Notes on when I saved CMSDOS, CMSBAM, CMSVSAM, and CMSAMS on VM/XA 2.1.
(These are all part of CMS, so do a SETUP CMS to see this stuff.)
 
  Refer to the VM/XA Installation and Service manual, SC23-0364.
Pretty much, I just followed their directions, but
 
1)  I didn't have the SAMPNSS EXEC.  The book described where it would
    create the segments, but for future reference, here's what they are.
    /* */
    'CP DEFSEG CMSDOS  B00-B0F SR            SPACE DOSBAM'
    'CP DEFSEG CMSBAM  B10-B3F SR            SPACE DOSBAM'
    'CP DEFSEG CMSAMS  B40-B9F SR A00-A2F EW SPACE DOSBAM'
    'CP DEFSEG CMSVSAM BA0-BFF SR A30-AFF EW SPACE DOSBAM'
 
>>>
>>>  See item 5) below !!!!  Ruth's version of VSAM didn't work on XA.
>>>
 
2)  I didn't have the VSAM program product tape, program product
    number 5746-AM2.  I called Ruth Chen and stole her copy of it and
    put it on PRODUCT's 321 disk.  Ruth had it on VSAM's 291 disk,
    with additional (replacement?) service on VSAM's 292 disk.
    It was 378 files which took up 92% of 15 3380 cylinders.
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
*-*-*    Brian Chan sent me a note on 7-27-92 asking about    *-*-*
*-*-*  PRODUCT's 321 disk.  Since this was a stolen,          *-*-*
*-*-*  bastardized version of VSAM, I told him to delete it.  *-*-*
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
 
3)  The VSAMGEN EXEC was trying to load stuff at the wrong address.
    It prompted for the beginning address, but then added a hard-coded
    x'60000 to that address for the shared code.  I looked at Ruth's
    CMSAMS & CMSVSAM SNT definitions and they matched her exec,
    that is,
 
        Page #   Use
        -------  -----------------
        B00-B5F  CMSAMS-Shared            Notice that the difference
        B60-B8F        -Nonshared         between the shared and
                                          non-shared code is indeed
    and B90-BEF  CMSVSAM-Shared           x'60000.  However, the XA
        BF0-BFF         -Nonshared        manual has both non-shared
                                          pieces in the A-meg.
    I changed the VSAMGEN EXEC to load the shared code in the correct
    places by subtracting x'140000 and x'170000 for CMSAMS and CMSVSAM
    respectively (d2x(x2d(B40000) - x2d(A00000)) = 140000 for CMSAMS
              and d2x(x2d(BA0000) - x2d(A30000)) = 170000 for CMSVSAM).
 
4)  The prompts in the book didn't quite match what the exec actually
    prompted me for.  It was missing some prompts.  I don't know why.
 
5)  Ruth's version of VSAM is the VM/SP version, which doesn't work on
    VM/XA.  One difference is x'DC' past the lowest address of the
    CMSAMS segment (the address returned by the diagnose 64 to load)
    is a pointer to the shared portion of CMSAMS.  If the shared
    piece is before the non-shared, this is 0.  In our case, the SR
    portion is after the EW portion, and this should be x'00B40000'.
    With Ruth's version, this was some kind of garbage, eventually
    causing a program check.  I got Bruce Merrill (BRUCEM at RHQVM16)
    from the DTN Network Systems Support group to send me his VSAM,
    which I installed and things worked fine after that.  Now
    location A000DC is indeed 00B40000.