Using C as a back end

[Stephen Weeks]

> In article <00-10-195@comp.compilers> you write:
> >I disagree with the first two points above.  Since SML has garbage
> >collection, we had to face the problem of implementing it... I wrote a simple
> >two-space copying gc... We went
> >through several iterations of the interface between the generated C
> >and the runtime (all basically different ways of telling the GC the
> >root set) and eventually found one we liked.
> 
> Have you published the details?  I hope you will.

It's doubtful we will at this point, since we've moved on to the native backend,
which gives much better performance.

> >As to exceptions, SML has them, and I don't think our implementation
> >is slow.  Raising an exception to an unknown handler involves a store,
> >an add, a dereference, a switch and possibly a trampoline, if the
> >destination is in another C procedure.  
> 
> Aha, I had forgotten about trampolines.  But it might be nice to know
> the details here.

Again, we've moved to the native backend, but in this case that didn't change
much, other than to replace the switch and the trampoline with a jump.  The
method is to store in a global (actually per thread) variable the offset of the
current exception handler (a code pointer) relative to the bottom of the stack.
Then, raising an exception involves
	store 		into a global of the value to raise
	add 		of the stack bottom to the global
	dereference 	of the code pointer
	jump 		to the code