Re: [PATCH net] mptcp: fix race in mptcp_pm_nl_flush_addrs_doit()

[Mat Martineau <martineau@kernel.org>]

[-- Attachment #1: Type: text/plain, Size: 3412 bytes --]

On Thu, 22 Jan 2026, Paolo Abeni wrote:

> On 1/22/26 2:54 PM, Eric Dumazet wrote:
>> On Thu, Jan 22, 2026 at 2:13 PM Eric Dumazet <edumazet@google.com> wrote:
>>>
>>> syzbot and Eulgyu Kim reported crashes in mptcp_pm_nl_get_local_id()
>>> and/or mptcp_pm_nl_is_backup()
>>>
>>> Root cause is list_splice_init() in mptcp_pm_nl_flush_addrs_doit()
>>> which is not RCU ready.
>>>
>>> list_splice_init_rcu() can not be called here while holding pernet->lock
>>> spinlock.
>>>
>>> Many thanks to Eulgyu Kim for providing a repro and testing our patches.
>>>
>>> Fixes: 141694df6573 ("mptcp: remove address when netlink flushes addrs")
>>> Signed-off-by: Eric Dumazet <edumazet@google.com>
>>> Reported-by: syzbot+5498a510ff9de39d37da@syzkaller.appspotmail.com
>>> Closes: https://lore.kernel.org/all/6970a46d.a00a0220.3ad28e.5cf0.GAE@google.com/T/
>>> Reported-by: Eulgyu Kim <eulgyukim@snu.ac.kr>
>>> Cc: Geliang Tang <geliang@kernel.org>
>>> ---
>>>  net/mptcp/pm_kernel.c | 14 +++++++++++---
>>>  1 file changed, 11 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/net/mptcp/pm_kernel.c b/net/mptcp/pm_kernel.c
>>> index 57570a44e4185370f531047fe97ce9f9fbd1480b..1a97d0eafa2b0c9e4275b90d4a576f837dc286a9 100644
>>> --- a/net/mptcp/pm_kernel.c
>>> +++ b/net/mptcp/pm_kernel.c
>>> @@ -1294,16 +1294,24 @@ static void __reset_counters(struct pm_nl_pernet *pernet)
>>>  int mptcp_pm_nl_flush_addrs_doit(struct sk_buff *skb, struct genl_info *info)
>>>  {
>>>         struct pm_nl_pernet *pernet = genl_info_pm_nl(info);
>>> -       LIST_HEAD(free_list);
>>> +       struct list_head free_list;
>>>
>>>         spin_lock_bh(&pernet->lock);
>>> -       list_splice_init(&pernet->endp_list, &free_list);
>>> +
>>> +       free_list = pernet->endp_list;
>>> +       INIT_LIST_HEAD_RCU(&pernet->endp_list);
>>> +
>>>         __reset_counters(pernet);
>>>         pernet->next_id = 1;
>>>         bitmap_zero(pernet->id_bitmap, MPTCP_PM_MAX_ADDR_ID + 1);
>>>         spin_unlock_bh(&pernet->lock);
>>> -       mptcp_nl_flush_addrs_list(sock_net(skb->sk), &free_list);
>>>         synchronize_rcu();
>>
>>
>>> +
>>> +       /* Adjust the pointers to free_list instead of pernet->endp_list */
>>> +       free_list.prev->next = &free_list;
>>> +       free_list.next->prev = &free_list;
>>
>>
>> We have to test if the list was empty, and avoid the synchronize_rcu
>> in this case.
>>
>> I will squash in V2, unless someone complains.
>>
>> diff --git a/net/mptcp/pm_kernel.c b/net/mptcp/pm_kernel.c
>> index 1a97d0eafa2b0c9e4275b90d4a576f837dc286a9..af23be6658ded4860133bb9495c7738014815d28
>> 100644
>> --- a/net/mptcp/pm_kernel.c
>> +++ b/net/mptcp/pm_kernel.c
>> @@ -1305,6 +1305,10 @@ int mptcp_pm_nl_flush_addrs_doit(struct sk_buff
>> *skb, struct genl_info *info)
>>         pernet->next_id = 1;
>>         bitmap_zero(pernet->id_bitmap, MPTCP_PM_MAX_ADDR_ID + 1);
>>         spin_unlock_bh(&pernet->lock);
>> +
>> +       if (free_list.next == &pernet->endp_list)
>> +               return 0;
>> +
>>         synchronize_rcu();
>>
>>         /* Adjust the pointers to free_list instead of pernet->endp_list */
>>
>
> LGTM, thanks Eric!

Likewise, I think the patch (plus squash) is good, no complaints. I'll 
watch for v2.

>
> Side note: I think is very busy elsewhere and this could go directly via
> the net tree.

I agree, straight to the net tree would be best in this case.


Thanks Eric and Paolo,


- Mat