grokfield miscompiled with -O9

[Alexandre Oliva]

On Jan  8, 2000, "Martin v. Loewis" <martin@loewis.home.cs.tu-berlin.de> wrote:

>> I wonder if this isn't the same bug as execute/991112-1.c.  That one has
>> to do with -O3 and not clearing an entire register, too.

> That could well be. Another commonality is that removing the unused,
> unrelated function corrects the code...

It is indeed caused by the same problem.  If only I had read this just
a second before checking in the testcase I've just posted...

Anyway, that testcase would risk hiding the bug if gcc got smarter
about inlining pointers to functions that can be proven constant.

-- 
Alexandre Oliva http://www.ic.unicamp.br/~oliva IC-Unicamp, Bra[sz]il
oliva@{lsd.ic.unicamp.br,guarana.{org,com}} aoliva@{acm,computer}.org
oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org}
** I may forward mail about projects to mailing lists; please use them