[Bug c++/30567] -fPIC -O3 optimizer bug (32-bit target only)

["dberlin at dberlin dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #13 from dberlin at gcc dot gnu dot org  2007-02-26 02:42 -------
Subject: Re:  -fPIC -O3 optimizer bug (32-bit target only)

On 26 Feb 2007 02:26:12 -0000, rwgk at yahoo dot com
<gcc-bugzilla@gcc.gnu.org> wrote:
>
>
> ------- Comment #12 from rwgk at yahoo dot com  2007-02-26 02:26 -------
> Daniel Berlin wrote:
> > This change actually made us more conservative with points-to, it certainly
> > won't cause *more* things to be optimized away.
>
> Was the change supposed to fix a certain problem?
Yes.
http://gcc.gnu.org/ml/gcc-patches/2006-12/msg00817.html
It also fixed vect-101 testcase, in addition to 464.h264ref from spec2006.
It initially caused a failure in 403.gcc, but this was later found to
be a latent bug elsewhere.
I imagine yours is the same thing (a latent bug elsewhere).

If you revert it, you will start failing these regression tests again.

> If not I suggest the two lines should be backed out since they clearly cause
> a regression on the 4.2 branch.
Then whatever backing that out papers over should be fixed.

> Here is my local patch to back out the two lines introduced in
> SVN revision 119791:

Sorry, but i'm not going to accept a patch that clearly does the wrong
thing, and will cause existing regression tests to fail, so that we
can hide some other latent bug.
If you want to try pointing out why you think what the patch does is
incorrect, i'm happy to work through it with you.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30567