<snip>
If I didn't have to save/restore R15 and R1, then I could totally get away
with something very close to standard linkage
</snip>

Huh? You can precisely use standard linkage. Standard linkage starts with 
"save all". It does not require "restore all".

If a service (or protocol) does not specifically tell you to preserve 
registers 15,0,1 then you do not to do so. 

As Tom Marchant mentioned, not saving all is simply short-sighted, since 
you might need that information for diagnostics.

Using BAKR/PR to save/restore is an easy way to avoid this problem.

The other part that goes with Ed Jaffe's astonishment about problem state, 
is that the ETXR is also entered in TCB key (even if the ATTACH was issued 
in some other key).
If Ed's astonishment results from a recent event, that indicates that he 
has not previously had much need for doing supervisor state stuff in 
ETXR's. 

z/OS (AKA OS/390 AKA MVS) presumably decided not to create a dependency 
for normal users with respect to how ETXR's work. 
Is it an IRB? Yes. Must it be an IRB? Not in theory. Maybe in practice 
now, due to compatibility concerns, it must be. 
Can you get away with saving nothing and just branching back to the 
address in reg 14? Today, yes. Maybe forever, yes. But that would be a 
silly thing to code to.

Peter Relson
z/OS Core Technology Design


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

Reply via email to