>I'm not really sure why IBM service would help someone debug an
application error, aside from on a for-fee basis, unless there is reason
to believe there could be a system problem.


I agree. But when all else fails, we might even try going down that route.



>23E reason 0: you have an error in the input to detach.  The system
program-checked referencing the data that you provided. Specifically, it
appears, loading the word pointed to by register 1 (where that word being
loaded is expected to contain the address of the TCB to detach).

>It is certainly true that if you issue DETACH after the freemain of the
TCB things will go poorly, but probably not 23E-00.

>If your ATTACH was not done with ECB or ETXR, it is wrong to issue DETACH
after the termination of the daughter task.




Yep, I understand we've got an error in the DETACH. The problem is we have no 
clue why. I suspect a storage overlay somewhere.


This is complex, multitasking code that has run for years without problem. The 
developer tried to change a tiny bit of code, and the errors startet do appear. 
The problem is that the code change has been backed out, but the problem still 
occurs in development. The only difference there should be between development 
and prod is that the development code has been recompiled/reassembled (it is 
maintained under ChangeMan). So recompilation/reassembly seems to lead to the 
error (I suspect it may be some AMODE24/31 issue). If the old code is activated 
(baseline), instead of what we think is the identical source, but recompiled, 
then the error does not appear.


I'm not in development, and I do not know the code. I cannot play with it on my 
own, I have to "trust" the developer (which I do).


Main reason I posted is I was hoping to get comments that would give me new 
ideas what to look for. And to understand some of the system trace entries I 
currently don't understand. Such as the TCB in Read/Write nucleus.




--
Peter Hunkeler



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

Reply via email to