>  > 
>  > hmm.  sigbus or not and message or not.  so processes on alphas don't
>  > get sigbus'ed at all (modulo the sysctl setting)?  would it
>  > "theoretically" be possible to do the same thing for other
>  > architectures?  i'm just curious...
> 
> Well, a NetBSD/alpha process will get SIGBUS'd if it does other SIGBUS'able
> access :-)  i.e. accessing a nonexistent (vs. unmapped) memory location.

Shouldn't that be a SIGSEGV?


> 
>  > >Also note that the NetBSD kernel will panic if unaligned access is
>  > >performed by the kernel, rather than by a user process.
>  > 
>  > certainly.  to where would it trap?
> 
> Same unaligned access fault handler, but if the processor is in kernel
> mode when the fault occurs, the fault handler decides to panic, instead.
> 

Historically SIGBUS was for memory actions that could not be handled due
to bus (i.e., 'path') problems. This includes alignment, but also width.
For example, a 32 bit access in VMED16 space should generate a SIGBUS.