Re: mm: deadlock between get_online_cpus/pcpu_alloc

[Michal Hocko]

On Tue 07-02-17 10:49:28, Vlastimil Babka wrote:
> On 02/07/2017 10:43 AM, Mel Gorman wrote:
> > If I'm reading this right, a hot-remove will set the pool POOL_DISASSOCIATED
> > and unbound. A workqueue queued for draining get migrated during hot-remove
> > and a drain operation will execute twice on a CPU -- one for what was
> > queued and a second time for the CPU it was migrated from. It should still
> > work with flush_work which doesn't appear to block forever if an item
> > got migrated to another workqueue. The actual drain workqueue function is
> > using the CPU ID it's currently running on so it shouldn't get confused.
> 
> Is the worker that will process this migrated workqueue also guaranteed
> to be pinned to a cpu for the whole work, though? drain_local_pages()
> needs that guarantee.

Yeah I guess you are right. This would mean that drain_local_pages_wq
should to preempt_{disable,enable} around drain_local_pages
> 
> > Tejun, did I miss anything? Does a workqueue item queued on a CPU being
> > offline get unbound and a caller can still flush it safely? In this
> > specific case, it's ok that the workqueue item does not run on the CPU it
> > was queued on.

I guess we need to do one more step and ensure that our (rebound) worker
doesn't race with the page_alloc_cpu_notify. I guess we can just cmpxchg
pcp->count in drain_pages_zone to ensure the exclusivity. Not as simple
as I originally thought but doable I guess and definitely better than
making a subtle dependency on the hotplug locks which is just a PITA to
maintain.

-- 
Michal Hocko
SUSE Labs