linux-kernel - Re: [PATCH v3 0/3] Implement numa node notifier

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: <aBb-yDDJLObXF8P5@localhost.localdomain>
Date: Sun, 4 May 2025 07:44:40 +0200
From: Oscar Salvador <osalvador@...e.de>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: David Hildenbrand <david@...hat.com>, linux-mm@...ck.org,
	linux-kernel@...r.kernel.org, Vlastimil Babka <vbabka@...e.cz>,
	Hyeonggon Yoo <42.hyeyoo@...il.com>,
	Jonathan Cameron <Jonathan.Cameron@...wei.com>,
	Rakie Kim <rakie.kim@...com>
Subject: Re: [PATCH v3 0/3] Implement numa node notifier

On Sat, May 03, 2025 at 08:03:34PM -0700, Andrew Morton wrote:
> Why is this a problem?  Is there some bug?  Are these notifications so
> frequent that there are significant inefficiencies here?

hi Andrew,

There is no bug, it is just suboptimal.

That the numa node state changes were tied to the memory notifier was
something hacky and that have us bugged for a while now.
Were mean to tidy that up but just never got around it.

Actually, first time I brought that up was when I reviewed the first implementation
of memory demotion (~ca 3-4 years ago now?).

With the addition of yet another consumer (auto-weitght mempolicy) that was only
interested in get notified on numa node changes, it became more clear that we
really want to split those up.

> Further down-thread, Gregory tells us that Dan's patch "seems to fix
> the underlying problem", but nobody (including Dan) told us about any
> "problem" at all.

That is related to auto-weight mempolicy patches, not to this one.
I _think_ Gregory means that I take it in as part of the series.

-- 
Oscar Salvador
SUSE Labs