From owner-freebsd-hackers Tue Jul 18 12:45:34 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id MAA09763 for hackers-outgoing; Tue, 18 Jul 1995 12:45:34 -0700 Received: from gndrsh.aac.dev.com (gndrsh.aac.dev.com [198.145.92.241]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id MAA09756 for ; Tue, 18 Jul 1995 12:45:32 -0700 Received: (from rgrimes@localhost) by gndrsh.aac.dev.com (8.6.11/8.6.9) id MAA05322; Tue, 18 Jul 1995 12:44:57 -0700 From: "Rodney W. Grimes" Message-Id: <199507181944.MAA05322@gndrsh.aac.dev.com> Subject: Re: A problem with disklabel & "use entire disk" on 2.0.5R. To: bde@zeta.org.au (Bruce Evans) Date: Tue, 18 Jul 1995 12:44:57 -0700 (PDT) Cc: imp@village.org, freebsd-hackers@freebsd.org In-Reply-To: <199507181830.EAA21989@godzilla.zeta.org.au> from "Bruce Evans" at Jul 19, 95 04:30:36 am X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2663 Sender: hackers-owner@freebsd.org Precedence: bulk > > >> Is there some way to write a valid partion table to BSD only disks? > >> Wouldn't that solve the problem, or am I dreaming... > > >Problem is that boot1 lives in sector 0 in this case, so anytime you > >install new boot blocks you end up with a 50K not quite right bogus > >partition table from the boot1 code. > > >Newfs needs fixed to deal with the fact it _must_ preserve the existing > >partition table if installing boot code to physical sector 0 of the disk. > > A partition that covers the whole disk must have offset 0 and cover the > MBR. Yes, that is true. > Such partitions can reasonably be considered as invalid. Such paritions have been considered valid for 386BSD 0.0, 0.1 all patchkits, FreeBSD 1.0, 1.1, and 2.0, NetBSD 0.8, 0.9 and as far as I know 1.0. I am not sure on the BSDI area, but I belive upto 1.0 this is true. They are considered valid foreign partitions by DOS 4.1, 5.0 and 6.x, Netware 3.12, DR DOS 6.0, DR DOS 7.0. I have not seen a fdisk or MBR utility thats says they are invalid. > Preserving > them would make no difference to FreeBSD and might not stop foreign > fdisks and installs from nuking them. You are correct in that preserving them makes no difference to FreeBSD itself. It _may_ make a difference to the BIOS and it _certainly_ makes a difference to any foreign fdisk. Up until recently I was in the habit of properly reseting the fdisk parameters after all disklabel installs of boot blocks on these types of disks to make sure that my disks would be impervious to likes of DOS 5.x and later install utilities that will splat them selves on any free disk space they see faster than you can say WTF. This was an important step to insure that one of my customers did not go booting a DOS install disk only to have it wipe his FreeBSD system from the disk (something that occured once before I took corrective action to make sure it never happened again). Something I still do on every machine that goes out of here, something I wasn't doing near as often on my own disks, but am now doing again since I don't use sysinstall and truly do use the ``whole'' disk for FreeBSD. I can garantee you that if the partition table in your MBR says that all blocks of the device are consumed by a valid parition entry these commercial foreign fdisks will not nuke your disk without first asking you. [I won't say the same for some pd software that will do anything you tell it to without a worry about confirmation or any form of validation :-)] -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation Company Reliable computers for FreeBSD