Anne & Lynn Wheeler wrote:
>following is on bitsavers, 1967, os (i.e. mvt) sup****t for 360/91:
>http://bitsavers.org/pdf/ibm/360/C28-6666-0_360-91_OSsup****t.pdf
>it has table on pg. 39 on what is precise and what is imprecise. for
>standard/precise interrupts, the ILC field in the program old psw, has
>the "length" of the instruction causing the interrupt. for imprecise
>interrupts, the ILC field is zero.
Yes, ILC zero is the indication for imprecise interrupts.
>on page 10, discussing program first-level interrupt handler, there is
>and added note (bar-code from TNL update, aka Page revised by TNL
>N28-2308, 1/31/68):
>| Note: When an imprecise or a multiple-imprecise program interruption
>| causes a task to be terminated, the completion code is "0C0" since the
>| last digit reflects the decimal content of bits 26-31 in the program
old
>| PSW.
>which is then translated to S0C0.
I had thought that for single imprecise interrupts it generated the
appropriate S0Cx code, with only S0C0 for multiple imprecise interrupts.
But I am remembering from much later, at least 1972. It could be that
the FLIH was changed, or that the PSW was decoded later in message
processing to generate the appropriate code. It could also be that
the system codes manual didn't agree with this note.
The ones I am remembering would have been generated by Fortran code.
In any case, I had thought that S0C0 was only for multiple imprecise
interrupts, and a previous post seems to agree. The OS Sup****t
manual from January 1968 doesn't seem to agree, but that is pretty
early in the life of the Model 91.
-- glen
|
