SEGV running gdb-6.7 on Tru64 UNIX 5.1

SEGV running gdb-6.7 on Tru64 UNIX 5.1

[Albert Chin]

Anyone running 6.7 successfully on Tru64 UNIX 5.1? gdb-6.6 Worked ok.

% cat a.c
#include <stdio.h>

int
main (void) {
  puts ("a");
}
% cc -g a.c
% gdb-6.6 a.out
gdb> list
...

% gdb-6.7 a.out
gdb> list
Memory fault - core dumped
% ladebug gdb-6.7 a.out
ladebug> where
>0  0x120125920 in symbol_found(funfirstline=1, canonical=0x0, copy=0x1401778d0="main", sym=0x14014d7c0, file_symtab=0x0, sym_symtab=0x0) "linespec.c":1786
#1  0x120125744 in decode_variable(copy=0x1401778d0="main", funfirstline=1, canonical=0x0, file_symtab=0x0, not_found_ptr=0x0) "linespec.c":1737
#2  0x120123a80 in decode_line_1(argptr=0x11fffb7a8, funfirstline=1, default_symtab=0x0, default_line=0, canonical=0x0, not_found_ptr=0x0) "linespec.c":859
#3  0x1200beb94 in decode_line_spec(string=0x1401260a4="", funfirstline=1) "symtab.c":4095
#4  0x120081a64 in select_source_symtab(s=0x0) "source.c":244
#5  0x1200818d4 in set_default_source_symtab_and_line() "source.c":180
#6  0x12008a008 in list_command(arg=0x0, from_tty=1) ".././gdb/cli/cli-cmds.c":735
#7  0x1200a1cd0 in do_cfunc(c=0x14013ee00, args=0x0, from_tty=1) ".././gdb/cli/cli-decode.c":60
#8  0x1200a5118 in cmd_func(cmd=0x14013ee00, args=0x0, from_tty=1) ".././gdb/cli/cli-decode.c":1663
#9  0x120085e00 in execute_command(p=0x1400e0484="", from_tty=1) "top.c":449
#10 0x1200a0560 in command_handler(command=0x1400e0480="list") "event-top.c":518
#11 0x1200a10dc in command_line_handler(rl=0x140177820="") "event-top.c":804
#12 0x3ffbffefe20 in rl_callback_read_char(...) in /opt/fsw/readline52/lib/libreadline.so.7

Any ideas?

-- 
albert chin (china@thewrittenword.com)

Re: SEGV running gdb-6.7 on Tru64 UNIX 5.1

[Joel Brobecker]

> Anyone running 6.7 successfully on Tru64 UNIX 5.1? gdb-6.6 Worked ok.

I am able to run it through your example. However, I did build the
executable using gcc, not cc. That's probably where the difference is.

> >0  0x120125920 in symbol_found(funfirstline=1, canonical=0x0, copy=0x1401778d0="main", sym=0x14014d7c0, file_symtab=0x0, sym_symtab=0x0) "linespec.c":1786

The problem here is that both file_symtab and sym_symtab are NULL.
Looking at the frame just above it:

> #1  0x120125744 in decode_variable(copy=0x1401778d0="main", funfirstline=1, canonical=0x0, file_symtab=0x0, not_found_ptr=0x0) "linespec.c":1737

Apparently, lookup_symbol found a symbol for "main", but did not
return the associated symtab.

You'll probably need to debug GDB to find out why
lookup_symbol_in_language set the symtab to NULL...

-- 
Joel

Re: SEGV running gdb-6.7 on Tru64 UNIX 5.1

[Albert Chin]

On Wed, Oct 17, 2007 at 08:19:54PM -0700, Joel Brobecker wrote:
> > Anyone running 6.7 successfully on Tru64 UNIX 5.1? gdb-6.6 Worked ok.
> 
> I am able to run it through your example. However, I did build the
> executable using gcc, not cc. That's probably where the difference is.

I built with gcc-3.4.3 and still get a SEGV with the 'list' command,
but the ladebug backtrace is different:
  >0  0x1200a802c in target_write_with_progress(ops=0x1401297c0, object=TARGET_OBJECT_SPU, annex=0x0, buf=0x0, offset=4831842700, len=4831819312, progress=0x11fffb6b0, baton=0x11fffb630) "target.c":1452
  #1  0x1200a7ff4 in target_write_with_progress(ops=0x1401297c0, object=TARGET_OBJECT_SPU, annex=0x0, buf=0x0, offset=4831842700, len=4831819312, progress=0x11fffb6b0, baton=0x11fffb630) "target.c":1439

> > >0  0x120125920 in symbol_found(funfirstline=1, canonical=0x0, copy=0x1401778d0="main", sym=0x14014d7c0, file_symtab=0x0, sym_symtab=0x0) "linespec.c":1786
> 
> The problem here is that both file_symtab and sym_symtab are NULL.
> Looking at the frame just above it:
> 
> > #1  0x120125744 in decode_variable(copy=0x1401778d0="main", funfirstline=1, canonical=0x0, file_symtab=0x0, not_found_ptr=0x0) "linespec.c":1737
> 
> Apparently, lookup_symbol found a symbol for "main", but did not
> return the associated symtab.
> 
> You'll probably need to debug GDB to find out why
> lookup_symbol_in_language set the symtab to NULL...

Ok, I'll have to do some digging. Thanks.

-- 
albert chin (china@thewrittenword.com)