Re: [PATCH] rcu: Safe access to rcu_node structure's->exp_tasks

[Paul E. McKenney]

On Sat, Dec 24, 2022 at 01:25:53PM +0800, Zqiang wrote:
> For kernels built with CONFIG_PREEMPT_RCU=y, the following scenario
> can result system oops.
> 
>            CPU1                                           CPU2
> rcu_preempt_deferred_qs_irqrestore                rcu_print_task_exp_stall
>   if (special.b.blocked)                            READ_ONCE(rnp->exp_tasks) != NULL
>     raw_spin_lock_rcu_node
>     np = rcu_next_node_entry(t, rnp)
>     if (&t->rcu_node_entry == rnp->exp_tasks)
>       WRITE_ONCE(rnp->exp_tasks, np)
>       ....
>       raw_spin_unlock_irqrestore_rcu_node
>                                                     raw_spin_lock_irqsave_rcu_node
>                                                     t = list_entry(rnp->exp_tasks->prev,
>                                                         struct task_struct, rcu_node_entry)
>                                                     (if rnp->exp_tasks is NULL
>                                                        will trigger oops)
> 
> This problem is that CPU2 accesses rcu_node structure's->exp_tasks
> without holding the rcu_node structure's ->lock and CPU2 did not
> observe CPU1's change to rcu_node structure's->exp_tasks in time,
> if rcu_node structure's->exp_tasks is set null pointer by CPU1, after
> that CPU2 accesses members of rcu_node structure's->exp_tasks will
> trigger oops.
> 
> This commit therefore allows rcu_node structure's->exp_tasks to be
> accessed while holding rcu_node structure's ->lock.
> 
> Signed-off-by: Zqiang <qiang1.zhang@xxxxxxxxx>

Apologies for the delay and thank you for the reminder!

Please check the wordsmithed version below, which I have queued.

						Thanx, Paul

------------------------------------------------------------------------

commit 389b0eafd72829fd63548f7ff4e8d6ac90fa1f98
Author: Zqiang <qiang1.zhang@xxxxxxxxx>
Date:   Sat Dec 24 13:25:53 2022 +0800

    rcu: Protect rcu_print_task_exp_stall() ->exp_tasks access
    
    For kernels built with CONFIG_PREEMPT_RCU=y, the following scenario can
    result in a NULL-pointer dereference:
    
               CPU1                                           CPU2
    rcu_preempt_deferred_qs_irqrestore                rcu_print_task_exp_stall
      if (special.b.blocked)                            READ_ONCE(rnp->exp_tasks) != NULL
        raw_spin_lock_rcu_node
        np = rcu_next_node_entry(t, rnp)
        if (&t->rcu_node_entry == rnp->exp_tasks)
          WRITE_ONCE(rnp->exp_tasks, np)
          ....
          raw_spin_unlock_irqrestore_rcu_node
                                                        raw_spin_lock_irqsave_rcu_node
                                                        t = list_entry(rnp->exp_tasks->prev,
                                                            struct task_struct, rcu_node_entry)
                                                        (if rnp->exp_tasks is NULL, this
                                                           will dereference a NULL pointer)
    
    The problem is that CPU2 accesses the rcu_node structure's->exp_tasks
    field without holding the rcu_node structure's ->lock and CPU2 did
    not observe CPU1's change to rcu_node structure's ->exp_tasks in time.
    Therefore, if CPU1 sets rcu_node structure's->exp_tasks pointer to NULL,
    then CPU2 might dereference that NULL pointer.
    
    This commit therefore holds the rcu_node structure's ->lock while
    accessing that structure's->exp_tasks field.
    
    Signed-off-by: Zqiang <qiang1.zhang@xxxxxxxxx>
    Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxx>

diff --git a/kernel/rcu/tree_exp.h b/kernel/rcu/tree_exp.h
index 7cc4856da0817..902e7c8709c7e 100644
--- a/kernel/rcu/tree_exp.h
+++ b/kernel/rcu/tree_exp.h
@@ -803,9 +803,11 @@ static int rcu_print_task_exp_stall(struct rcu_node *rnp)
 	int ndetected = 0;
 	struct task_struct *t;
 
-	if (!READ_ONCE(rnp->exp_tasks))
-		return 0;
 	raw_spin_lock_irqsave_rcu_node(rnp, flags);
+	if (!READ_ONCE(rnp->exp_tasks)) {
+		raw_spin_unlock_irqrestore_rcu_node(rnp, flags);
+		return 0;
+	}
 	t = list_entry(rnp->exp_tasks->prev,
 		       struct task_struct, rcu_node_entry);
 	list_for_each_entry_continue(t, &rnp->blkd_tasks, rcu_node_entry) {