optimization/10185
Janis Johnson
janis187@us.ibm.com
Sat Mar 22 01:28:00 GMT 2003
On Sat, Mar 22, 2003 at 12:57:32AM +0100, Steven Bosscher wrote:
> Op za 22-03-2003, om 00:42 schreef Janis Johnson:
> > On Fri, Mar 21, 2003 at 05:29:38PM -0600, Wolfgang Bangerth wrote:
> > >
> > > Actually, this bug was introduced between 2002-07-15 and 2002-08-01. Maybe
> > > this short-cuts your search a little, Janis.
> >
> > Yes, that ought to save an hour or two. The hunt is on and
> > has already verified that it's between "2002-02-25 00:00 UTC"
> > and "2002-12-13 00:00 UTC", when 3.1 and 3.3 branched from
> > mainline.
> >
> > These hunts are easiest for me to run on weekends, so feel
> > free to let me know if there are others that would be useful.
>
> Hmm... Let's concentrate on poorly analyzed 3.3 regressions; it'd be
> really cool if you have time to hunt these down:
>
> opt/10087 [3.3/3.4 regression] optimizer produces wrong code when indexing 2D array
I found this last Saturday, but my mail was rejected as spam for one
address and didn't show up in the audit trail of the PR. It's in
http://gcc.gnu.org/ml/gcc-bugs/2003-03/msg00942.html. There was a
patch submitted today to fix this.
> c++/10047 [3.3 regression] -fno-default-inline produces bogous warnings
This is reproducible on my laptop so I'll hunt for it this weekend.
> opt/9123 [3.2/3.3/3.4 regression ] Internal compiler error in do_SUBST at combine.c:434
I tried this one last weekend using the smaller C version of the test,
but that one often passes when the C++ version fails. I'll try again
with the C++ test.
> They have not been analyzed beyond "confirmed" (which should have its
> own state BTW). If you'll take a shot at some of these, please let me
> know and I'll add a note to the audit trails of the PRs.
This time I'll verify that my messages show up in the PRs.
Janis
More information about the Gcc-bugs
mailing list