Under what circumstances can the JS TCB DEB pointer (TCBDEB) be zero?
I have at least one open DCB, and previously I've seen a DEB on that
chain for each open DCB or ACB.

Put another way, how can I reliably find an open DCB/ACB for a given
DDNAME? I've previously (while the program is running) gone through
the DEB chain, DEB->DCB, DCB->TIOT entry, and match the DDNAME.
Clearly if there's no DEB chain this won't work.

This is a dump where one TCB in this multi-task STC is abending
(S878); could I/O cleanup have already zeroed TCBDEB by this point,
even though the DCB is still open and the DEB still seems to exist?

The open DCB in this case is for a subsystem dataset (JES3 SYSOUT), so
I suppose it logically doesn't really need a DEB, but I've never seen
it not there before. Maybe a JES3 thing? And yes, I know there'll be
an ACB in the middle of things in this case.  I found the DCB in
question by brute-force search, and its DEB pointer DCBDEBAD points to
a DEB-looking thing that passes several tests: it points to the TCB,
DEBECBAD points back to the DCB, and DEBDCBAD points to an ACB. But if
I hadn't been able to find the DCB on my own...

Trying another route, I went to look at the DEB table from JSCBDBTB.
But this seems to have gone OCO at some point, and certainly isn't PI.
And it doesn't look like the simple table of 4-byte pointers that it
once did. :-(  Is there anything in IPCS to format the DEB table, or
to some extent duplicate the DEBCHK service?

And yet another route, SCTSIOT/IEFDDSUM from the "6i" option in IPCS
show three allocated DDs, all of which I expect to be open. Is there a
path from the SIOT to the DCB/ACB? (In passing, the SIOT addresses
shown by IEFDDSUM are rejected when used as input to SIOTPLUS/IEFSIOTP
in the same panel. Looks as though the eyecatcher validation is
failing, but they look right in storage. Strange...)

Thanks for any (heh) pointers...

Tony H.

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

Reply via email to