[patch] Fix -EPERM returned by kernel_thead() if traced...

[Chuck Ebbert]

Bernhard Kaindl wrote:

> -     task_lock(task);
> -     if (task->ptrace) {
> -             task_unlock(task);
> -             return -EPERM;
> -     }
> -
>       old_task_dumpable = task->task_dumpable;
> +     /* prevent ptrace_attach() on the new task: */
>       task->task_dumpable = 0;
> -     task_unlock(task);

  Is it safe to remove the task_lock()?

>  So I really have to assume that CLONE_PTRACE is never passed
>  to create a kernel thread. If you are paranoid, you could just
>  unmask it in kernel_thread() if you want.

  I would mask it if it's a possible problem.  Someone might
try to pass that option in future code...