[Simh] error

[Johnny Billquist]

Hi.

On 2016-02-09 23:43, Mark Pizzolato wrote:
> On Tuesday, February 9, 2016 at 2:16 PM, Johnny Billquist wrote:
>> On 2016-02-09 22:58, Mark Pizzolato wrote:
>>> You may recall that older computers didn’t have a way to
>>> programmatically power themselves down.
>>
>> Well, this isn't about powering down, but halting...
>
> Well, not really.  The VAX has a HALT instruction and if that was all that was needed, it could be executed.

You forced me to read some manuals and drag up long unused memories. 
Thanks. :-)
You are right, there is a HALT. I probably was confusing this with the 
non-existant WAIT in the VAX. Sorry about that.

But the issue really is about halting, and not power down...

> However nothing defines HALT to do anything but stop executing instructions.  Well sitting in an infinite loop is sort of like that, but to save power someone might want to turn the system off since none of the VAX model systems that simh simulators for has any way to programmatically influence power.

If only things were that easy. :-)
Let's start by agreeing that VMS, when you shut the system down in an 
orderly fashion, ends up in a tight loop. Same with Ultrix. And NetBSD...

Now, your argument would be that this is because people should go and 
turn off power. Fair enough. But nothing would prevent you from HALTing 
just as well as going into a tight loop, for this. I mean, after all, 
once you printed out the message, what difference would it be to HALT 
instead of just spinning?
So, why do all these OSes go into a spin instead of halting? The reason 
is that after a HALT, you do not have any control of what happens. 
Depending on the model, the front end takes over, or some firmware, or 
god knows what else. And, as is the case with the large old models, you 
have a couple of switches that is set which tells what the machine will 
do at this moment.
So it might halt, it might reboot... The thing is, you do not have any 
way of controlling this from software. So, if you really want to stop 
the machine, you cannot rely on HALT on a VAX. So the only way is to spin.
And then the operator can manually halt the system in which ever way 
works for that model. If he then boots something else, or powers the 
system down, or what not, is the operators problem. Not the OS.

> The original poster executed the operating command to shutdown the operating system.   That would imply that he didn't want it running anymore.  If he wanted it to continue running after shutting down, the operating system command to reboot the system could have been executed.

Yes, he obviously did not want the system running anymore. It is just a 
peculiar quirk that no OS on the VAX actually halts the CPU, but instead 
they all just go into a spin.

If he wanted to continue after shutting down, on a real machine, he 
would have to first halt the machine, and then reboot it, yes.

>>> Would you want it to behave differently?
>>
>> I think simh does a reasonable thing, if you ask me. It was a mistake that the
>> VAX didn't standardize how to halt or reboot the machine, but that is not
>> simhs fault.
>
> There are absolutely mechanisms to reboot each of the VAX systems which
> simh has simulators for and reboot does work fine on these simulators.

simh do not work the same as the actual machines it tries to emulate... 
The 8600 is a fine example. Rebooting the 8600 actually requires sending 
specific commands to the FE before doing a HALT. There are more 
functions and features available over that communications channel. simh 
do not implement most of this. Also note that rebooting is done 
differently on the different models simh even do emulate. But yes, the 
specific commands used to reboot are recognized by simh, so that detail 
to work.

In short - VAX do not have a standardized way to halt or reboot the 
system. I didn't say there were no mechanisms. I said there was no 
standardized mechanisms... :-)

And for the 8600 there is no sure way to HALT the machine. There is a 
sure way to reboot it, but none to halt it. If you execute a HALT, 
something happens, but it might not be what you expected/thought.

> As it turns out, these mechanisms usually use the HALT instruction which
> interacts with some model specific features of the model specific console
> subsystem to effect the reboot.  So, clearly the message which VMS issues
> "USE THE CONSOLE TO HALT THE SYSTEM" is actually using the word HALT
> to mean 'turn off' rather than anything which relates to the halt instruction.

No. HALT means HALT. Really. It does not mean turn off. Most common 
reason for shutting down was actually not to power down the machine, but 
to run PM on it.

>> And some VAXen do have the ability to power themselves down as well...
>> But it's all very model specific.
>
> If we had a simulator for a VAX model system which had that functionality
> it would be supported.

I wonder what powering down would imply for simh? Just exiting simh, or 
should it power down the host? :-)

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol