It's worse than that. Programs that do an ATTACH can test dynamically, but not 
programs that do a CALL to a routine in the same program object. The call needs 
to know whether to use XPLINK.

-- 
Shmuel (Seymour J.) Metz
http://mason.gmu.edu/~smetz3
עַם יִשְׂרָאֵל חַי
נֵ֣צַח יִשְׂרָאֵ֔ל לֹ֥א יְשַׁקֵּ֖ר




________________________________________
From: IBM Mainframe Discussion List <[email protected]> on behalf of 
Paul Gilmartin <[email protected]>
Sent: Saturday, August 9, 2025 9:41 PM
To: [email protected] <[email protected]>
Subject: Re: RMODE64 services


External Message: Use Caution


On Thu, 7 Aug 2025 08:35:07 -0400, Joseph Reichman wrote:

>If you run a windows app using visual studio debugger you will see all code
>run in 64 bits
>
Good example of courageous design.  Look forward,
not back.

>On Thu, Aug 7, 2025 at 8:30 AM Tom Harper wrote:
>
>> I have a considerable amount of experience with this issue on z/OS (not
>> other z operating systems). Since z/OS introduced 64-bit support nearly
>> twenty-five years ago, it has gradually and ever so slowly been filling
>> in the gaps.
>>

Is IBM in a quandary?  For over a half century it has been
true that (almost) IBM languages used a uniform call
interface, that generated by the CALL macro and the initiator
used the same interface so any program used as a job b step
could equally be called as a subroutine.  Suddenly there are
two interfaces which must be supported: PLIST8=YES for
support of data in any location, and PLIST8=NO for support
of programs using the traditional interface. The choice
can not be made statically as there is no way to discern
ANODE of a subroutine at translation time.  Dynamically,
the decision can be made during execution by examining
the AMODE of the called program object.  The initiator
*could* do that.  The BPX1-BPX4- replication is a desperate
accommodation.


>> And one of the last gaps to fill in is RMODE(64).
>>
>> As others have pointed out, the documentation about RMODE(64)  is
>> somewhat lacking.
>>
>> There are a number of what I think are curious issues, like 64-bit latch
>> support, which allows 64-bit parameters, and AMODE(64), but no RMODE(64)
>> support. Like most services invoked via an instruction other than an SVC
>> or a PC, support has not been added for returning to an RMODE(64) bit
>> caller. I believe most of these could be changed with little effort and
>> testing, but IBM has other priorities like AI and containers. And most
>> services that do invoke via a PC or SVC do not document RMODE(64)
>> support which means it is not supported.
>>
>> Others are more subtle, like not allowing SPLIT load modules to be
>> loaded into ECSA and above the bar storage.  Another is lack of 64-bit
>> PC support using ETDEF. Recovery is another area where RMODE(64) bit
>> support is lacking. IPCS does support 64-bit storage access, however
>> awkwardly, but none of the IPCS services may be invoked in AMODE(64) nor
>> RMODE(64).
>>
>> To address these issues, I have submitted a significant number of IBM
>> "Ideas", and after getting some additional questions on them from
>> clueless IBM responders, have finally had most of them with a status
>> changed to "Future Consideration", the queen of vague terms. Still, I'm
>> an optimistic person and have hope that these will be changed in future
>> releases.
>>
>> At the last SHARE meeting at the closed IBM session, I did vigorously
>> point out that their RMODE(64) support was sorely lacking. I could see
>> them taking furious notes, but I'm too old to be intimidated. Hopefully
>> the notes were to take back to their development teams.  I pointed out
>> that many of these could likely be resolved by changing a single
>> instruction.
>>
>> At any rate, in the fullness of time, perhaps we will see some attention
>> paid to this issue by IBM.
>>
>> So, if this issue interests you, go to the IBM "Ideas" portal and vote
>> if you think these ideas should be considered. I personally believe that
>> RMODE(64) support will enhance the value of z/OS going forward. The
>> hardware people have done their job well. Let's hope the software side
>> can finish the task at hand.

--
gil

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN




----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to