Date: Wed, 20 Jan 1999 23:00:27 -0800 From: Mike Smith <mike@smith.net.au> To: "John Saunders" <john.saunders@nlc.net.au> Cc: Mike Smith <mike@smith.net.au>, freebsd-current@FreeBSD.ORG Subject: Re: Booting -current with new loader Message-ID: <199901210700.XAA06653@dingo.cdrom.com> In-Reply-To: Your message of "Thu, 21 Jan 1999 11:58:13 %2B1100." <Pine.LNX.3.95.990121114034.29987E-100000@nhj.nlc.net.au>
next in thread | previous in thread | raw e-mail | index | archive | help
> > > I'm not sure about anybody else here, but to my mind a "Winblows on C: > > > drive and let's try out FreeBSD on the second disk" configuration should > > > really be supported seamlessly. > > > > You're more than welcome to propose a technical solution that solves > > the problem. > > OK, the technical solution that I propose is an alternative (optional > because I want to retain the current semantics) config semantics. > > controller wdc0 at isa? port "IO_WD1" bio irq 14 flags 0xa0ffa0ff > controller wdc1 at isa? port "IO_WD2" bio irq 15 flags 0xa0ffa0ff > disk wd? at wdc? > > This would act identical to the way the SCSI system works by assigning > names to disks as they get probed. It would also have the advantage of > assigning disk names in the same order as the BIOS (and thereby the boot > loader), hence there would be a 1 to 1 mapping between what the boot > loader uses and what the FreeBSD kernel detects. > > I believe the current problem stems from the BIOS (and thereby the boot > loader) and the FreeBSD kernel assigning names using a different algorithm > and then getting out of step when the kernel skips a name. You're spot on the money here. If you have any good ideas about guaranteeing the probe order (and there are some other issues here, like what about systems where the BIOS doesn't know about the IDE disks?) matches up, I'm all ears. Later versions of the EDD spec offer a technique for geting a device path for a given ID, but support for that seems to be *very* flaky. > If anybody knows how to do this quickly then I am happy to help test it. Right now I'm looking at stamping the disklabel which contains the designated root filesystem and passing the stamp into the kernel, which will then go look for it. There are still gremlins with this approach, but I can't think of anything better. 8( -- \\ Sometimes you're ahead, \\ Mike Smith \\ sometimes you're behind. \\ mike@smith.net.au \\ The race is long, and in the \\ msmith@freebsd.org \\ end it's only with yourself. \\ msmith@cdrom.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199901210700.XAA06653>