[Bug tree-optimization/99079] New: Maybe a wrong code since r6-1462-g4ab1e111ef0669bb

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

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99079

            Bug ID: 99079
           Summary: Maybe a wrong code since r6-1462-g4ab1e111ef0669bb
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: marxin at gcc dot gnu.org
                CC: rguenth at gcc dot gnu.org
  Target Milestone: ---

Since the revision, we optimize the following in a different way:

$ cat x.cpp
unsigned long var = 31;
unsigned long arr;

static void test()
{
  unsigned long s = 1 << var;
  arr = 4897637220ULL % s;
}

int main()
{
  test();
  __builtin_printf ("arr: %ld\n", arr);
  if (arr != 4897637220ULL)
    __builtin_abort();
  return 0;
}

$ g++ x.cpp && ./a.out 
arr: 4897637220
$ g++ x.cpp && ./a.out 
arr: 4897637220

Clang is fine with the test-case:
$ clang++ x.cpp -O3 && ./a.out 
arr: 4897637220

and ICC as well.