We've been seeing this failure in the gdb_netstep test lately.
09:07:59.624655 km_gdb_set_thread_vc 1648 memchec Assertion `vcpu->state != IN_GUEST' failed: No such file or directory
The only recent change "seemingly" related to this was a change where we save the value of errno for use later in km_vcpu_one_kvm_run(). That change shifted the setting of state to IN_GUEST one statement further away from the ioctl( KVM_RUN ) call.
This change is more an experiment to see if it "helps" fix the assertion failure. If it does "fix" the problem, we will know there is a race here that we've only made less likely but probably not fixed.
We've been seeing this failure in the gdb_netstep test lately.
09:07:59.624655 km_gdb_set_thread_vc 1648 memchec Assertion `vcpu->state != IN_GUEST' failed: No such file or directory
The only recent change "seemingly" related to this was a change where we save the value of errno for use later in km_vcpu_one_kvm_run(). That change shifted the setting of state to IN_GUEST one statement further away from the ioctl( KVM_RUN ) call.
This change is more an experiment to see if it "helps" fix the assertion failure. If it does "fix" the problem, we will know there is a race here that we've only made less likely but probably not fixed.