From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andi Kleen Subject: Re: [net-next-2.6 PATCH 1/5] ixgbe: Allocate driver resources per NUMA node Date: Tue, 19 Jan 2010 20:34:00 +0100 Message-ID: <20100119193400.GB24652@basil.fritz.box> References: <20100107044741.28605.31414.stgit@localhost.localdomain> <87y6k1dk5t.fsf@basil.nowhere.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Andi Kleen , "Kirsher, Jeffrey T" , "davem@davemloft.net" , "netdev@vger.kernel.org" , "gospo@redhat.com" To: "Waskiewicz Jr, Peter P" Return-path: Received: from one.firstfloor.org ([213.235.205.2]:52198 "EHLO one.firstfloor.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751431Ab0ASTeJ (ORCPT ); Tue, 19 Jan 2010 14:34:09 -0500 Content-Disposition: inline In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: > >The problem of doing this is that the node might be full or have > >no memory and k*alloc_node will fail then. > > > >So you would need a fallback to be reliable (we probably should have a > >generic utility function for this somewhere, but we don't currently) > > So you'd rather see us call kzalloc() if kzalloc_node() fails, instead of immediately failing out to err_out? Yes. After all it's only an optimization to place memory, it's not a functional requirement. -Andi -- ak@linux.intel.com -- Speaking for myself only.