From owner-freebsd-questions Mon Mar 11 13:32:43 1996 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id NAA20029 for questions-outgoing; Mon, 11 Mar 1996 13:32:43 -0800 (PST) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id NAA20017 Mon, 11 Mar 1996 13:32:41 -0800 (PST) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id OAA04059; Mon, 11 Mar 1996 14:30:55 -0700 From: Terry Lambert Message-Id: <199603112130.OAA04059@phaeton.artisoft.com> Subject: Re: VNODE Pager Errors To: gpalmer@freebsd.org (Gary Palmer) Date: Mon, 11 Mar 1996 14:30:54 -0700 (MST) Cc: nc@ai.net, questions@freebsd.org In-Reply-To: <2443.826572729@palmer.demon.co.uk> from "Gary Palmer" at Mar 11, 96 07:32:09 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-questions@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > Network Coordinator wrote in message ID > <199603111815.NAA25619@aries.ai.net>: > > /kernel: vnode_pager_input: I/O read error > > /kernel: vm_fault: pager input (probably hardware error, PID 287 failure. > > Having seen this on my own machine, it normally means that the system > gets a read failure from the swap device, meaning that it can't page > the binary back into memory... Specifically, a fault on a copyin of a user page that isn't present (unlikely) or a fault on a copyout to a user page that isn't present (very likely, especially on a loaded system) when using a swapfile on a file of the file system from which the uiomove() in vfs_subr.c is being called. John has recently done a lot of work in this area; even though I disagree with some of the VM architectural choices because of their ramifications for file system (specifically VFS-to-VOP layer) complexity, John does damn good work. I think that -current as of a couple of days ago fixed this problem 100%. Now if only I could convince him to change cache indexing so that I can murder the vclean() and make the VOP_LOCK() code advisory and thus shared between all file systems... 8-). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.