22.06.2021 01:33, Jim Mattson пишет:
Maybe what you want is run->ready_for_interrupt_injection? And, if
I implemented this suggestion with
the patch below:
---
--- x86.c.old 2021-03-20 12:51:14.000000000 +0300
+++ x86.c 2021-06-26 16:51:17.366592310 +0300
@@ -4109,7 +4109,9 @@
static int kvm_vcpu_ready_for_interrupt_injection(struct kvm_vcpu *vcpu)
{
return kvm_arch_interrupt_allowed(vcpu) &&
- kvm_cpu_accept_dm_intr(vcpu);
+ kvm_cpu_accept_dm_intr(vcpu) &&
+ !vcpu->arch.exception.pending &&
+ !vcpu->arch.exception.injected;
}
static int kvm_vcpu_ioctl_interrupt(struct kvm_vcpu *vcpu,
---
With that change I indeed can
look into run->ready_for_interrupt_injection
and avoid the race.
That means a cpu-specific work-around
in my code, but at least that works.
But without this change,
run->ready_for_interrupt_injection
just lies.
Does this bring us any closer to the
understanding of what's going on?
If not, what should I find out next?
