + * Clear pointers to escalation interrupt ESB. > + * This is safe because the vcpu->mutex is held, preventing > + * any other CPU from concurrently executing a KVM_RUN ioctl. > + */ > + vcpu->arch.xive_esc_vaddr = 0; > + vcpu->arch.xive_esc_raddr = 0; > } > > void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu) >
diff --git a/arch/powerpc/kvm/book3s_xive.c b/arch/powerpc/kvm/book3s_xive.c index 4280cd8..4953957 100644 --- a/arch/powerpc/kvm/book3s_xive.c +++ b/arch/powerpc/kvm/book3s_xive.c @@ -1096,6 +1096,21 @@ void kvmppc_xive_disable_vcpu_interrupts(struct kvm_vcpu *vcpu) arch_spin_unlock(&sb->lock); } } + + /* Disable vcpu's escalation interrupt */ + if (vcpu->arch.xive_esc_on) { + __raw_readq((void __iomem *)(vcpu->arch.xive_esc_vaddr + + XIVE_ESB_SET_PQ_01)); + vcpu->arch.xive_esc_on = false; + } + + /* + * Clear pointers to escalation interrupt ESB. + * This is safe because the vcpu->mutex is held, preventing + * any other CPU from concurrently executing a KVM_RUN ioctl. + */ + vcpu->arch.xive_esc_vaddr = 0; + vcpu->arch.xive_esc_raddr = 0; } void kvmppc_xive_cleanup_vcpu(struct kvm_vcpu *vcpu)
This adds code to ensure that after a XIVE or XICS-on-XIVE KVM device is closed, KVM will not try to enable or disable any of the escalation interrupts for the VCPUs. We don't have to worry about races between clearing the pointers and use of the pointers by the XIVE context push/pull code, because the callers hold the vcpu->mutex, which is also taken by the KVM_RUN code. Therefore the vcpu cannot be entering or exiting the guest concurrently. Signed-off-by: Paul Mackerras <paulus@ozlabs.org> --- arch/powerpc/kvm/book3s_xive.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+)