[Bug c++/30252] [4.2 regression] miscompilation of sigc++-2.0 based code with -fstrict-aliasing

["rguenther at suse dot de" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #41 from rguenther at suse dot de  2007-06-06 08:49 -------
Subject: Re:  [4.2 regression] miscompilation of sigc++-2.0
 based code with -fstrict-aliasing

On Wed, 5 Jun 2007, dberlin at dberlin dot org wrote:

> q_2 = q_1 + 1
> q_3 = q_2 + 1
> q_4 = q_3 + 1
> q_5 = q_4 + 1
> p2 = q_5
> 
> Will become the set of constraints
> q_2 = q_1
> q_2 = q_1 + 1
> q_3 = q_2
> q_3 = q_2 + 1
> q_4 = q_3
> q_4 = q_3 + 1
> q_5 = q_4
> q_5 = q_4 +1
> 
> Hmmm.
> You are right, we should give up here and collapse the variable.
> Sigh.
> 
> Well, we can always just make it give up on all direct pointer
> arithmetic, rather than try to handle it at all :)
> 
> > Somehow it
> > doesn't cause problems to "fall back" to &p[0] for &p[0] + 1B,
> > but it would be nice to have some more confidence in that parts :)
> 
> TBQH, the only way i'll ever have confidence in these parts is if we
> make it stop trying to come up with everything a pointer to a
> structure could deref.

Now, this is of course only because of aliasing.  It should be
completely irrelevant for points-to.  So, if points-to comes up with
"only" &struct as a solution we would need some intermediate processing
to translate it to the correct alias set, possibly looking at the
type hierarchy again.

> Note that this also prevents us from doing cool things.

Too bad ;)

> In particular, because we say a pointer to struct p points to all it's
> members, we can never simply propagate the object it does point to if
> it points to a single object.

So we seem to agree the current solution is ugly and doesn't really
work.  Enough time to fix it for 4.3 then ;)

Btw, how does a patch to disable field sensitivity look like for the 4.2
branch?

Richard.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30252