Date: Mon, 23 Jun 2014 01:16:20 +0200 From: =?windows-1250?Q?Pawe=B3_Tyll?= <ofca@ofca.me> To: Adrian Chadd <adrian@freebsd.org> Cc: FreeBSD Net <freebsd-net@freebsd.org>, noc@hostelnet.ru Subject: Re: FreeBSD 9 as PPPoE BRAS(mpd 5.7) kernel panic Message-ID: <2510165566.20140623011620@ofca.me> In-Reply-To: <CAJ-Vmo=5YuSybyJMQtv6PUwXAGWyAxzvAZs4SgZyKwYS9aDMCw@mail.gmail.com> References: <53A719C3.3040002@hostelnet.ru> <CAJ-Vmo=5YuSybyJMQtv6PUwXAGWyAxzvAZs4SgZyKwYS9aDMCw@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello Adrian, Sunday, June 22, 2014, 8:14:22 PM, you wrote: > They're NULL pointer derferences, so it's likely a race condition with > some other thread destroying something and setting the pointer value > to NULL somewhere. > I thought this was a reasonably well known problem? Was it ever fixed > in 10/head? It probably wasn't, since I had similar issues on 10-STABLE r266523. It's caused (most often) by packet returning from traffic shaping queues, when in the meantime ng interface got destroyed by mpd. That's why the sleep hack works; mpd destroys interface 1s later that usual. Curious though, that hostelnet uses ng_car for traffic shaping, yet things still panic, even though whole ordeal happens inside netgraph. Lots of entry-points for a fix :)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?2510165566.20140623011620>