Re: mm: deadlock between get_online_cpus/pcpu_alloc

[Michal Hocko <mhocko@kernel.org>]

On Tue 07-02-17 13:03:50, Mel Gorman wrote:
> On Tue, Feb 07, 2017 at 12:43:27PM +0100, Michal Hocko wrote:
> > > Right. The unbind operation can set a mask that is any allowable CPU and
> > > the final process_work is not done in a context that prevents
> > > preemption.
> > > 
> > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > index 3b93879990fd..7af165d308c4 100644
> > > --- a/mm/page_alloc.c
> > > +++ b/mm/page_alloc.c
> > > @@ -2342,7 +2342,14 @@ void drain_local_pages(struct zone *zone)
> > >  
> > >  static void drain_local_pages_wq(struct work_struct *work)
> > >  {
> > > +	/*
> > > +	 * Ordinarily a drain operation is bound to a CPU but may be unbound
> > > +	 * after a CPU hotplug operation so it's necessary to disable
> > > +	 * preemption for the drain to stabilise the CPU ID.
> > > +	 */
> > > +	preempt_disable();
> > >  	drain_local_pages(NULL);
> > > +	preempt_enable_no_resched();
> > >  }
> > >  
> > >  /*
> > [...]
> > > @@ -6711,7 +6714,16 @@ static int page_alloc_cpu_dead(unsigned int cpu)
> > >  {
> > >  
> > >  	lru_add_drain_cpu(cpu);
> > > +
> > > +	/*
> > > +	 * A per-cpu drain via a workqueue from drain_all_pages can be
> > > +	 * rescheduled onto an unrelated CPU. That allows the hotplug
> > > +	 * operation and the drain to potentially race on the same
> > > +	 * CPU. Serialise hotplug versus drain using pcpu_drain_mutex
> > > +	 */
> > > +	mutex_lock(&pcpu_drain_mutex);
> > >  	drain_pages(cpu);
> > > +	mutex_unlock(&pcpu_drain_mutex);
> > 
> > You cannot put sleepable lock inside the preempt disbaled section...
> > We can make it a spinlock right?
> > 
> 
> The CPU down callback can hold a mutex and at least he SLUB callback
> already does so. That gives
> 
> page_alloc_cpu_dead
>   mutex_lock
>     drain_pages
>   mutex_unlock
> 
> drain_all_pages
>   mutex_lock
>     queue workqueue
>       drain_local_pages_wq
>         preempt_disable
>         drain_local_pages
>         drain_pages
>         preempt_enable
>    flush queues
>  mutex_unlock
> 
> I must be blind or maybe it's rushing between multiple concerns but which
> sleepable lock is of concern?

I thought the cpu hotplug callback was non-preemptible. This is not the
case as mentioned in other reply. The pcpu_drain_mutex in the hotplug
callback is alright. Sorry about the confusion! I am still wondering
whether the lock is really needed. See the other reply.

-- 
Michal Hocko
SUSE Labs