Potential issue with stack splat in exec16 from *RUN <addr>

[jackokring]

N.B. Is there a stack splat in misc.asm within _exec16 by not setting sp
to some value before call.is? Not for $B0000 loading, but for say
$A0000 loading, as it might need a temp save, set sp, fix splat,
do stuff, put stack back, ret.lis. It would likely be with some recusion
and an almost at the limit 16 bit binary, assuming a high memory stack.
Check splat in mos.c function mos_cmdRUN before exec16?

I mean a $B0000 is likely fine, but a potential future issue. Not really a bug yet.

[jackokring]

#9 FIX Perhaps a ld.sis sp, 0 or ld.sis sp, $8000 before the CALL.IS (DE): RET
based on the A segment being $0B to prevent global (SPL) clobber by the two byte (SPS) write of call.is?

This maximizes expected global size with confirmed max 32kB *plugin including stack, while making a logical aligned 64 kB segment except in $B0000. cp $0B: jr z, ....

Yes I assume the IX to SP restore is a stack frame thing.

[breakintoprogram]

I'll double check - the stacks * should * be set up with sensible defaults but that doesn't necessarily mean I've covered all the bases.