[gclist] memory protections and system calls

[David Chase]

> > .. stop in the middle of a system call and let a user process fix a problem,

> I agree with Paul. It seems from David Chase's note about the register
> spilling code in SunOS/Solaris that this mechanism is _exactly_ what
> was intended there (and would have been provided were it not for a
> minor bug).

Not quite so.  All the OS calls, curse their peanut brains, indicate 
access violations with bad return codes instead of signals.  It might 
well be slower to do it that way (nowadays) but that is the interface. 
The code I refer to was only intended to support an especially interesting 
and somewhat hairy user-controlled-stack register window spill/fill 
case.

David Chase