[Bug optimization/10679] [3.3/3.4] --param min-inline-insns not honoured

["rguenth at tat dot physik dot uni-tuebingen dot de" <gcc-bugzilla at gcc dot gnu dot org>]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10679



------- Additional Comments From rguenth at tat dot physik dot uni-tuebingen dot de  2003-07-21 09:39 -------
Subject: Re:  [3.3/3.4] --param min-inline-insns not
 honoured

On 21 Jul 2003, pinskia at physics dot uc dot edu wrote:

> Since it sounds like you have a testcase, a POOMA based one, can you attach it here?

I tried to verify my POOMA testcase with todays mainline and were not able
to reproduce the problems I have seen (and still see with 3.3.1). So the
new counting seems to be a huge improvement, though I still think the
check in tree-inline.c should be removed. Consider a not empty function
like

void foo(int i)
{
   switch (i) {
   case 1:
     bar();
     break;
   default:;
   }
}

If you have lots of calls to foo(x) with x!=1 you're going to hit this
limit even though the optimizer should be able to optimize away the
inlined function.

At the very last, if the check is not removed, the documentation needs to
be updated to honour it.

Richard.

--
Richard Guenther <richard dot guenther at uni-tuebingen dot de>
WWW: http://www.tat.physik.uni-tuebingen.de/~rguenth/