[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190725080551.GB2708@suse.de>
Date: Thu, 25 Jul 2019 09:05:51 +0100
From: Mel Gorman <mgorman@...e.de>
To: Mike Kravetz <mike.kravetz@...cle.com>
Cc: linux-mm@...ck.org, linux-kernel@...r.kernel.org,
Hillf Danton <hdanton@...a.com>,
Michal Hocko <mhocko@...nel.org>,
Vlastimil Babka <vbabka@...e.cz>,
Johannes Weiner <hannes@...xchg.org>,
Andrew Morton <akpm@...ux-foundation.org>
Subject: Re: [RFC PATCH 1/3] mm, reclaim: make should_continue_reclaim
perform dryrun detection
On Wed, Jul 24, 2019 at 10:50:12AM -0700, Mike Kravetz wrote:
> From: Hillf Danton <hdanton@...a.com>
>
> Address the issue of should_continue_reclaim continuing true too often
> for __GFP_RETRY_MAYFAIL attempts when !nr_reclaimed and nr_scanned.
> This could happen during hugetlb page allocation causing stalls for
> minutes or hours.
>
> Restructure code so that false will be returned in this case even if
> there are plenty of inactive pages.
>
> Need better description from Hillf Danton
> Need SOB from Hillf Danton
Agreed that the description could do with improvement. However, it
makes sense that if compaction reports it can make progress that it is
unnecessary to continue reclaiming. There might be side-effects with
allocation latencies in other cases but that would come at the cost of
potential premature reclaim which has consequences of itself. Overall I
think it's an improvement so I'll ack it once it has a S-O-B.
--
Mel Gorman
SUSE Labs
Powered by blists - more mailing lists