skmsg: Move sk_redir from TCP_SKB_CB to skb

Currently TCP_SKB_CB() is hard-coded in skmsg code, it certainly
does not work for any other non-TCP protocols. We can move them to
skb ext, but it introduces a memory allocation on fast path.

Fortunately, we only need to a word-size to store all the information,
because the flags actually only contains 1 bit so can be just packed
into the lowest bit of the "pointer", which is stored as unsigned
long.

Inside struct sk_buff, '_skb_refdst' can be reused because skb dst is
no longer needed after ->sk_data_ready() so we can just drop it.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Acked-by: Jakub Sitnicki <jakub@cloudflare.com>
Link: https://lore.kernel.org/bpf/20210223184934.6054-5-xiyou.wangcong@gmail.com

include/linux/skbuff.h

@@ -755,6 +755,9 @@ struct sk_buff {
 			void		(*destructor)(struct sk_buff *skb);
 		};
 		struct list_head	tcp_tsorted_anchor;
+#ifdef CONFIG_NET_SOCK_MSG
+		unsigned long		_sk_redir;
+#endif
 	};
 
 #if defined(CONFIG_NF_CONNTRACK) || defined(CONFIG_NF_CONNTRACK_MODULE)

include/linux/skmsg.h

@@ -455,4 +455,42 @@ static inline bool sk_psock_strp_enabled(struct sk_psock *psock)
 		return false;
 	return !!psock->saved_data_ready;
 }
+
+#if IS_ENABLED(CONFIG_NET_SOCK_MSG)
+
+/* We only have one bit so far. */
+#define BPF_F_PTR_MASK ~(BPF_F_INGRESS)
+
+static inline bool skb_bpf_ingress(const struct sk_buff *skb)
+{
+	unsigned long sk_redir = skb->_sk_redir;
+
+	return sk_redir & BPF_F_INGRESS;
+}
+
+static inline void skb_bpf_set_ingress(struct sk_buff *skb)
+{
+	skb->_sk_redir |= BPF_F_INGRESS;
+}
+
+static inline void skb_bpf_set_redir(struct sk_buff *skb, struct sock *sk_redir,
+				     bool ingress)
+{
+	skb->_sk_redir = (unsigned long)sk_redir;
+	if (ingress)
+		skb->_sk_redir |= BPF_F_INGRESS;
+}
+
+static inline struct sock *skb_bpf_redirect_fetch(const struct sk_buff *skb)
+{
+	unsigned long sk_redir = skb->_sk_redir;
+
+	return (struct sock *)(sk_redir & BPF_F_PTR_MASK);
+}
+
+static inline void skb_bpf_redirect_clear(struct sk_buff *skb)
+{
+	skb->_sk_redir = 0;
+}
+#endif /* CONFIG_NET_SOCK_MSG */
 #endif /* _LINUX_SKMSG_H */

include/net/tcp.h

@@ -883,30 +883,11 @@ struct tcp_skb_cb {
 			struct inet6_skb_parm	h6;
 #endif
 		} header;	/* For incoming skbs */
-		struct {
-			__u32 flags;
-			struct sock *sk_redir;
-		} bpf;
 	};
 };
 
 #define TCP_SKB_CB(__skb)	((struct tcp_skb_cb *)&((__skb)->cb[0]))
 
-static inline bool tcp_skb_bpf_ingress(const struct sk_buff *skb)
-{
-	return TCP_SKB_CB(skb)->bpf.flags & BPF_F_INGRESS;
-}
-
-static inline struct sock *tcp_skb_bpf_redirect_fetch(struct sk_buff *skb)
-{
-	return TCP_SKB_CB(skb)->bpf.sk_redir;
-}
-
-static inline void tcp_skb_bpf_redirect_clear(struct sk_buff *skb)
-{
-	TCP_SKB_CB(skb)->bpf.sk_redir = NULL;
-}
-
 extern const struct inet_connection_sock_af_ops ipv4_specific;
 
 #if IS_ENABLED(CONFIG_IPV6)

net/core/skmsg.c

@@ -525,7 +525,8 @@ static void sk_psock_backlog(struct work_struct *work)
 		len = skb->len;
 		off = 0;
 start:
-		ingress = tcp_skb_bpf_ingress(skb);
+		ingress = skb_bpf_ingress(skb);
+		skb_bpf_redirect_clear(skb);
 		do {
 			ret = -EIO;
 			if (likely(psock->sk->sk_socket))
@@ -631,7 +632,12 @@ void __sk_psock_purge_ingress_msg(struct sk_psock *psock)
 
 static void sk_psock_zap_ingress(struct sk_psock *psock)
 {
-	__skb_queue_purge(&psock->ingress_skb);
+	struct sk_buff *skb;
+
+	while ((skb = __skb_dequeue(&psock->ingress_skb)) != NULL) {
+		skb_bpf_redirect_clear(skb);
+		kfree_skb(skb);
+	}
 	__sk_psock_purge_ingress_msg(psock);
 }
 
@@ -754,7 +760,7 @@ static void sk_psock_skb_redirect(struct sk_buff *skb)
 	struct sk_psock *psock_other;
 	struct sock *sk_other;
 
-	sk_other = tcp_skb_bpf_redirect_fetch(skb);
+	sk_other = skb_bpf_redirect_fetch(skb);
 	/* This error is a buggy BPF program, it returned a redirect
 	 * return code, but then didn't set a redirect interface.
 	 */
@@ -804,9 +810,10 @@ int sk_psock_tls_strp_read(struct sk_psock *psock, struct sk_buff *skb)
 		 * TLS context.
 		 */
 		skb->sk = psock->sk;
-		tcp_skb_bpf_redirect_clear(skb);
+		skb_dst_drop(skb);
+		skb_bpf_redirect_clear(skb);
 		ret = sk_psock_bpf_run(psock, prog, skb);
-		ret = sk_psock_map_verd(ret, tcp_skb_bpf_redirect_fetch(skb));
+		ret = sk_psock_map_verd(ret, skb_bpf_redirect_fetch(skb));
 		skb->sk = NULL;
 	}
 	sk_psock_tls_verdict_apply(skb, psock->sk, ret);
@@ -818,7 +825,6 @@ EXPORT_SYMBOL_GPL(sk_psock_tls_strp_read);
 static void sk_psock_verdict_apply(struct sk_psock *psock,
 				   struct sk_buff *skb, int verdict)
 {
-	struct tcp_skb_cb *tcp;
 	struct sock *sk_other;
 	int err = -EIO;
 
@@ -830,8 +836,7 @@ static void sk_psock_verdict_apply(struct sk_psock *psock,
 			goto out_free;
 		}
 
-		tcp = TCP_SKB_CB(skb);
+		skb_bpf_set_ingress(skb);
-		tcp->bpf.flags |= BPF_F_INGRESS;
 
 		/* If the queue is empty then we can submit directly
 		 * into the msg queue. If its not empty we have to
@@ -892,9 +897,10 @@ static void sk_psock_strp_read(struct strparser *strp, struct sk_buff *skb)
 	skb_set_owner_r(skb, sk);
 	prog = READ_ONCE(psock->progs.skb_verdict);
 	if (likely(prog)) {
-		tcp_skb_bpf_redirect_clear(skb);
+		skb_dst_drop(skb);
+		skb_bpf_redirect_clear(skb);
 		ret = sk_psock_bpf_run(psock, prog, skb);
-		ret = sk_psock_map_verd(ret, tcp_skb_bpf_redirect_fetch(skb));
+		ret = sk_psock_map_verd(ret, skb_bpf_redirect_fetch(skb));
 	}
 	sk_psock_verdict_apply(psock, skb, ret);
 out:
@@ -1011,9 +1017,10 @@ static int sk_psock_verdict_recv(read_descriptor_t *desc, struct sk_buff *skb,
 	skb_set_owner_r(skb, sk);
 	prog = READ_ONCE(psock->progs.skb_verdict);
 	if (likely(prog)) {
-		tcp_skb_bpf_redirect_clear(skb);
+		skb_dst_drop(skb);
+		skb_bpf_redirect_clear(skb);
 		ret = sk_psock_bpf_run(psock, prog, skb);
-		ret = sk_psock_map_verd(ret, tcp_skb_bpf_redirect_fetch(skb));
+		ret = sk_psock_map_verd(ret, skb_bpf_redirect_fetch(skb));
 	}
 	sk_psock_verdict_apply(psock, skb, ret);
 out:

net/core/sock_map.c

@@ -657,7 +657,6 @@ const struct bpf_func_proto bpf_sock_map_update_proto = {
 BPF_CALL_4(bpf_sk_redirect_map, struct sk_buff *, skb,
 	   struct bpf_map *, map, u32, key, u64, flags)
 {
-	struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
 	struct sock *sk;
 
 	if (unlikely(flags & ~(BPF_F_INGRESS)))
@@ -667,8 +666,7 @@ BPF_CALL_4(bpf_sk_redirect_map, struct sk_buff *, skb,
 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
 		return SK_DROP;
 
-	tcb->bpf.flags = flags;
+	skb_bpf_set_redir(skb, sk, flags & BPF_F_INGRESS);
-	tcb->bpf.sk_redir = sk;
 	return SK_PASS;
 }
 
@@ -1250,7 +1248,6 @@ const struct bpf_func_proto bpf_sock_hash_update_proto = {
 BPF_CALL_4(bpf_sk_redirect_hash, struct sk_buff *, skb,
 	   struct bpf_map *, map, void *, key, u64, flags)
 {
-	struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
 	struct sock *sk;
 
 	if (unlikely(flags & ~(BPF_F_INGRESS)))
@@ -1260,8 +1257,7 @@ BPF_CALL_4(bpf_sk_redirect_hash, struct sk_buff *, skb,
 	if (unlikely(!sk || !sock_map_redirect_allowed(sk)))
 		return SK_DROP;
 
-	tcb->bpf.flags = flags;
+	skb_bpf_set_redir(skb, sk, flags & BPF_F_INGRESS);
-	tcb->bpf.sk_redir = sk;
 	return SK_PASS;
 }