Trunk merged into fortran-exp branch

[Jerry DeLisle]

On 01/26/2010 04:02 PM, Dominique Dhumieres wrote:
> With gfc_constructor_append_expr (&tmp_sym->value->value.constructor, ...
> I get a lot of failures while regtesting, the few I have investigated
> gives "f951: internal compiler error: Segmentation fault" and the backtrace
> is:
>
> Program received signal EXC_BAD_ACCESS, Could not access memory.
> Reason: KERN_INVALID_ADDRESS at address: 0x0000000000000018
> 0x0000000100095b45 in generate_isocbinding_symbol (mod_name=0x100973649 "__iso_c_binding", s=<value temporarily unavailable, due to optimizations>, local_name=<value temporarily unavailable, due to optimizations>) at ../../cons_clean/gcc/fortran/symbol.c:3692
> 3692	  c->expr = gfc_get_expr ();
> (gdb) bt
> #0  0x0000000100095b45 in generate_isocbinding_symbol (mod_name=0x100973649 "__iso_c_binding", s=<value temporarily unavailable, due to optimizations>, local_name=<value temporarily unavailable, due to optimizations>) at ../../cons_clean/gcc/fortran/symbol.c:3692
> #1  0x000000010005d2f2 in gfc_use_module () at ../../cons_clean/gcc/fortran/module.c:5201
> #2  0x0000000100062365 in accept_statement (st=<value temporarily unavailable, due to optimizations>) at ../../cons_clean/gcc/fortran/parse.c:1528
> #3  0x00000001000658d7 in parse_spec (st=<value temporarily unavailable, due to optimizations>) at ../../cons_clean/gcc/fortran/parse.c:2548
> #4  0x0000000100068cce in gfc_parse_file () at ../../cons_clean/gcc/fortran/parse.c:3944
> #5  0x00000001000a120c in gfc_be_parse_file (set_yydebug=<value temporarily unavailable, due to optimizations>) at ../../cons_clean/gcc/fortran/f95-lang.c:239
> #6  0x00000001006d02fa in toplev_main (argc=2, argv=0x7fff5fbfecb8) at ../../cons_clean/gcc/toplev.c:1053
> #7  0x0000000100001924 in start ()
>
> TIA
>
> Dominique
>
Dominique,

Thanks for testing.  If you see a problem like this after you have made a manual 
change, please post the svn diff if you can so we can see exactly what you have.

Locally, I have fixed the warning glitch and also re-merged trunk to catch the 
42858 patch.  No warnings when I "make install 1>/dev/null" and no regressions here.

I will bootstrap one more time, retest, and commit.

Jerry