On Tue, 12 Aug 2025 14:32:43 -0500 Jon Perryman <[email protected]> wrote:

:>On Tue, 12 Aug 2025 08:56:15 +0300, Binyamin Dissen 
<[email protected]> wrote:

:>>On Mon, 11 Aug 2025 18:32:38 -0500 Jon Perryman <[email protected]> wrote:
:>>:>This is how to crash a system 

:>>Services document what registers are required. 
:>>For example IEANTCR has documented 
:>>ETCON documents "If the caller has issued

:>It sounds like you don't deal with abends caused by 3rd party storage 
overlays. Automation deals with every product and as such must debug many 
problems not caused by automation. 

Your assumption is wrong.

:>Most if not all professional products have an infrastructure that 
automatically guarantees a valid R13. Most services (e.g. IEANTCR & ETCON) will 
have a valid R13 despite it not being necessary. There is only a small handful 
of programs where R13 must be a consideration and more important when running 
KEY0.

Only if documented as such. A high performance path that can piggyback on
storage available but not the 18/36 words should not do the STORAGE call
simply to look nice.

:>The OP did a storage obtain for a workarea that did not include a new 
savearea. I would like to repeat, take great care on the very rare occasion 
that R13 is known to be invalid. 

Granted that if he already had to go to the effort to GETMAIN he may ass well
have getmained a savearea (depending on the number of concurrent threads -
even 72/144 bytes add up, which is why ATTACH(X) has the SVAREA parameter). 

But it is possible to have tricky code that uses 12 out of the 24 bytes that
SETFRR gives, which would appear to be adequate for the call if not for the
R13 requirement. A lot depends on how high performance your code needs to be.

--
Binyamin Dissen <[email protected]>
http://www.dissensoftware.com

Director, Dissen Software, Bar & Grill - Israel

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

Reply via email to