Re: [PATCH v6 3/6] bpf: add BPF_F_ADJ_ROOM_DECAP_* flags for tunnel decapsulation

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The documented API behavior fails to clear tunnel modifier GSO flags, leading to SKB GSO state corruption.
--

commit b28df908f04d4d037671bd0f85e9d43e6c3ddef9
Author: Nick Hudson <nhudson@akamai.com>

bpf: add BPF_F_ADJ_ROOM_DECAP_* flags for tunnel decapsulation

This commit adds new bpf_skb_adjust_room() decapsulation flags to describe
which tunnel layer is being removed. This allows later changes to update
tunnel-related GSO state accordingly during decapsulation.

> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
[ ... ]
> + *		* **BPF_F_ADJ_ROOM_DECAP_L4_UDP**:
> + *		  Clear UDP tunnel GSO flags (SKB_GSO_UDP_TUNNEL and
> + *		  SKB_GSO_UDP_TUNNEL_CSUM) when decapsulating a UDP tunnel.

If a tunnel packet has remote checksum offload, does this documentation
imply that tunnel modifier flags like SKB_GSO_TUNNEL_REMCSUM and
SKB_GSO_PARTIAL are left intact when the base UDP flags are cleared?

[ ... ]
> + *		When using the decapsulation flags above, the skb->encapsulation
> + *		flag is automatically cleared if all tunnel-specific GSO flags
> + *		(SKB_GSO_UDP_TUNNEL, SKB_GSO_UDP_TUNNEL_CSUM, SKB_GSO_GRE,
> + *		SKB_GSO_GRE_CSUM, SKB_GSO_IPXIP4, SKB_GSO_IPXIP6) have been
> + *		removed from the packet. This handles cases where all tunnel
> + *		layers have been decapsulated.

If the base tunnel flags are cleared but tunnel modifier flags like
SKB_GSO_TUNNEL_REMCSUM remain, will the packet be left in an invalid
state where skb->encapsulation is 0 but tunnel offload features are
still present? This might cause the network stack's segmentation code to
process a packet with tunnel offload features but no actual tunnel.

Also, SKB_GSO_ESP appears to be missing from the list of tunnel-specific
GSO flags that prevent clearing skb->encapsulation. Could this result in
skb->encapsulation being cleared prematurely if an ESP tunnel is still
present?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260504101759.3319427-1-nhudson@akamai.com?part=3