patch for i960 'not a lvalue' bug

[Geoff Keating]

The i960 backend seems to firmly believe that symbol_next
is just a macro...  This works around it.  I didn't want
to add backpointers just for this one case.

OK to commit?

-- 
- Geoffrey Keating <geoffk@geoffk.org>

===File ~/patches/cygnus/nortel-boutlvalue.patch============
2000-11-13  Geoffrey Keating  <geoffk@redhat.com>

	* config/obj-bout.c (obj_crawl_symbol_chain): Don't take
	the address of a function result.

Index: gas/config/obj-bout.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gas/config/obj-bout.c,v
retrieving revision 1.30
diff -u -p -r1.30 obj-bout.c
--- obj-bout.c	2000/09/03 17:33:30	1.30
+++ obj-bout.c	2000/11/14 05:29:04
@@ -310,7 +310,7 @@ obj_crawl_symbol_chain (headers)
 	    }
 	  else			/* .Stabd case.  */
 	    symbolP->sy_name_offset = 0;
-	  symbolPP = &(symbol_next (symbolP));
+	  symbolPP = &(symbolP->sy_next);
 	}
       else
 	{
============================================================