limit checks

[Stephen Weeks]

> What was specific to vliw that caused this limit-check bug?  From
> your description it seems the way limit-check amounts were computed
> would have undercounted on any control-flow graph where there are
> multiple paths to the same entry point.  Was it just serendipity that
> this didn't show up earlier?

Yep.  Nothing specific to vliw.

> Also, I'm curious about the relationship between limitchecks of size 0, 
> and stack-slot allocation.  What is the relationship between limit
> checks (esp. of size 0), and liveness?

I don't understand the question.