[MiNT] Current beta problems - solved

[Joerg_Westheide@su.maus.de (Jörg Westheide)]

Hi Michael!

MS>I would not take too much care about that -saving  A2 may be OK, but
MS>saving all registers may be difficult or impossible,
a2 is not destroyed, and I can make a1 unmodifed too

MS>However, when introducing that new handler, I produced a debug version
MS>which clears those spilled registers to known values before returning
MS>from the trap
That's no problem

MS>Setting A0-A2 to some address which is guaranteed to cause a bus
MS>error (even when used with a moderate offset), will catch these
MS>programs quite fast and protect the kernel.
Sugestions for such an address?

MS>BTW: unintended passing back intermediate kernel variables in those
MS>registers can be a security problem.
Right, that shouldn't be done.
In this case it's the address of a kernel internal function depending on the
function you called (i.e. if you call Cconin() you will get the address of
c_conin()). I will change that

Bye

                Joerg