> 
> My take on SEGV and BUS has been as follows:

I believe that referencing an invalid (not in your address space) or
not proper access (such as writing a read only page) is supposed to generate
SIGSEGV still.

For most machines I have worked on (Sun, HP, RS/6000, 68k),
SIGBUS is delivered for things
like misaligned memory accesses and such and not much else.

I think SIGBUS for dereferencing a NULL pointer is wrong, and in fact I 
never have seen such behavior in a commercial Unix I've worked on.
(Not counting funky SysV x86 legacies that give you a null string ;-) sigh).
--
Mark