On Tue Aug 31 19:29:14 1999, Ross Harvey wrote:
> > From: Howard Leadmon <howardl@abs.net>
[..]
> > First, when I boot from SRM to the hard-drive, everything seems like it's 
> > going great.  I see the primary loader run, then the secondary, and it
> > initializes all the devices.  Finally it goes to mount the root filesystem
> > and I get a message like telling me it can't. here is the msg from the 
> > logs:
> >
> > WARNING: can't figure what device matches "IDE 0 108 0 0 0 0 0"
> 
> Sigh, SRM has gone back and forth on how it identifies IDE boot devices
> for that board. Some SRM versions call them SCSI, some IDE, and now it
> looks like sometimes the device is in the 100s digit and sometimes in the
> 1000s digit. Maybe the SCSI-happy SRM uses the 4-digit and the IDE-
> happy SRM uses the 3? Or maybe my quick guess is just wrong...

A SRM before v5.3 use that on my 164LX:

SCSI 0 11 0 0 0 0 0

v5.3 or higher use:

IDE 0 11 0 0 0 0 0

> Either way, this is an easy kernel fix, except that it looks like the
> obvious fix will break platforms it presently works fine on. So, either
> say "root on wd0" in your kernel config, or hack (if the board is system
> type 26, "eb164", I'm not sure) dec_eb164_device_register() in
> /sys/arch/alpha/alpha/dec_eb164.c to recognize the `108' properly.

We don't look for the number, we just check for BOOTP, SCSI or IDE.

> Actually, one of us (Veego?!) can do that for you, if you will send me or
> post your dmesg. And it would be really cool if you could s/#if 0/#if 1/
> around the printf's in that above module first, and then use _that_ kernel
> for the dmesg you send us.

Yup, I can help when I get the dmesg output.

Maybe I can put an current kernel this weekend on an ftp server which contains
the debug code.

Bernd