[Bug tree-optimization/19484] [4.0 Regression] function pointer propagation fails for noreturn functions

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-01-17 16:24 -------
Yes you are right in that we don't re-update the CFG at all when we proagate the function call.
  # BLOCK 1
  # PRED: 0 [1.0%]  (true,exec)
<L0>:;
  #   VUSE <nD.1112_3>;
  foo ();
  # SUCC: 2 [100.0%]  (fallthru,exec)

That should not be a SUCC of 2 but of EXIT.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19484