Re: [PATCH 7/8] Scanner changes needed to implement per-container scanner

From: Pavel Emelianov <xemul@openvz.org>
To: balbir@linux.vnet.ibm.com
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Pavel Emelianov <xemul@openvz.org>,
	Paul Menage <menage@google.com>, Balbir Singh <balbir@in.ibm.com>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	devel@openvz.org, Kirill Korotaev <dev@openvz.org>
Subject: Re: [PATCH 7/8] Scanner changes needed to implement per-container scanner
Date: Fri, 01 Jun 2007 11:40:09 +0400	[thread overview]
Message-ID: <465FCD59.7060301@openvz.org> (raw)
In-Reply-To: <465FC19B.8090905@linux.vnet.ibm.com>

Balbir Singh wrote:
> Andrew Morton wrote:
>>> +		nr_reclaimed += shrink_zones(priority, zones, sc);
>>> +		if (sc->cnt == NULL)
>>> +			shrink_slab(sc->nr_scanned, gfp_mask, lru_pages);
>> We don't we shrink slab if called to shrink a container.
>>
>> This is a fundamental design decision, and a design shortcoming.  A full
>> discussion of this is absolutely appropriate to the patch changelog. 
>> Please don't just hide stuff like this in the patch and leave people
>> wondering, or ignorant.
> 
> Yes, we don't because we do not account for slab usage right now. We account
> only for memory allocated to user space. A good fat comment will help here.
> 
> 

I have already added the comment. But the problem is not in that we
do not account for kernel memory. Shrinking slabs won't unmap any
pages from user-space and thus won't help user to charge more. This
will only make kernel suffer from re-creation of objects.

Thanks,
Pavel.