I'd still like Wayne to post the untruncated CSV031I message...
(or re-post it if he had done so but I missed amidst the many appends).
<snip>
Check out the reason code of the S106-xx message. It could be that the
module it's trying to load is in LLA, but LLA REFRESH was not done after
replacing the module.
</snip>
LLA is involved (the DDNAME of *VLF* shows that). The 106-0C is due to an
obtain of storage being unsuccessful. I can't think of a way that LLA
REFRESH could have an impact (whether done or not done).
<snip>
S106 - 0C - NOT ENOUGH STORAGE WAS AVAILABLE FOR FETCH TO DO A GETMAIN
FOR THE MODULE OR CONTROL BLOCKS. CHECK REGISTER 0:
</snip>
If you're going to post something like this, please post from the official
documentation. That documentation is "Not enough storage was available for
FETCH to get storage for the module or control blocks.". There is no
information in register 0. The specific piece of data happens to be in
register 4 (feel free to submit an RCF to ask that this be included in the
documentation). As shown in one of the posts, the value in reg 4 at time
of the abend 106-0C is x'14'.
The module appears to be small (less than x'C000') and has no relocation
needs. Since there was no relocation, the only reason for the 106-0C for
x'14' for *VLF* is that LLA's attempt to obtain storage for that amount
was not successful. For all I know, the utility did some variable-length
GETMAIN and ate up all the private storage below 16M. Take a dump of the
address space and see. And look at the system trace entry for the
completion of the conditional getmain (an SSRV entry) which would contain
things such as the size requested (in case there was some weirdness going
on between the "right size" and the "requested size").
You could always try things like stopping LLA, or not having LLA manage
the CEE.SCEERUN (perhaps named SYS1.CEE.SCEERUN) data set. That would keep
LLA from trying to obtain the storage, but then normal module fetch would
try to obtain the same amount.
I wonder if the difference between releases is that the module in question
(which sounds like it's CEEBINIT) is now eligible for LLA caching (but
doesn't work for some reason once it is cached), whereas it did not used
to be.
Peter Relson
z/OS Core Technology Design
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
