skb_transport_offset() can silently return a negative value when the transport_header becomes stale after tunnel decapsulation. A negative offset is never valid — it means transport_header points before skb->data, which will cause unsigned wraparound in any caller that assigns the result to an unsigned variable. Add a DEBUG_NET_WARN_ON_ONCE(off < 0) check so that such cases are caught early in CONFIG_DEBUG_NET=y builds (e.g., syzkaller, kernel test bots) with a full stack trace pointing to the caller, rather than silently propagating a bogus offset until something crashes downstream. Signed-off-by: Jiayuan Chen --- include/linux/skbuff.h | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index 2bcf78a4de7b..0b1aeacc25f7 100644 --- a/include/linux/skbuff.h +++ b/include/linux/skbuff.h @@ -3234,7 +3234,10 @@ static inline unsigned char *skb_checksum_start(const struct sk_buff *skb) static inline int skb_transport_offset(const struct sk_buff *skb) { - return skb_transport_header(skb) - skb->data; + int off = skb_transport_header(skb) - skb->data; + + DEBUG_NET_WARN_ON_ONCE(off < 0); + return off; } static inline u32 skb_network_header_len(const struct sk_buff *skb) -- 2.43.0