Subject: RE: NetBSD Hang after disklabel, xsm crash
To: None <harald@bitcon.no>
From: Bernd Salbrechter <salb@hell.gud.siemens.at>
List: port-i386
Date: 07/30/1998 12:22:57
To: Harald Bjoroy <harald@bitcon.no>
In-Reply-To: <Pine.SGI.3.96.980728163412.27835B-100000@box.argonet.co.uk>
Sory if this comes a seconde time, but I didn't found the reply in the
mail-list after one day.
Jasper Wallace wrote:
>
> On Tue, 28 Jul 1998, Harald Bjoroy wrote:
>
>
> > Just tell me how to get passed this, and I will try again. Someone
> > suggested that I should vipe the disk - point me to the tools and I will
> > get right at it.
>
> # dd if=/dev/zero of=/dev/rwd1c bs=1024k count=<size of discs in Mb>
> ^-- this is where the 'whole disk' patririon
> comes in usefull
...
As far I know "d" is the whol disk and the kernel dose not look in
the disklabel if you access the raw devive (/dev/rwd1d). Also I think
zeroing the first few sectors will do the job. At least I have used
this to wipe out several MS-DOS and Windos-NT disks before returning
it for recycling and it worked. But be aware that it overwites also non
NetBSD parts of the disk!
BTW. Dose someone have an idea, why Linux writes so much faser the whole
disk than NetBSD at least with the default parameter for "dd". I got
only a reasonable speed after seting the blocksize to -bs=64k. Also it
behaves bad, if you omit -count after reaching the end of the disk, you
get an error message and no information how any blocks are successfully
written, so i calculate the number of blocks I need befor and set it.
Anyway, because this is not the main function of NetBSD, I not vote for
optimising for this usage ;-).
Bernd Salbrechter