Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Tue, 31 Jul 2018 13:27:01 -0700
From:   Stephen Hemminger <stephen@networkplumber.org>
To:     "Jason A. Donenfeld" <Jason@zx2c4.com>
Cc:     linux-kernel@vger.kernel.org, netdev@vger.kernel.org,
        davem@davemloft.net, Greg KH <gregkh@linuxfoundation.org>
Subject: Re: [PATCH v1 3/3] net: WireGuard secure network tunnel
Message-ID: <20180731132701.522b55e2@xeon-e3>
In-Reply-To: <20180731191102.2434-4-Jason@zx2c4.com>
References: <20180731191102.2434-1-Jason@zx2c4.com>
        <20180731191102.2434-4-Jason@zx2c4.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, 31 Jul 2018 21:11:02 +0200
"Jason A. Donenfeld" <Jason@zx2c4.com> wrote:

> +#define push(stack, p, len) ({ \
> +	if (rcu_access_pointer(p)) { \
> +		BUG_ON(len >= 128); \
> +		stack[len++] = rcu_dereference_protected(p, lockdep_is_held(lock)); \
> +	} \
> +	true; \
> +})
> +static void free_root_node(struct allowedips_node __rcu *top, struct mutex *lock)
> +{
> +	struct allowedips_node *stack[128], *node;
> +	unsigned int len;
> +
> +	for (len = 0, push(stack, top, len); len > 0 && (node = stack[--len]) && push(stack, node->bit[0], len) && push(stack, node->bit[1], len);)
> +		call_rcu_bh(&node->rcu, node_free_rcu);
> +}

This looks like you are doing traversal to free a tree.  The stack is there so that you do the rcu callbacks
in the proper order. Won't this create an lot of RCU work at once?