bpf, sockmap: Fix FIONREAD for sockmap

[ Upstream commit 929e30f9312514902133c45e51c79088421ab084 ]

A socket using sockmap has its own independent receive queue: ingress_msg.
This queue may contain data from its own protocol stack or from other
sockets.

Therefore, for sockmap, relying solely on copied_seq and rcv_nxt to
calculate FIONREAD is not enough.

This patch adds a new msg_tot_len field in the psock structure to record
the data length in ingress_msg. Additionally, we implement new ioctl
interfaces for TCP and UDP to intercept FIONREAD operations.

Note that we intentionally do not include sk_receive_queue data in the
FIONREAD result. Data in sk_receive_queue has not yet been processed by
the BPF verdict program, and may be redirected to other sockets or
dropped. Including it would create semantic ambiguity since this data
may never be readable by the user.

Unix and VSOCK sockets have similar issues, but fixing them is outside
the scope of this patch as it would require more intrusive changes.

Previous work by John Fastabend made some efforts towards FIONREAD support:
commit e5c6de5fa025 ("bpf, sockmap: Incorrectly handling copied_seq")
Although the current patch is based on the previous work by John Fastabend,
it is acceptable for our Fixes tag to point to the same commit.

                                                      FD1:read()
                                                      --  FD1->copied_seq++
                                                          |  [read data]
                                                          |
                                   [enqueue data]         v
                  [sockmap]     -> ingress to self ->  ingress_msg queue
FD1 native stack  ------>                                 ^
-- FD1->rcv_nxt++               -> redirect to other      | [enqueue data]
                                       |                  |
                                       |             ingress to FD1
                                       v                  ^
                                      ...                 |  [sockmap]
                                                     FD2 native stack

Fixes: 04919bed948dc ("tcp: Introduce tcp_read_skb()")
Signed-off-by: Jiayuan Chen <jiayuan.chen@linux.dev>
Reviewed-by: Jakub Sitnicki <jakub@cloudflare.com>
Link: https://lore.kernel.org/r/20260124113314.113584-3-jiayuan.chen@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

1 files changed, 66 insertions, 2 deletions

diff --git a/include/linux/skmsg.h b/include/linux/skmsg.h
index 5c5a2d65184c..e923f1c24ce4 100644
--- a/include/linux/skmsg.h
+++ b/include/linux/skmsg.h
@@ -93,6 +93,8 @@ struct sk_psock {
 	struct sk_buff_head		ingress_skb;
 	struct list_head		ingress_msg;
 	spinlock_t			ingress_lock;
+	/** @msg_tot_len: Total bytes queued in ingress_msg list. */
+	u32				msg_tot_len;
 	unsigned long			state;
 	struct list_head		link;
 	spinlock_t			link_lock;
@@ -312,6 +314,27 @@ static inline void sock_drop(struct sock *sk, struct sk_buff *skb)
 	kfree_skb(skb);
 }
 
+static inline u32 sk_psock_get_msg_len_nolock(struct sk_psock *psock)
+{
+	/* Used by ioctl to read msg_tot_len only; lock-free for performance */
+	return READ_ONCE(psock->msg_tot_len);
+}
+
+static inline void sk_psock_msg_len_add_locked(struct sk_psock *psock, int diff)
+{
+	/* Use WRITE_ONCE to ensure correct read in sk_psock_get_msg_len_nolock().
+	 * ingress_lock should be held to prevent concurrent updates to msg_tot_len
+	 */
+	WRITE_ONCE(psock->msg_tot_len, psock->msg_tot_len + diff);
+}
+
+static inline void sk_psock_msg_len_add(struct sk_psock *psock, int diff)
+{
+	spin_lock_bh(&psock->ingress_lock);
+	sk_psock_msg_len_add_locked(psock, diff);
+	spin_unlock_bh(&psock->ingress_lock);
+}
+
 static inline bool sk_psock_queue_msg(struct sk_psock *psock,
 				      struct sk_msg *msg)
 {
@@ -320,6 +343,7 @@ static inline bool sk_psock_queue_msg(struct sk_psock *psock,
 	spin_lock_bh(&psock->ingress_lock);
 	if (sk_psock_test_state(psock, SK_PSOCK_TX_ENABLED)) {
 		list_add_tail(&msg->list, &psock->ingress_msg);
+		sk_psock_msg_len_add_locked(psock, msg->sg.size);
 		ret = true;
 	} else {
 		sk_msg_free(psock->sk, msg);
@@ -336,18 +360,25 @@ static inline struct sk_msg *sk_psock_dequeue_msg(struct sk_psock *psock)
 
 	spin_lock_bh(&psock->ingress_lock);
 	msg = list_first_entry_or_null(&psock->ingress_msg, struct sk_msg, list);
-	if (msg)
+	if (msg) {
 		list_del(&msg->list);
+		sk_psock_msg_len_add_locked(psock, -msg->sg.size);
+	}
 	spin_unlock_bh(&psock->ingress_lock);
 	return msg;
 }
 
+static inline struct sk_msg *sk_psock_peek_msg_locked(struct sk_psock *psock)
+{
+	return list_first_entry_or_null(&psock->ingress_msg, struct sk_msg, list);
+}
+
 static inline struct sk_msg *sk_psock_peek_msg(struct sk_psock *psock)
 {
 	struct sk_msg *msg;
 
 	spin_lock_bh(&psock->ingress_lock);
-	msg = list_first_entry_or_null(&psock->ingress_msg, struct sk_msg, list);
+	msg = sk_psock_peek_msg_locked(psock);
 	spin_unlock_bh(&psock->ingress_lock);
 	return msg;
 }
@@ -511,6 +542,39 @@ static inline bool sk_psock_strp_enabled(struct sk_psock *psock)
 	return !!psock->saved_data_ready;
 }
 
+/* for tcp only, sk is locked */
+static inline ssize_t sk_psock_msg_inq(struct sock *sk)
+{
+	struct sk_psock *psock;
+	ssize_t inq = 0;
+
+	psock = sk_psock_get(sk);
+	if (likely(psock)) {
+		inq = sk_psock_get_msg_len_nolock(psock);
+		sk_psock_put(sk, psock);
+	}
+	return inq;
+}
+
+/* for udp only, sk is not locked */
+static inline ssize_t sk_msg_first_len(struct sock *sk)
+{
+	struct sk_psock *psock;
+	struct sk_msg *msg;
+	ssize_t inq = 0;
+
+	psock = sk_psock_get(sk);
+	if (likely(psock)) {
+		spin_lock_bh(&psock->ingress_lock);
+		msg = sk_psock_peek_msg_locked(psock);
+		if (msg)
+			inq = msg->sg.size;
+		spin_unlock_bh(&psock->ingress_lock);
+		sk_psock_put(sk, psock);
+	}
+	return inq;
+}
+
 #if IS_ENABLED(CONFIG_NET_SOCK_MSG)
 
 #define BPF_F_STRPARSER	(1UL << 1)