This issue exists in kernel version 6.3-rc1 or above. The issue is
introduced by the commit 8e6ed96cdd50 ("KVM: x86: fire timer when it is
migrated and expired, and in oneshot mode"). The issue occurs on Intel
platform which APIC virtualization and posted interrupt processing.
The issue is first discovered when running the Android Emulator which
is based on QEMU 2.12. I can reproduce the issue with QEMU 8.0.4 in
Debian 12.
With the above commit, the timer gets fired immediately inside the
KVM_SET_LAPIC call when loading the snapshot. On such Intel platforms,
this eventually leads to setting the corresponding PIR bit. However,
the whole PIR bits get cleared later in the same KVM_SET_LAPIC call.
Missing lapic timer interrupt freeze the guest kernel.
Signed-off-by: Haitao Shan <hshan@xxxxxxxxxx>
---
arch/x86/kvm/lapic.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kvm/lapic.c b/arch/x86/kvm/lapic.c
index a983a16163b1..6f73406b875a 100644
--- a/arch/x86/kvm/lapic.c
+++ b/arch/x86/kvm/lapic.c
@@ -2977,14 +2977,14 @@ int kvm_apic_set_state(struct kvm_vcpu *vcpu, struct kvm_lapic_state *s)
apic_update_lvtt(apic);
apic_manage_nmi_watchdog(apic, kvm_lapic_get_reg(apic, APIC_LVT0));
update_divide_count(apic);
- __start_apic_timer(apic, APIC_TMCCT);
- kvm_lapic_set_reg(apic, APIC_TMCCT, 0);
kvm_apic_update_apicv(vcpu);
if (apic->apicv_active) {
static_call_cond(kvm_x86_apicv_post_state_restore)(vcpu);
static_call_cond(kvm_x86_hwapic_irr_update)(vcpu, apic_find_highest_irr(apic));
static_call_cond(kvm_x86_hwapic_isr_update)(apic_find_highest_isr(apic));
}
+ __start_apic_timer(apic, APIC_TMCCT);
+ kvm_lapic_set_reg(apic, APIC_TMCCT, 0);
kvm_make_request(KVM_REQ_EVENT, vcpu);
if (ioapic_in_kernel(vcpu->kvm))
kvm_rtc_eoi_tracking_restore_one(vcpu);
--
2.42.0.283.g2d96d420d3-goog
