Re: mm: deadlock between get_online_cpus/pcpu_alloc

[Michal Hocko]

On Tue 07-02-17 12:54:48, Vlastimil Babka wrote:
> On 02/07/2017 12:43 PM, Michal Hocko wrote:
> > > 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?
> 
> Could we do flush_work() with a spinlock? Sounds bad too.

We surely cannot. I thought the lock would be gone in drain_all_pages,
we would deadlock with the lock there anyway... But it is true that we
would need a way to only allow one caller to get in. This is getting
messier and messier...

> Maybe we could just use the fact that the whole drain happens with disabled
> irq's and obtain the current cpu under that protection?

preempt_disable should be enough, no? The CPU callback is not called
from an IRQ context, right?
---
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 1ee49474207e..4a9a65479435 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2343,7 +2343,16 @@ void drain_local_pages(struct zone *zone)
 
 static void drain_local_pages_wq(struct work_struct *work)
 {
+	/*
+	 * drain_all_pages doesn't use proper cpu hotplug protection so
+	 * we can race with cpu offline when the WQ can move this from
+	 * a cpu pinned worker to an unbound one. We can operate on a different
+	 * cpu which is allright but we also have to make sure to not move to
+	 * a different one.
+	 */
+	preempt_disable();
 	drain_local_pages(NULL);
+	preempt_enable();
 }
 
 /*
@@ -2379,12 +2388,6 @@ void drain_all_pages(struct zone *zone)
 	}
 
 	/*
-	 * As this can be called from reclaim context, do not reenter reclaim.
-	 * An allocation failure can be handled, it's simply slower
-	 */
-	get_online_cpus();
-
-	/*
 	 * We don't care about racing with CPU hotplug event
 	 * as offline notification will cause the notified
 	 * cpu to drain that CPU pcps and on_each_cpu_mask
@@ -2423,7 +2426,6 @@ void drain_all_pages(struct zone *zone)
 	for_each_cpu(cpu, &cpus_with_pcps)
 		flush_work(per_cpu_ptr(&pcpu_drain, cpu));
 
-	put_online_cpus();
 	mutex_unlock(&pcpu_drain_mutex);
 }
 
-- 
Michal Hocko
SUSE Labs