[MI400] alchss.

["Steve Richter" <srichter@xxxxxxxxxxxxx>]

This is a multi-part message in MIME format.
--
[ Picked text/plain from multipart/alternative ]
an fyi.

While running some code that uses ALCHSS a coding error passed an incorrect 
value as the heap identifier to the instruction. It took a while to get this 
bug under control, so ALCHSS was being run a number of times with the garbage 
heap id value ( a bin(4) value ). Most of the time an MCH exception resulted. 
But one time the job dumped and ended without the MCH exception. A 2nd time, 
the job dumped and then froze up. EndJob would not kill it, EndJobAbn would not 
end it either.

The job stayed in an *END status for a number of days, retaining some LOCKSL 
locks as a side affect. An ipl finally got rid of the job, but even the 
pwrdwnsys *immed took some extra time, waiting for 10 to 15 minutes at one 
point, me sweating bullets thinking of what to tell the dp manager,  as I guess 
the system waited for the job to end.

I dont have much specific info on this, but if you want to recreate the error, 
maybe just call ALCHSS with every possible bin(4) heap id value :).

The os release is V4R4.

Steve Richter




--