[Bug ipa/86132] Failure to elide condition known to be non-null

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86132

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |missed-optimization
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2018-06-14
                 CC|                            |hubicka at gcc dot gnu.org,
                   |                            |marxin at gcc dot gnu.org,
                   |                            |rguenth at gcc dot gnu.org
          Component|middle-end                  |ipa
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
I see calls to want_foo () which can clobber 'object':

  <bb 3> [local count: 187153200]:
  _8 = single_slow ();
  object = _8;

  <bb 4> [local count: 1073741824]:
  # prephitmp_16 = PHI <_8(3), object.1_7(2)>
  want_foo (prephitmp_16);
  object.1_4 = object;
  if (object.1_4 == 0B)
    goto <bb 5>; [17.43%]
  else
    goto <bb 6>; [82.57%]

that's the first issue we hit.  Then I would expect jump threading to
eventually optimize this.  I'm not sure returns_nonnull is handled
in the required places in optimizers though.

I'm not sure why IPA reference doesn't do its job here, it seems to be
confused by inlining?