netdev - Re: [PATCH v2] net: fix a concurrency bug in l2tp_tunnel

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] [day] [month] [year] [list]

Message-ID: <9860EA3C-374D-41D6-B0E6-58B448D4348A@purdue.edu>
Date:   Wed, 21 Apr 2021 19:21:20 +0000
From:   "Gong, Sishuai" <sishuai@...due.edu>
To:     Cong Wang <xiyou.wangcong@...il.com>
CC:     David Miller <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        "tparkin@...alix.com" <tparkin@...alix.com>,
        Matthias Schiffer <mschiffer@...verse-factory.net>,
        Linux Kernel Network Developers <netdev@...r.kernel.org>
Subject: Re: [PATCH v2] net: fix a concurrency bug in l2tp_tunnel_register()

On Apr 16, 2021, at 1:21 PM, Cong Wang <xiyou.wangcong@...il.com> wrote:
> 
> On Thu, Apr 15, 2021 at 7:18 AM Sishuai Gong <sishuai@...due.edu> wrote:
>> diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
>> index 203890e378cb..879f1264ec3c 100644
>> --- a/net/l2tp/l2tp_core.c
>> +++ b/net/l2tp/l2tp_core.c
>> @@ -1478,6 +1478,9 @@ int l2tp_tunnel_register(struct l2tp_tunnel *tunnel, struct net *net,
>>        tunnel->l2tp_net = net;
>>        pn = l2tp_pernet(net);
>> 
>> +       sk = sock->sk;
>> +       tunnel->sock = sk;
>> +
>>        spin_lock_bh(&pn->l2tp_tunnel_list_lock);
>>        list_for_each_entry(tunnel_walk, &pn->l2tp_tunnel_list, list) {
>>                if (tunnel_walk->tunnel_id == tunnel->tunnel_id) {
>> @@ -1490,9 +1493,7 @@ int l2tp_tunnel_register(struct l2tp_tunnel *tunnel, struct net *net,
>>        list_add_rcu(&tunnel->list, &pn->l2tp_tunnel_list);
>>        spin_unlock_bh(&pn->l2tp_tunnel_list_lock);
>> 
>> -       sk = sock->sk;
>>        sock_hold(sk);
>> -       tunnel->sock = sk;
> 
> I think you have to hold this refcnt before making tunnel->sock visible
> to others.
> 
> Why not just move this together and simply release the refcnt on error
> path?
Thanks. It totally makes sense. I am going to send the next version.
> 
> diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
> index 203890e378cb..8eb805ee18d4 100644
> --- a/net/l2tp/l2tp_core.c
> +++ b/net/l2tp/l2tp_core.c
> @@ -1478,11 +1478,15 @@ int l2tp_tunnel_register(struct l2tp_tunnel
> *tunnel, struct net *net,
>        tunnel->l2tp_net = net;
>        pn = l2tp_pernet(net);
> 
> +       sk = sock->sk;
> +       sock_hold(sk);
> +       tunnel->sock = sk;
> +
>        spin_lock_bh(&pn->l2tp_tunnel_list_lock);
>        list_for_each_entry(tunnel_walk, &pn->l2tp_tunnel_list, list) {
>                if (tunnel_walk->tunnel_id == tunnel->tunnel_id) {
>                        spin_unlock_bh(&pn->l2tp_tunnel_list_lock);
> -
> +                       sock_put(sk);
>                        ret = -EEXIST;
>                        goto err_sock;
>                }
> @@ -1490,10 +1494,6 @@ int l2tp_tunnel_register(struct l2tp_tunnel
> *tunnel, struct net *net,
>        list_add_rcu(&tunnel->list, &pn->l2tp_tunnel_list);
>        spin_unlock_bh(&pn->l2tp_tunnel_list_lock);
> 
> -       sk = sock->sk;
> -       sock_hold(sk);
> -       tunnel->sock = sk;
> -
>        if (tunnel->encap == L2TP_ENCAPTYPE_UDP) {
>                struct udp_tunnel_sock_cfg udp_cfg = {
>                        .sk_user_data = tunnel,