CLONE_PARENT shouldn't allow to set ->exit_signal

The child must not control its ->exit_signal, it is the parent who
decides which signal the child should use for notification.

This means that CLONE_PARENT should not use "clone_flags & CSIGNAL",
the forking task is the sibling of the new process and their parent
doesn't control exit_signal in this case.

This patch uses ->exit_signal of the forking process, but perhaps
we should simply use SIGCHLD.

We read group_leader->exit_signal lockless, this can race with the
ORIGINAL_SIGNAL -> SIGCHLD transition, but this is fine.

Potentially this change allows to kill self_exec_id/parent_exec_id.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

kernel/fork.c

@@ -1340,7 +1340,13 @@ static struct task_struct *copy_process(unsigned long clone_flags,
 	clear_all_latency_tracing(p);
 
 	/* ok, now we should be set up.. */
-	p->exit_signal = (clone_flags & CLONE_THREAD) ? -1 : (clone_flags & CSIGNAL);
+	if (clone_flags & CLONE_THREAD)
+		p->exit_signal = -1;
+	else if (clone_flags & CLONE_PARENT)
+		p->exit_signal = current->group_leader->exit_signal;
+	else
+		p->exit_signal = (clone_flags & CSIGNAL);
+
 	p->pdeath_signal = 0;
 	p->exit_state = 0;