------- Additional Comments From dave at hiauly1 dot hia dot nrc dot ca
2005-05-19 23:10 -------
Subject: Re: FAIL: gfortran.dg/eoshift.f90 -O0 execution test
> > I have no idea why it would only fail at -O0 though, the .original dump on
> > i586-linux looks fine, and there's nothing strongly system specific in it.
>
> This may be due to an uninitialized variable.
As of last night, it failed on hppa-linux:
May 19 01:37:57 gsyprf11 kernel: do_page_fault() pid=29787 command='eoshift.exe'
type=15 address=0x00000000
Failure is caused by dereferencing a null pointer.
(gdb) r
Starting program: /home/dave/gcc-4.1/objdir/gcc/testsuite/eoshift.xg
Program received signal SIGSEGV, Segmentation fault.
0x000109a8 in MAIN__ ()
at /home/dave/gcc-4.1/gcc/gcc/testsuite/gfortran.dg/eoshift.f90:5
5 write (line,'(I4)') eoshift((/1, 3/), 3)
Current language: auto; currently fortran
(gdb) bt
#0 0x000109a8 in MAIN__ ()
at /home/dave/gcc-4.1/gcc/gcc/testsuite/gfortran.dg/eoshift.f90:5
#1 0x00010a00 in main (argc=Variable "argc" is not available.
) at ../../../gcc/libgfortran/fmain.c:18
Dave
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21594
