linux-kernel - Re: [RFC PATCH] mm: migrate: Add new node demotion strategy

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: <87tugrxqks.fsf@yhuang6-desk2.ccr.corp.intel.com>
Date:   Fri, 05 Nov 2021 10:51:31 +0800
From:   "Huang, Ying" <ying.huang@...el.com>
To:     Dave Hansen <dave.hansen@...el.com>,
        Baolin Wang <baolin.wang@...ux.alibaba.com>
Cc:     <akpm@...ux-foundation.org>, <dave.hansen@...ux.intel.com>,
        <ziy@...dia.com>, <osalvador@...e.de>, <shy828301@...il.com>,
        <linux-mm@...ck.org>, <linux-kernel@...r.kernel.org>
Subject: Re: [RFC PATCH] mm: migrate: Add new node demotion strategy

Dave Hansen <dave.hansen@...el.com> writes:

> On 11/4/21 2:13 AM, Baolin Wang wrote:
>> +What:		/sys/kernel/mm/numa/demotion_mode
>> +Date:		November 2021
>> +Contact:	Linux memory management mailing list <linux-mm@...ck.org>
>> +Description:	Set the demotion mode when enabling demoting pages during reclaim
>
> I don't think we need a tunable for this.  The existing behavior is just
> stupid for your hardware and can be replaced.

Yes.  I think so too.  I don't think DIRECT_DEMOTION is reasonable for your system.

> Let's also try to do it with the existing node_demotion[] data
> structure before we go adding more.

To avoid cache ping-pong, I guess some kind of per-CPU data structure
may be more suitable for interleaving among multiple nodes.

Best Regards,
Huang, Ying