From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S265531AbUG1Ftz (ORCPT ); Wed, 28 Jul 2004 01:49:55 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S265810AbUG1Ftz (ORCPT ); Wed, 28 Jul 2004 01:49:55 -0400 Received: from smtp107.mail.sc5.yahoo.com ([66.163.169.227]:17007 "HELO smtp107.mail.sc5.yahoo.com") by vger.kernel.org with SMTP id S265531AbUG1Ftx (ORCPT ); Wed, 28 Jul 2004 01:49:53 -0400 Message-ID: <41073A6C.1050606@yahoo.com.au> Date: Wed, 28 Jul 2004 15:32:28 +1000 From: Nick Piggin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7) Gecko/20040707 Debian/1.7-5 X-Accept-Language: en MIME-Version: 1.0 To: Avi Kivity CC: Ulrich Weigand , linux-kernel@vger.kernel.org Subject: Re: [PATCH] Deadlock during heavy write activity to userspace NFS References: <200407280232.EAA14567@faui1m.informatik.uni-erlangen.de> <41073710.2020306@exanet.com> In-Reply-To: <41073710.2020306@exanet.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Avi Kivity wrote: > Ulrich Weigand wrote: > >> Avi Kivity wrote: >> >> >> >>> In our case, all block I/O is done using unbuffered I/O, and all >>> memory is preallocated, so we don't need kswapd at all, just that >>> small bit of memory that syscalls consume. >>> >> >> >> Does your userspace process need to send/receive network packets >> in order to perform a write-out? > > Yes. > >> If so, how can you make sure your >> incoming packets aren't thrown away in out-of-memory situations? >> (Outgoing packets can use PF_MEMALLOC memory I guess, but incoming >> ones aren't associated to any process yet ...) >> >> >> > I did nothing to address this. So far it works well, even under heavy > load. I guess a general solution needs to address this as well. > > The kernel NFS client (which kswapd depends on) has the same issue. Has > anyone ever observed kswapd deadlock due to imcoming or outgoing NFS > packets being discarded due to oom? > Yes this has been observed. alloc_skb on the client needs to somehow know that traffic coming from the server is "MEMALLOC" and allowed to use memory reserves.