Before this patch we incorrectly enter the guest without requesting an
interrupt window if the IRQ chip is split between user space and the
kernel.
Because lapic_in_kernel no longer implies the PIC is in the kernel, this
patch tests pic_in_kernel to determining whether an interrupt window
should be requested when entering the guest.
If the APIC is in the kernel and we request an interrupt window the
guest will return immediately. If the APIC is masked the guest will not
not make forward progress and unmask it, leading to a loop when KVM
reenters and requests again. This patch adds a check to ensure the APIC
is ready to accept an interrupt before requesting a window.
Reviewed-by: Andy Honig <ahonig@xxxxxxxxxx>
Signed-off-by: Matt Gingell <gingell@xxxxxxxxxx>
---
arch/x86/kvm/x86.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 456973b..fd64dee 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -6258,10 +6258,14 @@ void kvm_arch_mmu_notifier_invalidate_page(struct kvm *kvm,
static int vcpu_enter_guest(struct kvm_vcpu *vcpu)
{
int r;
- bool req_int_win = !lapic_in_kernel(vcpu) &&
+ bool req_int_win = !pic_in_kernel(vcpu->kvm) &&
vcpu->run->request_interrupt_window;
+
bool req_immediate_exit = false;
+ if (irqchip_split(vcpu->kvm))
+ req_int_win = req_int_win && kvm_apic_accept_pic_intr(vcpu);
+
if (vcpu->requests) {
if (kvm_check_request(KVM_REQ_MMU_RELOAD, vcpu))
kvm_mmu_unload(vcpu);
--
2.6.0.rc2.230.g3dd15c0
--
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
