Subject: Re: Installing problems...
To: RiscBSD Mailing List <port-arm32@NetBSD.ORG>
From: Robert Black <r.black@ic.ac.uk>
List: port-arm32
Date: 01/27/1997 11:42:26
On Jan 26,  9:13pm, Kjetil B. Thomassen wrote:
> Subject: Re: Installing problems...
> On Sun 26 Jan, Scott Ashcroft wrote:
> > On Sat 25 Jan, Mark Brinicombe wrote:
> >
> >> Ok the kernel is finding a drive but the drive appears to actually respond
> >> as drive 1 rather than drive 0. This may just be a case of a switch on the
> >> drive.
> >
> >
> > Does the kernel wire up the drive to /dev/fd1* ?
> >
> > If not why not?
>
> If I have understood BSD correctly, it will start on the lowest
> numbered devices and allocate the ones found successively.
>
> That means that if you only have one device of a type, that will
> always end up as device 0 no matter which physical number it
> actually is.

Nope. For some devices you can configure the kernel to do this. The SCSI bus is
normally configured like this because some ports have (or had?) a restriction
on the number of characters in a device name. Since the Machine Independant
devices are named globally this means that the device naming could get ugly if
you wanted to support a lot of SCSI buses. The GENERIC kernel is supposed to
support all configurations (almost) so it has to have this type of allocation.
It is perfectly possible to configure a kernel which has fixed device numbers
for fixed SCSI ids. Looking at the floppy configuration in the GENERIC config
it looks as if the numbers should be allocated sequentially starting at zero. I
suspect that something somewhere is making an assumption that it shouldn't,
probably because of the driver's origins in the i386 port.


Cheers

Rob