[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 7 Jul 2016 10:56:13 +0100
From: Mel Gorman <mgorman@...hsingularity.net>
To: Minchan Kim <minchan@...nel.org>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Linux-MM <linux-mm@...ck.org>, Rik van Riel <riel@...riel.com>,
Vlastimil Babka <vbabka@...e.cz>,
Johannes Weiner <hannes@...xchg.org>,
LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 08/31] mm, vmscan: simplify the logic deciding whether
kswapd sleeps
On Thu, Jul 07, 2016 at 02:51:21PM +0900, Minchan Kim wrote:
> > It becomes difficult to tell the difference between "no wakeup and init to
> > zone 0" and "wakeup and reclaim for zone 0". At least that's the problem
> > I ran into when I tried before settling on -1.
>
> Sorry for bothering you several times. I cannot parse what you mean.
> I didn't mean -1 is problem here but why do we need below two lines
> I removed?
>
What you have should be fine. The hazard initially was that both
classzone_idx and kswapd_classzone_idx are enum and the signedness of
enum is implementation-dependent. Using max_t avoids that but it's a
subtle. I prefer the obvious check of kswapd_classzone_idx == 1 because
it is clearer that we're checking for an initialised value instead of
depending on a side-effect of the casting in max_t to do the right thing.
I can apply it if you wish, I just don't think it helps.
--
Mel Gorman
SUSE Labs
Powered by blists - more mailing lists