From owner-freebsd-hackers Mon Feb 16 17:02:20 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id RAA23470 for freebsd-hackers-outgoing; Mon, 16 Feb 1998 17:02:20 -0800 (PST) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from smtp03.primenet.com (smtp03.primenet.com [206.165.6.133]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id RAA23232 for ; Mon, 16 Feb 1998 17:01:49 -0800 (PST) (envelope-from tlambert@usr06.primenet.com) Received: (from daemon@localhost) by smtp03.primenet.com (8.8.8/8.8.8) id RAA25073; Mon, 16 Feb 1998 17:55:20 -0700 (MST) Received: from usr06.primenet.com(206.165.6.206) via SMTP by smtp03.primenet.com, id smtpd025060; Mon Feb 16 17:55:19 1998 Received: (from tlambert@localhost) by usr06.primenet.com (8.8.5/8.8.5) id RAA28812; Mon, 16 Feb 1998 17:55:16 -0700 (MST) From: Terry Lambert Message-Id: <199802170055.RAA28812@usr06.primenet.com> Subject: Re: VM messed: vm_page_free panic problem To: mike@smith.net.au (Mike Smith) Date: Tue, 17 Feb 1998 00:55:15 +0000 (GMT) Cc: Tor.Egge@idi.ntnu.no, mike@smith.net.au, toor@dyson.iquest.net, hackers@FreeBSD.ORG In-Reply-To: <199802162358.PAA08389@dingo.cdrom.com> from "Mike Smith" at Feb 16, 98 03:58:17 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > [... explanation ...] > > It is not obvious to me how using the DISABLE_PSE option fixes this > > problem. > > Nor to me. At that point I was happy that it resolved the immediate > issue. I originated the DISABLE_PSE fix. It works because there is a spanning page mapping which takes place for an image when PSE is enabled. The mapping reverts to the old vn pager code (and call VOP_READ). All I could see from the backtrace without a system in hand was that the pager was puking. Clearly to get to the point where it could panic, the first set of pages containing the panic code, among other things, must have been loaded. I surmised (correctly, it seems) that the first non-kernel 4M mapping, which is where the init program would run, was failing. The initial kernel/MFS mapping was implemented in the kernel startup, and so could not be at fault. Unlike Tor, I didn't track this down to the missing vop_getpages; I just noted that it was possible to revert the pager behaviour by disabling the extension. I didn't know whether Mike was right about the mapping being damaged, or not, only that I could revert the behaviour with DISABLE_PSE. In any case, Tor made a very good catch; a fix is always better than a workaround. Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message