linux-kernel - Re: [PATCH] mm: fix LRU balancing effect of new transparent huge pages

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: <CALvZod7848_BETonZJFxXN1n1E7KtZ4DzT1-BeaN6hvBysN8WQ@mail.gmail.com>
Date:   Mon, 11 May 2020 14:38:23 -0700
From:   Shakeel Butt <shakeelb@...gle.com>
To:     Andrew Morton <akpm@...ux-foundation.org>
Cc:     Mel Gorman <mgorman@...e.de>, Johannes Weiner <hannes@...xchg.org>,
        Roman Gushchin <guro@...com>, Michal Hocko <mhocko@...nel.org>,
        Minchan Kim <minchan@...nel.org>,
        Rik van Riel <riel@...riel.com>,
        Linux MM <linux-mm@...ck.org>,
        LKML <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] mm: fix LRU balancing effect of new transparent huge pages

On Mon, May 11, 2020 at 2:11 PM Andrew Morton <akpm@...ux-foundation.org> wrote:
>
> On Sat,  9 May 2020 07:19:46 -0700 Shakeel Butt <shakeelb@...gle.com> wrote:
>
> > Currently, THP are counted as single pages until they are split right
> > before being swapped out. However, at that point the VM is already in
> > the middle of reclaim, and adjusting the LRU balance then is useless.
> >
> > Always account THP by the number of basepages, and remove the fixup
> > from the splitting path.
>
> Confused.  What kernel is this applicable to?

It is still applicable to the latest Linux kernel. Basically
lruvec->reclaim_stat->recent_[scanned|rotated] counters are used as
heuristic in get_scan_count() to measure how much file and anon LRUs
should be scanned by the current reclaim. Previously huge pages are
treated as single page while updating the recent_[scanned|rotated]
counters in swap.c while vmscan.c correctly updates them as huge
pages.