After some further experimentation, I have determined that when a disk =
read error occurs that cannot be corrected with an ecc, something =
(either MSCP, the disk controller or BSD) marks it as bad and does not =
allow anymore writes to the sector.  Anyway, to work around the problem, =
I filled up the disk with a large number of small files and then read =
each file (cat x > /dev/null) and this seemed to mark most of them as =
bad.. Interestingly, the format program on the disk controller (a Dilog =
DQ696) does not catch these sectors, It does catch and redirect other =
errors.  I think the disk might be going bad 1 sector at a time...

  --bryan

----------
From: 	Anders Magnusson[SMTP:ragge@ludd.luth.se]
Sent: 	Tuesday, April 23, 1996 10:31 AM
To: 	Clingman, Bryan
Cc: 	port-vax@NetBSD.ORG
Subject: 	Re: Bad Blocks

>=20
> I'm trying to put NetBSD on my uVaxII with good success until I start =
to =3D
> write the usr.*.tar files.  Unfortunantly, this area of the disk I'm =
=3D
> installing on has some errors.  Should the controller mask these out =
=3D
> (like on SCSI drives), or is there something else I have to do.  The =
=3D
> edlabel instructions say that the badblock stuff hasn't been =3D
> implemented, but is netbsd able to read the Bad Block list off of the =
=3D
> drive (the format software for my controller can write the bad block =
=3D
> list)
>=20
There are no working bad block replacement system on MSCP disks,=20
and I have no idea about how it should work (I have no docs on
the MSCP protocol). What should work anyway is the program badsect;
which creates a file of the bad blocks mentioned. Try that.

-- Ragge