Re: [PATCH v2] mm: Only re-generate demotion targets when a numa node changes its N_CPU state

From: Oscar Salvador <osalvador@suse.de>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: Dave Hansen <dave.hansen@linux.intel.com>,
	"Huang, Ying" <ying.huang@intel.com>,
	Abhishek Goel <huntbag@linux.vnet.ibm.com>,
	Baolin Wang <baolin.wang@linux.alibaba.com>,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] mm: Only re-generate demotion targets when a numa node changes its N_CPU state
Date: Fri, 11 Mar 2022 10:23:44 +0100	[thread overview]
Message-ID: <YisVIIGcSgi7NL8N@localhost.localdomain> (raw)
In-Reply-To: <20220310183951.cb713c6ae926ea6ea8489a71@linux-foundation.org>

On Thu, Mar 10, 2022 at 06:39:51PM -0800, Andrew Morton wrote:
> What I'm not getting here (as so often happens) is a sense of how badly
> this affects our users.  Does anyone actually hotplug frequently enough
> to care?

Well, I would not say it is critical, it just regresses the time
hotplug operations take. How bad is that? I guess it depends.
Memory hotplug operations are already slow per se, so I would say
expectations are not that high.

But it speeds up the process, that is for sure.

e.g: In a system with 144 CPUs and 2 Numa-nodes,
set_migration_target_nodes()
gets called exactly 144 times at boot time(one per every time a CPU is
brought up), where only 2 calls would suffice, so you can get an idea.

> If "yes" then I'm inclined to merge this up for 5.18 with a cc:stable. 
> Not for 5.17 because it's late and things are looking rather creaky
> already.
> 
> And I'll add a
> 
> Fixes: 884a6e5d1f93b ("mm/migrate: update node demotion order on hotplug events")
> 
> which is that patch's fourth such bouquet.

Thanks Andrew!

-- 
Oscar Salvador
SUSE Labs