[Bug tree-optimization/80980] -Os generates larger code than -O1 because loop header copying is not performend and thus loop is not removed

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80980

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |missed-optimization
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2017-06-06
                 CC|                            |rguenth at gcc dot gnu.org
            Version|unknown                     |8.0
            Summary|-Os generates larger code   |-Os generates larger code
                   |than -O1 because loop is    |than -O1 because loop
                   |not removed                 |header copying is not
                   |                            |performend and thus loop is
                   |                            |not removed
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
This is loop header copying not being run on this loop -Os and thus niter
analysis "failing" and ivcanon not removing the loop.

The reason is that inside main() this loop is predicted cold and thus

  if (optimize_loop_for_size_p (loop))
    {
      if (dump_file && (dump_flags & TDF_DETAILS))
        fprintf (dump_file,
                 "  Not duplicating bb %i: optimizing for size.\n",
                 header->index);
      return false;
    }

triggers.

I suppose instead factoring the above into the duplication size estimation
cap instead would "fix" this particular case.