[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4DFF8207.5080700@redhat.com>
Date: Tue, 21 Jun 2011 01:23:19 +0800
From: Cong Wang <amwang@...hat.com>
To: Dave Hansen <dave@...ux.vnet.ibm.com>
CC: linux-kernel@...r.kernel.org, akpm@...ux-foundation.org,
Andrea Arcangeli <aarcange@...hat.com>,
Mel Gorman <mel@....ul.ie>,
Benjamin Herrenschmidt <benh@...nel.crashing.org>,
Rik van Riel <riel@...hat.com>,
Johannes Weiner <hannes@...xchg.org>,
KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
linux-mm@...ck.org
Subject: Re: [PATCH 2/3] mm: make the threshold of enabling THP configurable
于 2011年06月21日 00:59, Dave Hansen 写道:
> On Tue, 2011-06-21 at 00:34 +0800, Amerigo Wang wrote:
>> +config TRANSPARENT_HUGEPAGE_THRESHOLD
>> + depends on TRANSPARENT_HUGEPAGE
>> + int "The minimal threshold of enabling Transparent Hugepage"
>> + range 512 8192
>> + default "512"
>> + help
>> + The threshold of enabling Transparent Huagepage automatically,
>> + in Mbytes, below this value, Transparent Hugepage will be disabled
>> + by default during boot.
>
> It makes some sense to me that there would _be_ a threshold, simply
> because you need some space to defragment things. But, I can't imagine
> any kind of user having *ANY* kind of idea what to set this to. Could
> we add some text to this? Maybe:
>
> Transparent hugepages are created by moving other pages out of
> the way to create large, contiguous swaths of free memory.
> However, some memory on a system can not be easily moved. It is
> likely on small systems that this unmovable memory will occupy a
> large portion of total memory, which makes even attempting to
> create transparent hugepages very expensive.
>
> If you are unsure, set this to the smallest possible value.
>
> To override this at boot, use the $FOO boot command-line option.
>
Yeah, I totally agree to improve the help message as you said,
please forgive a non-English speaker. ;)
> I'm also not sure putting a ceiling on this makes a lot of sense.
> What's the logic behind that? I know it would be a mess to expose it to
> users, but shouldn't this be a per-zone limit, logically? Seems like a
> 8GB system would have similar issues to a two-numa-node 16GB system.
>
I am not sure about this, since I am new to THP, I just replaced
the hard-code 512 with a Kconfig var. But I am certainly open
to improve this as you said if Andrea agrees.
Thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists