[Bug tree-optimization/81162] [8 Regression] UBSAN switch triggers incorrect optimization in SLSR

["wschmidt at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81162

--- Comment #7 from Bill Schmidt <wschmidt at gcc dot gnu.org> ---
This case comes up when we're going to replace a NEGATE_EXPR with a PLUS_EXPR
or MINUS_EXPR.  This is another case of an unprofitable replacement that should
be avoided anyway.  So I think the following fix is what we need:

Index: gcc/gimple-ssa-strength-reduction.c
===================================================================
--- gcc/gimple-ssa-strength-reduction.c (revision 249846)
+++ gcc/gimple-ssa-strength-reduction.c (working copy)
@@ -2082,13 +2082,14 @@ replace_mult_candidate (slsr_cand_t c, tree basis_
      types but allows for safe negation without twisted logic.  */
   if (wi::fits_shwi_p (bump)
       && bump.to_shwi () != HOST_WIDE_INT_MIN
-      /* It is not useful to replace casts, copies, or adds of
+      /* It is not useful to replace casts, copies, negates, or adds of
         an SSA name and a constant.  */
       && cand_code != SSA_NAME
       && !CONVERT_EXPR_CODE_P (cand_code)
       && cand_code != PLUS_EXPR
       && cand_code != POINTER_PLUS_EXPR
-      && cand_code != MINUS_EXPR)
+      && cand_code != MINUS_EXPR
+      && cand_code != NEGATE_EXPR)
     {
       enum tree_code code = PLUS_EXPR;
       tree bump_tree;

This fixes the particular test.  I have to head out on vacation now, but I will
regstrap this and submit it when I get back if there are no objections.