This patch allows to remove TUNNEL_KEY from the tunnel flags bitmap when using bpf_skb_set_tunnel_key by providing a BPF_F_NO_TUNNEL_KEY flag. On egress, the resulting tunnel header will not contain a tunnel key if the protocol and implementation supports it.
At the moment bpf_tunnel_key wants a user to specify a numeric tunnel key. This will wrap the inner packet into a tunnel header with the key bit and value set accordingly. This is problematic when using a tunnel protocol that supports optional tunnel keys and a receiving tunnel device that is not expecting packets with the key bit set. The receiver won't decapsulate and drop the packet.
RFC 2890 and RFC 2784 GRE tunnels are examples where this flag is useful. It allows for generating packets, that can be decapsulated by a GRE tunnel device not operating in collect metadata mode or not expecting the key bit set.
Signed-off-by: Christian Ehrig cehrig@cloudflare.com --- include/uapi/linux/bpf.h | 4 ++++ net/core/filter.c | 5 ++++- tools/include/uapi/linux/bpf.h | 4 ++++ 3 files changed, 12 insertions(+), 1 deletion(-)
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h index 464ca3f01fe7..bc1a3d232ae4 100644 --- a/include/uapi/linux/bpf.h +++ b/include/uapi/linux/bpf.h @@ -2001,6 +2001,9 @@ union bpf_attr { * sending the packet. This flag was added for GRE * encapsulation, but might be used with other protocols * as well in the future. + * **BPF_F_NO_TUNNEL_KEY** + * Add a flag to tunnel metadata indicating that no tunnel + * key should be set in the resulting tunnel header. * * Here is a typical usage on the transmit path: * @@ -5764,6 +5767,7 @@ enum { BPF_F_ZERO_CSUM_TX = (1ULL << 1), BPF_F_DONT_FRAGMENT = (1ULL << 2), BPF_F_SEQ_NUMBER = (1ULL << 3), + BPF_F_NO_TUNNEL_KEY = (1ULL << 4), };
/* BPF_FUNC_skb_get_tunnel_key flags. */ diff --git a/net/core/filter.c b/net/core/filter.c index 929358677183..c746e4d77214 100644 --- a/net/core/filter.c +++ b/net/core/filter.c @@ -4615,7 +4615,8 @@ BPF_CALL_4(bpf_skb_set_tunnel_key, struct sk_buff *, skb, struct ip_tunnel_info *info;
if (unlikely(flags & ~(BPF_F_TUNINFO_IPV6 | BPF_F_ZERO_CSUM_TX | - BPF_F_DONT_FRAGMENT | BPF_F_SEQ_NUMBER))) + BPF_F_DONT_FRAGMENT | BPF_F_SEQ_NUMBER | + BPF_F_NO_TUNNEL_KEY))) return -EINVAL; if (unlikely(size != sizeof(struct bpf_tunnel_key))) { switch (size) { @@ -4653,6 +4654,8 @@ BPF_CALL_4(bpf_skb_set_tunnel_key, struct sk_buff *, skb, info->key.tun_flags &= ~TUNNEL_CSUM; if (flags & BPF_F_SEQ_NUMBER) info->key.tun_flags |= TUNNEL_SEQ; + if (flags & BPF_F_NO_TUNNEL_KEY) + info->key.tun_flags &= ~TUNNEL_KEY;
info->key.tun_id = cpu_to_be64(from->tunnel_id); info->key.tos = from->tunnel_tos; diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h index 464ca3f01fe7..bc1a3d232ae4 100644 --- a/tools/include/uapi/linux/bpf.h +++ b/tools/include/uapi/linux/bpf.h @@ -2001,6 +2001,9 @@ union bpf_attr { * sending the packet. This flag was added for GRE * encapsulation, but might be used with other protocols * as well in the future. + * **BPF_F_NO_TUNNEL_KEY** + * Add a flag to tunnel metadata indicating that no tunnel + * key should be set in the resulting tunnel header. * * Here is a typical usage on the transmit path: * @@ -5764,6 +5767,7 @@ enum { BPF_F_ZERO_CSUM_TX = (1ULL << 1), BPF_F_DONT_FRAGMENT = (1ULL << 2), BPF_F_SEQ_NUMBER = (1ULL << 3), + BPF_F_NO_TUNNEL_KEY = (1ULL << 4), };
/* BPF_FUNC_skb_get_tunnel_key flags. */