namelist tests are failing...

[Steve Kargl]

On Sat, Nov 26, 2005 at 03:38:57PM +0100, Andreas Jaeger wrote:
> 
> Running namelist_12.f under gdb gives the following backtrace:
> 
> Program received signal SIGABRT, Aborted.
> 0x00002aaaaafff43a in raise () from /lib64/tls/libc.so.6
> (gdb) bt
> #0  0x00002aaaaafff43a in raise () from /lib64/tls/libc.so.6
> #1  0x00002aaaab000870 in abort () from /lib64/tls/libc.so.6
> #2  0x00002aaaaac27fee in _gfortran_abort () at /cvs/gcc-svn/trunk/libgfortran/intrinsics/abort.c:39
> #3  0x0000000000401415 in MAIN__ () at /cvs/gcc-svn/trunk/gcc/testsuite/gfortran.dg/namelist_12.f:54
> #4  0x000000000040142e in main (argc=<value optimized out>, argv=<value optimized out>) at /cvs/gcc-svn/trunk/libgfortran/fmain.c:18
> 
> So, line 54 gives an error, the line is:
>     54       if (xx.ne.42) call abort ()
>  
> xx has a content of 0 at this line,
> 
> If I compile the program with -m32, it succeeds, so only the 64-bit
> program is broken.
> 

There were only two commits concerning namelist in the last
few days.  Can you revert these changes to see which may
be the guilty party? 

2005-11-25  Richard Henderson  <rth@redhat.com>

        * io/list_read.c (nml_parse_qualifier): Use memcpy to extract
        values from dtp->u.p.value.
        * io/io.h (struct st_parameter_dt): Change reversion_flag, first_item,
        seen_dollar, sf_seen_eor, eor_condition, no_leading_blank, char_flag,
        input_complete, at_eol, comma_flag, namelist_mode, nml_read_error to
        single-bit fields.  Move value to where it'll be at least pointer
        aligned.


2005-11-22  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

        PR libgfortran/24794
        * io/list_read.c (read_character): Add auto completion on short
        namelist reads.




-- 
Steve