netdev - Re: [PATCH net-next 4/4] net, neigh: Add NTF_MANAGED flag for managed neighbor entries

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]

Date:   Mon, 31 Jan 2022 22:17:53 +0100
From:   Daniel Borkmann <daniel@...earbox.net>
To:     Eric Dumazet <eric.dumazet@...il.com>, davem@...emloft.net,
        kuba@...nel.org
Cc:     roopa@...dia.com, dsahern@...nel.org, m@...bda.lt,
        john.fastabend@...il.com, netdev@...r.kernel.org,
        bpf@...r.kernel.org
Subject: Re: [PATCH net-next 4/4] net, neigh: Add NTF_MANAGED flag for managed
 neighbor entries

On 1/31/22 9:43 PM, Eric Dumazet wrote:
[...]
>> @@ -1539,6 +1564,20 @@ int neigh_direct_output(struct neighbour *neigh, struct sk_buff *skb)
>>   }
>>   EXPORT_SYMBOL(neigh_direct_output);
>> +static void neigh_managed_work(struct work_struct *work)
>> +{
>> +    struct neigh_table *tbl = container_of(work, struct neigh_table,
>> +                           managed_work.work);
>> +    struct neighbour *neigh;
>> +
>> +    write_lock_bh(&tbl->lock);
>> +    list_for_each_entry(neigh, &tbl->managed_list, managed_list)
>> +        neigh_event_send(neigh, NULL);
> 
> neigh_event_send() can need to lock tbl->lock, leading to a deadlock ?

Thanks for forwarding the syzbot report! I'll take a look.

> __raw_write_lock_bh include/linux/rwlock_api_smp.h:202 [inline]
>   _raw_write_lock_bh+0x2f/0x40 kernel/locking/spinlock.c:334
>   ___neigh_create+0x9e1/0x2990 net/core/neighbour.c:652
>   ip6_finish_output2+0x1070/0x14f0 net/ipv6/ip6_output.c:123
>   __ip6_finish_output net/ipv6/ip6_output.c:191 [inline]
>   __ip6_finish_output+0x61e/0xe90 net/ipv6/ip6_output.c:170
>   ip6_finish_output+0x32/0x200 net/ipv6/ip6_output.c:201
>   NF_HOOK_COND include/linux/netfilter.h:296 [inline]
>   ip6_output+0x1e4/0x530 net/ipv6/ip6_output.c:224
>   dst_output include/net/dst.h:451 [inline]
>   NF_HOOK include/linux/netfilter.h:307 [inline]
>   ndisc_send_skb+0xa99/0x17f0 net/ipv6/ndisc.c:508
>   ndisc_send_ns+0x3a9/0x840 net/ipv6/ndisc.c:650
>   ndisc_solicit+0x2cd/0x4f0 net/ipv6/ndisc.c:742
>   neigh_probe+0xc2/0x110 net/core/neighbour.c:1040
>   __neigh_event_send+0x37d/0x1570 net/core/neighbour.c:1201
>   neigh_event_send include/net/neighbour.h:470 [inline]
>   neigh_managed_work+0x162/0x250 net/core/neighbour.c:1574
>   process_one_work+0x9ac/0x1650 kernel/workqueue.c:2307
>   worker_thread+0x657/0x1110 kernel/workqueue.c:2454
>   kthread+0x2e9/0x3a0 kernel/kthread.c:377
>   ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:295
> 
>> +    queue_delayed_work(system_power_efficient_wq, &tbl->managed_work,
>> +               NEIGH_VAR(&tbl->parms, DELAY_PROBE_TIME));
>> +    write_unlock_bh(&tbl->lock);
>> +}
>> +
>>   static void neigh_proxy_process(struct timer_list *t)
>>   {
>>       struct neigh_table *tbl = from_timer(tbl, t, proxy_timer);
>> @@ -1685,6 +1724,8 @@ void neigh_table_init(int index, struct neigh_table *tbl)
>>       INIT_LIST_HEAD(&tbl->parms_list);
>>       INIT_LIST_HEAD(&tbl->gc_list);
>> +    INIT_LIST_HEAD(&tbl->managed_list);
>> +