On Jan 1, 2013, at 1:07 PM, "Jon Perryman" <[email protected]> wrote:
> Below, you say that ALET's for other address spaces is not supported. Does > this > mean if I have ALESERV ADD in a non-pc program that switches to AR mode and > setting the AR to this ALET is not supported? Is it somehow now a requirement > that we use the special ALET's for PASN, SASN and HASN? Whatever I said was obviously not sufficiently clear. A suitably authorized program CAN get an ALET for another address space, but that can only occur if the target address space is non-swappable - usually implying that it is also some sort of server address space. So basically you would be making the server address space's private area addressable from some other address space without its overt cooperation. Bad idea. Outside of some BCP components it is hard to come up with a scenario where that makes sense - hence the restrictions and limitations. The CHKEAX=NO option on the macro is a dodge to get around the normal rules. I suspect its original intended use was internal and it "escaped". Nevertheless, if you lie to it and you don't have EAX authority to that space (or AX=1) when you try to USE the ALET you're going to be disappointed. See pops for all the gory details. Aside from all that, the usual thing people want to accomplish by getting an ALET for another address space is to go rummaging around inside that other address space. I can't count the number of times that certain monitoring product guys (you know who you are ;-) have asked me to provide that capability. There's a good reason they never got their wish. With respect to the "special" ALETs 1 and 2 (SASN and HASN respectively) you can use 2 any time you like (ignoring corner cases like holding certain locks) but 1 is tricky. The reason is that a PC instruction always changes the SASN and, without some fiddling around, the guy on the right hand side of the PC can't tell what "1" meant to the PC caller. Even worse, in some cases that SASN is no longer addressable at all except with considerable skulduggery. And since many system services are ultimately reached via PC calls, the BCP would have to indulge in said skulduggery to make sense of it. Now perhaps it DOES, in which case "my bad", but ISTR that ALET 1 doesn't really work for parameters passed to BCP services. CC
