W dniu 11.12.2025 o 17:57, Jon Perryman pisze:
On Thu, 11 Dec 2025 11:23:57 +0100, Radoslaw Skorupka <[email protected]>
wrote:
Same FMID, same set of PTFs, but results are different - load module
created in V1 consist of few MODs, while second case causes to not link
several modules and resulting LMOD is smaller.
Many LMODs are affected.
Have you tested both versions of the install? Both should be functionally the
same.
To investigate this difference, use AMBLIST LISTLOAD and compare the
differences in the load module. Most likely some csects will occur multiple
times or csects have been orphaned (missing from V2). As Kurt says, the most
likely cause would be CSECTS that SMP/e does not know about.
If multiple versions of a csect are occurring, then I'm guessing that binder
resolves addresses using the last occurrence which should be the most recent
version.
Another possibility is the CSECT alignment which I believe is double word.
Different module sequence could cause this difference.
V1 should have every module in V2 with a matching length.
Unfortunately in the LMODs are *not* the same. Different size: 3FF0 vs
DB0. Different set of modules inside - quite obvious.
Orphaned or repeated CSECTs? Well, how it could happen? In both cases it
is the same, original (not modified) IBM material.
--
Radoslaw Skorupka
Lodz, Poland
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
