[Bug tree-optimization/82090] Bogus warning: ‘magic_p’ may be used uninitialized in this function [-Wmaybe-uninitialized]

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82090

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |missed-optimization
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2017-09-04
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed (at -O1 even).

We fail to jump-thread oldmem_5(D) == 0 and get

  <bb 2> [100.00%]:
  if (oldmem_5(D) == 0B)
    goto <bb 3>; [4.07%]
  else
    goto <bb 7>; [95.93%]

  <bb 7> [95.93%]:
  goto <bb 4>; [100.00%]

  <bb 3> [4.07%]:

  <bb 4> [100.00%]:
  # magic_p_17 = PHI <magic_p_18(D)(3), oldmem_5(D)(7)>
  __asm__ __volatile__("cmpl $0, __libc_multiple_threads(%%rip)
...
  if (oldmem_5(D) == 0B)
    goto <bb 5>; [0.04%]
  else
    goto <bb 6>; [99.96%]

  <bb 5> [0.04%]:
  malloc_printerr ("realloc(): invalid pointer");

  <bb 6> [99.96%]:
  _2 = *magic_p_17;
  _3 = ~_2;
  *magic_p_17 = _3;
  return oldmem_5(D);

and the late uninit pass is not able to "prove" we're not using magic_p_18(D).

Jump threading probably gives up because of the asm ().