Date: Thu, 2 Oct 2003 20:54:06 -0400 (EDT) From: Robert Watson <rwatson@FreeBSD.org> To: Mark Woodson <mwoodson@sricrm.com> Cc: current@FreeBSD.org Subject: Re: lor on boot Message-ID: <Pine.NEB.3.96L.1031002205221.29639C-100000@fledge.watson.org> In-Reply-To: <200310021731.44227.mwoodson@sricrm.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 2 Oct 2003, Mark Woodson wrote: > On Thursday 02 October 2003 05:07 pm, Robert Watson wrote: > > On Thu, 2 Oct 2003, Mark Woodson wrote: > > > I'm getting a lor on a system just upgraded to sources from this > > > morning. The systems been running fine for the past month or so. > > > > What version of src/sys/net/netisr.c are you running with? > > * $FreeBSD: src/sys/net/netisr.c,v 1.4 2003/10/01 21:31:09 rwatson > Exp $ > > That's what I'm showing. So it's from last night then. Ah. Ok, this is because the if_rl driver holds the driver mutex across a call to the interface input routine, resulting in holding the mutex across a call into the remainder of the network stack. The reason this showed up for you now is that I temporarily enabled direct dispatch of the isr code directly from the driver interrupt threads for an hour or so last night, and you updated during that time. I backed it out to work on two issues -- one the possible reordering of packets (patch now bing reviewed), and the other that a few drivers currently hold their lock over the call into the remainder of the stack, which needs to be fixed. If you cvsup, the problem should go away. Robert N M Watson FreeBSD Core Team, TrustedBSD Projects robert@fledge.watson.org Network Associates Laboratories
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.NEB.3.96L.1031002205221.29639C-100000>