This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug fortran/18977] LAPACK test xeigtsts segfaults with optimization
- From: "Thomas dot Koenig at online dot de" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 6 Jan 2005 17:01:50 -0000
- Subject: [Bug fortran/18977] LAPACK test xeigtsts segfaults with optimization
- References: <20041214104244.18977.Thomas.Koenig@online.de>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Additional Comments From Thomas dot Koenig at online dot de 2005-01-06 17:01 -------
(In reply to comment #5)
> I'm no Fortran guru, but could be this related to PR 17675?
I don't think this is an alignment problem. Apparently,
ia64-unknown-linux-gnu sets up the processor differently from
HP-UX.
Test case:
$ cat equiv.f
integer a(5)
double precision d1,d2
equivalence (a(1),d1), (a(4),d2)
call foo(d1,d2)
print *,d1,d2
end
subroutine foo(d1,d2)
double precision d1,d2
d1 = 3.14D0
d2 = 2.71D0
end
$ gfortran equiv.f
$ ./a.out
3.14000000000000 2.71000000000000
$ ifort equiv.f
fortcom: Warning: equiv.f, line 2: Alignment of variable or array is
inconsistent with its data type. [D2]
double precision d1,d2
--------------------------^
$ ./a.out
3.14000000000000 2.71000000000000
$ g77 equiv.f
Can't place `d1' as directed by EQUIVALENCE due to alignment restrictions
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18977