From owner-freebsd-current Mon Sep 28 14:08:24 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id OAA24193 for freebsd-current-outgoing; Mon, 28 Sep 1998 14:08:24 -0700 (PDT) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from heathers.stdio.com (heathers.stdio.com [199.89.192.5]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id OAA24121 for ; Mon, 28 Sep 1998 14:08:04 -0700 (PDT) (envelope-from lile@stdio.com) Received: (from lile@localhost) by heathers.stdio.com (8.8.8/8.8.8) id QAA13018; Mon, 28 Sep 1998 16:56:54 -0400 (EDT) Date: Mon, 28 Sep 1998 16:56:53 -0400 (EDT) From: "Larry S. Lile" To: Poul-Henning Kamp cc: Eivind Eklund , Jeremy Lea , current@FreeBSD.ORG Subject: Re: Current is Really Broken(tm) In-Reply-To: <16399.907012303@critter.freebsd.dk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Mon, 28 Sep 1998, Poul-Henning Kamp wrote: > > >[...] > > >Which is the command that you use to generate bootable media compelete > >with kernel and a small ram filesystem to contain the stuff which is > >neccessary to boot. [...] > > > >It works but if you ever have to build or repair one by hand you will > >regret it, except in trivial cases of disk, AIX F.0.0 and one platform. > >In fact I should get back to my Post Traumatic AIX Therapy now. > > But you have to admit that it is a smarter way to boot :-) AIX got the idea right but the implementation is sometimes bothersome. You have different bootFS's (s/ramFS/MFS) for each way you boot (tape, ethernet, token-ring, disk, ROS, ...) and different platform specifiec bootcode. With SP2's it gets even worse, but I won't get into that. Now if you could get neccessary MFS's built easily or canned then you might try something like this: /boot/os-version/bootMFS/common <-common boot tools /ethernet <-ethernet specific /disk ... /token-ring ... /platform/i386 <-Intel bootstraps /disk /tape /net ... /i386-mp <-Intel Muti-processor bootstraps ... /alpha <-alpha bootstrap ... And then you would need a tool to generate the MFS from the info in bootMFS/common and the specific bootMFS you are interested in and add the apropriate bootstrap code and dump it all to the media. Also you would need some configuration tools to set things like disk number to boot from or network boot server (and/or dhcp server) in the boot image after it is built or while it is starting a machine. All in all it can be really nice to have the flexability that it provides. You can build foriegn bootstraps from a working machine or boot a machine to a working recovery state and rebuild the system or fix problems without having loaded your kernel from the real / of your system. It would probably be useful for building new bootstraps for new platforms, or could be extended to boot other operating systems over the network ie serving NeXT/OpenStep, AIX, FreeBSD or even old IBM xstations. Personally, implementing it is beyond my capabilities but I would like to see it :) But I'm easy I just want the new 3-stage boot loader and an elf-kernel right now, or to get my token-ring driver to transmit without causing an adapter check. Larry Lile lile@stdio.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message