https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127356
--- Comment #2 from GCC Commits <cvs-commit at gcc dot gnu.org> --- The trunk branch has been updated by Andrea Pinski <[email protected]>: https://gcc.gnu.org/g:edb9556068ae47e1a3cbad69249424ab7cce5b7c commit r17-4378-gedb9556068ae47e1a3cbad69249424ab7cce5b7c Author: Andrea Pinski <[email protected]> Date: Fri Sep 11 17:58:42 2026 -0700 match: `minmax (a - c, b) + c -> minmax (a, b + c)` in some cases [PR127103] This adds a pattern to optimize `minmax (a - c, b) + c` into `minmax (a, b + c)` when `b + c` is known not to overflow. This is a simplified version from previous one where we only handle the `a - c` case rather than the generic This shows up in std::vector::push_back : ``` # RANGE [irange] const size_type [0, 2147483647] MASK 0x7fffffff VALUE 0x0 __room_33 = 2147483647 - _32; ... # RANGE [irange] long unsigned int [1, 2147483646] MASK 0x7fffffff VALUE 0x0 _35 = MIN_EXPR <__room_33, _34>; # RANGE [irange] long unsigned int [1, 4294967292] _36 = _32 + _35; ``` _32 has a range of `[0, 2147483647]`, _34 has a range of `[1, 2147483646]`. This is transformed into: _t = _32 + _34; _36 = MIN_EXPR<_t, 2147483647> This then allows to remove a condition that was never could be executed too. This is currently limited to unsigned types as overflow_free_p does not tell me if we are going to introduce new undefined overflow; it will return true for all TYPE_OVERFLOW_UNDEFINED types. I filed PR 127356 to add a modified version of overflow_free_p. But since the types in push_back is unsigned this is enough for now. When that goes in; I will combine r17-4210-g9d734bb468e714 with this pattern and remove the special case of 0. Changes since v1: * v2: Only handle the case where we have `a - c` as the operand to minmax. Bootstrapped and tested on x86_64-linux-gnu. PR tree-optimization/127103 gcc/ChangeLog: * match.pd (`minmax (a - c, b) + c`): New pattern. gcc/testsuite/ChangeLog: * gcc.dg/tree-ssa/minmax-plus-1.c: New test. Signed-off-by: Andrea Pinski <[email protected]>
