Have your memory dumper dump the RB chain of the task.

Determine if there is a RB lower than the RB that issued the WAIT.

On Tue, 5 Nov 2024 15:42:15 +0100 Massimo Biancucci
<[email protected]> wrote:

:>Hi everybody,

:>DB2 can provide information to a caller via IFI interface.
:>During the startup of the process, the caller does a START TRACE providing
:>the ECB to post when the internal DB2 buffer has been filled up to a
:>certain extent.

:>The main program is PLI and it uses a routine written in assembler that
:>issues the WAIT macro.
:>Main program shares three ECBs:
:>1. For an attached routine that waits for xx seconds
:>2. For an attached routine that waits for a wallclock (this case it's not
:>used)
:>3. For DB2 post (it's the ECB communicated to DB2)

:>Routine uses QEDIT to get console command too.
:>Routine attaches the two timer programs.

:>Finally, the routine has a four fullwords ECBs list, the three mentioned
:>before and the fourth that is ECB from QEDIT.
:>The last address has been HighBit to on to say it's the last one.

:>When I invoke the WAIT, it awakes both for the first ECB or the fourth (end
:>of wait or command received from console). The second is not used in this
:>case.
:>It does not awake for DB2 post.
:>I can see the first byte of the DB2 ECB set to x'40' when the routine gives
:>control to the main so it seems DB2 posted me correctly. The same is live
:>with a memory dumper, I see the ECB to x'40something' but the WAIT does not
:>terminate until the timer has expired or command has been sent.
:>If it was relevant, the assembler routine runs in the same TCB of the PLI
:>main.

--
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