From owner-freebsd-hackers Mon Dec 15 10:34:33 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id KAA23094 for hackers-outgoing; Mon, 15 Dec 1997 10:34:33 -0800 (PST) (envelope-from owner-freebsd-hackers) Received: from smtp01.primenet.com (smtp01.primenet.com [206.165.6.131]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id KAA23087 for ; Mon, 15 Dec 1997 10:34:28 -0800 (PST) (envelope-from tlambert@usr05.primenet.com) Received: (from daemon@localhost) by smtp01.primenet.com (8.8.8/8.8.8) id LAA10900; Mon, 15 Dec 1997 11:33:55 -0700 (MST) Received: from usr05.primenet.com(206.165.6.205) via SMTP by smtp01.primenet.com, id smtpd010874; Mon Dec 15 11:33:49 1997 Received: (from tlambert@localhost) by usr05.primenet.com (8.8.5/8.8.5) id LAA19433; Mon, 15 Dec 1997 11:33:46 -0700 (MST) From: Terry Lambert Message-Id: <199712151833.LAA19433@usr05.primenet.com> Subject: Re: Bus/Processor specific I/O methods - was Re: Beginning SPARC port To: avalon@coombs.anu.edu.au (Darren Reed) Date: Mon, 15 Dec 1997 18:33:46 +0000 (GMT) Cc: gurney_j@resnet.uoregon.edu, avalon@coombs.anu.edu.au, hackers@FreeBSD.ORG In-Reply-To: <199712150750.XAA01781@hub.freebsd.org> from "Darren Reed" at Dec 15, 97 06:50:36 pm X-Mailer: ELM [version 2.4 PL23] Content-Type: text Sender: owner-freebsd-hackers@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Well, the Solaris2 kernel is 600k (/platform/sun4m/kernel/unix) for 2.5.1, > but /platform/sun4m is 4.5M in size with another 1.5M in /usr/kernel. > Personally, I find as the number of files required to boot into single > user increases, the greater the chance of shit happening on a bad crash > and the system becoming unable to boot itself. Personally, I think that > the boot device and root fs should always be "in" the kernel so that if > someone has nuked all your modules, you can still get up into single user > mode. Mount the root read-only, and you will have the same guarantees. Implement soft updates, and you will have the same guarantees. The statement "shit happens" only applies when you let your guard down; you have to make a conscious decision to do this. Let's note, also, that Solaris can be recovered, easily, from CDROM in case of this type of Spamming of the kernel... > I think that whilst the above goal is interesting (and perhaps worthwhile), > there are already problems which need to be solved for LKM's which aren't, > as yet, such as merging LKM symbols so that gdb on the kernel is sane and > fixing up crash dumps.. Both of these are relatively easy to accomplish using John Polstra's ELF implementation. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.