Subject: Re: Problems with TK70
To: None <port-vax@netbsd.org>
From: Carl Lowenstein <cdl@mpl.ucsd.edu>
List: port-vax
Date: 02/04/1999 13:41:05
> From: Andreas.Borsodi@hydro.com
> To: port-vax@netbsd.org
> Date: Thu, 4 Feb 1999 13:56:35 +0100
> Subject: Re: Problems with TK70
> 
> 
> Hi,
> I had the same problem awhile back. The first and second QBUS slots are
> occupied by a KA640 CPU and a 8Mb RAM card.
> The TK70 controller card where placed in the upper fifth QBUS slot. The
> SHOW DEV command displayed the controller,
> but not the tape drive and it made a high pitch sound with short intervals.
> I moved the TK70 controller card to upper third QBUS slot and the sounds
> disappeared. The SHOW DEV now correctly
> displays the TK70 and I can boot from it. The sounds are probably some sort
> of warning sound.

Grant Continuity.

With empty slots between the CPU and the controller, any bus requests
made by the controller are never granted.  That means that it can't
make interrupts or do DMA data transfers.  And in fact such an action
by the controller will put the CPU into some kind of abnormal state.
No doubt, the SHOW DEV command just checks for the presence of device
registers using programmed I/O.

In a horizontal BA23 backplane, the first three slots are Qbus on the
left side, CD interconnect on the right side.  A BA123 backplane is
rotated 90 degrees with respect to this, and the first four slots are
Qbus on the upper half and CD interconnect on the lower half.  After
this, the Qbus grant wiring proceeds in a serpentine fashion.

Short summary:  If all you have is CPU and memory and one controller,
put the controller as close to the memory as you can, in the AB slot,
which is left or upper, depending.  I'm sure that this is in an FAQ.

    carl

        carl lowenstein         marine physical lab     u.c. san diego
        {decvax|ucbvax} !ucsd!mpl!cdl                 cdl@mpl.ucsd.edu
                                                  clowenstein@ucsd.edu