[Bug fortran/42958] Weird temporary array allocation

[rguenth at gcc dot gnu dot org]

------- Comment #11 from rguenth at gcc dot gnu dot org  2010-03-27 19:12 
-------
The middle-end ends up with the following optimized:

  D.1776_95 = D.1775_94 - D.1591_90;
  if (D.1776_95 <= 0)
    goto <bb 34>;
  else
    goto <bb 33>;

<bb 33>:
  D.1620_204 = __builtin_malloc (D.1795_11);
  D.1796_281 = iy.dim[1].stride;
...
  goto <bb 35>;

<bb 34>:
  D.1620_135 = __builtin_malloc (1);
  goto <bb 38>;

<bb 35>:
   the loop code

<bb 38>:
Invalid sum of incoming frequencies 1062, should be 900
  # D.1620_374 = PHI <D.1620_204(37), D.1620_135(34)>
  if (D.1620_374 != 0B)
    goto <bb 39>;
  else
    goto <bb 40>;

<bb 39>:
  __builtin_free (D.1620_374);


so we are able to prove that if we end up allocating 1 byte only then
the loop isn't executed and we just free it (another missed middle-end
optimization, p = malloc(1); free (p); isn't optimized away).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42958