Date: Tue, 9 May 2017 20:36:12 +0000 From: Colin Percival <cperciva@tarsnap.com> To: =?UTF-8?Q?Roger_Pau_Monn=c3=a9?= <royger@FreeBSD.org> Cc: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: Re: svn commit: r301198 - head/sys/dev/xen/netfront Message-ID: <0100015beeed2bab-9d1e0344-c22c-4414-bfe2-d671747906d1-000000@email.amazonses.com> In-Reply-To: <20170509100912.h3ylwugahvfi5cc2@dhcp-3-128.uk.xensource.com> References: <201606021116.u52BGajD047287@repo.freebsd.org> <0100015bccba6abc-4c3b1487-25e3-4640-8221-885341ece829-000000@email.amazonses.com> <20170509100912.h3ylwugahvfi5cc2@dhcp-3-128.uk.xensource.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 05/09/17 03:09, Roger Pau Monn� wrote: > On Wed, May 03, 2017 at 05:13:40AM +0000, Colin Percival wrote: >> On 06/02/16 04:16, Roger Pau Monn� wrote: >>> Author: royger >>> Date: Thu Jun 2 11:16:35 2016 >>> New Revision: 301198 >>> URL: https://svnweb.freebsd.org/changeset/base/301198 >> >> I think this commit is responsible for panics I'm seeing in EC2 on T2 family >> instances. [...] >> but under high traffic volumes I think a separate thread can already be >> running in xn_rxeof, having dropped the RX lock while it passes a packet >> up the stack. This would result in two different threads trying to process >> the same set of responses from the ring, with (unsurprisingly) bad results. > > Hm, right, xn_rxeof drops the lock while pushing the packet up the stack. > There's a "XXX" comment on top of that, could you try to remove the lock > dripping and see what happens? > > I'm not sure there's any reason to drop the lock here, I very much doubt > if_input is going to sleep. Judging by $ grep -R -B 1 -A 1 if_input /usr/src/sys/dev I'm pretty sure that we do indeed need to drop the lock. If it's possible to enter if_input while holding locks, there are a *lot* of network interface drivers which are dropping locks unnecessarily... >> 3. Why xn_ifinit_locked is consuming ring responses. > > There might be pending RX packets on the ring, so netfront consumes them and > signals netback. In the unlikely event that the RX ring was full when > xn_ifinit_locked is called, not doing this would mean the RX queue would get > stuck forever, since there's no guarantee netfront will receive event channel > notifications. In that case, I'm guessing it would be safe to skip this if another thread is already running xn_rxeof and chewing through the packets on the ring? It would be easy to set a flag in xn_rxeof before we drop locks. -- Colin Percival Security Officer Emeritus, FreeBSD | The power to serve Founder, Tarsnap | www.tarsnap.com | Online backups for the truly paranoid
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?0100015beeed2bab-9d1e0344-c22c-4414-bfe2-d671747906d1-000000>