[Bug testsuite/86461] [9 regression] test case gcc.c-torture/execute/builtins/strlen-3.c fails starting with r262522

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86461

Martin Sebor <msebor at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2018-07-10
          Component|middle-end                  |testsuite
           Assignee|unassigned at gcc dot gnu.org      |msebor at gcc dot gnu.org
     Ever confirmed|0                           |1

--- Comment #1 from Martin Sebor <msebor at gcc dot gnu.org> ---
r262522 tightens up the test to verify that all calls to strlen() are
eliminated.  This is also what happens at all optimization levels for the
x86_64-linux target but not for powerpc64 where at -Og the optimized dump
shows:

  <bb 8> [local count: 214748365]:
  _10 = _3 + 1;
  x = _10;
  x.9_11 = (unsigned int) _3;
  _20 = x.9_11 & 7;
  _12 = (sizetype) _20;
  _13 = &larger + _12;
  _14 = strlen (_13);
  if (_14 != 5)
    goto <bb 9>; [0.00%]
  else
    goto <bb 10>; [99.96%]

  <bb 9> [count: 0]:
  abort ();

Besides fixing the test (the runtime test harness ignored dg-options in these
tests) I need to figure out what prevents the folding at -Og.