There is a timeout mechanism in the CMS side outside the IUCV part. I
believe it is currently set at 15 seconds. That should be plenty of time
for the TPF machine to field a connection request, if it gets the
interrupt, that is.

Just got TPF trace info:

 > #CP TRACE EXT 4000 RUN and #CP TRACE IUCV RUN on the TPF machine is
appropriate. 

I've already been running with the EXT 4000 interrupt trace - no hits.
The following hits on the IUCV trace are seen during restart, but
nothing when TPF is sitting there idle or during an attempt to connect
from CMS:

   -> 0000000000068808"  IUCV  B2F00000    0000000000000000     CC 0
QUERY
   -> 000000002549A0EC'  IUCV  B2F02000    02CF5A28    CC 0  DCLBFR



Regards, 
Richard Schuh 


-----Original Message-----
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Alan Altmark
Sent: Friday, July 20, 2007 1:18 PM
To: [email protected]
Subject: Re: IUCV Interruptions

On Friday, 07/20/2007 at 04:04 EDT, "Schuh, Richard" <[EMAIL PROTECTED]> 
wrote:
> Worse than you imagine, a time-out.

Timeouts are not part of IUCV so that doesn't have a lot of meaning for 
me.  It would make sense to get CC=0 and then wait for the Connection 
Complete interrupt, but did the CMS program actually get CC=0 on the
IUCV 
CONNECT?

Have you witnessed the PSW and CR0 in the TPF machine?  Have you
examined 
the IUCV CONNECT IPARML to be absolutely certain that it has the correct

values?

BE the CSI.  Gather hard evidence of the failure.

Alan Altmark
z/VM Development
IBM Endicott

Reply via email to