RX50 on RQDX3 on 2.11BSD

[Steven M. Schultz]

Hi -

> From: "Jason T. Miller" <jasomill at shaffstall.com>
> > 	Wow, quite a bit of interest in 2.11 these days - I suppose I should

> It's the first PDP operating system I've enjoyed working with, and one of
> the coolest UNIX implementations I've had the pleasure of working with.

	Ah, thanks!   I can't claim _all_ the credit but 2.10.1 was more or
	less directly my "fault" and 2.11 was all set to be called 2.11SMS
	until one of the CSRG folks intervened and gave me the BSD imprimateur.

> I unearthed two Teac 55-G series floppy drives, and they're both broken
> (won't format w/verify) -- the RX50 isn't the only flakey floppy. I've got

	Sigh. 

	Anyhow, to the problem you observed dd'ing data to an RX50 and the
	ensuing compare error.

	I'm using an RX33 (well, mod'd Teac 5.25" drive) on a RQDX3.

	I freshly formatted a floppy.  That's one nice thing about the RX33,
	the RQDX3 can format floppies using ZRQF?? - RX50's meant getting 
	preformat'd media or a Rainbow to do the formatting from what I
	remember.

	Then before doing anything I enabled a bit of extended logging from
	the MSCP driver with

		sysctl -w machdep.mscp.printf=9

	The  first access to the drive ("disklabel ra9") elicited a 
	"ra9a=entire disk: no disk label" message.  This is expected and
	correct - the kernel saw there was a corrupt/missing label and came
	up with a label that spanned the 2400 sectors of the drive using the
	'a' partition.

	Next a 1.2mb file (sector 0 having zeroes, sector 1 having ones, etc)
	was dd'd:

		dd if=/tmp/data of=/dev/rra9a

	and almost immediately dd reported:

	write: Read-only file system
	2+0 records in
	2+0 records out

	That probably should have been 2+0 and 1+0 since dd read two sectors but
	only successfully wrote one.   A bug in 'dd' perhaps that it doesn't
	decrement the output count on a write error.

	At any rate you should error out if the label area is not write
	enabled.  The 'disklabel' program automatically enables and disables
	the writeprotect when writing the label in case you were wondering
	about that ;)

	After doing the "disklabel -W ra9" the "dd" works fine and the floppy
	compares identical to the input file.

	The MSCP driver hasn't changed in quite a while so if you retrieved
	2.11 fairly recently the problem's not a bug in ra.c that I can
	see (or if it is, it's particular to the RX50 somehow).

	Why 'ra9' (I hear you ask)?   Well, the system is currently booted
	from a different controller (Emulex UC08).  The boot controller is
	*always* 'ra0 thru ra7' no matter what the CSR is.   The secondary
	controller (the RQDX3 in this case) is always 'ra8 thru ra15'.  The
	RD54 is 'ra8' (first drive on the 2nd controller) and the RX33 is
	ra9 (second drive on the second controller).

> > 	Oh, for debugging purposes you can enable more or all of the MSCP
> > 	messages with the 'sysctl' command:
> > 
> > 		sysctl -w machdep.mscp.printf=X
> 
> I will. Thanks. I didn't even know 2.11 had 'sysctl'. Cool.

	One more thing I stuffed into the system.  You'll also find 
	"sigaction" and friends along with RTS/CTS flowcontrol (for devices
	which support it), and numerous other goodies imported from 4.4BSD (the
	latest addition was 'pselect(2)' just a couple months ago).


	Steven Schultz
	sms at moe.2bsd.com

Received: (from major at localhost)
	by minnie.cs.adfa.edu.au (8.9.3/8.9.3) id PAA25267
	for pups-liszt; Thu, 8 Jun 2000 15:43:47 +1000 (EST)
	(envelope-from owner-pups at minnie.cs.adfa.edu.au)