Messages in this thread Patch in this message | | | Date | Mon, 20 Jun 2011 22:16:03 +0200 | From | Oleg Nesterov <> | Subject | Re: [PATCH 7/7] ptrace: s/tracehook_tracer_task()/ptrace_parent()/ |
| |
On 06/17, Tejun Heo wrote: > > tracehook.h is on the way out. Rename tracehook_tracer_task() to > ptrace_parent() and move it from tracehook.h to ptrace.h.
I am a bit surpised you decided to keep this helper. Can't we simply kill it?
OK, we will see. I guess this change is mostly needed to remove yet another function from tracehook.h.
> @@ -216,7 +216,7 @@ static struct mm_struct *__check_mem_permission(struct task_struct *task) > if (task_is_stopped_or_traced(task)) { > int match; > rcu_read_lock(); > - match = (tracehook_tracer_task(task) == current); > + match = (ptrace_parent(task) == current); > rcu_read_unlock(); > if (match && ptrace_may_access(task, PTRACE_MODE_ATTACH))
All we need
if (task_is_traced(task) && task->parent == current) { if (ptrace_may_access() return mm; }
Of course I do not blame this patch, my only point is that this helper only adds more confusion imho.
> @@ -67,7 +67,7 @@ static int may_change_ptraced_domain(struct task_struct *task, > int error = 0; > > rcu_read_lock(); > - tracer = tracehook_tracer_task(task); > + tracer = ptrace_parent(task); > if (tracer) { > /* released below */ > cred = get_task_cred(tracer);
Hmm. And then this task_struct is used after we dropped rcu_read_lock().
John, is this correct?
Oleg.
| |