From owner-freebsd-current Sat Nov 7 16:30:26 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id QAA00569 for freebsd-current-outgoing; Sat, 7 Nov 1998 16:30:26 -0800 (PST) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from ceia.nordier.com (m1-1-dbn.dial-up.net [196.34.155.1]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id QAA00550 for ; Sat, 7 Nov 1998 16:30:17 -0800 (PST) (envelope-from rnordier@nordier.com) Received: (from rnordier@localhost) by ceia.nordier.com (8.8.7/8.6.12) id CAA00278; Sun, 8 Nov 1998 02:28:27 +0200 (SAT) From: Robert Nordier Message-Id: <199811080028.CAA00278@ceia.nordier.com> Subject: Re: Boot Loader question In-Reply-To: from ADRIAN Filipi-Martin at "Nov 7, 98 04:40:03 pm" To: adrian@ubergeeks.com Date: Sun, 8 Nov 1998 02:28:23 +0200 (SAT) Cc: rnordier@nordier.com, freebsd-current@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG ADRIAN Filipi-Martin wrote: > On Sat, 7 Nov 1998, Robert Nordier wrote: > > > I have substantially revised boot1 and should be committing this > > within the next day or two. Until now, we've been making use of > > a new set of BIOS extended disk services, but it has become painfully > > evident that a few BIOSes have problems with this approach, and they > > really weren't much of a win, anyway. > > When do you expect to commit your changes. I bagan hacking on the > stage 1 boot block a few days ago to try and make it behave properly when > using a serial console. Around late Sunday, most likely; I'm still testing here at the moment. > The -current ones end up with garbage in the boot prompt after it > reads a boot.config with just a '-h' in it. After it switches to the > serial console, I have to backspace and type 'kernel', and it boots just > fine. If boot.config is empty, it boots fine but of course, I don't get a > chance to boot an alternate kernel since it doesn't switch to the serial > console until the kernel starts probing. The boot.config is currently being parsed twice if it boots with no intervention. This became a problem when the option behavior changed from `or' to `xor'. The patch below gives a temporary fix (the printf statement is changed just to make space). I'm not sure about the garbage, though, I'm not seeing it here. What kind of garbage (like random line noise or more like a garbled prompt)? Index: boot2.c =================================================================== RCS file: /home/ncvs/src/sys/boot/i386/boot2/boot2.c,v retrieving revision 1.13 diff -u -r1.13 boot2.c --- boot2.c 1998/10/27 20:16:36 1.13 +++ boot2.c 1998/11/07 23:45:08 @@ -157,9 +157,10 @@ readfile(PATH_HELP, help, sizeof(help)); readfile(PATH_CONFIG, cmd, sizeof(cmd)); if (*cmd) { - printf("%s: %s", PATH_CONFIG, cmd); + printf("%s", cmd); if (parse(cmd)) autoboot = 0; + *cmd = 0; } if (autoboot && !*kname) { if (autoboot == 2) { > If you need a tester for the serial functionality, I'm willing. I > have 6 rackmounted machines with serial consoles and I will be fixing this > on my own if it isn't already fixed. Thanks, testing is always very welcome; and any fixes that work for you would be handy too, particularly if it is something that can't be reproduced here. > Also, are the boot blocks going to be built using elf or aout in > the long run? I'm just getting into -current and I figired out that I > need tom compile them as aout or they don't fit. Definitely as ELF: aout has never been an option (IIRC, sio.s shouldn't even assemble with the aout version of gas). They should fit, though; they fit here, and would breaking "make world" for a lot of folks otherwise. -- Robert Nordier To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message