- Upgrade to GDB 6.5. Drop redundant patches, forward-port remaining ones.

gdb-6.3-ia64-corefile-fix-20050127.patch

@@ -4,20 +4,23 @@
 	unwind table syscall to support backtracing from syscalls in
 	a corefile.
 
---- gdb-6.3/gdb/ia64-tdep.c.fix	Thu Jan 27 12:03:40 2005
-+++ gdb-6.3/gdb/ia64-tdep.c	Thu Jan 27 12:06:23 2005
-@@ -2357,8 +2357,14 @@ static int
- getunwind_table (void *buf, size_t len)
- {
-   LONGEST x;
--  x = target_read_partial (&current_target, TARGET_OBJECT_UNWIND_TABLE, NULL,
--			   buf, 0, len);
-+
+Index: gdb-6.5/gdb/ia64-tdep.c
+===================================================================
+--- gdb-6.5.orig/gdb/ia64-tdep.c	2006-04-18 16:20:06.000000000 -0300
++++ gdb-6.5/gdb/ia64-tdep.c	2006-07-07 02:18:40.000000000 -0300
+@@ -2470,8 +2470,17 @@ getunwind_table (void *buf, size_t len)
+      we want to preserve fall back to the running kernel's table, then
+      we should find a way to override the corefile layer's
+      xfer_partial method.  */
++#if 0
+   x = target_read_partial (&current_target, TARGET_OBJECT_UNWIND_TABLE, NULL,
+ 			   buf, 0, len);
++#endif
 +  /* FIXME: This is a temporary solution to backtracing syscalls in corefiles.
 +            To do this properly, the AUXV section should be used.  This
 +	    fix will work as long as the kernel used to generate the corefile
 +	    is equivalent to the kernel used to debug the corefile.  */
-+  x = ia64_linux_xfer_unwind_table (&current_target, 
++  x = ia64_linux_xfer_unwind_table (&current_target,
 +		  		    TARGET_OBJECT_UNWIND_TABLE, NULL,
 +			   	    buf, NULL, 0, len);