> >> After it showed up on a customer's STROBE report, we switched over to
> >> using LPSW instead...
> > If you're unlucky enough to hit the edge case, you will lose the PER 
bit
> > by doing so.
> 
> Actually, that's a good point. If we get interrupted after the STOSM and 

> before the LPSW, and the system decides to turn on/off PER bit, etc. on 
> the very next dispatch, we would have the wrong value until it was 
> corrected on the next dispatch.
> 
>       STOSM PSWMASK,0               Retain the current system mask
>       MVZ   PSWBYTE1,TSOPKF         Set Key
>       OI    PSWBYTE1,PSWPROB        Problem state
>       LPSW  PSW                     Switch to TCB key and prob state
> 
> Not sure if we would ever care about such a thing, but I moved the STOSM 

> so the routine now looks as follows. That should reduce the likelihood 
> of that happening to even nearer zero than it already was...
> 
>       MVZ   PSWBYTE1,TSOPKF         Set Key
>       OI    PSWBYTE1,PSWPROB        Problem state
>       STOSM PSWMASK,0               Retain the current system mask
>       LPSW  PSW                     Switch to TCB key and prob state

 It won't get corrected on the next dispatch, and will continue to have
the wrong value.  z/OS does not recompute the setting of the PSW PER
bit on every dispatch. 

  If that caused someone's SLIP trap to not match, they would likely
care a lot about that, and open a PMR.  And if we were very lucky
and a dump was taken very shortly after the SLIP got enabled,
we might see in system trace where the EXT  EMS that SLIP used to 
enable for PER has an external old PSW that points
to the LPSW in your code, and know that the problem was due to your
code defect.  More likely, we would not be so lucky, and that PMR
would go unsolved, after spending a lot of time looking for a 
nonexistent bug in SLIP.

  Your code must be changed from STOSM PSWMASK,0 to 
STNSM PSWMASK,X'FC' .  And that means that PSW and your
code need to be in fixed or DREF storage.

Jim Mulder z/OS Diagnosis, Design, Development, Test  IBM Corp. 
Poughkeepsie NY



----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to