On Sat, 23 Nov 2002, M. Warner Losh wrote:

> Ah, I was confused by netbsd.ecoff and wondered where it should go.
> Too many years of doing OpenBSD/arc and putting the file in /dos to
> get it loaded. :-)
>
> : I guess you have tried that already hence the reason for asking here. If
> : you are still having problems it would pay to provide more info and/or
> : screen output.
>
> I hadn't tried that.  Is there a way to break into the boot process if
> this new kernel is no good?


The boot loader will fall back to onetbsd on failure (plus the usual
list of suspects)

If you get yourself into real trouble you can use the following process:

1)  Press ^C during the NetBSD primary/secondary bootloader process
    which should return you to the '>>' console prompt if not
    there already....

2)  boot -f dksd(,#,8)boot dksd(,#,8)oldkernel

    (Where # is the SCSI ID of the NetBSD disk, or empty for 0)

    (The -f dksd(,,8)boot step can be removed if you have set it
     in the 'bootfile' environment variable)

If you need to pass arguments to the NetBSD kernel, the console will
attempt to interpret it and you will loose the battle getting it to the
kernel.
The correct process is to place another dash in front.  Here is a complete
example for starting NetBSD in single user mode:

	setenv bootfile dksd(,,8)boot
	boot --s


-- 
Wayne Knowles			NetBSD/mipsco port maintainer
wdk@netbsd.org			http://www.netbsd.org