------- Comment #8 from pinskia at gcc dot gnu dot org 2006-04-03 17:42 ------- Just a note here, we really always want to deal with a + CST and not worry about if CST is negative or not, I had a patch to do but there was a testsuite regression because of VRP, see PR 25148. In fact currently we don't optimize g++.dg/tree-ssa/pr18178.C when supplying -fwrapv even though we should be able to, and more to the point PR 18178 was created for the -fwrapv case in the first place :).
-- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=26998
