* Guest hangs after some stress tests
@ 2012-05-17 19:49 Sasha Levin
2012-05-17 20:05 ` Gleb Natapov
0 siblings, 1 reply; 5+ messages in thread
From: Sasha Levin @ 2012-05-17 19:49 UTC (permalink / raw)
To: Gleb Natapov, Avi Kivity, Marcelo Tosatti; +Cc: kvm, Dave Jones
Hi all,
I've observed a guest hang after putting a little stress testing on
it. It basically hangs in ~10 secs after running several trinity
instances inside a KVM tools guest.
The hang is easy to describe and reproduce:
1. No userspace exits observed
2. kvm_stat looks as follows (with all other fields at 0):
kvm_entry 3631869 11819
kvm_apic_accept_irq 1078688 11819
kvm_exit 3631844 11819
I've bisected it down to:
9b72d3b07dd99ac8ab2b84de5004a295af460536 is the first bad commit
commit 9b72d3b07dd99ac8ab2b84de5004a295af460536
Author: Gleb Natapov <gleb@redhat.com>
Date: Mon Apr 30 14:45:49 2012 +0300
KVM guest: make kvm_para_available() check hypervisor bit reading cpuid leaf
This cpuid range does not exist on real HW and Intel spec says that
"Information returned for highest basic information leaf" will be
returned. Not very well defined.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: Guest hangs after some stress tests 2012-05-17 19:49 Guest hangs after some stress tests Sasha Levin @ 2012-05-17 20:05 ` Gleb Natapov 2012-05-17 20:14 ` Sasha Levin 0 siblings, 1 reply; 5+ messages in thread From: Gleb Natapov @ 2012-05-17 20:05 UTC (permalink / raw) To: Sasha Levin; +Cc: Avi Kivity, Marcelo Tosatti, kvm, Dave Jones On Thu, May 17, 2012 at 09:49:08PM +0200, Sasha Levin wrote: > Hi all, > > I've observed a guest hang after putting a little stress testing on > it. It basically hangs in ~10 secs after running several trinity > instances inside a KVM tools guest. > And do you set hypervisor bit in KVM tool? > The hang is easy to describe and reproduce: > > 1. No userspace exits observed > 2. kvm_stat looks as follows (with all other fields at 0): > > kvm_entry 3631869 11819 > kvm_apic_accept_irq 1078688 11819 > kvm_exit 3631844 11819 > > I've bisected it down to: > > 9b72d3b07dd99ac8ab2b84de5004a295af460536 is the first bad commit > commit 9b72d3b07dd99ac8ab2b84de5004a295af460536 > Author: Gleb Natapov <gleb@redhat.com> > Date: Mon Apr 30 14:45:49 2012 +0300 > > KVM guest: make kvm_para_available() check hypervisor bit reading cpuid leaf > > This cpuid range does not exist on real HW and Intel spec says that > "Information returned for highest basic information leaf" will be > returned. Not very well defined. > > Signed-off-by: Gleb Natapov <gleb@redhat.com> > Signed-off-by: Avi Kivity <avi@redhat.com> -- Gleb. ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Guest hangs after some stress tests 2012-05-17 20:05 ` Gleb Natapov @ 2012-05-17 20:14 ` Sasha Levin 2012-05-17 20:19 ` Gleb Natapov 0 siblings, 1 reply; 5+ messages in thread From: Sasha Levin @ 2012-05-17 20:14 UTC (permalink / raw) To: Gleb Natapov; +Cc: Avi Kivity, Marcelo Tosatti, kvm, Dave Jones On Thu, May 17, 2012 at 10:05 PM, Gleb Natapov <gleb@redhat.com> wrote: > On Thu, May 17, 2012 at 09:49:08PM +0200, Sasha Levin wrote: >> Hi all, >> >> I've observed a guest hang after putting a little stress testing on >> it. It basically hangs in ~10 secs after running several trinity >> instances inside a KVM tools guest. >> > And do you set hypervisor bit in KVM tool? We're basically forwarding the cpuid of the host, so I assume that no? ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Guest hangs after some stress tests 2012-05-17 20:14 ` Sasha Levin @ 2012-05-17 20:19 ` Gleb Natapov 2012-05-25 15:37 ` Marcelo Tosatti 0 siblings, 1 reply; 5+ messages in thread From: Gleb Natapov @ 2012-05-17 20:19 UTC (permalink / raw) To: Sasha Levin; +Cc: Avi Kivity, Marcelo Tosatti, kvm, Dave Jones On Thu, May 17, 2012 at 10:14:35PM +0200, Sasha Levin wrote: > On Thu, May 17, 2012 at 10:05 PM, Gleb Natapov <gleb@redhat.com> wrote: > > On Thu, May 17, 2012 at 09:49:08PM +0200, Sasha Levin wrote: > >> Hi all, > >> > >> I've observed a guest hang after putting a little stress testing on > >> it. It basically hangs in ~10 secs after running several trinity > >> instances inside a KVM tools guest. > >> > > And do you set hypervisor bit in KVM tool? > > We're basically forwarding the cpuid of the host, so I assume that no? You really should, but hang shouldn't happen anyway. -- Gleb. ^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: Guest hangs after some stress tests 2012-05-17 20:19 ` Gleb Natapov @ 2012-05-25 15:37 ` Marcelo Tosatti 0 siblings, 0 replies; 5+ messages in thread From: Marcelo Tosatti @ 2012-05-25 15:37 UTC (permalink / raw) To: Gleb Natapov; +Cc: Sasha Levin, Avi Kivity, kvm, Dave Jones On Thu, May 17, 2012 at 11:19:04PM +0300, Gleb Natapov wrote: > On Thu, May 17, 2012 at 10:14:35PM +0200, Sasha Levin wrote: > > On Thu, May 17, 2012 at 10:05 PM, Gleb Natapov <gleb@redhat.com> wrote: > > > On Thu, May 17, 2012 at 09:49:08PM +0200, Sasha Levin wrote: > > >> Hi all, > > >> > > >> I've observed a guest hang after putting a little stress testing on > > >> it. It basically hangs in ~10 secs after running several trinity > > >> instances inside a KVM tools guest. > > >> > > > And do you set hypervisor bit in KVM tool? > > > > We're basically forwarding the cpuid of the host, so I assume that no? > You really should, but hang shouldn't happen anyway. > > -- > Gleb. If kvmclock is not enabled a number of problems arise (miscalibrated loops per jiffy, sched_clock using tsc's out of sync on smp). Should set the hypervisor cpuid bit (see 49ab56ac6e1b907b7da). ^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2012-05-25 15:45 UTC | newest] Thread overview: 5+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2012-05-17 19:49 Guest hangs after some stress tests Sasha Levin 2012-05-17 20:05 ` Gleb Natapov 2012-05-17 20:14 ` Sasha Levin 2012-05-17 20:19 ` Gleb Natapov 2012-05-25 15:37 ` Marcelo Tosatti
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox