Patch for obj-coff.c

Patch for obj-coff.c

[Mark E.]

Hello Ian,

This patch seems to fix the debugging problem under BFD assember caused by not 
creating an endndx AUX record. I believe the Win32 ports use stabs by default, so 
it's no wonder no one ever noticed.

1999-07-30 Mark Elbrecht <snowball3@bigfoot.com>

* config/obj-coff.c (coff_frob_symbol): Permit set_end to be set for C_EFCN symbols.


*** config/obj-coff.c.orig	Sat Jun 26 15:35:44 1999
--- config/obj-coff.c	Fri Jul 30 12:43:22 1999
*************** coff_frob_symbol (symp, punt)
*** 1092,1098 ****
      }
  
    if (next_set_end != NULL
!       && ! *punt)
      set_end = next_set_end;
  
    if (! *punt
--- 1092,1098 ----
      }
  
    if (next_set_end != NULL
!       && ((! *punt) || (S_GET_STORAGE_CLASS (symp) == C_EFCN)))
      set_end = next_set_end;
  
    if (! *punt

--- 
Mark Elbrecht, snowball3@bigfoot.com
http://snowball.frogspace.net/

Re: Patch for obj-coff.c

[Ian Lance Taylor]

   From: "Mark E." <snowball3@bigfoot.com>
   Date: Fri, 30 Jul 1999 12:50:19 -0400

   This patch seems to fix the debugging problem under BFD assember caused by not 
   creating an endndx AUX record. I believe the Win32 ports use stabs by default, so 
   it's no wonder no one ever noticed.

Thanks for tracking this down.

Looking at this code, I don't why see it needs to ever check *punt.
Could you try out this patch and see if you ever get the warning?

Ian

--- config/obj-coff.c.~4~	Sat Jun 26 22:32:49 1999
+++ config/obj-coff.c	Fri Jul 30 13:02:26 1999
@@ -1091,9 +1091,13 @@ coff_frob_symbol (symp, punt)
       set_end = NULL;
     }
 
-  if (next_set_end != NULL
-      && ! *punt)
-    set_end = next_set_end;
+  if (next_set_end != NULL)
+    {
+      if (set_end != NULL)
+	as_warn ("Warning: internal error: forgetting to set endndx of %s",
+		 S_GET_NAME (set_end));
+      set_end = next_set_end;
+    }
 
   if (! *punt
       && S_GET_STORAGE_CLASS (symp) == C_FCN