11/93 Rebuild - SCSI HD now boots RT11

[Paul Koning]

> On May 31, 2019, at 2:40 PM, allison via cctalk <cctalk at classiccmp.org> wrote:
> 
> On 05/31/2019 02:04 PM, Rod Smallwood via cctalk wrote:
>> Hi
>> 
>>     Well I now have a bootable SCSI drive on my 11/93. Its not RSTS/E
>> (yet) but it is RT 11 and reliable.
>> 
>> Its a bit baseline but it runs.
>> 
>> So next up was to see if we could get the RQDX3 to co-exist with the
>> SCSI controller.
>> 
>> I switched the base address to 160336 and it does not stop the SCSI
>> drive booting as DU0.
>> 
>> Had the RQDX3 been on the normal base address I think you would get the
>> HD as DU0 and the two halves of an RX50 as the next two drives.
>> 
>> But what happens to the RX50's when you move the RQDX3 to 160336 ?
>> 
>> Rod
>> 
>> 
>> 
> Under RT-11 ...
> 
> RSTS/E the conventions for non standard device addresses are different
> but there is a mechanism for addressing that.  I've not used that.

On RSTS:

DU is one of the fairly common kinds of PDP-11 devices that has a fixed CSR assignment for the first controller, and uses floating CSRs for the others.  160336 doesn't sound like a standard float address (it isn't a multiple of the CSR size).  

RSTS will recognize devices automatically if they have the standard CSR (fixed or float).  Anything else you have to give it the CSR address: boot it, at the "Start timesharing" prompt say "NO" then you're in the INIT program.  Option "HARDWARE" lets you do stuff like set non-standard CSR addresses; it will also show the devices it has found.

MSCP devices have programmable vectors so that's not a concern, RSTS will assign a vector.

For most RSTS disks, only one controller is supported.  For MSCP, it supports up to four.  But unlike VMS, RSTS refers to disks by device type and unit number (such as "DU2:").  So the rule for multiple MSCP controller is that the unit numbers must be distinct.  Also, they must be below 16, RSTS does not allow larger unit numbers than that.

	paul