netdev - Re: Netlink socket leaks

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CANaxB-zvU0St+TkniHu9AKsW0Pd_j3BK17EfzCB+ghmDC0U22g@mail.gmail.com>
Date:	Sat, 2 May 2015 02:31:09 +0300
From:	Andrey Wagin <avagin@...il.com>
To:	netdev@...r.kernel.org, Herbert Xu <herbert@...dor.apana.org.au>
Subject: Re: Netlink socket leaks

2015-05-01 13:22 GMT+03:00 Andrey Wagin <avagin@...il.com>:
> Hi Herbert,
>
> I've found that netlink sockets start leaking after your patch:
> commit c428ecd1a21f1457ca3beb4df71b8a079c410e41
> Author: Herbert Xu <herbert@...dor.apana.org.au>
> Date:   Fri Mar 20 21:57:01 2015 +1100
>
>     netlink: Move namespace into hash key
>
> Could you look at this?

A socket leaks if it is released by sk_release_kernel(). The problem
is that netlink_insert() and netlink_remove() is called when a socket
has different values of sk->sk_net.

The following patch fixes this problem, but I am not sure that it
breaks nothing else.
diff --git a/net/core/sock.c b/net/core/sock.c
index e891bcf..292f422 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -1474,8 +1474,8 @@ void sk_release_kernel(struct sock *sk)
                return;

        sock_hold(sk);
-       sock_net_set(sk, get_net(&init_net));
        sock_release(sk->sk_socket);
+       sock_net_set(sk, get_net(&init_net));
        sock_put(sk);
 }
 EXPORT_SYMBOL(sk_release_kernel);

>
> It can be reproduced by the following script:
> ss -a | wc -l; ip netns add test && ip netns exec test ip a && ip
> netns del test ; ss -a | wc -l
>
> [root@...kins ~]# uname -a
> Linux jenkins.criu.org 4.1.0-rc1-00117-g4a152c3 #46 SMP Fri May 1
> 14:09:02 MSK 2015 x86_64 x86_64 x86_64 GNU/Linux
>
>
> I have applied a patch with debug messages:
> diff --git a/net/netlink/af_netlink.c b/net/netlink/af_netlink.c
> index 72c6b55..79aece4 100644
> --- a/net/netlink/af_netlink.c
> +++ b/net/netlink/af_netlink.c
> @@ -1081,7 +1081,9 @@ static int netlink_insert(struct sock *sk, u32 portid)
>                 if (err == -EEXIST)
>                         err = -EADDRINUSE;
>                 sock_put(sk);
> +               printk("%s:%d: %p\n", __func__, __LINE__, sk);
>         }
> +       printk("%s:%d: %p\n", __func__, __LINE__, sk);
>
>  err:
>         release_sock(sk);
> @@ -1097,7 +1099,9 @@ static void netlink_remove(struct sock *sk)
>                                     netlink_rhashtable_params)) {
>                 WARN_ON(atomic_read(&sk->sk_refcnt) == 1);
>                 __sock_put(sk);
> +               printk("%s:%d: %p\n", __func__, __LINE__, sk);
>         }
> +       printk("%s:%d: %p\n", __func__, __LINE__, sk);
>
>         netlink_table_grab();
>         if (nlk_sk(sk)->subscriptions) {
>
> And here is the output
> [ 3868.202260] netlink_insert:1086: ffff8800d8c14000
> [ 3868.202962] netlink_insert:1086: ffff8800d8c15800
> [ 3868.202995] netlink_insert:1086: ffff8800d8c12000
> [ 3868.203228] netlink_insert:1086: ffff8800d8c16000
> [ 3868.203360] netlink_insert:1086: ffff8800d8c11000
> [ 3868.203920] netlink_insert:1086: ffff8800d8c13000
> [ 3868.203951] netlink_insert:1086: ffff8800d8c11800
> [ 3868.204216] netlink_insert:1086: ffff8800d8c17000
> [ 3868.205846] netlink_remove:1102: ffff8800d8c14000
> [ 3868.205849] netlink_remove:1104: ffff8800d8c14000
> [ 3868.208270] netlink_insert:1086: ffff8800d81d3800
> [ 3868.215363] netlink_remove:1102: ffff8800d81d3800
> [ 3868.215368] netlink_remove:1104: ffff8800d81d3800
> [ 3868.216465] netlink_insert:1086: ffff8800d8c0a000
> [ 3868.226320] netlink_remove:1102: ffff8800d8c0a000
> [ 3868.226324] netlink_remove:1104: ffff8800d8c0a000
> [ 3868.230662] netlink_insert:1086: ffff8800d8c0e800
> [ 3868.235906] netlink_remove:1102: ffff8800d8c0e800
> [ 3868.235910] netlink_remove:1104: ffff8800d8c0e800
> [ 3868.280231] netlink_remove:1104: ffff8800d8c17000
> [ 3868.280283] netlink_remove:1104: ffff8800d8c11800
> [ 3868.286152] netlink_remove:1104: ffff8800d8c13000
> [ 3868.318148] netlink_remove:1104: ffff8800d8c11000
> [ 3868.318185] netlink_remove:1104: ffff8800d8c16000
> [ 3868.318241] netlink_remove:1104: ffff8800d8c12000
> [ 3868.318255] netlink_remove:1104: ffff8800d8c15800
>
> So it looks like
>
> netlink_insert() returns zero, but then rhashtable_remove_fast()
> returns non-zero code
>
> Thanks,
> Andrew
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html