Re: PATCH: gdb/corefile.c (0401 snap on HP-UX).

[Kevin Buettner <kevinb@cygnus.com>]

On Apr 9, 11:04pm, RDBrown@mira.net wrote:

> gdb/corefile.c was failing to compile with a missing declaration for
> `symfile_objfile'. This allows gdb to build, but still hangs when 
> starting the child process.
> 
> 2001-04-09  Rodney Brown  <RBrown64@csc.com.au>
> 
> 	* corefile.c: Include symfile.h and objfiles.h to declare
> 	symfile_objfile when HPUXHPPA.
> 
> --- gdb/corefile.c.orig	Mon Apr  9 18:45:13 2001
> +++ gdb/corefile.c	Mon Apr  9 18:02:21 2001
> @@ -34,6 +34,10 @@
>  #include "gdbcore.h"
>  #include "dis-asm.h"
>  #include "gdb_stat.h"
> +#ifdef HPUXHPPA
> +#include "symfile.h"
> +#include "objfiles.h"
> +#endif
>  
>  /* Local function declarations.  */

I'd prefer to see a somewhat different solution.

The reason that symfile_objfile is undefined for HP is because of the
following code in core_file_command():

	/* Yes, we were given the path of a core file.  Do we already
	   have a symbol file?  If not, can we determine it from the
	   core file?  If we can, do so.
	 */
#ifdef HPUXHPPA
	if (symfile_objfile == NULL)
	  {
	    char *symfile;
	    symfile = t->to_core_file_to_sym_file (filename);
	    if (symfile)
	      {
		char *symfile_copy = xstrdup (symfile);

		make_cleanup (xfree, symfile_copy);
		symbol_file_add_main (symfile_copy, from_tty);
	      }
	    else
	      warning ("Unknown symbols for '%s'; use the 'symbol-file' command.", filename);
	  }
#endif

The code controlled by the above ifdef looks pretty generic and it seems
to me that it could be useful on targets other than HP.  I suggest that
we do one of two things:

    1) Enable it for all targets and add Rodney's #include statments
       unconditionally.

    2) Remove it entirely.

I'm not familiar enough with the code in question to know what the best
course of action is.

Kevin