Re: [PATCH] tracing/function-return-tracer: don't trace kfree while it frees the return stack

[Ingo Molnar <mingo@elte.hu>]

* Ingo Molnar <mingo@elte.hu> wrote:

>  void ftrace_retfunc_exit_task(struct task_struct *t)
>  {
> -	kfree(t->ret_stack);
> +	struct ftrace_ret_stack	*ret_stack = t->ret_stack;
> +
>  	t->ret_stack = NULL;
> +	/* NULL must become visible to IRQs before we free it: */
> +	barrier();
> +
> +	kfree(ret_stack);
>  }

hm, this reminds me - this is the wrong place for the callback - the 
call stack is freed too early.

instead of freeing it in do_exit() (like it's done right now), it 
should be freed when the task struct and thread info is freed: in 
kernel/fork.c:free_task() - okay?

The difference is minor but could allow more complete tracing: as 
right now we'll skip the entries that get generated when we schedule 
away from a dead task.

	Ingo