diff options
| author | Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> | 2011-08-24 09:54:24 -0700 |
|---|---|---|
| committer | Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com> | 2011-08-24 09:54:24 -0700 |
| commit | f1c39625d63c9f8eba8f036429c10a9cb9e32920 (patch) | |
| tree | 8ca13e0d5f9c0db22f9933c07a170d865a582fe3 /arch | |
| parent | fcb8ce5cfe30ca9ca5c9a79cdfe26d1993e65e0c (diff) | |
| download | op-kernel-dev-f1c39625d63c9f8eba8f036429c10a9cb9e32920.zip op-kernel-dev-f1c39625d63c9f8eba8f036429c10a9cb9e32920.tar.gz | |
xen: use non-tracing preempt in xen_clocksource_read()
The tracing code used sched_clock() to get tracing timestamps, which
ends up calling xen_clocksource_read(). xen_clocksource_read() must
disable preemption, but if preemption tracing is enabled, this results
in infinite recursion.
I've only noticed this when boot-time tracing tests are enabled, but it
seems like a generic bug. It looks like it would also affect
kvm_clocksource_read().
Reported-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Avi Kivity <avi@redhat.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Diffstat (limited to 'arch')
| -rw-r--r-- | arch/x86/xen/time.c | 5 |
1 files changed, 3 insertions, 2 deletions
diff --git a/arch/x86/xen/time.c b/arch/x86/xen/time.c index 5158c50..163b467 100644 --- a/arch/x86/xen/time.c +++ b/arch/x86/xen/time.c @@ -168,9 +168,10 @@ cycle_t xen_clocksource_read(void) struct pvclock_vcpu_time_info *src; cycle_t ret; - src = &get_cpu_var(xen_vcpu)->time; + preempt_disable_notrace(); + src = &__get_cpu_var(xen_vcpu)->time; ret = pvclock_clocksource_read(src); - put_cpu_var(xen_vcpu); + preempt_enable_notrace(); return ret; } |
