[Info-vax] OT: Halt and Catch Fire

glen herrmannsfeldt gah at ugcs.caltech.edu
Thu Jun 4 07:55:08 EDT 2015


Neil Rieck <n.rieck at sympatico.ca> wrote:
> On Tuesday, June 2, 2015 at 11:25:53 PM UTC-4, JF Mezei wrote:
>> Yeah, I know , "HCF, Halt and Catch Fire"  was an IBM 370 instruction in
>> many "hacked" university documents.
 
(snip) 
> As others have already pointed out, there was no HCF instruction 
> but I have been told by people who worked on those machines that 
> you could not stop the CPU on some IBM machines: Halt really meant 
> the CPU went into a tight NOP loop (which is were the catch-fire 
> myth comes from).

You stop the CPU from executing instructions by setting the WAIT
bit in the PSW.  360's were leased, with the charge depending on
how much you used the system. There is a usage meter that runs when
the CPU is running, that is, when the WAIT bit is not set. 

When there is nothing else to do, the CPU goes to a WAIT state, until
the next interrupt. 

There are some error conditions, mostly during IPL when things
aren't going well, called disabled WAIT states.  If you set WAIT
and disable all interrupts, then the system stops.

-- glen



More information about the Info-vax mailing list