[Bug tree-optimization/126459] [17 Regression] Wrong code with x != CST1 ? x + CST2 : CST3 fold due to overflow since r17-126

[jakub at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126459

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I had
--- a/gcc/match.pd      2026-10-03 21:28:13.092320517 +0200
+++ b/gcc/match.pd      2026-10-06 21:36:22.644983570 +0200
@@ -5369,8 +5369,11 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
     tree cst1 = uniform_integer_cst_p (@1);
     tree cst2 = uniform_integer_cst_p (@3);
     tree cst3 = uniform_integer_cst_p (@4);
+    wi::overflow_type ovf;
   }
-  (if (wi::to_wide (cst1) + wi::to_wide (cst2) == wi::to_wide (cst3))
+  (if (wi::add (wi::to_wide (cst1), wi::to_wide (cst2),
+               TYPE_SIGN (TREE_TYPE (cst1)), &ovf) == wi::to_wide (cst3)
+       && (!ovf || TYPE_OVERFLOW_WRAPS (TREE_TYPE (cst1))))
    @2))))

 /* Simplifications of shift and rotates.  */
for this.