https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113716
--- Comment #4 from Andrew Macleod <amacleod at redhat dot com> --- (In reply to Jakub Jelinek from comment #3) > Maybe ranger could figure out that at least one of the multiplication > operands is zero in this case, because the second one is non-zero only if > the first one is zero? _2 = b_6(D) == 0; _3 = (int) _2; _4 = _3 * b_6(D); well, its kind of sort of predicate based. I mean, the information is calculable if one thought it was worth it. ie at the multiply, we know _3 is [0, 1] and we know b_6 is VARYING because there are only 2 ranges in _3, you could calculate using GORI "if _3 is [0,0], what is b_6?" (~[0, 0]) and "if _3 is [1,1], whats b_6?" [0,0]. do the 2 multiplies and union the results.. you'd get [0,0] We might need a slight API tweak to be able to ask that. we have fold_stmt routines for the folding, but I dont think I have added any "what if" hooks to GORI yet.
