http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55478
Jan Hubicka <hubicka at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
--- Comment #2 from Jan Hubicka <hubicka at gcc dot gnu.org> 2012-11-27
00:00:52 UTC ---
OK,
the inliner recognizes that the call is going to be inlined.
At -O2 this is however ignored because the caller is too large and not declared
inline. I am not sure if we really want to enforce inlining in this case
At -O3 this is ignored because the function is called just once and thus call
is cold. This is also sort of correct, but a testcase that will do the same in
a large loop will suffer from the same problem. I guess we should introduce
the notion of heavy calls like open64 does and inline in such case (i.e. inline
into functions called once when some of inline hints are met). I will look
into this.
