4 files changed, 36 insertions, 34 deletions

diff --git a/net/core/dev.c b/net/core/dev.c
index 74fd402d26dd..7ee9fecd3aff 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -6923,12 +6923,16 @@ EXPORT_SYMBOL(napi_disable);
  */
 void napi_enable(struct napi_struct *n)
 {
-	BUG_ON(!test_bit(NAPI_STATE_SCHED, &n->state));
-	smp_mb__before_atomic();
-	clear_bit(NAPI_STATE_SCHED, &n->state);
-	clear_bit(NAPI_STATE_NPSVC, &n->state);
-	if (n->dev->threaded && n->thread)
-		set_bit(NAPI_STATE_THREADED, &n->state);
+	unsigned long val, new;
+
+	do {
+		val = READ_ONCE(n->state);
+		BUG_ON(!test_bit(NAPI_STATE_SCHED, &val));
+
+		new = val & ~(NAPIF_STATE_SCHED | NAPIF_STATE_NPSVC);
+		if (n->dev->threaded && n->thread)
+			new |= NAPIF_STATE_THREADED;
+	} while (cmpxchg(&n->state, val, new) != val);
 }
 EXPORT_SYMBOL(napi_enable);
 
diff --git a/net/core/netclassid_cgroup.c b/net/core/netclassid_cgroup.c
index b49c57d35a88..1a6a86693b74 100644
--- a/net/core/netclassid_cgroup.c
+++ b/net/core/netclassid_cgroup.c
@@ -71,11 +71,8 @@ static int update_classid_sock(const void *v, struct file *file, unsigned n)
 	struct update_classid_context *ctx = (void *)v;
 	struct socket *sock = sock_from_file(file);
 
-	if (sock) {
-		spin_lock(&cgroup_sk_update_lock);
+	if (sock)
 		sock_cgroup_set_classid(&sock->sk->sk_cgrp_data, ctx->classid);
-		spin_unlock(&cgroup_sk_update_lock);
-	}
 	if (--ctx->batch == 0) {
 		ctx->batch = UPDATE_CLASSID_BATCH;
 		return n + 1;
@@ -121,8 +118,6 @@ static int write_classid(struct cgroup_subsys_state *css, struct cftype *cft,
 	struct css_task_iter it;
 	struct task_struct *p;
 
-	cgroup_sk_alloc_disable();
-
 	cs->classid = (u32)value;
 
 	css_task_iter_start(css, 0, &it);
diff --git a/net/core/netprio_cgroup.c b/net/core/netprio_cgroup.c
index 99a431c56f23..8456dfbe2eb4 100644
--- a/net/core/netprio_cgroup.c
+++ b/net/core/netprio_cgroup.c
@@ -207,8 +207,6 @@ static ssize_t write_priomap(struct kernfs_open_file *of,
 	if (!dev)
 		return -ENODEV;
 
-	cgroup_sk_alloc_disable();
-
 	rtnl_lock();
 
 	ret = netprio_set_prio(of_css(of), dev, prio);
@@ -221,12 +219,10 @@ static ssize_t write_priomap(struct kernfs_open_file *of,
 static int update_netprio(const void *v, struct file *file, unsigned n)
 {
 	struct socket *sock = sock_from_file(file);
-	if (sock) {
-		spin_lock(&cgroup_sk_update_lock);
+
+	if (sock)
 		sock_cgroup_set_prioidx(&sock->sk->sk_cgrp_data,
 					(unsigned long)v);
-		spin_unlock(&cgroup_sk_update_lock);
-	}
 	return 0;
 }
 
@@ -235,8 +231,6 @@ static void net_prio_attach(struct cgroup_taskset *tset)
 	struct task_struct *p;
 	struct cgroup_subsys_state *css;
 
-	cgroup_sk_alloc_disable();
-
 	cgroup_taskset_for_each(p, css, tset) {
 		void *v = (void *)(unsigned long)css->id;
 
diff --git a/net/core/sock.c b/net/core/sock.c
index 62627e868e03..512e629f9780 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -3179,17 +3179,15 @@ EXPORT_SYMBOL(sock_init_data);
 
 void lock_sock_nested(struct sock *sk, int subclass)
 {
+	/* The sk_lock has mutex_lock() semantics here. */
+	mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_);
+
 	might_sleep();
 	spin_lock_bh(&sk->sk_lock.slock);
 	if (sk->sk_lock.owned)
 		__lock_sock(sk);
 	sk->sk_lock.owned = 1;
-	spin_unlock(&sk->sk_lock.slock);
-	/*
-	 * The sk_lock has mutex_lock() semantics here:
-	 */
-	mutex_acquire(&sk->sk_lock.dep_map, subclass, 0, _RET_IP_);
-	local_bh_enable();
+	spin_unlock_bh(&sk->sk_lock.slock);
 }
 EXPORT_SYMBOL(lock_sock_nested);
 
@@ -3227,24 +3225,35 @@ EXPORT_SYMBOL(release_sock);
  */
 bool lock_sock_fast(struct sock *sk) __acquires(&sk->sk_lock.slock)
 {
+	/* The sk_lock has mutex_lock() semantics here. */
+	mutex_acquire(&sk->sk_lock.dep_map, 0, 0, _RET_IP_);
+
 	might_sleep();
 	spin_lock_bh(&sk->sk_lock.slock);
 
-	if (!sk->sk_lock.owned)
+	if (!sk->sk_lock.owned) {
 		/*
-		 * Note : We must disable BH
+		 * Fast path return with bottom halves disabled and
+		 * sock::sk_lock.slock held.
+		 *
+		 * The 'mutex' is not contended and holding
+		 * sock::sk_lock.slock prevents all other lockers to
+		 * proceed so the corresponding unlock_sock_fast() can
+		 * avoid the slow path of release_sock() completely and
+		 * just release slock.
+		 *
+		 * From a semantical POV this is equivalent to 'acquiring'
+		 * the 'mutex', hence the corresponding lockdep
+		 * mutex_release() has to happen in the fast path of
+		 * unlock_sock_fast().
 		 */
 		return false;
+	}
 
 	__lock_sock(sk);
 	sk->sk_lock.owned = 1;
-	spin_unlock(&sk->sk_lock.slock);
-	/*
-	 * The sk_lock has mutex_lock() semantics here:
-	 */
-	mutex_acquire(&sk->sk_lock.dep_map, 0, 0, _RET_IP_);
 	__acquire(&sk->sk_lock.slock);
-	local_bh_enable();
+	spin_unlock_bh(&sk->sk_lock.slock);
 	return true;
 }
 EXPORT_SYMBOL(lock_sock_fast);