Displaying 2 results from an estimated 2 matches for "leave_loop".
2016 Dec 31
2
SCCP is not always correct in presence of undef (+ proposed fix)
>
>
> >
> > You need to distinguish between not visited and visited but undef.
>
> What I'm getting at is, if you're implementing something close to the
> paper mentioned, I can't think of a case where you'd care to
> distinguish between "not visited" and "visited but undef". That is,
> instead of starting from "map each
2016 Dec 31
0
SCCP is not always correct in presence of undef (+ proposed fix)
...o `a`, and then "try to
see" if other edges are viable or not. You're allowed to incorrectly
optimistic results in the midst of the algorithm run (by design).
Otherwise we won't be able to get cases like:
loop:
%x = phi [ 0 , entry ], [ 1, loop ]
if (x != 0) br loop else br leave_loop
leave_loop;
=>
loop:
br leave_loop
leave_loop;
As far as I can tell, the <unknown> has the semantic "given the viable
set of edges so far, there is no value assigned to this slot by the
program". This has the same rewrite rules as undef as far as I can
tell.
However, if...