[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <65f318db70f3f_3f8f5b2945b@willemb.c.googlers.com.notmuch>
Date: Thu, 14 Mar 2024 11:33:47 -0400
From: Willem de Bruijn <willemdebruijn.kernel@...il.com>
To: Eric Dumazet <edumazet@...gle.com>,
"David S . Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>,
Paolo Abeni <pabeni@...hat.com>
Cc: netdev@...r.kernel.org,
eric.dumazet@...il.com,
Eric Dumazet <edumazet@...gle.com>,
syzbot+c669c1136495a2e7c31f@...kaller.appspotmail.com,
Willem de Bruijn <willemdebruijn.kernel@...il.com>
Subject: Re: [PATCH net] packet: annotate data-races around ignore_outgoing
Eric Dumazet wrote:
> ignore_outgoing is read locklessly from dev_queue_xmit_nit()
> and packet_getsockopt()
>
> Add appropriate READ_ONCE()/WRITE_ONCE() annotations.
>
> syzbot reported:
>
> BUG: KCSAN: data-race in dev_queue_xmit_nit / packet_setsockopt
>
> write to 0xffff888107804542 of 1 bytes by task 22618 on cpu 0:
> packet_setsockopt+0xd83/0xfd0 net/packet/af_packet.c:4003
> do_sock_setsockopt net/socket.c:2311 [inline]
> __sys_setsockopt+0x1d8/0x250 net/socket.c:2334
> __do_sys_setsockopt net/socket.c:2343 [inline]
> __se_sys_setsockopt net/socket.c:2340 [inline]
> __x64_sys_setsockopt+0x66/0x80 net/socket.c:2340
> do_syscall_64+0xd3/0x1d0
> entry_SYSCALL_64_after_hwframe+0x6d/0x75
>
> read to 0xffff888107804542 of 1 bytes by task 27 on cpu 1:
> dev_queue_xmit_nit+0x82/0x620 net/core/dev.c:2248
> xmit_one net/core/dev.c:3527 [inline]
> dev_hard_start_xmit+0xcc/0x3f0 net/core/dev.c:3547
> __dev_queue_xmit+0xf24/0x1dd0 net/core/dev.c:4335
> dev_queue_xmit include/linux/netdevice.h:3091 [inline]
> batadv_send_skb_packet+0x264/0x300 net/batman-adv/send.c:108
> batadv_send_broadcast_skb+0x24/0x30 net/batman-adv/send.c:127
> batadv_iv_ogm_send_to_if net/batman-adv/bat_iv_ogm.c:392 [inline]
> batadv_iv_ogm_emit net/batman-adv/bat_iv_ogm.c:420 [inline]
> batadv_iv_send_outstanding_bat_ogm_packet+0x3f0/0x4b0 net/batman-adv/bat_iv_ogm.c:1700
> process_one_work kernel/workqueue.c:3254 [inline]
> process_scheduled_works+0x465/0x990 kernel/workqueue.c:3335
> worker_thread+0x526/0x730 kernel/workqueue.c:3416
> kthread+0x1d1/0x210 kernel/kthread.c:388
> ret_from_fork+0x4b/0x60 arch/x86/kernel/process.c:147
> ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:243
>
> value changed: 0x00 -> 0x01
>
> Reported by Kernel Concurrency Sanitizer on:
> CPU: 1 PID: 27 Comm: kworker/u8:1 Tainted: G W 6.8.0-syzkaller-08073-g480e035fc4c7 #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 02/29/2024
> Workqueue: bat_events batadv_iv_send_outstanding_bat_ogm_packet
>
> Fixes: fa788d986a3a ("packet: add sockopt to ignore outgoing packets")
> Reported-by: syzbot+c669c1136495a2e7c31f@...kaller.appspotmail.com
> Closes: https://lore.kernel.org/netdev/CANn89i+Z7MfbkBLOv=p7KZ7=K1rKHO4P1OL5LYDCtBiyqsa9oQ@mail.gmail.com/T/#t
> Signed-off-by: Eric Dumazet <edumazet@...gle.com>
> Cc: Willem de Bruijn <willemdebruijn.kernel@...il.com>
> ---
> net/core/dev.c | 2 +-
> net/packet/af_packet.c | 4 ++--
> 2 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/net/core/dev.c b/net/core/dev.c
> index 0766a245816bdf70f6609dc7b6d694ae81e7a9e5..722787c3275527f1652ec98623f61500ee753b45 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -2245,7 +2245,7 @@ void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev)
> rcu_read_lock();
> again:
> list_for_each_entry_rcu(ptype, ptype_list, list) {
> - if (ptype->ignore_outgoing)
> + if (READ_ONCE(ptype->ignore_outgoing))
> continue;
>
> /* Never send packets back to the socket
> diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
> index 61270826b9ac73e66f9011c3230d4668f0bf7c77..7cfc7d301508fcead214fbdb4e962b0553a17916 100644
> --- a/net/packet/af_packet.c
> +++ b/net/packet/af_packet.c
> @@ -4000,7 +4000,7 @@ packet_setsockopt(struct socket *sock, int level, int optname, sockptr_t optval,
> if (val < 0 || val > 1)
> return -EINVAL;
>
> - po->prot_hook.ignore_outgoing = !!val;
> + WRITE_ONCE(po->prot_hook.ignore_outgoing, !!val);
Should we also include a WRITE_ONCE on the fanout prot_hook:
match->prot_hook.ignore_outgoing = type_flags & PACKET_FANOUT_FLAG_IGNORE_OUTGOING;
> return 0;
> }
> case PACKET_TX_HAS_OFF:
> @@ -4134,7 +4134,7 @@ static int packet_getsockopt(struct socket *sock, int level, int optname,
> 0);
> break;
> case PACKET_IGNORE_OUTGOING:
> - val = po->prot_hook.ignore_outgoing;
> + val = READ_ONCE(po->prot_hook.ignore_outgoing);
> break;
> case PACKET_ROLLOVER_STATS:
> if (!po->rollover)
> --
> 2.44.0.278.ge034bb2e1d-goog
>
Powered by blists - more mailing lists