[PATCH] fix crash in kdba bt command

[linas at austin.ibm.com]

Anton,

Please apply attached patch to ameslab.  Note, this is just the first,
there will probably be some more patches coming in the next few days.

This patch fixes a minor bug in th backtrace command; bt <addr>
was failing to actually trace the stack at addr.

--linas

-------------- next part --------------
--- kdba_bt.c.orig	2004-04-16 14:45:44.000000000 -0500
+++ kdba_bt.c	2004-04-16 15:29:42.000000000 -0500
@@ -126,7 +126,7 @@ kdba_bt_stack_ppc(struct pt_regs *regs,
 	if (!addr)
 		addr = (kdb_machreg_t *)p->thread.ksp;

-	if (addr && !task_curr(p)) {
+	if (addr && (!p || !task_curr(p))) {
 		eip = 0;
 		esp = *addr;
 		ebp = 0;
@@ -183,7 +183,8 @@ kdba_bt_stack_ppc(struct pt_regs *regs,
 		/*		kdbnearsym(eip, &symtab); */
 		kdba_find_tb_table(eip, &tbtab);

-		/*		sym = symtab.sym_name ? &symtab : &tbtab.symtab; *//* use fake symtab if necessary */
+		/* sym = symtab.sym_name ? &symtab : &tbtab.symtab; */
+		/* use fake symtab if necessary */
 		name = NULL;
 		if (esp >= PAGE_OFFSET) {
 			/*if ((sym) )*/
@@ -288,8 +289,8 @@ kdba_bt_stack(struct pt_regs *regs, kdb_
 /*
  * kdba_bt_address
  *
- *	Do a backtrace starting at a specified stack address.  Use this if the
- *	heuristics get the i386 stack decode wrong.
+ *	Do a backtrace starting at a specified stack address.  Handy
+ *	if the stack is somwhere unexpected/unusual.
  *
  * Inputs:
  *	addr	Address provided to 'bt' command.
@@ -300,9 +301,6 @@ kdba_bt_stack(struct pt_regs *regs, kdb_
  *	zero for success, a kdb diagnostic if error
  * Locking:
  *	none.
- * Remarks:
- *	mds %esp comes in handy when examining the stack to do a manual
- *	traceback.
  */

 int