From owner-freebsd-hackers Mon Jan 23 13:21:55 1995 Return-Path: hackers-owner Received: (from root@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id NAA15207 for hackers-outgoing; Mon, 23 Jan 1995 13:21:55 -0800 Received: from cs.weber.edu (cs.weber.edu [137.190.16.16]) by freefall.cdrom.com (8.6.9/8.6.6) with SMTP id NAA15200 for ; Mon, 23 Jan 1995 13:21:52 -0800 Received: by cs.weber.edu (4.1/SMI-4.1.1) id AA09719; Mon, 23 Jan 95 14:16:03 MST From: terry@cs.weber.edu (Terry Lambert) Message-Id: <9501232116.AA09719@cs.weber.edu> Subject: Re: writing bootcode; need a method To: mmead@goof.com (matthew c. mead) Date: Mon, 23 Jan 95 14:16:03 MST Cc: hackers@FreeBSD.org In-Reply-To: <199501231923.OAA03201@goof.com> from "matthew c. mead" at Jan 23, 95 02:23:28 pm X-Mailer: ELM [version 2.4dev PL52] Sender: hackers-owner@FreeBSD.org Precedence: bulk > > For right now, you are required to have a disklabel before proceeding. > > This unfortunately isn't the problem. The problem lies in the fact > that the sysinstall does not require BSD partitions to begin and end on > cylinder boundaries, whereas the disklabel program does. Thus, the disklabel > program, when told to write boot code, finds what it considers to be errors > with the disklabel, because sysinstall has made partitions that don't begin/end > on cylinder boundaries, and does not write out the new boot code, thus > hd(1,a)/kernel can NOT be made the default, again, simply because disklabel > never writes the boot code. Oh. Well. That's very different. 8-). I don't know about the ending BS, but the beginning BS is correct. The MBR may or may not be able to load a second stage boot that doesn't fit in a track, since it might be incapable of sector I/O to do the deed. Now, judging this on the basis of a BSD-apparent instead of a BIOS apparent starting location, that *is* a mistake. And that's what disklabel seems to be doing. It's a wonder anyone with a translated geomtery drive can install at all. -- BEGIN WORKAROUND -- What *should* work as a workaround is telling the machine the BIOS apparent geometry so it gets the same idea about cylinder start in BIOS and in BSD (sysinstall and disklabel). -- END WORKAROUND -- Meanwhile, anyone want to cut a new sysinstall AND disklabel program? Or just a disklabel hacked to not care and a warning in the install instructions about picking a cylinder boundry in sysinstall? You *can* write the boot block out there with dd or even cat by shoving it directly out the 0c device (or 1c for disk 1, etc.), but it won't help you, since you will still be screwed when the boot block goes looking for a valid disklabel. Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.