Re: set spcp to pgm assoc space

From: Steve Richter <srichter@AutoCoder.com>

Maybe I was a bit hasty. The situation is like this:
I tried the following program "TEST" (I'm running at level 50).

DCL DD RESOLVE CHAR(34);
...
DCL SYSPTR .SYSPGM;
DCL SPCPTR .SPCPGM;
DCL DD PGAS CHAR(1024) BAS(.SPCPGM);

    CPYBLA       RESOLVE-TYPE, X'0201';
    CPYBLAP     RESOLVE-NAME, "TEST", " ";
    RSLVSP      .SYSPGM, RESOLVE, *, *;
    SETSPPFP .SPCPGM, .SYSPGM;
    BRK "1";
    RTX               *;

Then
CRTMIPGM TEST
STRDBG TEST
ADDBKP STMT(1) PGMVAR(('PGAS' ())) OUTFMT(*HEX)
CALL TEST

and got:
CPF1989  Domain violation occurred for variable PGAS.

Then
CALL MAKESYS TEST       (don't try this; it is my homegrown program to make
system state)

then
CALL TEST

This time no error message.

This was my rationale for my statement that your program needed
to be system state. But I was not happy with this. In fact, the actual
domains for both the program and its associated state were both
USER, so what was going on?

Next step, change the program as follows:

DCL DD PGAS CHAR(1024) BAS(.SPCPGM);
DCL DD DATA CHAR(1024);

    CPYBLA       RESOLVE-TYPE, X'0201';
    CPYBLAP     RESOLVE-NAME, "TEST", " ";
    RSLVSP      .SYSPGM, RESOLVE, *, *;
    SETSPPFP .SPCPGM, .SYSPGM;
    CPYBLA       DATA, PGAS;
    BRK "1";
    RTX               *;

compile, start the debugger, then
ADDBKP STMT(1) PGMVAR(('DATA' ())) OUTFMT(*HEX)
CALL TEST

NO ERROR ! The debugger shows what is in DATA.
This is somewhat puzzling (a bug, I think, although I'm
waiting to hear the WAD argument).

So, it seems that accessing PGAS directly leads to an error, but
copying PGAS into DATA, and then accessing DATA is OK.

In any case the SETSPPFP instruction is not the problem, so
you can get a pointer to PGAS even at level 50.