Subject: Disklabel problems under 1.3.3
To: None <port-sparc@netbsd.org>
From: Jukka Marin <jmarin@kyyhky.jmp.fi>
List: port-sparc
Date: 01/19/1999 18:18:14
Hello,

I had a disk crash (those wonderful quantum crap disks once again) and
am trying to install NetBSD 1.3.3 on SS1+ with a 2.2 GB Quantum Viking
(never can get rid of these - when a quantum dies, they send me another
which dies in a year or so).

Anyway, I can't label the new disk properly.  I've been trying for about
4 hours now.  Whenever the total disk size (tracks*sects*cyls*512)
exceeds 2 GB, disklabel prints:

disklabel: ioctl DIOCWDINFO: Label magic number or checksum is wrong!
(disklabel or kernel is out of date?)

If I change number of sectors or cylinders so that the result is below
2 GB, everything works just fine.

I would really like to use _all_ of my disks, not just the first <2 GB.

Am I doing something wrong or is there a BUG somewhere?

Here's my disklabel (from disktab) - this one works:

Viking:\
        :ty#winchester:nt#4:ns#180:nc#5820:\
        :pa#131072:oa#0:ba#4096:fa#512:ta=4.2BSD:\
        :pb#262144:ob#131072:tb=swap:\
        :pc#4190400:oc#0:\      
        :pd#983040:od#393216:bd#4096:fd#512:td=4.2BSD:

If I change nc to 5826, I get the error above.  (I'd like to set nc to
6144, btw).

Here's the command I use to write the label:

# disklabel -w -r sd3 Viking

Please e-mail to me directly (to jmarin@kyyhky.jmp.fi) because my
regular mail system is down (until I can figure out how to label the
disk).

I don't have this problem under i386, some maybe this is a sparc
specific thing?  The error message I get is caused by a failing
ioctl, so the problem is in kernel (right?).

Please reply soon.. need to get my mail working again :-I

Thanks,

  -jm