Bob;

I sincerely apologize if I have insulted your skills as a programmer,
computer user, or as a individual. That was never my intention. My only
intention was to document the steps necessary for me to solve this problem,
and to possibly help other users who have encountered these errors, solve
them.

On re-reading my post, I agree that, I too, might feel that the author (me)
has slighted Mkfs.

I hope that others reading my past posting, will not take this hastily
composed statement as a negative viewing of the Mkfs program.

Mike


mwalker <mwalker@en.com> wrote:
>
>>Specifically, I found that there seems to be a problem with Mfks to deal
>>with very large drives, and the NetBSD partitions that reside in the upper
>>portions of these drive. Apparently, this just might be caused by Mkfs'
>>failure to recognize partitions in the upper portions of large drives,
>>although there seems to be a bit of a disagreement over this. Secondly, I
>>also determined, mostly through my own trial and error, that Silverlining
>>does not do a good job with A/UX partitioning.
>
>*sigh*  I really wish you hadn't posted this to the mailing list without
>first checking all your facts.  The disk I/O routines used in Mkfs are
>the same ones used in the Installer. They were modified a few years ago
>to handle large disk drive addressing.  Prior to those changes Mkfs and
>the Installer were unable to properly handle disk partitions above the 1
>Gig limit.
>
>Mkfs may indeed have a problem with initialization of some disks, but
>it's not an I/O problem and it's not strictly a problem with large disks.
> The whole reason I became involved in working on Mkfs was that the old
>version was corrupting structures on my 1.6Gig disk. Since then I've
>successfully used it on disks as large as 8.9 Gig.
>
>If there is a problem with Mkfs it's in the underlying "newfs" code which
>is quite old having been lifted from NetBSD 0.9.  Many changes have been
>made to the NetBSD code since then, but none of those changes have been
>ported into Mkfs.
>
>Finally, there are some known problems with some SCSI disks that are not
>completely understood yet.  These problems affect disks under NetBSD
>running both NCR and SBC drivers. In fact the whole reason for the SBC
>driver in NetBSD is to provide something that might work for users when
>the NCR dirver doesn't.  More recently we've seen some problems with SCSI
>disk I/O while running under MacOS.  These may or may not be related to
>the problems we're tring to isolate in NetBSD.  Some users have reported
>varying degrees of success by disabling the SCSI disk on-board buffering.
> Apparently this is an option in some disk formatters.  As always,
>dealing with SCSI chains is a bit of a black art.  Most people don't
>understand or appreciate the need for good connections, organization and
>termination.  Larger disks almost always benefit from Active Termination.
> In fact on my 8.9 Gig drive I can only format and use the first few 100
>Meg if I don't have it terminated with an Active Terminator.
>
>Anyway, now that you've posted that Mkfs has a problem, maybe you're also
>volunteering to fix it?  After all, you're the one with the dirve that
>doesn't work with Mkfs.
>
>-bob