Hi~ @list I'm on my way to jump into the kernel network stack;) And I find the usage of neigh_tbl_lock might not make sense in following functions static int neigh_delete(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg) static int neigh_add(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg) They both have the following code snippets read_lock(&neigh_tbl_lock); --L1 for (tbl = neigh_tables; tbl; tbl = tbl->next) { --L2 struct neighbour *neigh; --L3 if (tbl->family != ndm->ndm_family) --L4 continue; --L5 read_unlock(&neigh_tbl_lock); --L6 if (nla_len(dst_attr) < tbl->key_len) --L7 goto out_dev_put; IMHO, there might be some race condition after L6. I mean when the L7 use the tbl, is it possible other processor releasing the table entry? Am I wrong? -- Yours sincerely Jammy