On Apr 12,  1:19pm, Olaf Seibert wrote:
> osymh@gemini.oscs.montana.edu (Michael L. Hitch) wrote:
> >   I've been playing with the driver a bit, and have gotten it to read
> > fine, but I've had problems when writing to the floppy.  I've even built
> > a very minimal boot floppy and booted NetBSD a couple of times using the
> > floppy as the root file system.  I had to modify the delay() routine
> > since a 33Mhz 68040 runs a bit faster than the 68030 systems.
> 
> Aha. That may be why I get so many MFM_TRACK messages (indicating that
> it read the wrong track). Also, so far I haven't succeeded in mounting
> a floppy (using friday's sources); I get something like GCIOCTL: wrong
> argument from fsck, while newfs complains about cg0. But tarring to the
> disk is reasonably successful (until I get the MFM_TRACK messages).

  I was getting lots of MFM_TRACK messages that I was finally able to get
rid of by increasing the delay times.  I'm using something like
	DELAY (delay_ms * 40000);
in the delay(delay_ms) routine at the moment.  That seems to let me do
random seeks when doing newfs and other operations.  However, I now get
MFM_DATA messages after doing some writes.

  I also noticed that the retries() routine can be called with no active
block buffer, so the references to bp->xxx can panic the system.  I put
a check in retries() and fdstate() for a NULL bp value.  I think I've only
seen in from the retries() routine so far.  It seems to occur when I get
timeouts on a write.

Michael

-- 
Michael L. Hitch			INTERNET:  osymh@montana.edu
Computer Consultant			BITNET:  OSYMH@MTSUNIX1.BITNET
Office of Systems and Computing Services
Montana State University	Bozeman, MT	USA

------------------------------------------------------------------------------