Delay struct net freeing while there's a sysfs instance refering to it

* new refcount in struct net, controlling actual freeing of the memory
	* new method in kobj_ns_type_operations (->drop_ns())
	* ->current_ns() semantics change - it's supposed to be followed by
corresponding ->drop_ns().  For struct net in case of CONFIG_NET_NS it bumps
the new refcount; net_drop_ns() decrements it and calls net_free() if the
last reference has been dropped.  Method renamed to ->grab_current_ns().
	* old net_free() callers call net_drop_ns() instead.
	* sysfs_exit_ns() is gone, along with a large part of callchain
leading to it; now that the references stored in ->ns[...] stay valid we
do not need to hunt them down and replace them with NULL.  That fixes
problems in sysfs_lookup() and sysfs_readdir(), along with getting rid
of sb->s_instances abuse.

	Note that struct net *shutdown* logics has not changed - net_cleanup()
is called exactly when it used to be called.  The only thing postponed by
having a sysfs instance refering to that struct net is actual freeing of
memory occupied by struct net.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

fs/sysfs/mount.c

@@ -95,6 +95,14 @@ static int sysfs_set_super(struct super_block *sb, void *data)
 	return error;
 }
 
+static void free_sysfs_super_info(struct sysfs_super_info *info)
+{
+	int type;
+	for (type = KOBJ_NS_TYPE_NONE; type < KOBJ_NS_TYPES; type++)
+		kobj_ns_drop(type, info->ns[type]);
+	kfree(info);
+}
+
 static struct dentry *sysfs_mount(struct file_system_type *fs_type,
 	int flags, const char *dev_name, void *data)
 {
@@ -108,11 +116,11 @@ static struct dentry *sysfs_mount(struct file_system_type *fs_type,
 		return ERR_PTR(-ENOMEM);
 
 	for (type = KOBJ_NS_TYPE_NONE; type < KOBJ_NS_TYPES; type++)
-		info->ns[type] = kobj_ns_current(type);
+		info->ns[type] = kobj_ns_grab_current(type);
 
 	sb = sget(fs_type, sysfs_test_super, sysfs_set_super, info);
 	if (IS_ERR(sb) || sb->s_fs_info != info)
-		kfree(info);
+		free_sysfs_super_info(info);
 	if (IS_ERR(sb))
 		return ERR_CAST(sb);
 	if (!sb->s_root) {
@@ -131,12 +139,11 @@ static struct dentry *sysfs_mount(struct file_system_type *fs_type,
 static void sysfs_kill_sb(struct super_block *sb)
 {
 	struct sysfs_super_info *info = sysfs_info(sb);
-
 	/* Remove the superblock from fs_supers/s_instances
 	 * so we can't find it, before freeing sysfs_super_info.
 	 */
 	kill_anon_super(sb);
-	kfree(info);
+	free_sysfs_super_info(info);
 }
 
 static struct file_system_type sysfs_fs_type = {
@@ -145,28 +152,6 @@ static struct file_system_type sysfs_fs_type = {
 	.kill_sb	= sysfs_kill_sb,
 };
 
-void sysfs_exit_ns(enum kobj_ns_type type, const void *ns)
-{
-	struct super_block *sb;
-
-	mutex_lock(&sysfs_mutex);
-	spin_lock(&sb_lock);
-	list_for_each_entry(sb, &sysfs_fs_type.fs_supers, s_instances) {
-		struct sysfs_super_info *info = sysfs_info(sb);
-		/*
-		 * If we see a superblock on the fs_supers/s_instances
-		 * list the unmount has not completed and sb->s_fs_info
-		 * points to a valid struct sysfs_super_info.
-		 */
-		/* Ignore superblocks with the wrong ns */
-		if (info->ns[type] != ns)
-			continue;
-		info->ns[type] = NULL;
-	}
-	spin_unlock(&sb_lock);
-	mutex_unlock(&sysfs_mutex);
-}
-
 int __init sysfs_init(void)
 {
 	int err = -ENOMEM;

fs/sysfs/sysfs.h

@@ -136,7 +136,7 @@ struct sysfs_addrm_cxt {
  * instance).
  */
 struct sysfs_super_info {
-	const void *ns[KOBJ_NS_TYPES];
+	void *ns[KOBJ_NS_TYPES];
 };
 #define sysfs_info(SB) ((struct sysfs_super_info *)(SB->s_fs_info))
 extern struct sysfs_dirent sysfs_root;

include/linux/kobject_ns.h

@@ -32,15 +32,17 @@ enum kobj_ns_type {
 
 /*
  * Callbacks so sysfs can determine namespaces
- *   @current_ns: return calling task's namespace
+ *   @grab_current_ns: return a new reference to calling task's namespace
  *   @netlink_ns: return namespace to which a sock belongs (right?)
  *   @initial_ns: return the initial namespace (i.e. init_net_ns)
+ *   @drop_ns: drops a reference to namespace
  */
 struct kobj_ns_type_operations {
 	enum kobj_ns_type type;
-	const void *(*current_ns)(void);
+	void *(*grab_current_ns)(void);
 	const void *(*netlink_ns)(struct sock *sk);
 	const void *(*initial_ns)(void);
+	void (*drop_ns)(void *);
 };
 
 int kobj_ns_type_register(const struct kobj_ns_type_operations *ops);
@@ -48,9 +50,9 @@ int kobj_ns_type_registered(enum kobj_ns_type type);
 const struct kobj_ns_type_operations *kobj_child_ns_ops(struct kobject *parent);
 const struct kobj_ns_type_operations *kobj_ns_ops(struct kobject *kobj);
 
-const void *kobj_ns_current(enum kobj_ns_type type);
+void *kobj_ns_grab_current(enum kobj_ns_type type);
 const void *kobj_ns_netlink(enum kobj_ns_type type, struct sock *sk);
 const void *kobj_ns_initial(enum kobj_ns_type type);
-void kobj_ns_exit(enum kobj_ns_type type, const void *ns);
+void kobj_ns_drop(enum kobj_ns_type type, void *ns);
 
 #endif /* _LINUX_KOBJECT_NS_H */

include/linux/sysfs.h

@@ -177,9 +177,6 @@ struct sysfs_dirent *sysfs_get_dirent(struct sysfs_dirent *parent_sd,
 struct sysfs_dirent *sysfs_get(struct sysfs_dirent *sd);
 void sysfs_put(struct sysfs_dirent *sd);
 
-/* Called to clear a ns tag when it is no longer valid */
-void sysfs_exit_ns(enum kobj_ns_type type, const void *tag);
-
 int __must_check sysfs_init(void);
 
 #else /* CONFIG_SYSFS */
@@ -338,10 +335,6 @@ static inline void sysfs_put(struct sysfs_dirent *sd)
 {
 }
 
-static inline void sysfs_exit_ns(int type, const void *tag)
-{
-}
-
 static inline int __must_check sysfs_init(void)
 {
 	return 0;

include/net/net_namespace.h

@@ -35,9 +35,12 @@ struct netns_ipvs;
 #define NETDEV_HASHENTRIES (1 << NETDEV_HASHBITS)
 
 struct net {
-	atomic_t		count;		/* To decided when the network
+	atomic_t		passive;	/* To decided when the network
 						 * namespace should be freed.
 						 */
+	atomic_t		count;		/* To decided when the network
+						 *  namespace should be shut down.
+						 */
 #ifdef NETNS_REFCNT_DEBUG
 	atomic_t		use_count;	/* To track references we
 						 * destroy on demand
@@ -154,6 +157,9 @@ int net_eq(const struct net *net1, const struct net *net2)
 {
 	return net1 == net2;
 }
+
+extern void net_drop_ns(void *);
+
 #else
 
 static inline struct net *get_net(struct net *net)
@@ -175,6 +181,8 @@ int net_eq(const struct net *net1, const struct net *net2)
 {
 	return 1;
 }
+
+#define net_drop_ns NULL
 #endif
 
 

lib/kobject.c

@@ -948,14 +948,14 @@ const struct kobj_ns_type_operations *kobj_ns_ops(struct kobject *kobj)
 }
 
 
-const void *kobj_ns_current(enum kobj_ns_type type)
+void *kobj_ns_grab_current(enum kobj_ns_type type)
 {
-	const void *ns = NULL;
+	void *ns = NULL;
 
 	spin_lock(&kobj_ns_type_lock);
 	if ((type > KOBJ_NS_TYPE_NONE) && (type < KOBJ_NS_TYPES) &&
 	    kobj_ns_ops_tbl[type])
-		ns = kobj_ns_ops_tbl[type]->current_ns();
+		ns = kobj_ns_ops_tbl[type]->grab_current_ns();
 	spin_unlock(&kobj_ns_type_lock);
 
 	return ns;
@@ -987,23 +987,15 @@ const void *kobj_ns_initial(enum kobj_ns_type type)
 	return ns;
 }
 
-/*
- * kobj_ns_exit - invalidate a namespace tag
- *
- * @type: the namespace type (i.e. KOBJ_NS_TYPE_NET)
- * @ns: the actual namespace being invalidated
- *
- * This is called when a tag is no longer valid.  For instance,
- * when a network namespace exits, it uses this helper to
- * make sure no sb's sysfs_info points to the now-invalidated
- * netns.
- */
-void kobj_ns_exit(enum kobj_ns_type type, const void *ns)
+void kobj_ns_drop(enum kobj_ns_type type, void *ns)
 {
-	sysfs_exit_ns(type, ns);
+	spin_lock(&kobj_ns_type_lock);
+	if ((type > KOBJ_NS_TYPE_NONE) && (type < KOBJ_NS_TYPES) &&
+	    kobj_ns_ops_tbl[type] && kobj_ns_ops_tbl[type]->drop_ns)
+		kobj_ns_ops_tbl[type]->drop_ns(ns);
+	spin_unlock(&kobj_ns_type_lock);
 }
 
-
 EXPORT_SYMBOL(kobject_get);
 EXPORT_SYMBOL(kobject_put);
 EXPORT_SYMBOL(kobject_del);

net/core/net-sysfs.c

@@ -1179,9 +1179,14 @@ static void remove_queue_kobjects(struct net_device *net)
 #endif
 }
 
-static const void *net_current_ns(void)
+static void *net_grab_current_ns(void)
 {
-	return current->nsproxy->net_ns;
+	struct net *ns = current->nsproxy->net_ns;
+#ifdef CONFIG_NET_NS
+	if (ns)
+		atomic_inc(&ns->passive);
+#endif
+	return ns;
 }
 
 static const void *net_initial_ns(void)
@@ -1196,22 +1201,13 @@ static const void *net_netlink_ns(struct sock *sk)
 
 struct kobj_ns_type_operations net_ns_type_operations = {
 	.type = KOBJ_NS_TYPE_NET,
-	.current_ns = net_current_ns,
+	.grab_current_ns = net_grab_current_ns,
 	.netlink_ns = net_netlink_ns,
 	.initial_ns = net_initial_ns,
+	.drop_ns = net_drop_ns,
 };
 EXPORT_SYMBOL_GPL(net_ns_type_operations);
 
-static void net_kobj_ns_exit(struct net *net)
-{
-	kobj_ns_exit(KOBJ_NS_TYPE_NET, net);
-}
-
-static struct pernet_operations kobj_net_ops = {
-	.exit = net_kobj_ns_exit,
-};
-
-
 #ifdef CONFIG_HOTPLUG
 static int netdev_uevent(struct device *d, struct kobj_uevent_env *env)
 {
@@ -1339,6 +1335,5 @@ EXPORT_SYMBOL(netdev_class_remove_file);
 int netdev_kobject_init(void)
 {
 	kobj_ns_type_register(&net_ns_type_operations);
-	register_pernet_subsys(&kobj_net_ops);
 	return class_register(&net_class);
 }

net/core/net_namespace.c

@@ -128,6 +128,7 @@ static __net_init int setup_net(struct net *net)
 	LIST_HEAD(net_exit_list);
 
 	atomic_set(&net->count, 1);
+	atomic_set(&net->passive, 1);
 
 #ifdef NETNS_REFCNT_DEBUG
 	atomic_set(&net->use_count, 0);
@@ -210,6 +211,13 @@ static void net_free(struct net *net)
 	kmem_cache_free(net_cachep, net);
 }
 
+void net_drop_ns(void *p)
+{
+	struct net *ns = p;
+	if (ns && atomic_dec_and_test(&ns->passive))
+		net_free(ns);
+}
+
 struct net *copy_net_ns(unsigned long flags, struct net *old_net)
 {
 	struct net *net;
@@ -230,7 +238,7 @@ struct net *copy_net_ns(unsigned long flags, struct net *old_net)
 	}
 	mutex_unlock(&net_mutex);
 	if (rv < 0) {
-		net_free(net);
+		net_drop_ns(net);
 		return ERR_PTR(rv);
 	}
 	return net;
@@ -286,7 +294,7 @@ static void cleanup_net(struct work_struct *work)
 	/* Finally it is safe to free my network namespace structure */
 	list_for_each_entry_safe(net, tmp, &net_exit_list, exit_list) {
 		list_del_init(&net->exit_list);
-		net_free(net);
+		net_drop_ns(net);
 	}
 }
 static DECLARE_WORK(net_cleanup_work, cleanup_net);