Got it: https://www.mail-archive.com/[email protected]/msg131045.html 

2011. As Kermit (the frog, not the protocol) says "Time's fun when you're 
having flies."

The point of TCBCMP versus STCBCMP is that TCBCMP may have an "intermediate" 
termination result, such as if

- user code calls (PC, SVC, whatever) some service routine with an invalid 
address
- service routine gets a S0C4
- service routine ABENDs the job or returns some other error with an ABEND code 
or RC that is the service routine's (presumably documented) error for "invalid 
address from user" (not S0C4).

At least at some points in time TCBCMP might contain the S0C4 while STCBCMP 
will contain the actual completion code.

I suspect I am good with using the subtask's ECB. 

Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf 
Of Tony Harminc
Sent: Saturday, December 12, 2015 5:26 PM
To: [email protected]
Subject: Re: How distinguish User ABEND from return code?

On 12 December 2015 at 10:24, Charles Mills <[email protected]> wrote:
> Is there a definitive "this task ABENDed" bit in the TCB? I see lots 
> of bits with "ABEND" in their description but I don't see one that 
> seems to definitively indicate simply "this task ABENDed."

I'm pretty sure you want TCBEndingAbnormally. This is defined, at least, when 
the TCB *has* ended abnormally, and the comments suggest that STCBCMP is then 
where you should be looking for the abend code, since TCBCMP may have been 
changed by that point.

There are detailed comments for several flags in the byte that holds 
TCBEndingAbnormally. IIRC, Peter Relson commented on this here a year or two 
ago.

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

Reply via email to