Re: [patch -mm 4/5] mm: test and set zone reclaim lock before starting reclaim

[Balbir Singh <balbir@linux.vnet.ibm.com>]

David Rientjes wrote:
> On Tue, 25 Sep 2007, Balbir Singh wrote:
> 
>>> +
>>> +	if (zone_test_and_set_flag(zone, ZONE_RECLAIM_LOCKED))
>>> +		return 0;
>> What's the consequence of this on the caller of zone_reclaim()?
>> I see that the zone is marked as full and will not be re-examined
>> again.
>>
> 
> It's only marked as full in the zonelist cache for the zonelist that 
> __alloc_pages() was called with, which is an optimization.  The zone is 
> already flagged as being in __zone_reclaim() so there's no need to 
> reinvoke it for this allocation attempt; that behavior is unchanged from 
> current behavior.
> 

OK

> One thing that has been changed in -mm with regard to my last patchset is 
> that kswapd and try_to_free_pages() are allowed to call shrink_zone() 
> concurrently.
> 

Aah.. interesting. Could you define concurrently more precisely,
concurrently as in the same zone or for different zones concurrently?

-- 
	Warm Regards,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL

--
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>