linux-kernel - Re: [PATCH 2/3] mm: make the threshold of enabling THP configurable

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20110622111525.GK9396@suse.de>
Date:	Wed, 22 Jun 2011 12:15:25 +0100
From:	Mel Gorman <mgorman@...e.de>
To:	Cong Wang <amwang@...hat.com>
Cc:	Mel Gorman <mel@....ul.ie>, linux-kernel@...r.kernel.org,
	akpm@...ux-foundation.org, Andrea Arcangeli <aarcange@...hat.com>,
	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

On Wed, Jun 22, 2011 at 06:46:39PM +0800, Cong Wang wrote:
> ??? 2011???06???22??? 17:16, Mel Gorman ??????:
> >
> >What I meant was that there is a rational reason why 512M is the
> >default for enabling THP by default. Tuning it lower than that by any
> >means makes very little sense. Tuning it higher might make some sense
> >but it is more likely that THP would simply be disabled via sysctl. I
> >see very little advantage to introducing this Kconfig option other
> >than as a source of confusion when running make oldconfig.
> >
> 
> The tunable range is (512, 8192), so 512M is the minimum.
> 
> Sure, I knew it can be disabled via /sys, actually we can do even
> more in user-space, that is totally move the 512M check out of kernel,
> why we didn't?
> 

Because the reason why 512M is the default is not obvious and there
was no guarantee all distros would chose a reasonable default for
an init script (or know that an init script was even necessary).
This is one of the few cases where there is a sensible default that
is the least surprising.

> In short, I think we should either remove the 512M from kernel, or
> make 512M to be tunable.
> 

That just hands them a different sort of rope to hang themselves with
where THP gets enabled on small machines or botting with mem=128M
and getting surprised later by the high min_free_kbytes.

At this point, I don't really care if the Kconfig entry exists or
not. I think it gains nothing but additional confusion for people
who write .config files but it's not a topic I want to discuss for
days either.

-- 
Mel Gorman
SUSE Labs
--
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/