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.2.02.1402191339000.31921@chino.kir.corp.google.com>
Date:	Wed, 19 Feb 2014 13:45:14 -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:

> Interesting. So is the PPC NUMA basically about local vs. very distant?

The point is that it's impossible to tell how distant they are from one 
NUMA domain to the next NUMA domain.

> Should REMOTE_DISTANCE reflect that as well? Or can we have 
> distance < REMOTE_DISTANCE and it would still make sense to have
> zone_reclaim enabled?
> 

Ppc doesn't want to allocate in a different NUMA domain unless required, 
the latency of a remote access is too high.  Everything that isn't in the 
same domain has a distance >10 and is setup as 2^(domain hops) * 10.  We 
don't have the ability like with a SLIT to define remote nodes to have 
local latency vs remote or costly latency so the safe setting is a 
RECLAIM_DISTANCE of 10.
--
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/