On 10/04/2016 08:13 AM, Peter Relson wrote:
And what the ECB contains is puzzling to me. Why would my WAIT
get posted, if the task in question had not finished?
It would not. It could not, unless you posted it yourself.
If you're saying that you looked at the ECB after the POST and at that
time it contains an RB address then either the ECB was not POSTed (and you
did not wake up because of the POST but because of something else) or
something else has subsequently waited on it.
Well, IDCAMS, the Attached program, ran to CC=0 according to the
SYSPRINT DD.
The ATTACHX is done, then in a bit, the WAIT on the named ECB is
done, immediately followed by a CALLDISP (NON-APF) to ensure that
the sub task has "terminated" so that the DETACH doesn't take an
ABEND (which I've had that happen as well).
I don't believe that. The CALLDISP cannot be useful let alone necessary.
DETACH will not abend at any point after the POST has occurred unless, I
suppose, you are issuing it a second time.
OK, but ATTACHX gave back R15=0.
And I do not use the POST macro. So, I'm not doing the POST.
I have taken the CALLDISP out, and still do not get posted with
CC=0. The last test showed S400.
BTW - I could see where CALLDISP gets done because I'm using the
CPU trace to see that I hit my SVC50s correctly and I get to see
all the other trace entries, including when dispatch kicked me
back off at the instruction following CALLDISP.
Back to this test I ran just before responding to this (which
does not have CALLDISP in it). There is no summary dump info in
the JES log for the IDCAMS TCB (again, SYSPRINT shows IDCAMS got
CC=0).
So, it didn't ABEND. But yet, the ECB shows a S400.
While this might be a non sequitur, I've already discovered a bug
in "SPIE" (OA51375). And I battled with that one for a week to
make sure it wasn't me doing something stupid.
The POST is done out of an internal call to DETACH after termination has
completed. When ECB (or ETXR) is used on the ATTACH(X), free-ing of the
TCB itself and de-chaining it from the TCB queues is not done until your
DETACH.
It may or may not need to be APF authorized.
APF authorization is not relevant to the discussion. That may affect the
result of the target task (success or failure) and thus the completion
code of the POST, but not when the ECB is posted.
I dunno, it was in answer to someone asking if IDCAMS was APF
authorized. I just used one posting to answer a few questions.
Peter Relson
z/OS Core Technology Design
<SNIPPAGE>
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
