bpf: support cloning sk storage on accept()

Add new helper bpf_sk_storage_clone which optionally clones sk storage
and call it from sk_clone_lock.

Cc: Martin KaFai Lau <[e-mail]>
Cc: Yonghong Song <[e-mail]>
Acked-by: Martin KaFai Lau <[e-mail]>
Acked-by: Yonghong Song <[e-mail]>
Signed-off-by: Stanislav Fomichev <[e-mail]>
Signed-off-by: Daniel Borkmann <[e-mail]>

include/net/bpf_sk_storage.h

···
 extern const struct bpf_func_proto bpf_sk_storage_get_proto;
 extern const struct bpf_func_proto bpf_sk_storage_delete_proto;
 
+#ifdef CONFIG_BPF_SYSCALL
+int bpf_sk_storage_clone(const struct sock *sk, struct sock *newsk);
+#else
+static inline int bpf_sk_storage_clone(const struct sock *sk,
+				       struct sock *newsk)
+{
+	return 0;
+}
+#endif
+
 #endif /* _BPF_SK_STORAGE_H */

include/uapi/linux/bpf.h

···
 #define BPF_F_RDONLY_PROG	(1U << 7)
 #define BPF_F_WRONLY_PROG	(1U << 8)
 
+/* Clone map from listener for newly accepted socket */
+#define BPF_F_CLONE		(1U << 9)
+
 /* flags for BPF_PROG_QUERY */
 #define BPF_F_QUERY_EFFECTIVE	(1U << 0)
 

net/core/bpf_sk_storage.c

···
 
 static atomic_t cache_idx;
 
+#define SK_STORAGE_CREATE_FLAG_MASK					\
+	(BPF_F_NO_PREALLOC | BPF_F_CLONE)
+
 struct bucket {
 	struct hlist_head list;
 	raw_spinlock_t lock;
···
 		kfree_rcu(sk_storage, rcu);
 }
 
-/* sk_storage->lock must be held and sk_storage->list cannot be empty */
 static void __selem_link_sk(struct bpf_sk_storage *sk_storage,
 			    struct bpf_sk_storage_elem *selem)
 {
···
 	return 0;
 }
 
-/* Called by __sk_destruct() */
+/* Called by __sk_destruct() & bpf_sk_storage_clone() */
 void bpf_sk_storage_free(struct sock *sk)
 {
 	struct bpf_sk_storage_elem *selem;
···
 
 	smap = (struct bpf_sk_storage_map *)map;
 
+	/* Note that this map might be concurrently cloned from
+	 * bpf_sk_storage_clone. Wait for any existing bpf_sk_storage_clone
+	 * RCU read section to finish before proceeding. New RCU
+	 * read sections should be prevented via bpf_map_inc_not_zero.
+	 */
 	synchronize_rcu();
 
 	/* bpf prog and the userspace can no longer access this map
···
 
 static int bpf_sk_storage_map_alloc_check(union bpf_attr *attr)
 {
-	if (attr->map_flags != BPF_F_NO_PREALLOC || attr->max_entries ||
+	if (attr->map_flags & ~SK_STORAGE_CREATE_FLAG_MASK ||
+	    !(attr->map_flags & BPF_F_NO_PREALLOC) ||
+	    attr->max_entries ||
 	    attr->key_size != sizeof(int) || !attr->value_size ||
 	    /* Enforce BTF for userspace sk dumping */
 	    !attr->btf_key_type_id || !attr->btf_value_type_id)
···
 	}
 
 	return err;
+}
+
+static struct bpf_sk_storage_elem *
+bpf_sk_storage_clone_elem(struct sock *newsk,
+			  struct bpf_sk_storage_map *smap,
+			  struct bpf_sk_storage_elem *selem)
+{
+	struct bpf_sk_storage_elem *copy_selem;
+
+	copy_selem = selem_alloc(smap, newsk, NULL, true);
+	if (!copy_selem)
+		return NULL;
+
+	if (map_value_has_spin_lock(&smap->map))
+		copy_map_value_locked(&smap->map, SDATA(copy_selem)->data,
+				      SDATA(selem)->data, true);
+	else
+		copy_map_value(&smap->map, SDATA(copy_selem)->data,
+			       SDATA(selem)->data);
+
+	return copy_selem;
+}
+
+int bpf_sk_storage_clone(const struct sock *sk, struct sock *newsk)
+{
+	struct bpf_sk_storage *new_sk_storage = NULL;
+	struct bpf_sk_storage *sk_storage;
+	struct bpf_sk_storage_elem *selem;
+	int ret = 0;
+
+	RCU_INIT_POINTER(newsk->sk_bpf_storage, NULL);
+
+	rcu_read_lock();
+	sk_storage = rcu_dereference(sk->sk_bpf_storage);
+
+	if (!sk_storage || hlist_empty(&sk_storage->list))
+		goto out;
+
+	hlist_for_each_entry_rcu(selem, &sk_storage->list, snode) {
+		struct bpf_sk_storage_elem *copy_selem;
+		struct bpf_sk_storage_map *smap;
+		struct bpf_map *map;
+
+		smap = rcu_dereference(SDATA(selem)->smap);
+		if (!(smap->map.map_flags & BPF_F_CLONE))
+			continue;
+
+		/* Note that for lockless listeners adding new element
+		 * here can race with cleanup in bpf_sk_storage_map_free.
+		 * Try to grab map refcnt to make sure that it's still
+		 * alive and prevent concurrent removal.
+		 */
+		map = bpf_map_inc_not_zero(&smap->map, false);
+		if (IS_ERR(map))
+			continue;
+
+		copy_selem = bpf_sk_storage_clone_elem(newsk, smap, selem);
+		if (!copy_selem) {
+			ret = -ENOMEM;
+			bpf_map_put(map);
+			goto out;
+		}
+
+		if (new_sk_storage) {
+			selem_link_map(smap, copy_selem);
+			__selem_link_sk(new_sk_storage, copy_selem);
+		} else {
+			ret = sk_storage_alloc(newsk, smap, copy_selem);
+			if (ret) {
+				kfree(copy_selem);
+				atomic_sub(smap->elem_size,
+					   &newsk->sk_omem_alloc);
+				bpf_map_put(map);
+				goto out;
+			}
+
+			new_sk_storage = rcu_dereference(copy_selem->sk_storage);
+		}
+		bpf_map_put(map);
+	}
+
+out:
+	rcu_read_unlock();
+
+	/* In case of an error, don't free anything explicitly here, the
+	 * caller is responsible to call bpf_sk_storage_free.
+	 */
+
+	return ret;
 }
 
 BPF_CALL_4(bpf_sk_storage_get, struct bpf_map *, map, struct sock *, sk,

net/core/sock.c

···
 			goto out;
 		}
 		RCU_INIT_POINTER(newsk->sk_reuseport_cb, NULL);
-#ifdef CONFIG_BPF_SYSCALL
-		RCU_INIT_POINTER(newsk->sk_bpf_storage, NULL);
-#endif
+
+		if (bpf_sk_storage_clone(sk, newsk)) {
+			sk_free_unlock_clone(newsk);
+			newsk = NULL;
+			goto out;
+		}
 
 		newsk->sk_err	   = 0;
 		newsk->sk_err_soft = 0;