Re: [PATCH bpf v3 1/2] bpf: reject incompatible socket assignments

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] bpf: bpf_sk_assign bypasses family check for encapsulated packets
--

commit bd75dbe60b27d785e5799901fce33eb038028e20
Author: Shihuang Liu <shlomojune6@gmail.com>

bpf: reject incompatible socket assignments

This commit fixes an out-of-bounds stack buffer copy by validating socket
and packet family compatibility in bpf_sk_assign(). It ensures IPv4/IPv6
socket mappings match the packet protocol, maintaining compatibility for
dual-stack and mapped sockets.

> diff --git a/net/core/filter.c b/net/core/filter.c
> index 70dc621672f2e..173349bd2a3f9 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c

[ ... ]

> @@ -3491,6 +3491,64 @@ static int bpf_skb_proto_xlat(struct sk_buff *skb, __be16 to_proto)
>  	return -ENOTSUPP;
>  }
>  
> +static bool bpf_sk_assign_family_ok_proto(const struct sock *sk, __be16 proto)
> +{
> +	const struct inet_connection_sock_af_ops *af_ops;
> +	unsigned short family;
> +
> +	switch (proto) {
> +	case htons(ETH_P_IP):
> +		family = AF_INET;
> +		break;
> +	case htons(ETH_P_IPV6):
> +		family = AF_INET6;
> +		break;
> +	default:
> +		return true;

[Severity: High]
Does this default case bypass the family check for encapsulated packets?

If a TC ingress BPF program uses bpf_sk_assign() on an encapsulated packet
(such as MPLS), the switch statement falls through to the default case
and unconditionally returns true because protocols like ETH_P_MPLS_UC are
unrecognized.

If this packet is later decapsulated (e.g., via the TC act_mpls module), its
protocol could change to IPv6 while retaining an assigned AF_INET socket. When
the networking stack processes this decapsulated IPv6 packet with the AF_INET
socket, it may interpret the IPv6 control block data as IPv4 metadata,
leading to an out-of-bounds stack access.

Can this be triggered remotely by sending encapsulated packets, defeating the
vulnerability fix for setups performing decapsulation?

> +	}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260930164020.41006-1-shlomojune6@gmail.com?part=1