On Tue, 2022-03-08 at 18:54 +0100, Paolo Bonzini wrote: > In the first case the timer has expired and the event has been injected > by the time the state is retrieved, so you'd get a double event. > > In the second the timer expires and the event is injected _on the > source_ only. This is required because, if the destination for > whatever reason aborts the migration, your change would introduced a > missed event. True. I'll change it to use hrtimer_active(). But first I'll concentrate on the one that was *really* the cause of the bug I've been chasing today.
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
