Re: [PATCH v2 MPTCP-net] pm: userspace: fix address ID overflow when all IDs exhausted

Hi luoqing,

(-cc David/Paolo)

Thank you for the patch.

Please only Cc the MPTCP mailing list.

Also, please send new versions in a new email thread, not as a reply to
a previous one.

One last thing: here, you can send the fix and the selftest as part of
the same series, not as separated patches, so the CI will validate the
new selftest with the fix. It is fine to have patches for -net and -next
in the same series. Also, please use 'mptcp-ne(x)t', no capital letters.

On 04/08/2026 04:23, luoqing wrote:
> From: Qing Luo <[email protected]>
> 
> When all MPTCP address IDs (1-255) are exhausted in the userspace PM,
> find_next_zero_bit() returns MPTCP_PM_MAX_ADDR_ID + 1 (256). This value
> overflows when stored in the u8 field e->addr.id, resulting in ID 0
> being stored and the entry being incorrectly added to the list.
> 
> ID 0 is reserved for the initial connection in MPTCP, so this overflow
> can cause address conflicts.
> 
> Note: the in-kernel PM already has an 'endpoints == MPTCP_PM_MAX_ADDR_ID'
> check in mptcp_pm_nl_append_new_local_addr() that returns -ERANGE before
> reaching find_next_zero_bit(), preventing this overflow. So this fix only
> addresses the userspace PM path.
> 
> Store the find_next_zero_bit() result in a temporary unsigned int, check
> against MPTCP_PM_MAX_ADDR_ID, and return -ENOSPC if all IDs are truly
> exhausted. Properly free the allocated entry with sock_kfree_s() on error.
> 
> Also modify the CSF (subflow create) handler to pass !entry.addr.id as
> the needs_id parameter. When no local ID is provided by user-space
> (entry.addr.id == 0), this triggers auto-allocation instead of silently
> using the reserved ID 0.
> 
> Fixes: 4638de5aefe5 ("mptcp: handle local addrs announced by userspace PMs")
> Assisted-by: LLM # review
> Signed-off-by: Qing Luo <[email protected]>
> ---
>  net/mptcp/pm_userspace.c | 18 +++++++++++++-----
>  1 file changed, 13 insertions(+), 5 deletions(-)
> 
> diff --git a/net/mptcp/pm_userspace.c b/net/mptcp/pm_userspace.c
> index 945aa5afc2dd..ada7d6cef625 100644
> --- a/net/mptcp/pm_userspace.c
> +++ b/net/mptcp/pm_userspace.c
> @@ -74,10 +74,17 @@ static int mptcp_userspace_pm_append_new_local_addr(struct mptcp_sock *msk,
>  			goto append_err;
>  		}
>  
> -		if (!e->addr.id && needs_id)
> -			e->addr.id = find_next_zero_bit(id_bitmap,
> -							MPTCP_PM_MAX_ADDR_ID + 1,
> -							1);
> +		if (!e->addr.id && needs_id) {
> +			unsigned int id = find_next_zero_bit(id_bitmap,
> +							     MPTCP_PM_MAX_ADDR_ID + 1,
> +							     1);
> +			if (id > MPTCP_PM_MAX_ADDR_ID) {
> +				sock_kfree_s(sk, e, sizeof(*e));
> +				ret = -ENOSPC;
> +				goto append_err;
> +			}
> +			e->addr.id = id;
> +		}
>  		list_add_tail_rcu(&e->list, &msk->pm.userspace_pm_local_addr_list);
>  		msk->pm.local_addr_used++;
>  		ret = e->addr.id;
> @@ -400,7 +407,8 @@ int mptcp_pm_nl_subflow_create_doit(struct sk_buff *skb, struct genl_info *info)
>  		goto create_err;
>  	}
>  
> -	err = mptcp_userspace_pm_append_new_local_addr(msk, &entry, false);
> +	err = mptcp_userspace_pm_append_new_local_addr(msk, &entry,
> +						       !entry.addr.id);

I don't think you can do that: it is valid to give an ID set to 0. But
it is different to give no ID.

I don't think this modification here should be part of this patch: it is
different from the overflow case you are trying to fix. It could be in
another patch, but that's changing the behaviour, and that's not a fix I
think.

Also, see my series:


https://lore.kernel.org/20260727-mptcp-pm-userspace-id0-case-v1-0-9877f02a9bae@kernel.org

*Maybe*, for -next, we could check if the ID was not set via Netlink
(and not set to 0), and pick an ID for it. *But*, it might not be a good
idea: the userspace daemon will not know the ID that has been picked,
and will need to get it via another request. That doesn't sound like a
good idea, no?

>  	if (err < 0) {
>  		NL_SET_ERR_MSG_ATTR(info->extack, laddr,
>  				    "did not match address and id");

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.