[patch] Fix crash in read_dbx_symtab

[patch] Fix crash in read_dbx_symtab

[Paul Pluzhnikov]

Greetings,

This checkin:

2009-03-31  Daniel Jacobowitz  <dan@codesourcery.com>
            Keith Seitz  <keiths@redhat.com>
            Jan Kratochvil  <jan.kratochvil@redhat.com>

        PR gdb/6817
        * Makefile.in (dbxread.o): Update.
        * dbxread.c (read_dbx_symtab): Use cp_canonicalize_string.
        ...

Introduced an alloca in a loop for each symbol, which causes GDB
to run out of stack and crash when stack is limited (e.g.
'ulimit -s' == 8192) and the executable has a lot of long STABs
(C++ N_LSYMs in my test case).

Here is a proposed fix.

Thanks,
-- 
Paul Pluzhnikov

2009-07-24  Paul Pluzhnikov  <ppluzhnikov@google.com>

	PR gdb/6817
	* dbxread.c (read_dbx_symtab): Reduce stack use.



Index: dbxread.c
===================================================================
RCS file: /cvs/src/src/gdb/dbxread.c,v
retrieving revision 1.104
diff -u -p -u -r1.104 dbxread.c
--- dbxread.c	29 Jun 2009 16:48:15 -0000	1.104
+++ dbxread.c	24 Jul 2009 16:39:03 -0000
@@ -1689,7 +1689,7 @@ pos %d"),
 	  sym_name = NULL;	/* pacify "gcc -Werror" */
  	  if (psymtab_language == language_cplus)
  	    {
- 	      char *new_name, *name = alloca (p - namestring + 1);
+ 	      char *new_name, *name = xmalloc (p - namestring + 1);
  	      memcpy (name, namestring, p - namestring);
  	      name[p - namestring] = '\0';
  	      new_name = cp_canonicalize_string (name);
@@ -1700,6 +1700,7 @@ pos %d"),
  					   &objfile->objfile_obstack);
  		  xfree (new_name);
  		}
+              xfree (name);
  	    }
 
  	  if (sym_len == 0)

Re: [patch] Fix crash in read_dbx_symtab

[Tom Tromey]

>>>>> "Paul" == Paul Pluzhnikov <ppluzhnikov@google.com> writes:

Paul> 2009-07-24  Paul Pluzhnikov  <ppluzhnikov@google.com>
Paul> 	PR gdb/6817
Paul> 	* dbxread.c (read_dbx_symtab): Reduce stack use.

This is ok.

There's also the maybe-related PR 10457.  If you're looking for
something to do ;)

Tom