[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 3 Mar 2016 08:31:35 +0100 (CET)
From: Thomas Gleixner <tglx@...utronix.de>
To: "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
cc: rafael.j.wysocki@...el.com, peterz@...radead.org, riel@...hat.com,
arjan@...ux.intel.com, rusty@...tcorp.com.au, tj@...nel.org,
srivatsa@....edu, torvalds@...ux-foundation.org,
bigeasy@...utronix.de, pjt@...gle.com, akpm@...ux-foundation.org,
hpa@...or.com, oleg@...hat.com, mingo@...nel.org,
linux-kernel@...r.kernel.org, rostedt@...dmis.org,
linux-tip-commits@...r.kernel.org
Subject: Re: [tip:smp/hotplug] rcu: Make CPU_DYING_IDLE an explicit call
On Wed, 2 Mar 2016, Paul E. McKenney wrote:
> > --- a/kernel/cpu.c
> > +++ b/kernel/cpu.c
> > @@ -762,6 +762,7 @@ void cpuhp_report_idle_dead(void)
> > BUG_ON(st->state != CPUHP_AP_OFFLINE);
> > st->state = CPUHP_AP_IDLE_DEAD;
> > complete(&st->done);
>
> Not to be repetitive or anything, but if we delay here, it can break
> RCU on a number of architectures. Either the CPU can be killed holding
> one of RCU's locks or RCU can wrongly see the CPU as still being alive.
> Either can prevent future RCU grace periods from ever completing, thus
> OOMing the system.
Thanks for the reminder. I wanted to fix that, but then forgot again. Fix is
on the way.
Thanks,
tglx
Powered by blists - more mailing lists