pidfd: don't do_notify_pidfd() if !thread_group_empty()

do_notify_pidfd() makes no sense until the whole thread group exits, change
do_notify_parent() to check thread_group_empty().

This avoids the unnecessary do_notify_pidfd() when tsk is not a leader, or
it exits before other threads, or it has a ptraced EXIT_ZOMBIE sub-thread.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Link: https://lore.kernel.org/r/20240127132407.GA29136@redhat.comReviewed-by: Tycho Andersen <tandersen@netflix.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>

kernel/signal.c

@@ -2050,9 +2050,11 @@ bool do_notify_parent(struct task_struct *tsk, int sig)
 
 	WARN_ON_ONCE(!tsk->ptrace &&
 	       (tsk->group_leader != tsk || !thread_group_empty(tsk)));
-
-	/* Wake up all pidfd waiters */
-	do_notify_pidfd(tsk);
+	/*
+	 * tsk is a group leader and has no threads, wake up the pidfd waiters.
+	 */
+	if (thread_group_empty(tsk))
+		do_notify_pidfd(tsk);
 
 	if (sig != SIGCHLD) {
 		/*