linux-kernel - Re: ppc: RECLAIM

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: <alpine.DEB.2.02.1402190017480.7280@chino.kir.corp.google.com>
Date:	Wed, 19 Feb 2014 00:20:21 -0800 (PST)
From:	David Rientjes <rientjes@...gle.com>
To:	Michal Hocko <mhocko@...e.cz>
cc:	Anton Blanchard <anton@...ba.org>,
	Benjamin Herrenschmidt <benh@...nel.crashing.org>,
	linuxppc-dev@...ts.ozlabs.org, linux-mm@...ck.org,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: ppc: RECLAIM_DISTANCE 10?

On Wed, 19 Feb 2014, Michal Hocko wrote:

> > I strongly suspect that the patch is correct since powerpc node distances 
> > are different than the architectures you're talking about and get doubled 
> > for every NUMA domain that the hardware supports.
> 
> Even if the units of the distance is different on PPC should every NUMA
> machine have zone_reclaim enabled? That doesn't right to me.
> 

In my experience on powerpc it's very correct, there's typically a 
significant latency in remote access and we don't have the benefit of a 
SLIT that actually defines the locality between proximity domains like we 
do on other architectures.  We have had significant issues with thp, for 
example, being allocated remotely instead of pages locally, much more 
drastic than on our x86 machines, particularly AMD machines.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/