From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Thu, 20 Mar 2008 23:07:30 -0700 (PDT) From: Christoph Lameter Subject: Re: [patch 5/9] slub: Fallback to minimal order during slab page allocation In-Reply-To: <1206076457.14496.85.camel@ymzhang> Message-ID: References: <20080317230516.078358225@sgi.com> <20080317230528.939792410@sgi.com> <1205989839.14496.32.camel@ymzhang> <1206060738.14496.66.camel@ymzhang> <1206076457.14496.85.camel@ymzhang> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-mm@kvack.org Return-Path: To: "Zhang, Yanmin" Cc: Pekka Enberg , Mel Gorman , Matt Mackall , linux-mm@kvack.org List-ID: On Fri, 21 Mar 2008, Zhang, Yanmin wrote: > On Thu, 2008-03-20 at 20:35 -0700, Christoph Lameter wrote: > > On Fri, 21 Mar 2008, Zhang, Yanmin wrote: > > > > > > However, its a division in a potentially hot codepath. > > > No as long as there is no allocation failure because of fragmentation. > > > > If its only used for the fallback path then the race condition is still > > there? > I can't understand your question. Does it means min_objects? It's not related > to the race. The fallback path also isn't related to the race. > > The race is when kernel runs in allocate_slab, just between fetching s->order and > s->objects,user might change s->order by sysfs. Right. That patch matters most and with the patch that I posted a few hours ago there is a common scheme that addresses both the race and the issue with min_objects (hopefully...). -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org