Re: [PATCH 2/3] mm/page_alloc: convert zone_pcp_update() to use on_each_cpu() instead of stop_machine()

[Cody P Schafer <cody@linux.vnet.ibm.com>]

On 04/08/2013 03:18 PM, KOSAKI Motohiro wrote:
> (4/8/13 3:49 PM), Cody P Schafer wrote:>
>> If this turns out to be an issue, schedule_on_each_cpu() could be an
>> alternative.
>
> no way. schedule_on_each_cpu() is more problematic and it should be removed
> in the future.
> schedule_on_each_cpu() can only be used when caller task don't have any lock.
> otherwise it may make deadlock.

I wasn't aware of that. Just to be clear, the deadlock you're referring 
to isn't the same one refered to in

commit b71ab8c2025caef8db719aa41af0ed735dc543cd
Author: Tejun Heo <tj@kernel.org>
Date:   Tue Jun 29 10:07:14 2010 +0200
workqueue: increase max_active of keventd and kill current_is_keventd()

and

commit 65a64464349883891e21e74af16c05d6e1eeb4e9
Author: Andi Kleen <ak@linux.intel.com>
Date:   Wed Oct 14 06:22:47 2009 +0200
HWPOISON: Allow schedule_on_each_cpu() from keventd

If you're referencing some other deadlock, could you please provide a 
link to the relevant discussion? (I'd really like to add a note to 
schedule_on_each_cpu()'s doc comment about it so others can avoid that 
pitfall).

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>