[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1228169385.18834.136.camel@lts-notebook>
Date: Mon, 01 Dec 2008 17:09:45 -0500
From: Lee Schermerhorn <Lee.Schermerhorn@...com>
To: Rik van Riel <riel@...hat.com>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Johannes Weiner <hannes@...urebad.de>,
torvalds@...ux-foundation.org, kosaki.motohiro@...fujitsu.com,
linux-mm@...ck.org, linux-kernel@...r.kernel.org
Subject: Re: [patch v2] vmscan: protect zone rotation stats by lru lock
On Mon, 2008-12-01 at 16:46 -0500, Rik van Riel wrote:
> Andrew Morton wrote:
> > On Mon, 01 Dec 2008 03:00:35 +0100
> > Johannes Weiner <hannes@...urebad.de> wrote:
> >
> >> The zone's rotation statistics must not be accessed without the
> >> corresponding LRU lock held. Fix an unprotected write in
> >> shrink_active_list().
> >>
> >
> > I don't think it really matters. It's quite common in that code to do
> > unlocked, racy update to statistics such as this. Because on those
> > rare occasions where a race does happen, there's a small glitch in the
> > reclaim logic which nobody will notice anyway.
> >
> > Of course, this does need to be done with some care, to ensure the
> > glitch _will_ be small.
>
> Processing at most SWAP_CLUSTER_MAX pages at once probably
> ensures that glitches will be small most of the time.
>
> The only way this could be a big problem is if we end up
> racing with the divide-by-two logic in get_scan_ratio,
> leaving the rotated pages a factor two higher than they
> should be.
>
> Putting all the writes to the stats under the LRU lock
> should ensure that never happens.
And he's not actually adding a lock. Just moving the exiting one up to
include the stats update. The intervening pagevec, pgmoved and lru
initializations don't need to be under the lock, but that's probably not
a big deal?
Lee
--
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