Messages in this thread | | | Date | Tue, 1 Jul 2008 11:58:54 +0530 | From | Dhaval Giani <> | Subject | Re: [PATCH] fix rcu vs hotplug race |
| |
On Tue, Jul 01, 2008 at 08:16:01AM +0200, Ingo Molnar wrote: > > * Gautham R Shenoy <ego@in.ibm.com> wrote: > > > So, it's the design of stop_machine() that's preventing someone from > > updating the cpu_online_map while force_quiescent_state() is > > performing the cpu_is_online() check. Becase we always call > > force_quiescent_state() with irqs disabled :) > > Paul, do you concur? I'll apply the commit in the form below to > tip/core/urgent if Paul agrees. > > Ingo
Ingo,
I believe Gautham's fix at http://lkml.org/lkml/2008/6/27/9 is better and also explains it better.
Thanks. -- regards, Dhaval
| |