1 files changed, 118 insertions, 15 deletions

diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c
index d5ea62b82bb8..8c89d0b0ca18 100644
--- a/net/smc/af_smc.c
+++ b/net/smc/af_smc.c
@@ -566,17 +566,115 @@ static void smc_stat_fallback(struct smc_sock *smc)
 	mutex_unlock(&net->smc.mutex_fback_rsn);
 }
 
+/* must be called under rcu read lock */
+static void smc_fback_wakeup_waitqueue(struct smc_sock *smc, void *key)
+{
+	struct socket_wq *wq;
+	__poll_t flags;
+
+	wq = rcu_dereference(smc->sk.sk_wq);
+	if (!skwq_has_sleeper(wq))
+		return;
+
+	/* wake up smc sk->sk_wq */
+	if (!key) {
+		/* sk_state_change */
+		wake_up_interruptible_all(&wq->wait);
+	} else {
+		flags = key_to_poll(key);
+		if (flags & (EPOLLIN | EPOLLOUT))
+			/* sk_data_ready or sk_write_space */
+			wake_up_interruptible_sync_poll(&wq->wait, flags);
+		else if (flags & EPOLLERR)
+			/* sk_error_report */
+			wake_up_interruptible_poll(&wq->wait, flags);
+	}
+}
+
+static int smc_fback_mark_woken(wait_queue_entry_t *wait,
+				unsigned int mode, int sync, void *key)
+{
+	struct smc_mark_woken *mark =
+		container_of(wait, struct smc_mark_woken, wait_entry);
+
+	mark->woken = true;
+	mark->key = key;
+	return 0;
+}
+
+static void smc_fback_forward_wakeup(struct smc_sock *smc, struct sock *clcsk,
+				     void (*clcsock_callback)(struct sock *sk))
+{
+	struct smc_mark_woken mark = { .woken = false };
+	struct socket_wq *wq;
+
+	init_waitqueue_func_entry(&mark.wait_entry,
+				  smc_fback_mark_woken);
+	rcu_read_lock();
+	wq = rcu_dereference(clcsk->sk_wq);
+	if (!wq)
+		goto out;
+	add_wait_queue(sk_sleep(clcsk), &mark.wait_entry);
+	clcsock_callback(clcsk);
+	remove_wait_queue(sk_sleep(clcsk), &mark.wait_entry);
+
+	if (mark.woken)
+		smc_fback_wakeup_waitqueue(smc, mark.key);
+out:
+	rcu_read_unlock();
+}
+
+static void smc_fback_state_change(struct sock *clcsk)
+{
+	struct smc_sock *smc =
+		smc_clcsock_user_data(clcsk);
+
+	if (!smc)
+		return;
+	smc_fback_forward_wakeup(smc, clcsk, smc->clcsk_state_change);
+}
+
+static void smc_fback_data_ready(struct sock *clcsk)
+{
+	struct smc_sock *smc =
+		smc_clcsock_user_data(clcsk);
+
+	if (!smc)
+		return;
+	smc_fback_forward_wakeup(smc, clcsk, smc->clcsk_data_ready);
+}
+
+static void smc_fback_write_space(struct sock *clcsk)
+{
+	struct smc_sock *smc =
+		smc_clcsock_user_data(clcsk);
+
+	if (!smc)
+		return;
+	smc_fback_forward_wakeup(smc, clcsk, smc->clcsk_write_space);
+}
+
+static void smc_fback_error_report(struct sock *clcsk)
+{
+	struct smc_sock *smc =
+		smc_clcsock_user_data(clcsk);
+
+	if (!smc)
+		return;
+	smc_fback_forward_wakeup(smc, clcsk, smc->clcsk_error_report);
+}
+
 static int smc_switch_to_fallback(struct smc_sock *smc, int reason_code)
 {
-	wait_queue_head_t *smc_wait = sk_sleep(&smc->sk);
-	wait_queue_head_t *clc_wait;
-	unsigned long flags;
+	struct sock *clcsk;
 
 	mutex_lock(&smc->clcsock_release_lock);
 	if (!smc->clcsock) {
 		mutex_unlock(&smc->clcsock_release_lock);
 		return -EBADF;
 	}
+	clcsk = smc->clcsock->sk;
+
 	smc->use_fallback = true;
 	smc->fallback_rsn = reason_code;
 	smc_stat_fallback(smc);
@@ -587,16 +685,22 @@ static int smc_switch_to_fallback(struct smc_sock *smc, int reason_code)
 		smc->clcsock->wq.fasync_list =
 			smc->sk.sk_socket->wq.fasync_list;
 
-		/* There may be some entries remaining in
-		 * smc socket->wq, which should be removed
-		 * to clcsocket->wq during the fallback.
+		/* There might be some wait entries remaining
+		 * in smc sk->sk_wq and they should be woken up
+		 * as clcsock's wait queue is woken up.
 		 */
-		clc_wait = sk_sleep(smc->clcsock->sk);
-		spin_lock_irqsave(&smc_wait->lock, flags);
-		spin_lock_nested(&clc_wait->lock, SINGLE_DEPTH_NESTING);
-		list_splice_init(&smc_wait->head, &clc_wait->head);
-		spin_unlock(&clc_wait->lock);
-		spin_unlock_irqrestore(&smc_wait->lock, flags);
+		smc->clcsk_state_change = clcsk->sk_state_change;
+		smc->clcsk_data_ready = clcsk->sk_data_ready;
+		smc->clcsk_write_space = clcsk->sk_write_space;
+		smc->clcsk_error_report = clcsk->sk_error_report;
+
+		clcsk->sk_state_change = smc_fback_state_change;
+		clcsk->sk_data_ready = smc_fback_data_ready;
+		clcsk->sk_write_space = smc_fback_write_space;
+		clcsk->sk_error_report = smc_fback_error_report;
+
+		smc->clcsock->sk->sk_user_data =
+			(void *)((uintptr_t)smc | SK_USER_DATA_NOCOPY);
 	}
 	mutex_unlock(&smc->clcsock_release_lock);
 	return 0;
@@ -2115,10 +2219,9 @@ out:
 
 static void smc_clcsock_data_ready(struct sock *listen_clcsock)
 {
-	struct smc_sock *lsmc;
+	struct smc_sock *lsmc =
+		smc_clcsock_user_data(listen_clcsock);
 
-	lsmc = (struct smc_sock *)
-	       ((uintptr_t)listen_clcsock->sk_user_data & ~SK_USER_DATA_NOCOPY);
 	if (!lsmc)
 		return;
 	lsmc->clcsk_data_ready(listen_clcsock);