From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: OOM when adding ipv6 route: How to make available more per-cpu memory? Date: Fri, 05 Nov 2010 23:11:43 +0100 Message-ID: <1288995103.2665.653.camel@edumazet-laptop> References: <4CD43C87.5040403@candelatech.com> <1288980361.2882.1070.camel@edumazet-laptop> <4CD449A5.5070305@candelatech.com> <1288988403.2665.268.camel@edumazet-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: NetDev , linux-kernel , Tejun Heo To: Ben Greear Return-path: Received: from mail-ww0-f44.google.com ([74.125.82.44]:48092 "EHLO mail-ww0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752641Ab0KEWLt (ORCPT ); Fri, 5 Nov 2010 18:11:49 -0400 In-Reply-To: <1288988403.2665.268.camel@edumazet-laptop> Sender: netdev-owner@vger.kernel.org List-ID: Le vendredi 05 novembre 2010 =C3=A0 21:20 +0100, Eric Dumazet a =C3=A9c= rit : > Your vmalloc space is very fragmented. pcpu_get_vm_areas() want > hugepages (4MB on your machine, 2MB on mine because I have > CONFIG_HIGHMEM64G=3Dy) Well, this is wrong. We use normal (4KB) pages, unfortunately. I have a NUMA machine, with two nodes, so pcpu_get_vm_areas() allocates two zones, one for each node, with a 'known' offset between them. Then, 4KB pages are allocated to populate the zone when needed. # grep pcpu_get_vm_areas /proc/vmallocinfo=20 0xffffe8ffa0400000-0xffffe8ffa0600000 2097152 pcpu_get_vm_areas+0x0/0x7= 40 vmalloc 0xffffe8ffffc00000-0xffffe8ffffe00000 2097152 pcpu_get_vm_areas+0x0/0x7= 40 vmalloc BTW, we dont have the number of pages currently allocated in each 'vmalloc' zone, and/or node information. Tejun, do you have plans to use hugepages eventually ? (and fallback to 4KB pages, but most percpu data are allocated right after boot) Thanks