Skip to contentSkip to conversation

Pending signals prevent entry to the guest

DISCUSSION Steven Stovall 3 participants 3 messages Jan 21, 2009 1 message not in the archive KVM

lore
Share & tools
lore

Steven Stovall

author
#1 lore raw
In reply to a message not in the archive

I posted about a week ago regarding a very different behavior with a host-side app vis a vis kvm-77 vs kvm-33. Pending signals (SIGIO and SIGALRM) never seem to be dequeued so that I never kvm_guest_enter()...it appears that only the dynticks alarm_timer gets enabled which only handles SIGALRM, while the rtc alarm_timer which handles SIGIO does not. I have tried to make sure both are enabled, but I still have perpetual pending signals...is this possible?? Or is there some corruption somewhere?

Steven

#2 lore raw
Quoted 2 lines · Steven Stovall

I posted about a week ago regarding a very different behavior with a host-side app vis a vis kvm-77 vs kvm-33. Pending signals (SIGIO and SIGALRM) never seem to be dequeued so that I never kvm_guest_enter()...it appears that only the dynticks alarm_timer gets enabled which only handles SIGALRM, while the rtc alarm_timer which handles SIGIO does not. I have tried to make sure both are enabled, but I still have perpetual pending signals...is this possible?? Or is there some corruption somewhere?

How many signal_exits do you see on kvm_stat?
What's your host kernel version?
It wasn't clean if you use dyn tick or rtc clock in qemu. Please post the command line.

Quoted 6 lines

Steven
--
To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#3 lore raw
In reply to Steven Stovall
Quoted 2 lines · Steven Stovall

I posted about a week ago regarding a very different behavior with a host-side app vis a vis kvm-77 vs kvm-33. Pending signals (SIGIO and SIGALRM) never seem to be dequeued so that I never kvm_guest_enter()...it appears that only the dynticks alarm_timer gets enabled which only handles SIGALRM, while the rtc alarm_timer which handles SIGIO does not. I have tried to make sure both are enabled, but I still have perpetual pending signals...is this possible?? Or is there some corruption somewhere?

You must call sigtimedwait() or one of its friends with the pending signal enabled after KVM_VCPU_RUN exits with -EINTR.

Signature · 1 line
error compiling committee.c: too many arguments to function