[PATCH] powerpc: fix support for latencytop

[Arnd Bergmann]

We need to pass the kernel stack pointer instead of the user space
stack pointer in save_stack_trace_tsk().

Signed-off-by: Arnd Bergmann <arnd at arndb.de>

On Wednesday 16 July 2008, Nathan Lynch wrote:
> Arnd Bergmann wrote:
> > Implement save_stack_trace_tsk on powerpc, so that we can run with
> > latencytop.
> 
> So I tried latencytop with linux-next and got the following oops, but
> I didn't really look into it yet.

Oh, I didn't even realize that benh had merged that patch of mine.
As I wrote in the description, it was entirely untested. You found
another obvious bug: The code was passing the user space stack
pointer instead of the kernel stack pointer.

Again, this patch is entirely untested, and I would not be at all
surprised to find other trivial bugs.

--- a/arch/powerpc/kernel/stacktrace.c
+++ b/arch/powerpc/kernel/stacktrace.c
@@ -59,6 +59,6 @@ EXPORT_SYMBOL_GPL(save_stack_trace);
 
 void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
 {
-	save_context_stack(trace, tsk->thread.regs->gpr[1], tsk, 0);
+	save_context_stack(trace, tsk->thread.ksp, tsk, 0);
 }
 EXPORT_SYMBOL_GPL(save_stack_trace_tsk);