From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tejun Heo Subject: Re: [Bug #14467] Linker errors on ia64 with NR_CPUS=4096 Date: Thu, 29 Oct 2009 15:48:03 +0100 Message-ID: <4AE9AB23.8010207@kernel.org> References: <6dRYo8ss7vL.A.nnF.Cre5KB@chimera> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: kernel-testers-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: text/plain; charset="us-ascii" To: Christoph Lameter Cc: Jiri Kosina , "Rafael J. Wysocki" , Linux Kernel Mailing List , Kernel Testers List , Ingo Molnar , Jeff Mahoney , "Luck, Tony" , Peter Zijlstra , Peter Zijlstra Christoph Lameter wrote: >> There already are two patches for this acked by Ingo/Tejun, which Tejun is >> going to take through his tree tomorrow. >> >> http://lkml.org/lkml/2009/10/27/132 > > per cpu alloc from an atomic context without passing gfp flags through to > the page allocator? That does not look right. Sure wish that the percpu > allocator would be working from atomic contexts for other cases. It's just for sched_init() which has irq off but is not really in atomic context and does GFP_KERNEL allocations. The following comment has been added to the first patch to explain it. + * allocations are done using GFP_KERNEL with pcpu_lock released. In + * general, percpu memory can't be allocated with irq off but + * irqsave/restore are still used in alloc path so that it can be used + * from early init path - sched_init() specifically. Thanks. -- tejun