On 21 August 2018 at 08:23, Peter Relson <[email protected]> wrote:
> 1) Is BRANCH=XM supported?
> Ans: Since it is not in the books, no. That is the stock answer for anything.

Yeah, I know... Though there has been the occasional thing that has
lacked only the doc, or is documented only in the macro, so worth
asking.

> If you need this function, then please submit an RFE.

It's a would-be-nice. I can program around it. OTOH if it's been there
for 15 years, is it really likely to go away?

> You can probably get away with checking the JSCBAUTH bit.
> There is actually a complex definition of the APF state, and that's what
> MODESET checks.
> Perhaps someone knows the history behind it. It appears to have allowed
> for a lot of future extension that never happened.

I do remember some decades ago speculating on the fact that while the
linkage editor AC(n) option supports only 0 and 1, the result does go
into a one-byte field in the LM/PO. Hints of Multics and multi-level
privileges...?

> 4) Design-wise, is there a reason that APF authorization of the caller
> is not a criterion that can be applied when the PC routine is set up
> with ETDEF?
> Ans: Yes (for the most part). APF authorization is not a hardware state.
> PC's (and things identified by ETDEF, in general) have no way of looking
> at software-defined structures.

You make a good point. I was thinking that SVCs can be set up in
advance to accept or reject non-APF callers, but that of course isn't
done by the hardware. And there is no "PC FLIH" analogous to the SVC
one. I suppose JSCBAUTH could be copied to some known-to-hardware
location, as various formerly software-only things have been over the
years. But not in my lifetime, I suspect.

Thanks, Peter.

Tony H.

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

Reply via email to