ftrace: Use asynchronous grace period for register_ftrace_direct()

commit 33f1371 upstream.

When running heavy test workloads with KASAN enabled, RCU Tasks grace
periods can extend for many tens of seconds, significantly slowing
trace registration.  Therefore, make the registration-side RCU Tasks
grace period be asynchronous via call_rcu_tasks().

Link: https://lore.kernel.org/linux-trace-kernel/ac05be77-2972-475b-9b57-56bef15aa00a@paulmck-laptop

Reported-by: Jakub Kicinski <kuba@kernel.org>
Reported-by: Alexei Starovoitov <ast@kernel.org>
Reported-by: Chris Mason <clm@fb.com>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Mahmoud Adam <mngyadam@amazon.com>

Author: paulmckrcu

kernel/trace/ftrace.c

@@ -5612,6 +5612,13 @@ static void remove_direct_functions_hash(struct ftrace_hash *hash, unsigned long
 	}
 }
 
+static void register_ftrace_direct_cb(struct rcu_head *rhp)
+{
+	struct ftrace_hash *fhp = container_of(rhp, struct ftrace_hash, rcu);
+
+	free_ftrace_hash(fhp);
+}
+
 /**
  * register_ftrace_direct_multi - Call a custom trampoline directly
  * for multiple functions registered in @ops
@@ -5709,10 +5716,8 @@ int register_ftrace_direct_multi(struct ftrace_ops *ops, unsigned long addr)
  out_unlock:
 	mutex_unlock(&direct_mutex);
 
-	if (free_hash && free_hash != EMPTY_HASH) {
-		synchronize_rcu_tasks();
-		free_ftrace_hash(free_hash);
-	}
+	if (free_hash && free_hash != EMPTY_HASH)
+		call_rcu_tasks(&free_hash->rcu, register_ftrace_direct_cb);
 
 	if (new_hash)
 		free_ftrace_hash(new_hash);