[Bug tree-optimization/56610] New: IPA(-CP) clone materialization fails to cleanup the CFG

[Bug tree-optimization/56610] New: IPA(-CP) clone materialization fails to cleanup the CFG

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56610

             Bug #: 56610
           Summary: IPA(-CP) clone materialization fails to cleanup the
                    CFG
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: rguenth@gcc.gnu.org
                CC: hubicka@gcc.gnu.org


__attribute__((noinline)) static int
foo (int x)
{
  int i = 1;
  if (x > 1)
    do
      bar (i);
    while (++i != x);
}

void
baz (void)
{
  foo (1);
  foo (1);
  foo (1);
}


We can see a

  if (1 == 0)

survive for a long time.  When IPA-CP get's the chance to look at the
clones body in ipcp_transform_function there is nothing that changes
the CFG and thus it doesn't queue a cfg-cleanup to fix up the above
(which is created by clone materialization itself).

Now we can just unconditionally schedule a cfg-cleanup after IPA-CP,
but really all those "todo"s should queue up for all IPA transforms
(including fixup-cfg, which should be a IPA transform-only pass in the
IPA pipeline).

[Bug tree-optimization/56610] IPA(-CP) clone materialization fails to cleanup the CFG

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56610

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |12.0
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2021-07-26
     Ever confirmed|0                           |1

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed.  Though I noticed the inlining will come in and clean it up which is
two passes after ipa-cp.