[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <5984EDF7.3060808@gmail.com>
Date: Fri, 04 Aug 2017 14:58:15 -0700
From: John Fastabend <john.fastabend@...il.com>
To: davem@...emloft.net, daniel@...earbox.net,
alexander.levin@...izon.com
CC: netdev@...r.kernel.org
Subject: Re: [net-next PATCH] net: devmap fix mutex in rcu critical section
On 08/04/2017 02:21 PM, John Fastabend wrote:
> Originally we used a mutex to protect concurrent devmap update
> and delete operations from racing with netdev unregister notifier
> callbacks.
>
[...]
> }
> @@ -396,22 +385,20 @@ static int dev_map_notification(struct notifier_block *notifier,
>
Daniel reminds me this is not in a rcu_read_lock/unlock() section as
needed, so v2 on its way. Thanks!
> switch (event) {
> case NETDEV_UNREGISTER:
> - mutex_lock(&dev_map_list_mutex);
rcu_read_lock();
> list_for_each_entry(dtab, &dev_map_list, list) {
> for (i = 0; i < dtab->map.max_entries; i++) {
> - struct bpf_dtab_netdev *dev;
> + struct bpf_dtab_netdev *dev, *odev;
>
> - dev = dtab->netdev_map[i];
> + dev = READ_ONCE(dtab->netdev_map[i]);
> if (!dev ||
> dev->dev->ifindex != netdev->ifindex)
> continue;
> - dev = xchg(&dtab->netdev_map[i], NULL);
> - if (dev)
> + odev = cmpxchg(&dtab->netdev_map[i], dev, NULL);
> + if (dev == odev)
> call_rcu(&dev->rcu,
> __dev_map_entry_free);
> }
> }
rcu_read_unlock();
> - mutex_unlock(&dev_map_list_mutex);
> break;
> default:
> break;
>
Powered by blists - more mailing lists