[Info-vax] DZQ11 on VAX/4000 Model 300

Stephen Hoffman seaohveh at hoffmanlabs.invalid
Mon Jan 14 17:02:39 EST 2019


On 2019-01-14 21:19:21 +0000, David Wade said:

> On 14/01/2019 17:42, Stephen Hoffman wrote:
>> On 2019-01-14 17:04:41 +0000, David Wade said:
>> 
>>>   I have just installed a DZQ11 in my VAX/4000 and once I figured out I 
>>> had by "A" and "B" pins mixed up on the BERG connector  (I had to 
>>> fabricate a lead as it came without any) on a brief test it seems to be 
>>> working just fine.
>>> 
>>> However when I removed it the VAX seemed to hang on boot. I popped it 
>>> back in and it seems to be working fine.
>> 
>>> So should I do anything to VMS to manage its configuration?
>> 
>> Nope.  OpenVMS is working exactly as expected.
>> 
> 
> Why does it hang when the card is removed? Thats not expected to me...
> .. if the CSRs worked before, surely they work the same afterwards....

You're probably thinking about how the somewhat modern buses work, like PCI.

That's not how CSRs and interrupt vectors work.  Q-bus is a much older design.

Q-Bus didn't have enough address space to assign everything its own 
unique CSR and vector, akin to what PCI offers with the PCIID.

Which meant some devices had fixed addresses, and some didn't.

Hence what's called the "floating vectors" and "floating addresses" on Q-Bus.

The CSR settings of each board are dependent on the other Q-bus boards present.

Add or remove one board, and the settings on other boards can require changes.

In a few cases, adding or removing a disk hanging off of a Q-bus board 
can require changes to other boards, too.

Incorrect CSR and vector settings will cause configuration errors and hangs.

OpenVMS VAX probes the bus at boot, and—based on the patterns of what 
memory is accessible where in I/O space—loads and connects different 
drivers.  OpenVMS VAX assumes the Q-bus is configured per the 
requirements, and both OpenVMS VAX SYSGEN and various recent consoles 
can feature a CONFIGURE command.  This CONFIGURE command accepts the 
list of devices present, and you then get to pull all the boards and 
set the switches to match the expectations.  When you mess up or when 
somebody pulls a board or otherwise messes with the configuration, the 
autoconfiguration may not detect the board, may misdetect the board, or 
may configure the wrong board.  And if the interrupt vectors are wrong, 
bus hardware interrupts won't arrive at the host as expected.

Search for and read the various previous Q-bus discussions here in 
comp.os.vms newsgroup, when modern assumptions slammed into old bus 
designs.    This has gotten explained on various occasions.

The VAXstation user's manuals had a decent description of the Q-bus.  
Most boards didn't explain the whole bus, just how to set the board CSR 
and (if present) vector settings to meet the current configuration 
requirements.

More reading, though the following is a bit more dense:
http://www.textfiles.com/bitsavers/pdf/dec/standards/EL-00160-00-0_A_DEC_STD_160_LSI-11_Bus_Specification_Sep91.pdf 

http://web.frainresearch.org:8080/projects/pdp-11/index.php


>> HPE is exiting the OpenVMS business.
> 
> Then why does it still have the documentation catalogue ...

Note my use of "exiting" and not "exited".  What they have posted is a 
small fraction of what was once posted, and it's only getting smaller.

Most of that content is now gone, though someone did archive the old 
OpenVMS Ask The Wizard area some years ago.  You can fetch the contents 
of the OpenVMS Ask The Wizard area archives from the OpenVMS Freeware.  
There are various copies of the Freeware around, including at the 
following server:
https://www.digiater.nl/openvms/freeware/v60/wizard/
Have a look at topic 1149 there, as that's related to some of the Q-bus 
discussions.

>> Use Google Groups search of the comp.os.vms newsgroup archives, search 
>> for the word "serpentine" (though that's less of an issue with BA213 
>> and BA215 and BA4xx series Q-bus boxes), and learn how the Q-bus and 
>> the Q-bus boards are configured.
>> 
> 
> The BA440 backplane isn't a serpentine plane is it?
> That is a Serpentine backplane isn't it? Frome reading the VAX/4000 
> manual the BA440 isn't a serpentine backplane. The CD slots are seem to 
> be generally only used for memory and there is no

Hence my "less of an issue", but that search keyword will get you to 
many of the discussions of the topic here in the comp.os.vms newsgroup. 
 You'll still have to understand how the DMA grant passes, and which 
CSRs and vectors are required.


-- 
Pure Personal Opinion | HoffmanLabs LLC 




More information about the Info-vax mailing list