26.06.2021 03:15, Jim Mattson пишет:
If the squashed exception was a trap, it's now lost.
OK, below are 2 more patches, each of
them alone is fixing the problem.
---
--- x86.c.old 2021-03-20 12:51:14.000000000 +0300
+++ x86.c 2021-06-27 00:38:45.547355116 +0300
@@ -9094,6 +9094,7 @@
if (req_immediate_exit)
kvm_make_request(KVM_REQ_EVENT, vcpu);
kvm_x86_ops.cancel_injection(vcpu);
+ kvm_clear_exception_queue(vcpu);
if (unlikely(vcpu->arch.apic_attention))
kvm_lapic_sync_from_vapic(vcpu);
out:
---
Or:
---
--- x86.c.old 2021-03-20 12:51:14.000000000 +0300
+++ x86.c 2021-06-27 00:47:06.958618185 +0300
@@ -1783,8 +1783,7 @@
bool kvm_vcpu_exit_request(struct kvm_vcpu *vcpu)
{
xfer_to_guest_mode_prepare();
- return vcpu->mode == EXITING_GUEST_MODE ||
kvm_request_pending(vcpu) ||
- xfer_to_guest_mode_work_pending();
+ return vcpu->mode == EXITING_GUEST_MODE ||
kvm_request_pending(vcpu);
}
EXPORT_SYMBOL_GPL(kvm_vcpu_exit_request);
---
Still not a clue?
