Re: [PATCH net v2] tcp: do not send a SYNACK to a broadcast or multicast address

[Jiayuan Chen <jiayuan.chen@linux.dev>]

On 10/8/26 7:49 PM, Theodor Arsenij Larionov Trichkine wrote:
> tcp_v4_conn_request() drops a SYN sent to a broadcast or multicast
> address. The SYNACK route has no such check. A SYN with a multicast
> source address can reach a listener when it is looped back with its
> dst attached: ip_rcv_finish_core() then skips ip_route_input_noref()
> and its martian source check. A raw IP_HDRINCL socket sending to a
> local address does this, as does re-injection by nft dup or the
> iptables TEE target.
>
> If the SYN destination is an address on a non-loopback device and the
> source is a group joined there (such as 224.0.0.1), the SYNACK route
> has RTCF_MULTICAST and RTCF_LOCAL set, and ip_build_and_send_pkt()
> sends it through ip_mc_output(). skb->sk of a SYNACK is the request
> socket, so sk_mc_loop() reads inet_flags past the end of it:
>
> BUG: KASAN: slab-out-of-bounds in sk_mc_loop+0x111/0x170
> Read of size 8 at addr ffff88800f2b7f10 by task repro-raw/470
> Call Trace:
>   <IRQ>
>   sk_mc_loop+0x111/0x170
>   ip_mc_output+0x355/0x930
>   ip_build_and_send_pkt+0xb87/0xc50
>   tcp_v4_send_synack+0x500/0x6f0
>   tcp_conn_request+0x2135/0x2d90
>   tcp_v4_conn_request+0xa5/0x210
>   tcp_rcv_state_process+0x136e/0x6920
>   tcp_v4_do_rcv+0x339/0xb10
>   tcp_v4_rcv+0x34ab/0x3ab0
>   ip_protocol_deliver_rcu+0x6e/0x3e0
>   ip_local_deliver_finish+0x34d/0x510
>   ip_local_deliver+0x1bc/0x310
>   ip_rcv+0x390/0x410
>   __netif_receive_skb_one_core+0x199/0x1e0
>   process_backlog+0x239/0x680
>   __napi_poll+0xb5/0x650
>   net_rx_action+0x980/0xd60
>   handle_softirqs+0x17f/0x590
>   do_softirq+0x3f/0x60
>   </IRQ>
>   <TASK>
>   __local_bh_enable_ip+0x66/0x80
>   __dev_queue_xmit+0xa65/0x3520
>   ip_finish_output2+0xb00/0x16f0
>   ip_output+0x2ad/0x4a0
>   raw_sendmsg+0x245b/0x28e0
>   inet_sendmsg+0x121/0x150
>   __sys_sendto+0x450/0x4e0
>   do_syscall_64+0xf6/0x500
>   </TASK>
>
> Allocated by task 470:
>   inet_reqsk_alloc+0x97/0x6f0
>   tcp_conn_request+0x4c6/0x2d90
>   tcp_v4_conn_request+0xa5/0x210
>
> The buggy address belongs to the object at ffff88800f2b7d60
>   which belongs to the cache request_sock_TCP of size 312
> The buggy address is located 120 bytes to the right of
>   allocated 312-byte region [ffff88800f2b7d60, ffff88800f2b7e98)
>
> Apply the same check to the SYNACK route in inet_csk_route_req().
>
> Fixes: ca6fb0651883 ("tcp: attach SYNACK messages to request sockets instead of listener")
> Suggested-by: Eric Dumazet <edumazet@kernel.org>
> Signed-off-by: Theodor Arsenij Larionov Trichkine <theodorlarionov@gmail.com>


Reviewed-by: Jiayuan Chen <jiayuan.chen@linux.dev>


> ---
> v2:
> - Reject broadcast/multicast SYNACK routes in inet_csk_route_req() (Eric Dumazet).
> - Describe how the SYN reaches the listener; add KASAN splat and repro.
> v1: https://lore.kernel.org/netdev/20261008100423.1256884-1-theodorlarionov@gmail.com/


I think SYN with 192.x.x.255 as source can also tirgger this warning 
when rp_filter is disabled.